WATER RESEARCH
WATER RESEARCH
WATER RESEARCH
You also want an ePaper? Increase the reach of your titles
YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.
Editor-in-Chief<br />
Mogens Henze<br />
Institute of Environment & Resources<br />
Technical University of Denmark<br />
Bygningstorvet<br />
DK-2800 KGS Lyngby<br />
Denmark<br />
Tel: 45 4525 1477<br />
Fax: 45 4593 2850<br />
E-mail: moh@er.dtu.dk<br />
Editors<br />
J. Block<br />
Université H. Poincaré, Nancy I<br />
France<br />
David Dixon<br />
University of Melbourne<br />
Australia<br />
Hiroaki Furumai<br />
The University of Tokyo<br />
Japan<br />
Gregory Korshin<br />
University of Washington<br />
USA<br />
Anna Ledin<br />
Technical University of Denmark<br />
Denmark<br />
Eberhard Morgenroth<br />
University of Illinois Urbana-Champaign<br />
USA<br />
W. Rauch<br />
University Innsbruck<br />
Austria<br />
Maria Reis<br />
Universidade Nova de Lisboa/FCT<br />
Portugal<br />
Hang-Shik Shin<br />
Korea Advanced Institute of Science<br />
and Technology<br />
Korea<br />
Mark van Loosdrecht<br />
Delft University of Technology<br />
The Netherlands<br />
Thomas Ternes<br />
Bundesanstalt für Gewässerkunde<br />
Germany<br />
Stefan Wuertz<br />
Univ. of California, Davis<br />
USA<br />
Hanqing Yu<br />
University of Science & Technology of China<br />
China<br />
Associate Editors<br />
Andrew Baker<br />
The University of Birmingham<br />
UK<br />
<strong>WATER</strong> <strong>RESEARCH</strong><br />
A Journal of the International Water Association<br />
Damien Batstone<br />
The University of Queensland<br />
Australia<br />
G-H. Chen<br />
The Hong Kong University of Science<br />
& Technology<br />
Hong Kong<br />
China<br />
Tom Curtis<br />
Univ. of Newcastle upon Tyne<br />
UK<br />
Ana Deletic<br />
Monash University<br />
USA<br />
Francis de los Reyes III<br />
North Carolina State University<br />
USA<br />
Rob Eldridge<br />
The University of Melbourne<br />
Australia<br />
Rosina Girones<br />
University of Barcelona<br />
Spain<br />
Stephen Gray<br />
Victoria University<br />
Australia<br />
Kate Grudpan<br />
Chiang Mai University<br />
Thailand<br />
Xiaodi Hao<br />
Beijing University of Civil Engineering<br />
and Architecture<br />
China<br />
E.E. Herricks<br />
University of Illinois - Urbana<br />
USA<br />
H-Y. Hu<br />
Tsinghua University<br />
China<br />
P.M. Huck<br />
University of Waterloo<br />
Canada<br />
Bruce Jefferson<br />
Cranfi eld University<br />
UK<br />
Sergey Kalyuzhnyi<br />
Moscow State University<br />
Russian Federation<br />
Jaehong Kim<br />
Georgia Institute of Technology<br />
USA<br />
Jes La Cour Jansen<br />
Lund Institute of Technology<br />
Sweden<br />
G. Langergraber<br />
BOKU/Univ. of Natural Res. and<br />
Applied Life Scs.<br />
Austria<br />
S-L. Lo<br />
National Taiwan University<br />
Taiwan<br />
Y. Matsui<br />
Hokkaido University<br />
Japan<br />
Max Maurer<br />
EAWAG<br />
Switzerland<br />
How Yong Ng<br />
National University of Singapore<br />
Singapore<br />
Satoshi Okabe<br />
Hokkaido University<br />
Japan<br />
S.L. Ong<br />
National University of Singapore<br />
Singapore<br />
Jong M. Park<br />
Pohang University of Science & Technology<br />
Korea<br />
Miguel Salgot<br />
University of Barcelona<br />
Spain<br />
David Sedlak<br />
University of California, Berkeley<br />
USA<br />
Jean-Philippe Steyer<br />
LBE-INRA<br />
France<br />
M. Takahashi<br />
Hokkaido University<br />
Japan<br />
Kai Udert<br />
EAWAG<br />
Switzerland<br />
V.P. Venugopalan<br />
BARC Facilities<br />
India<br />
E. von Sperling<br />
Federal University of Minas Gerais<br />
Brazil<br />
A.I. Zouboulis<br />
Aristotle University<br />
Greece<br />
Editorial Offi ce<br />
Mailing address<br />
Water Research<br />
Radarweg 29<br />
1043 NX Amsterdam<br />
The Netherlands<br />
Tel: 31 20 485 2701<br />
Fax: 31 20 485 2521; 31 20 485 3264<br />
E-mail: wr-eo@elsevier.com<br />
Publication information: Water Research (ISSN 0043-1354). For 2010, volume 44 is scheduled for publication. Subscription prices are available upon request from the<br />
publisher or from the Elsevier Customer Service Department nearest you or from this journal’s website (http://www.elsevier.com/locate/watres). Further information<br />
is available on this journal and other Elsevier products through Elsevier’s website (http://www.elsevier.com). Subscriptions are accepted on a prepaid basis<br />
only and are entered on a calendar year basis. Issues are sent by standard mail (surface within Europe, air delivery outside Europe). Priority rates are available upon<br />
request. Claims for missing issues should be made within six months of the date of dispatch.<br />
Author Enquiries: Please visit this journal’s homepage at http://www.elsevier.com/locate/watres. You can track accepted articles at http://www.elsevier.com/<br />
trackarticle and set up e-mail alerts to inform you of when an article’s status has changed. Also accessible from here is information on copyright, frequently asked<br />
questions and more. Contact details for questions arising after acceptance of an article, especially those relating to proofs, will be provided by the publisher.<br />
Orders, claims, and journal enquiries: Please contact the Elsevier Customer Service Department nearest you:<br />
St. Louis: Elsevier Customer Service Department, 11830 Westline Industrial Drive, St. Louis, MO 63146, USA; phone: (877) 8397126 [toll free within the USA];<br />
(+1) (314) 4537076 [outside the USA]; fax: (+1) (314) 5235153; e-mail: JournalCustomerService-usa@elsevier.com<br />
Oxford: Elsevier Customer Service Department, The Boulevard, Langford Lane, Kidlington OX5 1GB, UK; phone: (+44) (1865) 843434; fax: (+44) (1865)<br />
843970; e-mail: JournalsCustomerServiceEMEA@elsevier.com<br />
Tokyo: Elsevier Customer Service Department, 4F Higashi-Azabu, 1-Chome Bldg, 1-9-15 Higashi-Azabu, Minato-ku, Tokyo 106-0044, Japan; phone: (+81) (3)<br />
5561 5037; fax: (+81) (3) 5561 5047; e-mail: JournalsCustomerServiceJapan@elsevier.com<br />
Singapore: Elsevier Customer Service Department, 3 Killiney Road, #08-01 Winsland House I, Singapore 239519; phone: (+65) 63490222; fax: (+65) 67331510;<br />
e-mail: JournalsCustomerServiceAPAC@elsevier.com<br />
Application for membership of International Water Association should be made to: Executive Director, IWA, Alliance House, 12 Caxton Street,<br />
London SW1H 0QS, U.K. (Tel.: +44 207 654 5500; Fax: +44 207 654 5555; e-mail: members@IWAhq.org.uk; website: http://www.IWAhq.org.uk).<br />
Registered Charity (England) No. 289269. Individual membership is available from £30 upwards.<br />
For details contact IWA.
Editorial<br />
Editorial to special issue in Water Research<br />
Emerging contaminants in water<br />
The chemical pollution of natural waters is one of the big<br />
challenges of the 21st century. Based on the rapid evolution in<br />
analytical chemistry, with the possibility to detect more polar<br />
compounds a whole new suite of ‘‘emerging contaminants’’<br />
such as pharmaceuticals, hormones and perfluorinated<br />
compounds has been identified in various compartments of<br />
the water cycle including both natural and technical aquatic<br />
systems. The discovery of these micropollutants in the aquatic<br />
environment has triggered research efforts to investigate<br />
sources and mitigation strategies with conventional and novel<br />
treatment processes and assess their importance with regard<br />
to ecotoxicology and human health. In this special issue on<br />
emerging contaminants, we have compiled 20 research articles<br />
and 2 reviews, which deal with these timely issues.<br />
As mentioned above, quantification of micropollutants in<br />
aquatic systems is a key requirement to assess their fate.<br />
Several studies in this issue address the determination of<br />
pharmaceuticals in archived biosolids (Halden et al.), in<br />
wastewater treatment (Lindberg et al.) and lagoon treatment<br />
(Wong et al.). Another study uses analytical data to evaluate<br />
the fraction of pharmaceutical residues in wastewater originating<br />
from hospitals (Ort et al.). Finally, a review covers the<br />
occurrence and fate of phytoestrogens in the environment<br />
(Liu et al.).<br />
To further elucidate the relevance of the micropollutants<br />
detected in various aquatic compartments, their (eco)toxicological<br />
potential has to be assessed. Several papers address<br />
related issues for individual compounds (lipid regulators,<br />
Fernandez-Pinas et al.) or classes of compounds (ionic liquids,<br />
Yun et al.) and for pharmaceuticals in advanced wastewater<br />
treatment systems such as powdered activated carbon and<br />
ozonation with in vivo and in vitro tests (Escher et al., Stalter<br />
et al.). Furthermore, one study is focused on toxicity nanotube<br />
suspensions (Tarabara et al.), which have been on the radar of<br />
emerging contaminants recently. Finally, one paper focuses<br />
on the toxicological relevance of emerging contaminants for<br />
drinking water (Schriks et al.).<br />
In recent years, municipal wastewater has been recognized<br />
as an important source of micropollutants to the<br />
water research 44 (2010) 351<br />
Available at www.sciencedirect.com<br />
journal homepage: www.elsevier.com/locate/watres<br />
receiving water bodies. Therefore, mitigation strategies for<br />
the minimization of the discharge of these compounds play<br />
an increasingly important role in the urban water management.<br />
In this special issue, the removal of benzotriazoles<br />
(Reemtsma et al.) and pharmaceuticals, caffeine and DEET (Yu<br />
et al.) and other emerging contaminants (Rosal et al.) has been<br />
investigated for conventional wastewater treatment.<br />
Numerous papers address the oxidative removal of micropollutants<br />
from wastewater with chlorine, chlorine dioxide,<br />
ferrate, ozone, advanced oxidation processes, such as UV/<br />
H2O2, (solar) photo-Fenton and non-thermal plasma (Lee<br />
et al., Malato et al., Reungoat et al., Gerrity et al., Mendez-<br />
Arriaga). Other options for removal of micropollutants<br />
include membrane processes such as reverse osmosis (Hu<br />
et al.) and nanofiltration (Yangali-Quintanilla et al.) as well as<br />
sorption on sludge (Carrere et al.). The challenges caused by<br />
harmful algae producing toxins for desalination operations<br />
were reviewed by David Caron and co-workers. Finally, mitigation<br />
of micropollutants may also occur during managed<br />
aquifer recharge (Drewes et al.) or in biological Fenton-like<br />
processes (Vicent et al.).<br />
In short, we are very happy to provide you this Theme Issue<br />
on Emerging Contaminants. We appreciate the contributions<br />
by the authors, reviewers and editorial staff of Water Research<br />
to this project.<br />
Thomas Ternes*<br />
Federal Institute of Hydrology (BFG),<br />
Am Mainzer Tor 1, 56068 Koblenz, Germany<br />
*Corresponding author. Tel.: þ49 261 1306 5560;<br />
fax: þ49 261 1306 5363.<br />
Urs von Gunten<br />
EAWAG, Ueberlandstrasse 133, Duebendorf CH-8600,<br />
Switzerland<br />
0043-1354/$ – see front matter<br />
ª 2010 Published by Elsevier Ltd.<br />
doi:10.1016/j.watres.2010.01.015
Review<br />
Environmental fate and toxicity of ionic liquids: A review<br />
Thi Phuong Thuy Pham a , Chul-Woong Cho a , Yeoung-Sang Yun a,b, *<br />
a Department of Bioprocess Engineering, Chonbuk National University, Jeonju, Chonbuk 561-756, Republic of Korea<br />
b Division of Semiconductor and Chemical Engineering and Research Institute of Industrial Technology, Chonbuk National University,<br />
Chonbuk 561-756, Republic of Korea<br />
article info<br />
Article history:<br />
Received 31 May 2009<br />
Received in revised form<br />
27 August 2009<br />
Accepted 12 September 2009<br />
Available online 24 September 2009<br />
Keywords:<br />
Ionic liquids<br />
Toxicity<br />
Degradation<br />
Biodegradation<br />
Environmental fate<br />
Sorption<br />
Contents<br />
water research 44 (2010) 352–372<br />
Available at www.sciencedirect.com<br />
journal homepage: www.elsevier.com/locate/watres<br />
abstract<br />
Ionic liquids (ILs) are organic salts with low melting point that are being considered as<br />
green replacements for industrial volatile organic compounds. The reputation of these<br />
solvents as ‘‘environmental friendly’’ chemicals is based primarily on their negligible vapor<br />
pressure. Nonetheless, the solubility of ILs in water and a number of literature<br />
documenting toxicity of ILs to aquatic organisms highlight a real cause for concern. The<br />
knowledge of ILs behavior in the terrestrial environment, which includes microbial<br />
degradation, sorption and desorption, is equally important since both soil and aquatic<br />
milieu are possible recipients of IL contamination. This article reviews the achievements<br />
and current status of environmental risk assessment of ILs, and hopefully provides<br />
insights into this research frontier.<br />
ª 2009 Elsevier Ltd. All rights reserved.<br />
1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 353<br />
2. Toxicological aspect of ILs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 354<br />
2.1. Effects of ILs in an enzyme level . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 354<br />
2.2. Antibacterial activity of ILs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 356<br />
2.3. Toxicity of ILs to algae . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 361<br />
2.4. Cytotoxicity of ILs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 361<br />
2.5. Phytotoxicity of ILs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 363<br />
2.6. Toxicity of ILs to invertebrates . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 363<br />
2.7. Inhibitory effects of ILs on vertebrates . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 364<br />
3. Environmental fate of ILs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 364<br />
3.1. Chemical degradation of ILs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 364<br />
3.2. Biodegradability of ILs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 365<br />
3.3. Sorption of ILs in environmental systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 367<br />
* Corresponding author. Division of Semiconductor and Chemical Engineering, Chonbuk National University, Jeonju, Chonbuk 561-756,<br />
Republic of Korea. Tel.: þ82 63 270 2308; fax: þ82 63 270 2306.<br />
E-mail address: ysyun@chonbuk.ac.kr (Y.-S. Yun).<br />
0043-1354/$ – see front matter ª 2009 Elsevier Ltd. All rights reserved.<br />
doi:10.1016/j.watres.2009.09.030
4. Concluding remarks and future directions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 368<br />
Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 369<br />
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 369<br />
1. Introduction<br />
Most of volatile organic compounds (VOCs) commonly used in<br />
industrial applications cause a major concern in the current<br />
chemical processing industry. The main problems are the<br />
toxicity of the organic solvents to both the process operators<br />
and the environment as well as the volatile and flammable<br />
nature of these solvents which make them a potential explosion<br />
hazard (Schmid et al., 1998). Recently, the deleterious<br />
effects of many solvents combined with serious<br />
environmental issues, such as atmospheric emissions and<br />
contamination of aqueous effluents are making their use<br />
prohibitive. Thus, many researchers have focused on the<br />
development of ‘‘green engineering’’ which represents<br />
research aimed at finding environmentally benign alternatives<br />
to harmful chemicals. Among the neoteric solvents applicable<br />
in ‘‘green technologies’’ ionic liquids (ILs) have garnered<br />
increasing attention over the others such as supercritical CO 2<br />
(Blanchard et al., 1999; Blanchard and Brennecke, 2001;<br />
Kazarian et al., 2000) and aqueous biphasic systems (Myasoedov<br />
et al., 1995; Rogers et al., 1995; Willauer et al., 1999).<br />
Ionic liquids, formerly known as molten salts, constitute<br />
one of the hottest areas in chemistry these days. Basically,<br />
they have melting points below 100 C, which can be achieved<br />
by incorporating a bulky asymmetric cation into the structure<br />
together with a weakly-coordinating anion (Ranke et al., 2004).<br />
The unique, highly solvating, yet non-coordinating environment<br />
of ILs provides an attractive medium for various types of<br />
chemical processes. Also, the physical properties of ILs can be<br />
tailored by a judicious variation in the length and branching of<br />
the alkyl chain and the anionic precursor (Fuler et al., 1997;<br />
Huddleston et al., 2001). In this way, ILs can be made taskspecific<br />
for a certain application. The almost limitless structural<br />
possibilities of ILs, as opposed to limited structural<br />
variations within molecular solvents, make them ‘‘designer<br />
water research 44 (2010) 352–372 353<br />
Fig. 1 – Applications of ionic liquids.<br />
solvents’’ (Marsh et al., 2004; McFarlane et al., 2005; Sheldon,<br />
2005). Some independent reports (Hagiwara and Ito, 2000;<br />
Olivier, 1999; Welton, 1999) and many reviews (Earle and<br />
Seddon, 2000; Rooney and Seddon, 2001) have highlighted ILs<br />
as representing a state-of-the-art, innovative approach to<br />
sustainable chemistry, with the argument that their vapor<br />
pressure is immeasurably low and they are not flammable.<br />
Recently, the application of these liquids as reaction media for<br />
organic synthesis, catalysis, or biocatalysis has been well<br />
documented (Earle and Seddon, 2000; Wasserscheid and<br />
Keim, 2000; Welton, 1999) (Fig. 1). Gordon (2001) pointed out<br />
that there is an obvious advantage in performing many reactions<br />
in ILs due to the improvement in process economics,<br />
reaction activity, selectivity and yield.<br />
Although ILs can lessen the risk of air pollution due to their<br />
insignificant vapor pressure, they do have significant solubility<br />
in water (Anthony et al., 2001; McFarlane et al., 2005;<br />
Wong et al., 2002). As a result, this is the most likely medium<br />
through which ILs will be released into the environment. Ionic<br />
liquids currently are not widely used in industrial applications;<br />
nonetheless, continued development and further use of<br />
these solvents may lead to accidental discharge and<br />
contamination. The properties that make them be the target<br />
of industrial interest (i.e. high chemical, thermal stability and<br />
non-volatility) suggest potential problems with degradation or<br />
persistence in the environment. In general, the deficiency of<br />
information and uncertainty surrounding the environmental<br />
impact of ILs is a major barrier to the utilization of these<br />
compounds by industry. Initial efforts have been made to<br />
overcome this drawback and offer a preliminary insight into<br />
the behavior of ILs in the aqueous environments. These<br />
studies provided extensive data sets, e.g. on (eco)toxicity,<br />
biodegradability, bioaccumulation and distribution of ILs in<br />
different environmental compartments. Therefore, it is<br />
necessary to consolidate all the available data in a single
354<br />
review to lay the groundwork for more comprehensive<br />
community and ecosystem investigations. The overall objective<br />
of this review is to systematically gather and interpret<br />
existing information about the fate, removal options and<br />
(eco)toxicological assessment strategies of ILs.<br />
2. Toxicological aspect of ILs<br />
The current literature represents a number of studies addressing<br />
the biological effects of ILs evaluated on the basis of<br />
toxicological test systems. The ILs toxicities towards these<br />
systems of different levels of biological complexity as well as<br />
several environmental compartments (Fig. 2) aresuccessively<br />
discussed in the following subsections. All the structures of<br />
IL compounds discussed in this review were listed in Table 1.<br />
The acronyms used for these substances were adapted<br />
from Ranke et al. (2007a). In this way, the cation head groups<br />
were abbreviated as ‘‘IM’’ for imidazolium, ‘‘Py’’ for pyridinium,<br />
‘‘Pyr’’ for pyrrolidinium, ‘‘Mor’’ for morpholinium, ‘‘Pip’’ for<br />
piperidinium, ‘‘Quin’’ for quinolinium, ‘‘N’’ for quaternary<br />
ammonium and ‘‘P’’ for quaternary phosphonium. The alkyl<br />
chains attached to the head group were given as numbers<br />
corresponding to the number of carbon in the alkyl residues. For<br />
example, the 1-butyl-3-methylimidazolium moiety was denoted<br />
as IM14. In case the carbon chain length equals or exceeds 10,<br />
the numbers were separated by a hyphen (e.g. IM1-10 indicated<br />
1-decyl-3-methylimidazolium). Particularly, for pyridinium<br />
water research 44 (2010) 352–372<br />
entities, the carbon-bound alkyl chains were appended to the<br />
head group at different positions and the abbreviation was<br />
made by noting the position of attachment and a symbol for the<br />
attached group (e.g. Py4-2Me for 1-butyl-2-methylpyridinium).<br />
The anionic components were shortened as they are in the<br />
periodic table for the halides. For tetrafluoroborate, hexafluorophosphate,<br />
bis(trifluoromethylsulfonyl)imide, dicyanamide<br />
and hydrogen sulfate the abbreviations were BF4, PF6,<br />
(CF 3SO 2) 2N, CN(N) 2 and HSO 4 in respective to their structural<br />
formula.<br />
2.1. Effects of ILs in an enzyme level<br />
Enzyme inhibition data by ILs include those of the acetylcholinesterase<br />
from electric eel (Electrophorus electricus)<br />
(Arning et al., 2008; Jastorff et al., 2005; Matzke et al., 2007;<br />
Ranke et al., 2007b; Stasiewicz et al., 2008; Stock et al., 2004;<br />
Torrecilla et al., 2009; Zhang and Malhotra, 2005), the AMP<br />
deaminase (Sk1adanowski et al., 2005) and the antioxidant<br />
enzyme system of mouse liver (Yu et al., 2009a). The enzyme<br />
acetylcholinesterase plays the most important role in nerve<br />
response and function. Also, acetylcholinesterase catalyzes<br />
the hydrolysis of acetylcholinesters with a relative specificity<br />
for acetylcholine, which is a neurotransmitter common to<br />
many synapses throughout mammalian nervous systems<br />
(Fulton and Key, 2001; Massoulié et al., 1993). Thus, an inhibition<br />
of acetylcholinesterase leads to various adverse effects<br />
in neuronal processes, such as heart diseases or myasthenia<br />
Fig. 2 – The flexible (eco)toxicological test battery considering aquatic and terrestrial compartments as well as different<br />
trophic levels including enzymes, luminescent marine bacteria, freshwater green algae, mammalian cells, duckweed,<br />
freshwater crustacean and zebrafish (Adapted from Matzke et al. (2007) by permission of the Royal Society of Chemistry).
Table 1 – Selection of cationic and anionic structures of commonly used ionic liquids.<br />
Cation<br />
water research 44 (2010) 352–372 355<br />
CH 3<br />
CH 3<br />
R 2<br />
Head group Side chain<br />
N N +<br />
R 1<br />
Imidazolium (IM)<br />
CH 3<br />
N +<br />
Pyridinium (Py)<br />
N +<br />
CH 3<br />
R 1<br />
Pyrrolidinium (Pyr)<br />
O<br />
N +<br />
CH 3<br />
R 1<br />
Morpholinium (Mor)<br />
N +<br />
CH3 Piperidinium (Pip)<br />
N +<br />
R 1<br />
R 1<br />
Quinolinium (Quin)<br />
N +<br />
R1 R2<br />
R 4<br />
R3<br />
Quaternary ammonium (N)<br />
P +<br />
R1 R2<br />
R 4<br />
R3<br />
Quaternary phosphonium (P (<br />
R 1<br />
R1 ¼ -C2H5, -C3H7, -C4H9,<br />
-C 5H 11, -C 6H 13, -C 7H 15,<br />
-C 8H 17, -C 9H 19, -C 10H 21,<br />
-C14H29, -C16H33, -C18H37,<br />
-C19H39<br />
R 2 ¼ -CH 3,-C 2H 5<br />
R1 ¼ -C2H5, -C3H7, -C4H9,<br />
-C 5H 11, -C 6H 13, -C 8H 17<br />
R1 ¼ -C4H9, -C6H13, -C8H17<br />
R 1 ¼ -C 4H 9<br />
R1 ¼ -C4H9<br />
R1 ¼ -C4H9, -C6H13, -C8H17<br />
R 1-4 ¼ -CH 3,-C 2H 5,-C 3H 7,<br />
-C4H9, -C6H13<br />
R 1-4 ¼ -C 4H 9,-C 6H 13, -C 14H 29<br />
(continued on next page)
356<br />
Table 1 (continued)<br />
in humans (Chemnitius et al., 1999; Pope et al., 2005). Ranke<br />
et al. (2007b) published a comprehensive collection of acetylcholinesterase<br />
inhibition values for 292 compounds covering<br />
a large variety of ILs and closely related salts. Among these<br />
data, only those of the commonly tested ILs are summarized<br />
in Table 2 for the ease of comparison of ILs toxicity from<br />
molecular up to organism levels of biological complexity.<br />
It was found that all observed inhibitory effects on the enzyme<br />
could be exclusively accounted for the cationic moiety (Arning<br />
et al., 2008). In particular, the IL with pyridinium as cationic<br />
core structure inhibited the enzyme slightly stronger than the<br />
imidazolium analogue whereas the compounds based on<br />
phosphonium was less inhibitory. All anion species exerted<br />
no effect on the enzyme activity with only exception of the<br />
fluoride anion and the fluoride containing [SbF6] and [PF6]<br />
species. Both species are known to readily undergo hydrolysis<br />
in contact with moisture and thus the fluoride seems to be the<br />
active compound. The non-inhibiting effects of anion might<br />
be explained by their limited interactions with the active site<br />
of this enzyme (Matzke et al., 2007). In addition, a correlation<br />
between an increasing chain length of the side chains<br />
connected to the cationic head groups and an enhanced<br />
inhibitory potential of the ILs was found. It is believed that the<br />
mechanism involves the similarity of the positively charged<br />
imidazolium or pyridinium to the choline part that binds to<br />
the anionic site of the enzyme, such that the longer alkyl<br />
chain results in an improved fit (Stock et al., 2004).<br />
Sk1adanowski et al. (2005) discussed the usefulness of in<br />
vitro AMP deaminase inhibition assay as a potential molecular<br />
method in prospective risk analysis of imidazolium-based ILs.<br />
The results revealed that IM14 salts associated with [PF 6] ,<br />
[BF 4] , p-tosylate and [Cl] demonstrated a dose-dependent<br />
inhibition of AMP deaminase activity. The IC50 values<br />
(concentration of ILs inhibiting 50% of enzyme activity) for<br />
those containing a fluorine compartment [PF6] and [BF4] are<br />
lower (5 mM) than those for [Cl] and p-tosylate (10 mM), which<br />
indicated the adverse effect of these fluoride-containing<br />
anions. The other study on enzyme inhibition assay dealt with<br />
Head group Side chain<br />
Anion Chloride Cl<br />
Bromide Br<br />
Tetrafluoroborate [BF4]<br />
Hexafluorophosphate [PF6]<br />
Bis(trifluoromethylsulfonyl)imide [(CF 3SO 2) 2N]<br />
Dicyanamide [(CN)2N]<br />
water research 44 (2010) 352–372<br />
the effects of acute exposure of intraperitoneal injection of<br />
aqueous IM18 Br on the antioxidant enzymes of the treated<br />
mouse liver (Yu et al., 2009a). The antioxidant enzymes tested<br />
included superoxide dismutase, catalase, glutathione peroxidase<br />
and glutathione-S-transferase. The results showed that<br />
administration of IM18 Br modified activities of these defense<br />
enzymes in mouse liver, and caused damage to livers of<br />
treated mice at median lethal dose (LD 50) of 35.7 mg/kg.<br />
Though data published by these authors did not cover<br />
a large variety of ILs, the enzyme inhibition assays suggest the<br />
trend in which cationic moiety is the dominating factor<br />
influencing the toxicity of ILs, especially when substituted<br />
with a long alkyl side chain. Regarding the anion types,<br />
perfluoronated ions are of toxicological interest due to<br />
hydrolysis resulting in HF formation, while the others cause<br />
less prominent effect.<br />
2.2. Antibacterial activity of ILs<br />
Bacteria serve as an ideal starting point for ILs toxicity<br />
estimations as they have short generation times. Preliminary<br />
toxicological investigations have shown quaternary ammonium<br />
and pyridinium compounds have critical inhibitory<br />
effects on a variety of bacteria and fungi (Babalola, 1998;<br />
Kelman et al., 2001; Li et al., 1998). In the studies of Pernak’s<br />
group (Cieniecka-Ros1onkiewicz et al., 2005; Pernak et al.,<br />
2001a; Pernak et al., 2001b; Pernak and Chwa1a, 2003; Pernak<br />
et al., 2003; Pernak et al., 2004a), they observed a trend of<br />
increasing toxicity with an increase in the alkyl chain length<br />
substituent in the pyridinium, imidazolium and quaternary<br />
ammonium salts to various bacteria including rods, cocci<br />
and fungi. As a measure of microbial activity of imidazolium<br />
and pyridinium ILs with varying alkyl chain lengths, Docherty<br />
and Kulpa (2005) also used a group of microorganisms<br />
possessing a variety of physiological and respiratory activities.<br />
It was found that imidazolium and pyridinium bromides<br />
incorporated hexyl- and octyl-chain had considerable antimicrobial<br />
effect to pure cultures of Escherichia coli,<br />
F<br />
F<br />
F<br />
B -<br />
F<br />
F<br />
P -<br />
F<br />
F<br />
F<br />
F<br />
F<br />
N -<br />
O O<br />
F3C CF3 S S<br />
O<br />
O<br />
N -<br />
N N
Table 2 – Toxicity of ILs to different levels of biological complexity including enzyme, bacteria, algae, rat cell line, human cell lines, duckweed and invertebrate.<br />
Compound Log10EC50 (mM) a<br />
IM12 Cl 2.06 14<br />
IM12 BF4 IM12 PF6<br />
2.05 14<br />
IM12 (CF3SO2)2N 2.03 14<br />
IM13 Cl 2.27 14<br />
IM13 BF4 2.28 0.03 18<br />
IM13 PF6<br />
2.22 14<br />
IM14 Cl 1.91 0.04 11<br />
Acetylcholin esterase Vibrio<br />
fischeri<br />
2.05 14<br />
IM14 Br 1.90 0.02 18<br />
IM14 BF 4 1.98 0.018 11<br />
Escherichia<br />
coli<br />
Pseudo kirchneriella<br />
subcapitata<br />
Scenedesmus<br />
vacuolatus<br />
IPC-81 HeLa MCF7 b<br />
Lemna<br />
minor<br />
4.55 10<br />
4.33 0.11 19<br />
N.A. N.A. 2.78 0.06 19<br />
N.A. N.A. N.A. N.A. N.A.<br />
N.A. 5.25 0.06 9<br />
N.A. N.A. 3.44 14<br />
4.00 0.04 20<br />
N.A. N.A. N.A.<br />
N.A. N.A. N.A. N.A. 3.92 14<br />
N.A. N.A. N.A. N.A.<br />
N.A. N.A. N.A. N.A. N.A. 3.26 0.04 20<br />
N.A. N.A. N.A.<br />
N.A. N.A. N.A. N.A. >4.30 14<br />
N.A. N.A. N.A. N.A.<br />
3.94 0.06 13<br />
N.A. N.A. N.A. 3.47 14<br />
N.A. N.A. N.A. N.A.<br />
N.A. N.A. N.A. N.A. >3.00 14<br />
N.A. N.A. N.A. N.A.<br />
3.71 0.14 4<br />
2.95 5<br />
3.34 0.13 6<br />
3.47 0.04 19<br />
4.01 0.05 4<br />
3.07 0.03 13<br />
3.35 5<br />
3.27 0.09 6<br />
3.55 0.04 13<br />
3.10 0.17 6<br />
3.12 0.35 16<br />
3.07 0.29 6<br />
N.A. 2.34 0.01 21<br />
N.A. 3.46 0.062 3<br />
4.60 0.02 9<br />
N.A. 2.11 11<br />
2.26 0.08 11<br />
3.55 14<br />
N.A. 3.43 14<br />
3.12 14<br />
N.A. N.A. 2.82 11<br />
3.44 0.11 20<br />
3.72 0.05 17<br />
3.66 0.08 20<br />
Daphnia<br />
magna c<br />
1.93 4<br />
1.93 0.06 1<br />
N.A. N.A. 1.57 4<br />
1.56 0.07 1<br />
1.85 0.06 22<br />
N.A. 2.49 7<br />
1.68 4<br />
1.67 0.11 1<br />
IM14 PF6 2.15 0.05 18<br />
4.15 0.06 9<br />
2.20 0.04 21<br />
N.A. 3.10 14<br />
4.14 0.22 17<br />
N.A. N.A. 1.85 4<br />
1.85 0.10 1<br />
IM14 (CF3SO2)2N 1.96 0.021 11<br />
3.39 0.08 4<br />
2.55 0.15 9<br />
1.80 0.07 12<br />
1.81 0.15 11<br />
2.68 14<br />
3.07 0.08 20<br />
N.A. 2.45 0.08 11<br />
N.A.<br />
IM14 (CN)2N 1.95 0.07 18<br />
3.67 0.10 4<br />
2.99 5<br />
N.A. N.A. N.A. 3.15 14<br />
N.A. N.A. N.A. N.A.<br />
IM15 Cl 1.96 14<br />
N.A. N.A. N.A. N.A. >3.00 14<br />
N.A. N.A. N.A. N.A.<br />
IM15 BF4<br />
1.86 14<br />
3.14 0.02 13<br />
N.A. N.A. N.A. >3.00 14<br />
N.A. N.A. N.A. N.A.<br />
IM15 PF6<br />
1.85 14<br />
N.A. N.A. N.A. N.A. >3.00 14<br />
N.A. N.A. N.A. N.A.<br />
IM16 Cl 1.92 14<br />
1.94 10<br />
2.32 0.16 6<br />
2.91 0.09 13<br />
N.A. 1.92 0.09 21<br />
0.08 19<br />
2.85 14<br />
N.A. N.A. N.A. N.A.<br />
IM16 Br N.A. 1.42 0.12 4<br />
0.81 5<br />
N.A. 2.57 0.15 2<br />
N.A. N.A. N.A. N.A. N.A. 0.78 4<br />
1.06 0.04 22<br />
IM16 BF4 1.88 14<br />
3.18 0.03 13<br />
N.A. N.A. N.A. 2.98 14<br />
N.A. N.A. N.A. N.A.<br />
IM16 PF6<br />
1.88 14<br />
2.17 0.06 6<br />
3.25 0.67 9<br />
N.A. N.A. 2.91 14<br />
N.A. N.A. N.A. N.A.<br />
IM16 (CF3SO2) 2N 2.15 14<br />
N.A. 2.53 0.15 9<br />
N.A. N.A. 2.24 14<br />
N.A. 2.81 8<br />
N.A. N.A.<br />
IM17 Cl 2.07 14<br />
N.A. N.A. N.A. N.A. 2.53 14<br />
N.A. N.A. N.A. N.A.<br />
IM17 BF4<br />
2.12 14<br />
2.44 0.06 13<br />
N.A. N.A. N.A. 2.58 14<br />
N.A. N.A. N.A. N.A.<br />
IM17 PF6<br />
1.91 14<br />
N.A. N.A. N.A. N.A. 2.30 14<br />
N.A. N.A. N.A. N.A.<br />
IM18 Cl 1.60 14<br />
1.19 0.11 6<br />
1.01 0.06 19<br />
N.A. 1.46 21<br />
2.67 0.37 19<br />
2.01 14<br />
N.A. N.A. N.A. N.A.<br />
IM18 Br N.A. 0.63 0.07 4<br />
0.07 5<br />
N.A. 1.65 0.25 2<br />
N.A. N.A. 2.48 0.04 20<br />
N.A. N.A. 1.33 4<br />
0.54 0.12 22<br />
(continued on next page)<br />
water research 44 (2010) 352–372 357
Table 2 (continued)<br />
Compound Log 10EC 50 (mM) a<br />
Acetylcholin esterase Vibrio<br />
fischeri<br />
Escherichia<br />
coli<br />
Pseudo kirchneriella<br />
subcapitata<br />
Scenedesmus<br />
vacuolatus<br />
IPC-81 HeLa MCF7 b<br />
IM18 BF4 1.53 0.025 11<br />
1.41 0.07 13<br />
N.A. N.A. 2.30 11<br />
1.59 14<br />
2.48 0.02 20<br />
2.84 8<br />
0.90 7<br />
N.A.<br />
IM18 PF6 2.03 14<br />
0.95 0.12 6<br />
2.64 0.15 9<br />
N.A. N.A. 1.96 14<br />
N.A. N.A. N.A. N.A.<br />
IM18 (CF3SO2)2N 2.03 14<br />
N.A. N.A. N.A. N.A. 1.64 14<br />
2.28 0.02 20<br />
N.A. N.A. N.A.<br />
IM19 BF4 N.A. 0.72 0.04 13<br />
N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A.<br />
IM1-10 Cl 1.09 14<br />
0.50 0.07 13<br />
0.23 0.06 19<br />
N.A. N.A. 3.57 0.06 19<br />
1.34 14<br />
N.A. N.A. N.A. N.A.<br />
IM1-10 BF4 1.10 0.04 18<br />
0.18 0.06 13<br />
N.A. N.A. N.A. 0.77 14<br />
N.A. N.A. N.A. N.A.<br />
IM1-10 PF6 1.68 14<br />
N.A. N.A. N.A. N.A. 1.50 14<br />
N.A. N.A. N.A. N.A.<br />
IM1-14 Cl 0.54 14<br />
0.15 0.07 19<br />
N.A. N.A. 2.48 0.2 19<br />
0.42 14<br />
N.A. N.A. N.A. N.A.<br />
IM1-16 Cl 0.68 14<br />
0.23 0.08 19<br />
N.A. N.A. > 2.00 19<br />
0.19 14<br />
N.A. N.A. N.A. N.A.<br />
IM1-18 Cl 0.96 14<br />
1.45 0.05 19<br />
N.A. N.A. > 2.00 19<br />
0.01 14<br />
N.A. N.A. N.A. N.A.<br />
IM1-19 Cl 1.36 14<br />
N.A. N.A. N.A. N.A. 1.40 14<br />
N.A. N.A. N.A. N.A.<br />
IM1-19 BF4 1.43 14<br />
N.A. N.A. N.A. N.A. 1.65 14<br />
N.A. N.A. N.A. N.A.<br />
IM1-19 PF6<br />
1.62 14<br />
N.A. N.A. N.A. N.A. 1.85 14<br />
N.A. N.A. N.A. N.A.<br />
IM22 Br 2.08 14<br />
N.A. N.A. N.A. N.A. >3.00 14<br />
N.A. N.A. N.A. N.A.<br />
IM23 Br 2.21 14<br />
N.A. N.A. N.A. N.A. >3.30 14<br />
N.A. N.A. N.A. N.A.<br />
IM24 BF4 2.03 0.01 18<br />
2.8 0.04 13<br />
N.A. N.A. N.A. 3.26 14<br />
4.36 0.09 17<br />
N.A. N.A. N.A.<br />
IM25 BF4 N.A. 3.14 13<br />
N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A.<br />
IM26 Br 1.77 14<br />
N.A. N.A. N.A. N.A. 2.01 14<br />
N.A. N.A. N.A. N.A.<br />
IM26 BF4 1.84 14<br />
2.15 0.05 13<br />
N.A. N.A. N.A. 2.26 14<br />
N.A. N.A. N.A. N.A.<br />
IM2-10 Br 0.92 14<br />
N.A. N.A. N.A. N.A. 0.53 14<br />
N.A. N.A. N.A. N.A.<br />
Py Cl >3.00 14<br />
N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A.<br />
Py2 Cl 2.10 14<br />
N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A.<br />
Py3 Br 2.22 14<br />
N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A.<br />
Py3 (CF3SO2)2N 2.21 14<br />
N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A.<br />
Py4 Cl 1.70 14<br />
3.41 0.08 4<br />
2.64 5<br />
3.18 0.06 19<br />
N.A. 2.57 0.06 21<br />
2.59 0.11 19<br />
N.A. N.A. N.A. 2.32 0.18 19<br />
N.A.<br />
Py4 Br 1.77 14<br />
3.40 0.01 4<br />
2.73 5<br />
N.A. N.A. N.A. 3.90 14<br />
3.50 0.07 20<br />
N.A. N.A. N.A.<br />
Py4 BF4 1.80 14<br />
N.A. N.A. N.A. N.A. 3.18 14<br />
N.A. N.A. N.A. N.A.<br />
Py4 PF6<br />
1.84 14<br />
N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A.<br />
Py4 (CN)2N N.A. 3.31 0.10 4<br />
N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A.<br />
2.61 5<br />
Py5 Br 1.52 14<br />
N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A.<br />
Py5 (CF3SO2)2N 1.55 14<br />
N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A.<br />
Py6 Cl 1.72 14<br />
N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A.<br />
Py6 Br N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. 1.07 4<br />
Py6 PF6 1.76 14<br />
N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A.<br />
Py6 (CF3SO2)2N 1.85 14<br />
N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A.<br />
Py8 Cl 1.60 14<br />
N.A. N.A. N.A. N.A. 1.27 14<br />
N.A. N.A. N.A. N.A.<br />
Py8 (CF3SO2) 2N 1.40 14<br />
N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A.<br />
Py4-2Me Cl 0.70 14<br />
N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A.<br />
Py4-2Me BF4<br />
0.82 14<br />
N.A. N.A. N.A. N.A. 3.25 14<br />
N.A. N.A. N.A. N.A.<br />
Lemna<br />
minor<br />
Daphnia<br />
magna c<br />
358<br />
water research 44 (2010) 352–372
Py4-3Me Cl 1.15 14<br />
N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A.<br />
Py4-3Me Br N.A. 2.75 0.13 4<br />
N.A. 3.46 0.062 3<br />
N.A. N.A. N.A. N.A. N.A. 1.76 4<br />
2.12 5<br />
Py4-3Me BF4 1.53 0.02 18<br />
N.A. N.A. N.A. N.A. 3.30 14<br />
N.A. N.A. N.A. N.A.<br />
Py4-3Me PF6 1.45 0.02 18<br />
N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A.<br />
Py4-3Me (CN) 2N 1.22 14<br />
2.66 0.05 4<br />
1.99 5<br />
N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A.<br />
Py6-3Me Cl 1.06 14<br />
1.44 10<br />
N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A.<br />
Py6-3Me Br N.A. 2.06 0.16 4<br />
N.A. N.A. N.A. N.A. N.A. N.A. N.A. 0.59 4<br />
1.48 5<br />
Py6-4Me Cl 1.44 14<br />
N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A.<br />
Py6-4Me BF4<br />
1.48 14<br />
N.A. N.A. N.A. N.A. 2.17 14<br />
N.A. N.A. N.A. N.A.<br />
Py8-3Me Cl 0.64 14<br />
N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A.<br />
Py8-3Me Br N.A. 0.79 0.05 4<br />
N.A. N.A. N.A. N.A. N.A. 1.00 8<br />
N.A. 0.40 4<br />
Py8-4Me Cl 1.11 14<br />
Py8-4Me BF4 1.22 14<br />
Pyr14 Cl 1.92 14<br />
Pyr14 Br 1.93 14<br />
0.25 5<br />
N.A. N.A. N.A. N.A. 1.63 14<br />
N.A. N.A. N.A. N.A. 1.49 14<br />
>4.30 19<br />
N.A. N.A. 3.37 0.10 19<br />
>4.30 14<br />
N.A. N.A. 3.67 0.28 3<br />
N.A. 3.77 14<br />
N.A. N.A. N.A. N.A.<br />
N.A. N.A. N.A. N.A.<br />
N.A. N.A. 2.16 0.25 19<br />
N.A.<br />
N.A. 4.64 0.02 15<br />
N.A. N.A.<br />
Pyr14 BF4 1.91 14<br />
N.A. N.A. N.A. N.A. 2.90 14<br />
N.A. N.A. N.A. N.A.<br />
Pyr14 (CF3SO2)2N 2.13 14<br />
N.A. N.A. >2.38 12<br />
2.53 19<br />
3.01 14<br />
N.A. 3.14 8<br />
2.98 0.32 19<br />
N.A.<br />
Pyr14 (CN)2N 1.98 14<br />
N.A. N.A. N.A. N.A. 4.23 14<br />
N.A. N.A. N.A. N.A.<br />
Pyr16 Cl 2.48 14<br />
2.99 10<br />
N.A. N.A. N.A. 2.91 14<br />
N.A. N.A. N.A. N.A.<br />
Pyr16 (CF3SO2) 2N 2.60 14<br />
N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A.<br />
Pyr18 Cl 2.36 14<br />
N.A. N.A. N.A. N.A. 2.59 14<br />
N.A. N.A. N.A. N.A.<br />
Pyr18 BF4<br />
2.02 14<br />
N.A. N.A. N.A. N.A. 1.82 14<br />
N.A. N.A. N.A. N.A.<br />
Pyr66 2.08 14<br />
N.A. N.A. N.A. N.A. 1.23 14<br />
N.A. N.A. N.A. N.A.<br />
Mor14 Cl N.A. >4.30 19<br />
N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A.<br />
Mor14 Br 2.71 14<br />
N.A. N.A. N.A. >4.00 19<br />
>4.30 14<br />
N.A. N.A. 3.11 0.13 19<br />
N.A.<br />
Mor14 (CF3SO2) 2N 2.78 14<br />
N.A. N.A. N.A. 2.00 19<br />
3.43 14<br />
N.A. N.A. 3.15 0.13 19<br />
N.A.<br />
Pip14 Br 1.83 14<br />
4.27 0.09 19<br />
N.A. N.A. 3.27 0.12 19<br />
4.03 14<br />
N.A. 4.15 0.04 15<br />
0.47 19<br />
N.A.<br />
Pip14 (CF3SO2)2N 1.78 14<br />
N.A. N.A. N.A. 2.08 19<br />
3.41 14<br />
N.A. 2.93 8<br />
2.85 0.07 19<br />
N.A.<br />
Quin4 Br 0.79 14<br />
N.A. N.A. N.A. N.A. 2.32 14<br />
N.A. N.A. N.A. N.A.<br />
Quin4 BF4 0.62 14<br />
N.A. N.A. N.A. N.A. 2.16 14<br />
N.A. N.A. N.A. N.A.<br />
Quin6 BF4<br />
0.48 14<br />
N.A. N.A. N.A. N.A. 1.07 14<br />
N.A. N.A. N.A. N.A.<br />
Quin8 Br N.A N.A. N.A. N.A. N.A. 0.03 14<br />
N.A. N.A. N.A. N.A.<br />
Quin8 BF4 0.30 14<br />
N.A. N.A. N.A. N.A. 0.17 14<br />
N.A. N.A. N.A. N.A.<br />
N1111 Br N.A. >5.00 4<br />
N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A.<br />
N1114 (CF3SO2)2N 2.60 14<br />
N.A. N.A. N.A. N.A. 3.61 14<br />
N.A. N.A. N.A. N.A.<br />
N1123 (CF3SO2) 2N 2.34 14<br />
N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A.<br />
N1124 Cl 2.06 14<br />
N.A. N.A. N.A. >4.00 19<br />
>4.30 14<br />
N.A. N.A. 0.83 0.67 19<br />
N.A.<br />
N1124 (CF3SO2)2N 2.03 14<br />
N.A. N.A. N.A. 1.78 0.17 19<br />
3.43 13<br />
N.A. N.A. N.A. N.A.<br />
N2222 Cl 2.80 14<br />
N.A. N.A. N.A. N.A. >3.48 14<br />
N.A. N.A. N.A. N.A.<br />
N2222 Br N.A. >5.00 4<br />
N.A. N.A. N.A. N.A. 4.26 0.04 20<br />
N.A. N.A. N.A.<br />
N2226 Br N.A. 2.46 0.16 4<br />
N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A.<br />
N4444 Br 2.30 14<br />
3.27 0.07 4<br />
N.A. N.A. N.A. 2.25 14<br />
N.A. N.A. N.A. 1.47 4<br />
P4444 Br 2.61 14<br />
2.71 4<br />
N.A. N.A. N.A. 1.66 14<br />
N.A. N.A. N.A. 0.95 4<br />
P666-14 Br 2.85 14<br />
3.41 0.02 4<br />
N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A.<br />
4.25 8<br />
4.15 8<br />
(continued on next page)<br />
water research 44 (2010) 352–372 359
360<br />
Table 2 (continued)<br />
Compound Log10EC50 (mM) a<br />
Daphnia<br />
magna c<br />
Lemna<br />
minor<br />
IPC-81 HeLa MCF7 b<br />
Scenedesmus<br />
vacuolatus<br />
Pseudo kirchneriella<br />
subcapitata<br />
Escherichia<br />
coli<br />
Acetylcholin esterase Vibrio<br />
fischeri<br />
N.A. N.A. N.A. N.A. 0.48 14<br />
N.A. N.A. N.A. N.A.<br />
N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A.<br />
N.A. N.A. N.A. N.A. 0.24 14<br />
1.90 0.05 20<br />
N.A. N.A. N.A.<br />
N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A.<br />
P666-14 BF4 3.47 0.08 18<br />
P666-14 PF6<br />
>3.30 17<br />
P666-14 (CF3SO2)2N >3.48 14<br />
P666-14 (CN) 2N 3.40 0.2 18<br />
References: 1 Bernot et al. (2005a); 2 Cho et al. (2007); 3 Cho et al. (2008a,b); 4 Couling et al. (2006); 5 Docherty and Kulpa (2005); 6 Garcia et al. (2005); 7 Jastorff et al. (2005); 8 Kumar et al. (2009); 9 Lee et al. (2005);<br />
10 11 12 13 14 15 16 17 18<br />
Luis et al. (2007); Matzke et al. (2007); Pretti et al. (2008); Ranke et al. (2004); Ranke et al. (2007b); Salminen et al. (2007); Samorì et al. (2007); Stepnowski et al. (2004); Stock et al. (2004);<br />
19 Stolte et al. (2007a); 20 Wang et al. (2007); 21 Wells and Coombe (2006); 22 Yu et al. (2009).<br />
a N.A. means not available (not determined).<br />
b Toxicity of ILs is expressed as log10IC50 (mM) in case of MCF7 cell line.<br />
c Toxicity of ILs is expressed as log10LC50 (mM) in case of D. magna.<br />
water research 44 (2010) 352–372<br />
Staphylococcus aureus, Bacillus subtilis, Pseudomonas fluorescens<br />
and Saccharomyces cerevisiae. The anion performed nearly no<br />
effect on antimicrobial activity in the case of imidazolium<br />
analogues (Docherty and Kulpa, 2005; Garcia et al., 2005; Lee<br />
et al., 2005; Pernak et al., 2003; Pernak et al., 2004a) whereas<br />
this was not the case for phosphonium salts. Within the group<br />
of alkyltrihexylphosphonium ILs in the study of Cieniecka-<br />
Ros1onkiewicz et al. (2005), both cation structure and the type<br />
of anion had effects on the biological activity.<br />
The antibacterial activity of ILs not only involves in<br />
hampering the growth rate of microbes but also interferes<br />
with their productivity. Matsumoto et al. (2004a) tested the<br />
toxicity of imidazolium-based ILs to lactic acid producing<br />
bacterium Lactobacillus rhamnosus to examine whether these<br />
compounds can replace conventional organic solvents in the<br />
extractive fermentation of lactate. The results showed that<br />
the bacterium L. rhamnosus grew, consumed glucose, and<br />
produced lactate in the presence of imidazolium-based ILs.<br />
A change of alkyl length in the imidazolium cation had little<br />
difference on the survival of the cells. In a similar study<br />
(Matsumoto et al., 2004b), they focused on hiochii bacteria,<br />
Lactobacillus homochiochii and Lactobacillus fructivorans and also<br />
found that the bacteria could produce lactic acid in the presence<br />
of ILs. Nonetheless, the lactic acid producing activities of<br />
these bacteria generally decreased with the extension of alkyl<br />
chain length in the imidazolium cation moiety.<br />
Water miscible ILs had various effects on the physiology of<br />
Clostridium sporogenes when tested as additives in culture<br />
media or reaction media for reduction of nitrobenzene<br />
(Dipeolu et al., 2008). In their study, 2-hydroxyethyltrimethylammonium<br />
dimethylphosphate and N,N-dimethylethanolammonium<br />
acetate increased the growth rate of C. sporogenes;<br />
by contrast, IM14 BF4 and IM12 EtSO 4 inhibited growth.<br />
Although IM12 EtSO 4 inhibited growth, it was sufficiently<br />
non-toxic to allow efficient reduction of nitrobenzene using<br />
harvested cells. Thus, it is recommended that both noninhibitory<br />
and partially inhibitory ILs should be screened for<br />
use in biotransformation. Nonetheless, Ganske and Bornscheuer<br />
(2006) referred that ILs could have substantial inhibitory<br />
effects on the growth of microorganisms when they<br />
explored the effects of the two most commonly used ILs IM14<br />
BF 4 and IM14 PF 6 on the growth of E. coli, Pichia pastoris and<br />
Bacillus cereus.<br />
Regarding inhibition assays used in assessment of environmental<br />
potential risk of a compound in aquatic milieu, the<br />
bioluminescence assay using Vibrio fischeri (formerly known as<br />
Photobacterium phosphoreum) is one of the most applied (Kaiser<br />
and Palabrica, 1991; Steinberg et al., 1995). This is a rapid, costeffective,<br />
and well-established method for toxicity determination<br />
focusing on environmental issues, and also a standard<br />
ecotoxicological bioassay in Europe (DIN EN ISO 11348). The<br />
published data on ILs toxicity towards V. fischeri were listed on<br />
Table 2 and were comprehensively interpreted in the study of<br />
Peraccini et al. (2007). Although it has been claimed that<br />
modifications of the anion lead to changes in chemical and<br />
physical properties of ILs (Sheldon, 2001), no clear increase in<br />
toxicity caused by the anion could be observed, and toxicity<br />
seemed to be determined mainly by the cationic component<br />
(Ranke et al., 2004). This is likely explained by the fact that<br />
lipophilic part of the molecules can be intercalated into the
membrane, whereas their ionic head group is at least partially<br />
solvated in the aqueous solution, as suggested by Austin et al.<br />
(1998). The ILs toxicity was also observed to correlate directly<br />
with the length of the n-alkyl residues in the methylimidazolium<br />
cation (Romero et al., 2008). Interestingly, Ranke<br />
et al. (2004) noted a slight hormetic effect at concentrations<br />
below inhibitory concentrations. Concerning the anionic<br />
influence, compounds with [PF6] were found to be slightly<br />
more toxic than compounds with other anions in their study<br />
(Ranke et al., 2004). The anion [(CF 3SO 2) 2N] showed no<br />
intrinsic toxicity to V. fischeri in the report of Matzke et al.<br />
(2007); in contrast, an increased in toxicity was found for all<br />
tested compounds combined with [(CF3SO2)2N] for V. fischeri<br />
(Stolte et al., 2007a). Couling et al. (2006) extended the bioluminescence<br />
inhibition assay to pyridinium derivatives and it<br />
was noted that the quaternary ammonium compounds<br />
seemed to be less toxic to V. fischeri than the pyridinium<br />
and imidazolium analogues. Also, the quantitative structureproperty<br />
relationship (QSPR) modeling suggested that imidazolium<br />
cations, with two nitrogen atoms, are predicted to<br />
be more toxic than pyridinium moieties, which only have<br />
one nitrogen atom in the structure. In addition, the QSPR<br />
correlation predicted that quaternary ammonium cations are<br />
less toxic than those with cations containing nitrogen-bearing<br />
rings, which was in agreement with the experimental results<br />
(Couling et al., 2006). However, in contrast to the cases of<br />
aromatic ILs and ammonium compounds, the authors were<br />
unsuccessful in modeling the behavior of phosphonium salts<br />
using the developed correlation.<br />
2.3. Toxicity of ILs to algae<br />
As algae are primary producers, either directly or indirectly, of<br />
organic matter required by animals in freshwater food chains,<br />
their ecology is crucial in providing the energy for sustaining<br />
other higher trophic levels. The ubiquity of algae makes these<br />
organisms ideal for toxicological studies and, because they<br />
have a short life cycle they can respond quickly to environmental<br />
change (Blaise, 1993; Lewis, 1995). To date, several<br />
groups have focused their attention on the use of algal<br />
primary producers to assess the effects of ILs to aquatic<br />
environments (Cho et al., 2007; Cho et al., 2008a,b,c;<br />
Grabinska-Sota and Kalka, 2006; Kulacki and Lamberti, 2008;<br />
Lata1a et al., 2005; Matzke et al., 2007; Matzke et al., 2008; Pham<br />
et al., 2008a,b; Pretti et al., 2009; Stolte et al., 2007a; Wells and<br />
Coombe, 2006). Cho and co-workers used Pseudokirchneriella<br />
subcapitata (formerly known as Selenastrum capricornutum) to<br />
study the effect of different head groups, side chains and<br />
anions of ILs on algal growth rate and photosynthetic activity.<br />
The data revealed that the toxic influence of ILs on growth<br />
rates were more significant than those of photosynthetic<br />
performance (Pham et al., 2008b). Once again, the trend of<br />
increasing toxicity with increasing alkyl chain length was<br />
observed in their reports (Cho et al., 2007; Pham et al., 2008b).<br />
Regarding the anionic effects, P. subcapitata was sensitive<br />
to the anion moieties in the order: [SbF6] > [PF6] ><br />
[BF4] > [CF3SO3] > [C8H17OSO3] > [Br] z [Cl] . In particularly,<br />
it was found that with respect to IL incorporating perfluorinated<br />
anion (i.e. IM14 BF4), EC50 values (concentrations<br />
which lead to a 50% reduction of the exposed organisms<br />
water research 44 (2010) 352–372 361<br />
relative to control) of the previously prepared stock solution<br />
(6 months prior to experiment) were significantly lower<br />
compared to those of the freshly made one (Pham et al.,<br />
2008a). This might be due to hydrolytic effects of IM14 BF4<br />
leading to fluoride formation, as confirmed by ion chromatography<br />
analysis. This implies that after ILs are released into<br />
the aqueous system; they can become more hazardous than<br />
expected by laboratory data with fresh ILs. In a detailed study<br />
on hydrolysis of fluoride-containing anions, Cho et al. (2008a)<br />
showed that IM14 SbF 6 generated a greater amount of fluoride<br />
compared to IM14 BF 4, but no fluoride formation occurred<br />
with the hexafluorophosphate. When only small amounts of<br />
fluoride ions were formed from IM14 SbF6 and IM14 BF4 within<br />
96 h, the formed fluoride ion did not affect the algal growth<br />
rate. Nevertheless, the fluoride ion formation from IM14 BF4<br />
increased with incubating time of the stock solution; thus, the<br />
toxicity might significantly increase according to the further<br />
formed fluoride ions. In view of cationic effect, Pyr14 Br was<br />
found to be the least toxic of all the ILs tested to P. subcapitata<br />
(Cho et al., 2008b). For the limnic green alga Scenedesmus<br />
vacuolatus, a severe toxicity was found for 1-butyl-4-(dimethylamino)pyridinium,<br />
whereas the quaternary ammonium<br />
and morpholinium compounds exhibited no toxicity (Stolte<br />
et al., 2007a). Despite the extensive studies on the toxicological<br />
impact of ILs towards freshwater phytoplankton, inhibition<br />
mechanism of both the growth rate and photosynthetic<br />
activity by ILs has not been described by the authors.<br />
Lata1a et al. (2005), who selected two marine algae Oocystis<br />
submarina (green algae) and Cyclotella meneghiniana (diatom) as<br />
testing organisms, found that the two species differed<br />
dramatically in their ability to recover from IL exposure.<br />
Additionally, it was discovered that IL toxicity declined with<br />
increasing salinity. The lower toxicity of IL in this case is<br />
probably due to the reduced permeability of IL cations through<br />
the algal cell walls. High amounts of chloride provide a good<br />
ion-pairing environment for imidazolium cations, which<br />
consequently compete with hydroxyl or silanol functional<br />
groups in the cell-wall structure of green alga and diatom,<br />
respectively. Though no information on EC50 values was<br />
described, the facts emerged from this work provide useful<br />
information in the further fate assessment of ILs in marine<br />
environments.<br />
2.4. Cytotoxicity of ILs<br />
As a cellular test system, promyelotic leukemia rat cell line<br />
IPC-81 has been frequently used in cytotoxicity assays of ILs,<br />
with the reduction of the WST-1 dye as an indicator of cell<br />
viability (Matzke et al., 2007; Ranke et al., 2004; Ranke et al.,<br />
2007a; Stasiewicz et al., 2008; Stolte et al., 2006; Stolte et al.,<br />
2007b; Torrecilla et al., 2009). It was observed that ILs with<br />
polar ether, hydroxyl and nitrile functional groups within the<br />
side chains exhibited low cytotoxicity compared to those<br />
incorporated with ‘‘simple’’ alkyl side chains (Kumar et al.,<br />
2009; Stasiewicz et al., 2008; Stolte et al., 2007b). Those functional<br />
groups were thought to impede cellular uptake by<br />
membrane diffusion and reduce lipophilicity based interactions<br />
with the cell membrane (Stolte et al., 2007b). Taking<br />
a closer look at the effects of sub-structural elements of ILs,<br />
[(CF3SO2)2N] anion and 4-(dimethylamino)pyridinium cation
362<br />
were described to have intrinsic effects of anion and head<br />
group on cytotoxicity, respectively (Stolte et al., 2007b). The<br />
well known side chain length effect (decrease in EC50 values<br />
with elongation of the alkyl side chain) could also be<br />
confirmed in these studies.<br />
To date many studies have analyzed the toxicity of ILs on<br />
human cell lines (Frade et al., 2007; García-Lorenzo et al.,<br />
2008; Hassoun et al., 2002; Kumar et al., 2009; Salminen et al.,<br />
2007; Stepnowski et al., 2004; Wang et al., 2007). These in vitro<br />
systems have been extremely beneficial in studying the<br />
molecular basis of chemical’s biological activity, including its<br />
toxic mode of action (Blaauboer et al., 1998) and could facilitate<br />
extrapolation of in vitro data with regard to possible<br />
effects on humans (Malich et al., 1997). Most of studies dealt<br />
with HeLa cells exemplifying prototypical cells of the human<br />
epithelium which is normally the site of first contact of an<br />
organism with toxicants. According to Stepnowski et al.<br />
(2004), the cytotoxicity data implied that effects of IM14<br />
cation coupled with chloride, tetrafluoroborate or hexafluorophosphate<br />
were probably dependent on the anionic<br />
moieties. The lowest effect concentrations for tetrafluoroborate<br />
species were found to be 0.63 mM, whereas<br />
hexafluorophosphate and chloride inhibited HeLa cell growth<br />
at comparably high concentrations of >10 mM. Surprisingly,<br />
when the anion effect was compared, the strongest inhibition<br />
was found for [PF6] . This might be due to hydrolysis<br />
affecting fluoride formation, thus causing serious toxicological<br />
consequences through the decomposition product.<br />
A similar phenomenon was observed by Ranke et al. (2004) in<br />
IPC-81 leukemia cells, where the lower toxicity of 1-n-butyl-3methylimidazolium<br />
hexafluorophosphate in comparison to<br />
the hexafluorophosphate anion alone was explained by<br />
reduced anion uptake due to the formation of an ion pair.<br />
The anion in this ion pair can, however, also be partially<br />
decomposed. This was shown in recent work by Swatloski<br />
et al. (2003), who identified traces of 1-n-butyl-3-methylimidazolium<br />
fluoride hydrate as a decomposition product<br />
formed during the purification of the 1-n-butyl-3-methylimidazolium<br />
hexafluorophosphate.<br />
As shown by Wang et al. (2007) the phosphonium<br />
bis(trifluoromethylsulfonyl)imide salts performed the highest<br />
inhibitory to HeLa cells, followed by alkylimidazolium,<br />
alkylpyridinium, alkyltriethylammonium and N-alkyl-N,Ndimethyl-N-(2-hydroxylethyl)ammonium<br />
salts, in decreasing<br />
order. For each cation class the toxicity increased with<br />
increasing chain length of the alkyl substituent for a given anion:<br />
1-ethyl-3-methylimidazolium bromide yielded an EC50 of<br />
8.4 mM, substituting the ethyl moiety for a butyl group led to an<br />
EC50 of 2.8 mM, and for an octyl moiety an EC50 of 0.3 mM. This<br />
result was consistent with what has been observed in other<br />
studies. Salts containing the tetrafluoroborate anion showed the<br />
highest EC50, followed closely by bromide and chloride. Bis(trifluoromethylsulfonyl)imide<br />
salts were significantly more toxic<br />
than their halide counterparts. However, the effect of changing<br />
the anion was smaller than that of changing the alkyl substituent,<br />
e.g. while 1-ethyl-3-methylimidazolium tetrafluoroborate<br />
was observed to have an EC 50 of 9.9 mM, the corresponding<br />
bromide and bis(trifluoromethylsulfonyl)imide salts had EC50 of<br />
8.4 and 1.8 mM, respectively – these all considerably less<br />
toxic than 1-octyl-3-methylimidazolium bromide.<br />
water research 44 (2010) 352–372<br />
The CaCo-2 cells were used in the study of García-Lorenzo<br />
et al. (2008) with the aim of a convenient screening method<br />
for obtaining first rough estimates for the toxic potential of<br />
ILs. The obtained data showed that in general, ILs with longer<br />
alkyl chains were more lipophilic than those with shorter<br />
alkyl chains. The former can be presumed to have a tendency<br />
to be incorporated into the phospholipid bilayers of biological<br />
membranes. In this respect, some authors have indicated<br />
that the increased toxicity of longer ILs can be accounted for<br />
enhanced membrane permeability altering the physical<br />
properties of the lipid bilayer (Lata1a et al., 2005; Ranke et al.,<br />
2004; Stepnowski et al., 2004). Additionally, it has been<br />
proposed that the mode of toxic action for ILs takes place<br />
through membrane disruption because of the structural<br />
similarity of imidazolium-based ILs to detergent, pesticides<br />
and antibiotics able to cause membrane-bound protein<br />
disturbance (Docherty and Kulpa, 2005). Recently, Ranke et al.<br />
(2007a,b) have demonstrated that lipophilicity of ILs dominates<br />
their in vitro cytotoxicity over a wide range of structural<br />
variations. The contribution of the anionic part of the ILs to<br />
the observed biological effect was evaluated by comparing<br />
the EC 50 values obtained for the cations IM16 and IM18,<br />
combined with two different anions [Cl] and [PF 6] . For both<br />
cations, a stronger toxic effect was found for chloride derivatives,<br />
but not for fluoride containing hexafluorophosphate.<br />
A similar result was reported by Stock et al. (2004) where the<br />
inhibitory effects of IM14 Cl and IM14 PF6 on the acetylcholinesterase<br />
activity were compared. In addition, slightly<br />
higher cytotoxicity for the chloride derivative has also been<br />
observed when the cytotoxicity of IM14 Cl and IM14 PF6 on<br />
HeLa cells was tested (Stepnowski et al., 2004). This implies<br />
the effect of perfluorinated ions is not drastic to all but vary<br />
according to species of organisms tested. Several authors<br />
have pointed out that altering the anion has only minimal<br />
effects on the toxicity of several imidazolium compounds<br />
(Bernot et al., 2005a; Garcia et al., 2005; Ranke et al., 2004).<br />
This indicates that ILs toxicity seems to be related to the alkyl<br />
chain branching and to the hydrophobicity of the imidazolium<br />
cation but not to the various anions. In this respect,<br />
a recent study using the IPC-81 rat leukemia cell line with<br />
a large pool of anions demonstrated that most of the<br />
commercially available anions showed no or only marginal<br />
cytotoxic effects. However, anionic compartments with<br />
lipophilic and hydrolysable structural elements are likely to<br />
be of considerable relevance with respect to the toxicity of ILs<br />
(Stolte et al., 2006).<br />
In a recent study (Frade et al., 2007), the human cell lines<br />
such as HT-29 and CaCo-2 cells were utilized to estimate the<br />
inhibitory effect of ILs with several types of cations and<br />
anions. In both cells, IM14, IM12OH (1-(2-hydroxyethyl)-3methylimidazolium),<br />
IM12O2O1 (1-(2-(2-methoxyethoxy)ethyl)-3-methylimidazolium)<br />
and cholines were the least<br />
toxic cations independently of the anion. Within the studied<br />
combinations, it can be noted that IM14 PF 6, IM14 acesulfame,<br />
IM12OH BF 4/PF 6, IM12O2O1 BF 4/PF 6, IM12OH acesulfame and<br />
IM12OH saccharine are not toxic and present good alternatives<br />
to organic solvents. Meanwhile, increasing the length of<br />
the substituent chain may contribute to a significant<br />
increasing of imidazolium toxicity. It was also noted that<br />
[(CF3SO2)2N] anion decreased the toxicity to a large extent,
independently of the cation and for both cell types, which was<br />
in accordance with Salminen et al. (2007).<br />
2.5. Phytotoxicity of ILs<br />
The studies on phytotoxic activity of ILs were conducted<br />
mostly on the duckweed, Lemna minor, a common aquatic<br />
vascular plant (Jastorff et al., 2005; Larson et al., 2008; Matzke<br />
et al., 2007; Stolte et al., 2007a). In general, 1-alkyl-3methylimidazolium<br />
compounds with longer alkyl chains were<br />
more toxic to L. minor than those with short alkyl chain<br />
lengths. Imidazolium and pyridinium cations with butyl<br />
groups had similar EC50s (the concentrations that produced<br />
a 50% reduction in root growth) (39.07 and 32.54 mM, respectively);<br />
while the equivalent ammonium cation had a much<br />
higher EC50 (101.48 mM; i.e., less toxic) (Larson et al., 2008). In<br />
consideration of anionic effect, [(CF 3SO 2) 2N] was found to<br />
cause moderate toxicity to this duckweed (EC 50 ¼ 6300 mM)<br />
(Matzke et al., 2007). On the other hand, this anion had no or<br />
even a positive influence on the observed effects on L. minor<br />
(Stolte et al., 2007a).<br />
Focusing on the terrestrial environment, Matzke et al.<br />
(2009a) investigated the influence of differently composed<br />
soils, with varying contents of the clay minerals smectite and<br />
kaolinite, on the toxicity of different anion species of imidazolium-based<br />
ILs towards the wheat Triticum aestivum. The<br />
data showed that IM14 (CF3SO2)2N appeared the most toxic,<br />
independently of the type and concentration of added clay.<br />
This is totally in contrast to the findings of Stolte et al. (2007a),<br />
who reported that [(CF 3SO 2) 2N] caused no harm to L. minor,<br />
indicating the toxic effect of this anion is different between<br />
certain plants. The toxicity of 1-butyl-3-methylimidazolium<br />
incorporated chloride, tetrafluoroborate and hydrogen sulfate<br />
was mainly controlled by the cationic moiety. The observed<br />
effects varied according to the added clay type and clay<br />
concentration. An increase of clay content resulted in less<br />
inhibitory effects of these substances. On the contrary, for<br />
IM14 combined with bis(trifluoromethylsulfonyl)imide the<br />
addition of clay minerals led to higher toxicity compared to<br />
the reference soil. Since results are contradictious further<br />
study is necessary to unravel the underlying mechanism.<br />
Moreover, a detailed study on the effect of IM14 BF 4 on the<br />
wheat T. aestivum seedlings (Wang et al., 2009) showed that<br />
IM14 BF4 was hazardous to the early development of wheat<br />
and had varying effects on different organs. At low concentrations,<br />
IM14 BF4 did not inhibit, and even promoted, wheat<br />
seedling growth. Nonetheless, at high concentrations, this IL<br />
inhibited wheat seedling growth significantly and decreased<br />
chlorophyll content, thereby reducing photosynthesis and<br />
plant growth. Therefore, the authors suggested that dilution<br />
could decrease the toxicity of IM14 BF 4 to plants and would be<br />
a good method for remediating IL-polluted environments.<br />
In another research, the phytotoxicity tests of chiral ILs<br />
containing (-)-nopyl derivatives were carried out in a plant<br />
house using spring barley (Hordeum vulgare) which is a monocotyledonous<br />
plant, and a common radish (Raphanus sativus L.<br />
subvar. radicula Pers.) which is a dicotyledonous plant (Ba1czewski<br />
et al., 2007). According to the data obtained, increasing<br />
the concentration of ILs resulted in a systematic decrease in<br />
the crop fresh weight of total sprouts and the crop fresh<br />
water research 44 (2010) 352–372 363<br />
weight per plant, both for spring barley and for common<br />
radish. It could also be noted that common barley was a more<br />
resistant plant which fairly well tolerates test IL concentrations<br />
up to 200 mg kg 1 of soil; whereas, for radish, the growth<br />
and development inhibiting concentration is 100 mg kg 1 of<br />
soil. Using the same target plant (H. vulgare), Pernak et al.<br />
(2004b) reported that the 1,3-dialkoxymethylimidazolium<br />
tetrafluoroborate salts introduced to the soil at concentration<br />
of 1,000 mg kg 1 , or 100 mg kg 1 dry mass of soil, were found<br />
to exert a phytotoxic effect on monocotyledonous plants.<br />
On the other hand, at a concentration of 10 mg kg 1 no such<br />
effect on the growth of the roots was notified. Concerning<br />
phytotoxicity of ILs to garden cress (Lepidium sativum L.) in soil<br />
environment, Studzińska and Buszewski (2009) have proved<br />
that hazardous effects of imidazolium ILs are closely<br />
connected with organic matter content in soil. Soil with more<br />
organic carbon was observed to sorb IL cations more extensively<br />
than soil with little or no organic matter; hence, the<br />
more fertile in soil, the lower probability of hazardous effect of<br />
ILs to plants. On the other hand, the hazardous character of<br />
analyzed ILs was strongly connected with their hydrophobicity,<br />
indicating that the more hydrophobic IL, the higher<br />
decrease of seed germination.<br />
Although intensive work has not been conducted on<br />
phytotoxic influence of ILs, the available data offer initial<br />
hints for environmental scientists dealing with the potential<br />
impact of ILs towards aqueous and terrestrial plants.<br />
2.6. Toxicity of ILs to invertebrates<br />
Ecotoxicological literature of ILs to invertebrates mainly focus<br />
on the use of Daphnia magna as a test organism (Bernot et al.,<br />
2005a; Couling et al., 2006; Garcia et al., 2005; Grabinska-Sota<br />
and Kalka, 2006; Luo et al., 2008; Nockemann et al., 2007; Pretti<br />
et al., 2009; Samorì et al., 2007; Wells and Coombe, 2006; Yu<br />
et al., 2009b). Daphnia is an important link between microbial<br />
and higher trophic levels (McQueen et al., 1986), and has been<br />
the subject of hundreds of intensive ecological studies. The<br />
results of all studies again observed the well-established link<br />
between toxicity and alkyl chain length of the tested ILs<br />
containing imidazolium, pyridinium or quaternary ammonium<br />
as counter cations. The most toxic compound towards<br />
D. magna was found to be IM18 Br whereas the least toxic one<br />
was IM14 Cl with log10EC50 values of 1.33 and 1.93, respectively<br />
(Table 2). Also, the nature of the anion was suggested to<br />
have smaller effects compared to those of the cation. In a<br />
recent study, Luo et al. (2008) investigated the developmental<br />
toxicity of IM18 Br on D. magna. It was found that this<br />
compound exhibited toxicity on the development of three<br />
generation of D. magna with the decrease of number of<br />
offspring and average brood size correlated to increasing IM18<br />
Br concentrations. This indicated that IM18 Br could cause<br />
deleterious effect to the population of Daphnia and indirectly<br />
disturb freshwater food webs. Couling et al. (2006) used<br />
experimental data to determine which part of the IL molecule<br />
is responsible for the observed toxic effects through a quantitative<br />
structure-property relationship (QSPR) modeling. In this<br />
respect, correlative and predictive equations were generated<br />
and proved that there was a distinct influence of the length of<br />
alkyl residues attached to the aromatic nitrogen atoms
364<br />
towards D. magna. Moreover, the models predicted that the<br />
toxicity increased slightly with increasing number of aromatic<br />
nitrogen atoms in cation ring. This implies that ammonium<br />
salts are less toxic than pyridinium salts, which in turn are<br />
less toxic than imidazolium moieties. Interestingly, it was<br />
noted that methylating the aromatic carbons could be<br />
effective in reducing toxicity to D. magna, indicating that 1-nbutylpyridinium<br />
bromide can be more toxic than 1-n-butyl-3methylpyridinium<br />
bromide, which is more toxic than 1-nbutyl-3,5-dimethylpyridinium<br />
analogue. The QSPR, though is<br />
still in its infancy, has contributed initial guidelines for<br />
a rationale design of a new category of ILs with an acceptable<br />
environmental profile.<br />
Other studies include data on the snail Physa acuta (Bernot<br />
et al., 2005b), the spring tail Folsomia candida (a soil invertebrate)<br />
(Matzke et al., 2007), Caenorhabditis elegans (a soil<br />
roundworm) (Swatloski et al., 2004) and Dreissena polymorpha<br />
(zebra mussel) (Costello et al., 2009). It was also demonstrated<br />
a positive relationship between alkyl chain length and toxicity<br />
in these reports. In the research of Bernot et al. (2005b), the<br />
estimated LC 50 (median lethal concentration) ranged from<br />
3.50 to 1799.8 mM (0.54 to 3.26 in the logarithmic form), which<br />
implied that P. acuta are less sensitive to ILs than are D. magna<br />
(log10LC50 (mM) ranging from 1.33 to 1.93 (Table 2)). Also, the<br />
authors observed that at low concentrations, the IL may<br />
suppress snail movement, but concentrations above this<br />
threshold level trigger an escape response, causing the<br />
organism to move faster. Grazing patterns, nonetheless,<br />
showed that snails grazed less at higher IL concentrations.<br />
Physa spp. are key components of freshwater food webs,<br />
because they graze algae and are themselves important prey<br />
for fish and invertebrate predators (Bernot and Turner, 2001;<br />
Osenberg and Mittelbach, 1989). Thus, nonlethal IL concentrations<br />
affected P. acuta behaviors, potentially influencing<br />
individual fitness and good web interactions.<br />
2.7. Inhibitory effects of ILs on vertebrates<br />
Zebrafish (Danio rerio) plays an important role in ecotoxicology<br />
as a prominent model vertebrate. Concerning toxicity of ILs to<br />
the zebrafish, Pretti et al. (2006) revealed that ILs may cause<br />
a completely different effect on fish according to their chemical<br />
structures. As imidazolium, pyridinium and pyrrolidinium<br />
showed a LC50 (lethal effect) >100 mg L 1 , they could be<br />
regarded as non-highly lethal towards zebrafish. On the other<br />
hand, the ammonium salts showed LC50 remarkably lower<br />
than that reported for organic solvents and tertiary amines.<br />
In general, these data referred that fish are less sensitive to ILs<br />
toxicity compared to other species belonging to lower trophic<br />
levels.<br />
In a recent report, Li et al. (2009) used the frog Rana nigromaculata<br />
as an amphibian model for toxicity testing.<br />
Amphibians are often the main vertebrate group prone to<br />
contaminant exposure in aquatic systems mostly because<br />
their larvae live in water (Lahr, 1997; Mann and Bidwell, 2000).<br />
In their study, they evaluated the toxic effects of IM18 Br on<br />
the early embryonic development of the frog R. nigromaculata.<br />
The results demonstrated that the highest embryonic<br />
mortality occurred in the neural plate stage, followed by the<br />
early gastrula and early cleavage stages with the LC50 values<br />
water research 44 (2010) 352–372<br />
being 42.4, 43.4 and 85.1 mg/L, respectively, indicating that the<br />
developmental toxicity of IM18 Br in the frog was stagesensitive.<br />
The number of dead embryos was also found to<br />
increase with the increasing concentrations of the IL IM18 Br.<br />
The developmental impact of IM18 Br was claimed not only in<br />
this finding but also in the work of Luo et al. (2008), who<br />
investigated on D. magna.<br />
Other work in the literature has focused on the acute<br />
toxicity of ILs on rats and mice (Bailey et al., 2008; Cheng et al.,<br />
2009; Landry et al., 2005; Pernak and Czepukowicz, 2001; Sipes<br />
et al., 2008). The values of acute toxicity of 3-hexyloxymethyl-<br />
1-methylimidazolium tetrafluoroborate were found to be<br />
LC50 ¼ 1400 and 1370 mg kg 1 for female and male Wistar rats,<br />
respectively (Pernak and Czepukowicz, 2001). Bailey et al.<br />
(2008) studied the effects of prenatal exposure of mice to IM14<br />
Cl due to the potential for human exposure as a result of water<br />
or soil contamination from industrial effluent or accidental<br />
spills. As shown in the experimental data, after being contacted<br />
to the IL, fetal weight was considerably reduced at the<br />
two highest concentrations (169 and 225 mg kg 1 d 1 ).<br />
Malformations were also somewhat more numerous at the<br />
highest dosage, suggesting that IM14 Cl may be teratogenic.<br />
Maternal toxicity was also present, indicating that IM14 Cl<br />
appeared to be developmentally toxic at maternally toxic<br />
dosages. Also, IM14 Cl has been shown to cause thermal irritation<br />
when applied topically to rats, but produced only<br />
minimal contact sensitization when evaluated in the mouse<br />
local lymph node assay (Landry et al., 2005). Additionally, in<br />
this report, it is worth noting that the transdermal toxicity of<br />
IM14 Cl was influenced by the vehicle of administration. Use<br />
of the organic solvent, dimethylformamide, accentuated the<br />
acute toxicity. Very high concentrations of IM14 Cl (up to 95%<br />
IM14 Cl in water) applied to the rat skin were markedly less<br />
acutely toxic. This result may have a practical guideline that to<br />
reduce the acute toxicity, ILs can be handled in pure form with<br />
water as a co-solvent.<br />
3. Environmental fate of ILs<br />
3.1. Chemical degradation of ILs<br />
Ionic liquids possess excellent chemical and thermal stability,<br />
which gives, unfortunately, a negative aspect for their treatment<br />
after usage prior to disposal. To assess the persistence of<br />
ILs in the environment as well as verify possibilities of their<br />
cleanup by chemical methods, several groups have focused<br />
their attention on oxidative and thermal degradation of ILs in<br />
aqueous media (Awad et al., 2004; Baranyai et al., 2004;<br />
Berthon et al., 2006; Itakura et al., 2008; Li et al., 2007; Morawski<br />
et al., 2005; Siedlecka and Stepnowski, 2009; Siedlecka<br />
et al., 2008a,b; Stepnowski and Zaleska, 2005). Pioneering work<br />
in the field of oxidative degradation was done by Stepnowski<br />
and Zaleska (2005) and Morawski et al. (2005) who showed that<br />
the greatest degradation efficiency for imidazolium ILs was<br />
achieved with a combination of UV light and a catalytic<br />
oxidant such as hydrogen peroxide or titanium dioxide.<br />
Subsequently, Li et al. (2007) studied the oxidative degradation<br />
of 1,3-dialkylimidazolium ILs in hydrogen peroxide/acetic acid<br />
medium assisted by ultrasonic chemical irradiation. It was
observed that 99% of tested compounds was degraded after<br />
72 h. In addition, advanced oxidative degradation in the<br />
presence of reactive peroxides generated by Fenton reagent<br />
has been applied for the removal of ILs from water (Siedlecka<br />
and Stepnowski, 2009; Siedlecka et al., 2008a,b). According to<br />
the results, in a Fenton system with 1 mM of Fe(III) and<br />
100 mM of H2O2, more than 97% of IM14 Cl was observed to<br />
degrade after 90 min. For Pyr14 Cl, IM16 Cl and IM18 Cl, the<br />
levels of degradation were 92%, 88% and 68%, respectively.<br />
Investigations of the degradation mechanisms indicated IM18<br />
Cl was more resistant to oxidation by OH radicals cleaved<br />
from H 2O 2, suggesting that the oxidation rates of imidazolium<br />
ILs by OH are structure-dependent (Siedlecka and Stepnowski,<br />
2009). The level of degradation was dependent on the alkyl<br />
chain length, consistent with Stepnowski and Zaleska (2005),<br />
who indicated that lengthening the alkyl chain lowered the<br />
rate of IL degradation. On contrast, the different length of the<br />
side chains and the type of anions did not affect the degradation<br />
process (Li et al., 2007). Regarding the thermal degradation<br />
studies of alkylimidazolium salts (Awad et al., 2004),<br />
extension of the alkyl chain enhanced the thermo-oxidative<br />
degradation of imidazolium salts. Interestingly, methyl<br />
substitution in the 2-position (i.e. between the two N atoms)<br />
was observed to decrease the oxidative decomposition of<br />
imidazolium ILs. The longer alkyl chain was also observed to<br />
induce an enhancement in photocatalytic decomposition of<br />
ILs (Morawski et al., 2005), which was not in the case of<br />
oxidative degradation (Stepnowski and Zaleska, 2005; Siedlecka<br />
and Stepnowski, 2009). Nonetheless, detailed account on<br />
the degradation of ILs by photocatalysis is required to verify<br />
this phenomenon.<br />
3.2. Biodegradability of ILs<br />
In contrast to chemical degradation, which requires the<br />
assistance of a certain oxidant for catalysis, biodegradation is<br />
the microbial breakdown of chemical compounds. Biodegradation<br />
seems to be more environmentally friendly compared<br />
to chemical decomposition process. The initial attempt to<br />
examine the degradation potential of different IM14 cations<br />
combined with [Br] , [BF 4] , [PF 6] , [N(CN) 2] , [(CF 3SO 2) 2N]<br />
and octylsulfate as the counter ion was done using the Sturm<br />
and Closed-Bottle test protocols by the group of Scammells<br />
(Garcia et al., 2005; Gathergood and Scammells, 2002; Gathergood<br />
et al., 2004; Gathergood et al., 2006). Nonetheless, no<br />
compound showed significant degree of biodegradation with<br />
the exception of the octylsulfate-containing IL. The next step<br />
study on the biodegradation of ILs involved the design of ILs<br />
containing biodegradable side chains (Gathergood and<br />
Scammells, 2002). The design was done according to the<br />
principles of Boethling (Boethling, 1994, 1996; Howard et al.,<br />
1991) who identified three important parameters including<br />
the potential sites of enzymatic hydrolysis (for example,<br />
esters and amides) and oxygen in form of hydroxyl, aldehyde<br />
or carboxylic acid groups as well as unsubstituted linear alkyl<br />
chains (especially 4 carbons) and phenyl rings, which<br />
represent possible sites for attack by oxygenases. However,<br />
for a balance between chemical properties and biodegradability,<br />
not all of these factors were suitable for ILs. The<br />
water research 44 (2010) 352–372 365<br />
addition of oxygen containing functional groups such as<br />
alcohols, aldehyde and carboxylic acids was reported to<br />
restrict the ILs performance as reaction media whereas the<br />
incorporation of phenyl rings was known to increase the<br />
melting points of IL solvents (McGuinness and Cavell, 2000).<br />
Therefore, ester or amide group was selected to be coupled in<br />
alkyl side chain of ILs. The introduction of ester groups<br />
derived from a C2 acid and C4 or higher alcohol in the<br />
3-N-substitutent was demonstrated to increase the biodegradation<br />
of imidazolium-based ILs (Gathergood et al., 2004). This<br />
can be explained by the fact that introduction of ester moiety<br />
probably provides a site susceptible to enzymatic attack<br />
(Gathergood et al., 2004; Gathergood et al., 2006) and hence,<br />
improves the biodegradation level. Though the addition of<br />
amide group is informed to improve the biodegradation of<br />
organic compounds (Boethling, 1994, 1996; Howard et al.,<br />
1991), no critical enhancement of biological degradation was<br />
noted when this group was appended into the imidazoliumbased<br />
ILs (Gathergood et al., 2004). However, no compound<br />
could be classified as ‘‘readily biodegradable’’ corresponding<br />
to Organization for Economic Cooperation and Development<br />
(OECD) standards (U.S. EPA, 1998), for which 60–70% or greater<br />
biodegradation by activated sludge microbial inoculate is<br />
required within a 10-day window in a 28-day period. Finally,<br />
the combination of the octylsulfate anion and imidazolium<br />
cation containing ester side chains resulted in readily biodegradable<br />
IL (Gathergood et al., 2006). Recently, Stolte et al.<br />
(2008) also paid their attention on investigation of functional<br />
groups incorporating alkyl chain ILs. Nonetheless, the introductions<br />
of terminal hydroxyl, carboxyl, ether and nitrile<br />
groups did not improve the biological degradation as<br />
expected.<br />
Kumar et al. (2006) investigated the fate of IM14 BF 4 when<br />
in contact with soil-microorganisms, wastewater microorganisms,<br />
Pseudomonas putida and E. coli. Although IM14 BF4<br />
was indicated to be recalcitrant in Sturm and Closed-Bottle<br />
test assays as mentioned above, it was observed in this study<br />
that P. putida was able to break down IM14 BF4 after 15 days of<br />
incubation. The breakdown products were monitored<br />
using GC-MS and identified to be 1-H-methylimidazole and<br />
1-H-butylimidazole, which were in consistent with the<br />
theoretical metabolism scheme proposed by Jastorff et al.<br />
(2003). In case of bacteria from soil and wastewater, the<br />
metabolic intermediates appeared on the 12th day. It was<br />
also noted that different intermediate peaks were observed at<br />
different retention time with different microbes, indicating<br />
that the degradation mechanism of IM14 BF4 may vary in<br />
correspondence to certain microbes and metabolic pathways.<br />
In another study, the biodegradation pathway of IM18 moiety<br />
(Fig. 3) was proposed based on intermediate products via<br />
HPLC-MS analysis after 24-day period of incubation with<br />
activated sludge (Stolte et al., 2008). The metabolism of IM18<br />
cation appeared to undergo oxidation reactions catalyzed<br />
probably by mono-oxygenases, e.g. the cytochrome P 450<br />
system on the terminal methyl group (u-oxidation). The<br />
alcohol formed was subsequently oxidized and converted<br />
into aldehydes, and then into carboxylic acids by dehydrogenases.<br />
The resulting carboxylic acids then might undergo<br />
b-oxidation and finally generated two carbon fragments that<br />
can enter the tricarboxylic acid cycle as acetyl Co-A (Fig. 3).
366<br />
retention time<br />
in min.<br />
8.9<br />
13.5 / 14.4<br />
12.2 / 12.7<br />
10.0 – 12.5<br />
19.5<br />
16.2<br />
26.5<br />
24.2<br />
26.7<br />
m/z +<br />
195<br />
211<br />
209<br />
225<br />
183<br />
197<br />
155<br />
169<br />
141<br />
intensity<br />
4*10 5<br />
3*10 5 / 2*10 5<br />
1*10 6 / 2*10 6<br />
2*10 4 – 6*10 4<br />
1*10 5<br />
3*10 5<br />
0.5*10 5<br />
4*10 6<br />
1*10 5<br />
N N +<br />
N N +<br />
N N +<br />
N N +<br />
N N +<br />
N N +<br />
N N +<br />
N N +<br />
O<br />
OH<br />
The proposed pathways provide basic information to both<br />
environmental scientists and chemical engineers; however,<br />
no studies have sought to examine the toxicity of metabolic<br />
products after degradation of ILs. This issue is of paramount<br />
importance since metabolism might not always end in less<br />
toxic products.<br />
Subsequently, Wells and Coombe (2006) extended the<br />
microbial degradation study with ammonium, imidazolium,<br />
phosphonium and pyridinium compounds by measuring the<br />
biological oxygen demand. The authors observed no biodegradability<br />
of cations incorporated short chains (C 4) within<br />
this test series, which was in agreement with Docherty et al.<br />
(2007) and Stolte et al. (2008). For longer alkyl chains (C 12, C 16<br />
and C 18) containing ILs, a strong inhibitory effect of these<br />
compounds on the inoculum used was found, indicating the<br />
active microbial consortium was significantly impacted by ILs<br />
toxicity. In recent studies (Docherty et al., 2007; Grabinska-<br />
Sota and Kalka, 2004; Harjani et al., 2008; Stasiewicz et al.,<br />
2008), pyridinium-based ILs were reported to be fully catabolized<br />
by microbial community in activated sludge. This can be<br />
inferred from the fact that degradation pathways for pyridine –<br />
water research 44 (2010) 352–372<br />
OH<br />
OH<br />
O<br />
OH<br />
O<br />
OH<br />
H<br />
N N +<br />
O<br />
O<br />
OH<br />
OH<br />
N N +<br />
N N +<br />
N N +<br />
the precursor of pyridinium-based compounds – under<br />
aerobic and anaerobic conditions were intensively investigated<br />
in the work of Kaiser et al. (1996). With respect to the<br />
common 1,3-dialkylpyridinium ILs, Pham et al. (2009) reported<br />
that after 21 days of incubation, microorganisms from activated<br />
sludge were able to break down Py4-3Me Br. Analyses of<br />
HPLC and MS/MS demonstrated that this biodegradation<br />
led to the formation of 1-hydroxybutyl-3-methylpyridinium,<br />
1-(2-hydroxybutal)-3-methylpyridinium, 1-(2-hydroxyethyl)-<br />
3-methylpyridinium and methylpyridine. Based on these<br />
intermediate products, biodegradation pathways were also<br />
suggested (Fig. 4), thereby providing the basic information<br />
which might be useful for assessing the factors related to the<br />
environmental fate and behavior of this commonly used<br />
pyridinium IL. Although this is the first report on biodegradation<br />
intermediates and pathway of pyridinium ILs, the<br />
authors have failed to systematically screen a single microorganism<br />
or a microbial consortium responsible for biodegradation<br />
of ILs. Therefore, it is needed to further investigate<br />
which type of microorganism is adaptable to ILs and which is<br />
responsible for degradation process. Also, the broken<br />
OH<br />
O<br />
OH O<br />
+<br />
Fig. 3 – Biodegradation pathways of 1-octyl-3-methylimidazolium by activated sludge microbial community (Reproduced<br />
from Stolte et al. (2008) by permission of the Royal Society of Chemistry).<br />
m/z +<br />
211<br />
209<br />
225
C<br />
H 3<br />
C<br />
H 3<br />
N +<br />
N +<br />
C<br />
H 3<br />
structures much further than methylpyridinium were not<br />
measured in the study. In particular, the possibility of<br />
cleavage of heterocyclic ring in ILs molecular structure (both<br />
pyridinium in this work and imidazolium in the study of Stolte<br />
et al. (2008)) has not been indentified. This issue should be<br />
clarified through further studies. Interestingly, Py4-3Me Br<br />
was not found to be metabolized by the activated sludge<br />
community (Docherty et al., 2007). This was attributed likely<br />
to the high IL concentration used in the study of Docherty’s<br />
group, which consequently inhibited the microbial consortium.<br />
Nonetheless, it was demonstrated that the structural<br />
manipulation of the pyridinium skeleton may lead to ILs with<br />
greater biodegradable extent compared to imidazolium-based<br />
compouds (Harjani et al., 2008).<br />
Concerning the anionic effect, ILs with halide counter ions<br />
were postulated to be more stable to degradation than<br />
perfluoronated ions (Awad et al., 2004; Gathergood and<br />
Scammells, 2002). In a preliminary study, Gathergood and<br />
Scammells (2002) confirmed this assumption and showed that<br />
the biodegradation efficiency decreased in the order<br />
[PF 6] > [BF 4] > [Br] with 60%, 59% and 48% of CO 2 evolution<br />
values, respectively. In a later study (Gathergood et al., 2006),<br />
it was found that the octylsulfate anion is considerably more<br />
biodegradable than the other commonly used anions.<br />
The alkyl chain with C4, C6 or C8 was found to increase the<br />
rate of degradation (Docherty et al., 2007; Stolte et al., 2008);<br />
nonetheless, further increasing the chain length to C12, C16 or<br />
C18 was noted to cause toxic effect towards inoculum (Wells<br />
and Coombe, 2006). However, it was stated that the long octyl<br />
side chain was not a compulsory factor for biodegradation, but<br />
more important is a certain overall lipophilicity of the<br />
compound (Stolte et al., 2008).<br />
I<br />
C<br />
H 3<br />
C<br />
H 3<br />
OH<br />
H<br />
+<br />
+<br />
N +<br />
N +<br />
N +<br />
water research 44 (2010) 352–372 367<br />
OH<br />
C<br />
H 3<br />
C<br />
H 3<br />
CH 3<br />
OH<br />
O<br />
O<br />
OH<br />
II<br />
C<br />
H 3<br />
N + C H3 CH3 OH<br />
N +<br />
OH<br />
+ C<br />
H 3<br />
Fig. 4 – Biodegradation pathways of 1-butyl-3-methylpyridinium entity by microorganisms from activated sludge (Reprinted<br />
with permission from Pham et al. (2009). Copyright 2009 American Chemical Society).<br />
C<br />
H 3<br />
N +<br />
H<br />
CH 3<br />
3.3. Sorption of ILs in environmental systems<br />
+<br />
C<br />
H 3<br />
Because the aquatic and terrestrial environments are possible<br />
recipients for contaminants, the distribution and behavior of<br />
ILs in soil are also extremely important. The retention and<br />
mobility of ILs in soils and sediments are strongly influenced by<br />
its tendency to be sorbed onto the various components of the<br />
soil matrix (Stepnowski, 2005). Since imidazolium-based ILs<br />
possess high electron acceptor potential of delocalized<br />
aromatic systems and hydrophobic components (e.g. the alkyl<br />
chain) (Stepnowski, 2005), they could be sorbed onto soils and<br />
sediments via several mechanisms. In a preliminary study,<br />
Stepnowski (2005) proved that electrostatic interactions<br />
contributed to the sorption of the imidazolium cations. Moreover,<br />
totally contrast trends were also observed demonstrating<br />
an extremely strong and practically irreversible sorption onto<br />
fine-textured marine sediments and a relatively weakly and<br />
reversibly binding to peaty soil (with the highest organic<br />
carbon content) (Stepnowski, 2005). This indicates the importance<br />
of the mineral component of the soil (sediment) in the<br />
sorption mechanism of ILs. Also, in this work, the author<br />
pointed out that compounds with longer alkyl chains were<br />
irreversibly bound to the soil component, which was in<br />
agreement with Stepnowski et al. (2007) and Matzke et al.<br />
(2009b). Interestingly, Beaulieu et al. (2008) did not find a positive<br />
effect of alkyl chain length on the sorption of alkylimidazolium-based<br />
ILs to aquatic sediments and suggested that<br />
hydrophobic interactions were not the most important sorption<br />
mechanism. The contrast results between these groups<br />
imply that sorption mechanisms of ILs may vary according to<br />
properties and composition of the environmental systems. The<br />
studies suggest that ILs may be retained by aquatic sediments;<br />
OH
368<br />
nonetheless, the toxic action of these sorbed materials towards<br />
aquatic organisms has not been addressed. Also, further<br />
efforts should be continued to elucidate the reversibility of IL<br />
sorption on these sediments to give a better understanding of<br />
the ILs fate after being released into aqueous environment.<br />
Matzke et al. (2009b) investigated the influences of the two<br />
different clay minerals kaolinite and smectite as well as of<br />
organic matter on the cation sorption and desorption behaviors<br />
of three imidazolium based ILs including IM14 BF 4, IM18<br />
BF 4 and IM14 (CF 3SO 2) 2N in soil. The addition of organic matter<br />
and clays was observed to increase the sorption/decrease the<br />
desorption of all ILs tested, and in particular smectite had<br />
striking effects on the sorption efficiency of all substances.<br />
It is worth noting that not only the cationic moiety with<br />
different alkyl side chain lengths but the anionic compartments<br />
can also play an importance role in the sorption/<br />
desorption processes. Imidazolium compounds with BF4 as<br />
a counter anion showed higher sorption capacity compared to<br />
that of [(CF3SO2)2N] , indicating the high potential of this type<br />
of IL to form ionic pairs in the soil matrix (Matzke et al., 2009b).<br />
However, further work with a variety of ILs incorporated<br />
different cationic and anionic moieties should be carried out<br />
to verify these phenomena.<br />
Gorman-Lewis and Fein (2004) examined the sorption<br />
behavior of IM14 Cl onto a range of surfaces which are<br />
commonly found in the near-surface environment. The results<br />
suggested that IM14 Cl could be minimally retarded by noninterlayer<br />
clay system and might lead to unimpeded transport<br />
through subsurface groundwater. Also, the adsorption<br />
capacity of this IL onto bacterial surfaces was not high, which<br />
might be due to the low hydrophobicity of IM14 Cl. Additionally,<br />
investigations of the adsorption of IM14 Cl towards<br />
different media carried out by our group (Vijayaraghavan et al.,<br />
2009) have shown that retardation of this compound was<br />
possible only by an ion-exchange resin and activated carbon,<br />
which was in consistency with the work of Anthony et al.<br />
(2001). However, no significant adsorption of IM14 Cl in the<br />
media of a fermentation waste (Corynebacterium glutamicum)<br />
and dried activated sludge was observed in our study.<br />
Conclusively, the data currently available demonstrated<br />
that ILs incorporated imidazolium cation can be sorbed to<br />
organic matter, whether found in aquatic sediments or<br />
terrestrial soils, and the presence of clays significantly<br />
enhanced the sorption capacity.<br />
4. Concluding remarks and future directions<br />
Ionic liquids, of which the most often cited attribute is their<br />
negligible vapor pressure, have been suggested as a green<br />
alternative to traditional organic solvents with the desire to<br />
minimize diffusion to the atmosphere. Low volatility, however,<br />
does not completely eliminate potential environmental<br />
hazards and might pose serious threats to aquatic and terrestrial<br />
ecosystems. The studies of environmental fate and<br />
toxicity of ILs have shown that the ILs commonly used to date<br />
are toxic in nature and their toxicities vary considerably across<br />
organisms and trophic levels. In general, the effect of anionic<br />
moieties is not drastic as the alkyl length effect except for the<br />
case of [(CF3SO2)2N] , which shows a clear (eco)toxicological<br />
water research 44 (2010) 352–372<br />
hazard potential. The other perfluorinated anions have been<br />
also proved to be hazardous due to hydrolytically unstable<br />
properties. In addition, the introduction of functional polar<br />
groups to the alkyl chain has been shown to reduce the toxicity<br />
of ILs and increase the biodegradation efficiency to some<br />
extent. This indicates the possibility of tailoring ILs by coupling<br />
suitable functional groups to their structure, which in turn<br />
leads to a more environmental friendly compound. The side<br />
chain length effect has been found to be consistent in all levels<br />
of biological complexity as well as different environmental<br />
compartments. Also, an increase in alkyl-chain length, or lipophilicity,<br />
was observed to be related to an increase in the rate<br />
of degradation as well as an increase in toxicity. This indicates<br />
a conflict of aims between minimizing the toxicity and maximizing<br />
the biodegradability of these neoteric solvents.<br />
Regarding the cationic compartment, pyridinium has been<br />
found to be more environmental friendly than imidazolium<br />
from both viewpoints of toxicology and microbial degradation.<br />
It can therefore be suggested that the structural manipulation<br />
of the pyridinium skeleton should be considered in design of<br />
a sustainable IL. From the currently available data, it is clear<br />
that some commonly used ILs are very far away from the image<br />
of green chemicals that are often cited in the literature. The<br />
uncertainties in their sustainable development hinder the<br />
applications of ILs under real conditions. Although some<br />
attempts have been made to give important hints in the<br />
prospective design and synthesis of inherently safer ILs,<br />
comprehensive studies dealing with the behaviors of ILs in<br />
aqueous media still await to be conducted. The important<br />
features required for the thorough insight into environmental<br />
fate of ILs include, but are not limited to:<br />
- Providing more fundamental understanding into the<br />
mechanism for IL-induced toxicity to different levels of<br />
biological complexity. The underlying mechanisms of IL<br />
toxicity have rarely been studied.<br />
- Assessing the biodegradability of cationic and anionic<br />
compartments and toxicity of their degradation intermediates.<br />
This may provide useful information in consciously<br />
designing safer chemicals.<br />
- Investigating the aerobic and anaerobic biodegradation of<br />
ILs, which would suggest initial guidelines for the treatment<br />
of ILs waste by using the existing aerobic and anaerobic<br />
wastewater treatment facilities. Especially, anaerobic<br />
degradation awaits to be investigated.<br />
- Defining which organisms or enzymes may promote<br />
degradation pathways and determining specific microbial<br />
consortium or cultivatable communities capable of<br />
biotransformation of ILs.<br />
- Performing the ecotoxicity and biodegradation tests in real<br />
environmental conditions instead of controlled conditions<br />
of laboratory experiments, which would be advantageous in<br />
understanding the fate and behavior of ILs under real<br />
conditions. For this, the potential toxicological effects at<br />
population level and community level should be addressed.<br />
It must be encouraged to use tools such as experimental<br />
mesocosms to study the effects of ILs at higher levels of<br />
organization.<br />
- Creating database of environmentally benign structure<br />
moieties of ILs based upon their toxicological and
iodegradation information, which would be practically<br />
useful as a reference for manufacturers and regulators to<br />
properly develop and regulate the use of ILs.<br />
Acknowledgements<br />
This work was supported by NRF Grant funded by the Korean<br />
Government (KRF-2007-521-D00106, NRL 2009-0083194, and in<br />
part WCU R31-2008-000-20029-0).<br />
references<br />
Anthony, J.L., Maginn, E.J., Brennecke, J.F., 2001. Solution<br />
thermodynamics of imidazolium-based ionic liquids and<br />
water. J. Phys. Chem. B 105, 10942–10949.<br />
Arning, J., Stolte, S., Böschen, A., Stock, F., Pitner, W.-R., Welz-<br />
Biermann, U., Jastorff, B., Ranke, J., 2008. Qualitative and<br />
quantitative structure activity relationships for the inhibitory<br />
effects of cationic head groups, functionalized side chains and<br />
anions of ionic liquids on acetylcholinesterase. Green Chem.<br />
10, 47–58.<br />
Austin, R.P., Barton, P., Davis, A.M., Manners, C.N., Stansfield, M.C.,<br />
1998. The effect of ionic strength on liposome-buffer and<br />
1-octanol-buffer distribution coefficients. J. Pharm. Sci. 87,<br />
599–607.<br />
Awad, W.H., Gilman, J.W., Nyden, M.,Harris,R.H.,Sutto,T.E.,<br />
Callahan, J., Trulove, P.C., DeLong, H.C., Fox, D.M., 2004.<br />
Thermal degradation studies of alkyl-imidazolium salts and<br />
theirapplicationinnanocomposites. Thermochim. Acta<br />
409, 3–11.<br />
Babalola, G.O., 1998. Anti-bacterial activity of synthetic<br />
N-heterocyclic oxidizing compounds. Lett. Appl. Microbiol. 26,<br />
43–46.<br />
Bailey, M.M., Townsend, M.B., Jernigan, P.L., Sturdivant, J., Hough-<br />
Troutman, W.L., Rasco, J.F., Swatloski, R.P., Rogers, R.D.,<br />
Hood, R.D., 2008. Developmental toxicity assessment of the<br />
ionic liquid 1-butyl-3-methylimidazolium chloride in CD-1<br />
mice. Green Chem. 10, 1213–1217.<br />
Ba1czewski, P., Bachowska, B., Bia1as, T., Biczak, R., Wieczorek, W.<br />
M., Balińska, A., 2007. Synthesis and phytotoxicity of new<br />
ionic liquids incorporating chiral cations and/or chiral anions.<br />
J. Agric. Food Chem. 55, 1881–1892.<br />
Baranyai, K.J., Deacon, G.B., MacFarlane, D.R., Pringle, J.M., Scott, J.L.<br />
, 2004. Thermal degradation of ionic liquids at elevated temperatures.<br />
Aust. J. Chem. 57, 145–147.<br />
Beaulieu, J.J., Tank, J.L., Kopacz, M., 2008. Sorption of imidazoliumbased<br />
ionic liquids to aquatic sediments. Chemosphere 70,<br />
1320–1328.<br />
Bernot, R.J., Turner, A.M., 2001. Predator identity and traitmediated<br />
indirect effects in a littoral food web. Oecologia 129,<br />
139–146.<br />
Bernot, R.J., Brueseke, M.A., Evans-White, M.A., Lamberti, G.A.,<br />
2005a. Acute and chronic toxicity of imidazolium-based<br />
ionic liquids on Daphnia magna. Environ. Toxicol. Chem. 24,<br />
87–92.<br />
Bernot, R.J., Kennedy, E.E., Lamberti, G.A., 2005b. Effects of ionic<br />
liquids on the survival, movement, and feeding behavior of<br />
the freshwater snail, Physa acuta. Environ. Toxicol. Chem. 24,<br />
1759–1765.<br />
Berthon, L., Nikitenko, S.I., Bisel, I., Berthon, C., Faucon, M.,<br />
Saucerotte, B., Zorz, N., Moisy, Ph, 2006. Influence of gamma<br />
water research 44 (2010) 352–372 369<br />
irradiation on hydrophobic room-temperature ionic liquids<br />
[BuMeIm] PF 6 and [BuMeIm] (CF 3SO 2) 2N. Dalton Trans, 2526–2534.<br />
Blaauboer, B.J., Balls, M., Barratt, M., Casati, S., Coecke, S.,<br />
Mohamed, M.K., Moore, J., Rall, D., Smith, K.R., Tennant, R.,<br />
Schwetz, B.A., Stokes, W.S., Younes, M., 1998. Alternative<br />
testing methodologies and conceptual issues. Environ. Health<br />
Perspect 106, 413–418.<br />
Blaise, C.R., 1993. In: Richardson, M.L. (Ed.), Ecotoxicology<br />
Monitoring. VCH, Weinheim, pp. 83–108.<br />
Blanchard, L.A., Hancu, D., Beckman, E.J., Brennecke, J.F., 1999.<br />
Green processing using ionic liquids and CO 2. Nature 399, 28–29.<br />
Blanchard, L.A., Brennecke, J.F., 2001. Recovery of organic<br />
products from ionic liquids using supercritical carbon dioxide.<br />
Ind. Eng. Chem. Res. 40, 287–292.<br />
Boethling, R.S., 1994. Cationic Surfactants, Surfactant Science<br />
Series, vol. 53. Marcel Dekker, New York, pp. 95–135.<br />
Boethling, R.S., 1996. Designing safer chemicals. ACS Symp. Ser.<br />
640, 156.<br />
Chemnitius, J.-M., Sadowski, R., Winkel, H., Zech, R., 1999.<br />
Organophosphate inhibition of human heart muscle<br />
cholinesterase isoenzymes. Chem. Biol. Interact 120, 183–192.<br />
Cheng, Y., Wright, S.H., Hooth, M.J., Sipes, I.G., 2009.<br />
Characterization of the disposition and toxicokinetics of<br />
N-butylpyridinium chloride in male F-344 rats and female<br />
B6C3F1 mice and its transport by organic cation transporter 2.<br />
Drug Metab. Dispos. 37, 909–916.<br />
Cho, C.-W., Pham, T.P.T., Jeon, Y.-C., Vijayaraghavan, K., Choe, W.-S.,<br />
Yun, Y.-S., 2007. Toxicity of imidazolium salt with anion bromide<br />
to a phytoplankton Selenastrum capricornutum:effectofalkylchain<br />
length. Chemosphere 69, 1003–1007.<br />
Cho, C.-W., Pham, T.P.T., Jeon, Y.-C., Yun, Y.-S., 2008a. Influence of<br />
anions on the toxic effects of ionic liquids to a phytoplankton<br />
Selenastrum capricornutum. Green Chem. 10, 67–72.<br />
Cho, C.-W., Jeon, Y.-C., Pham, T.P.T., Vijayaraghavan, K., Yun, Y.-S.,<br />
2008b. The ecotoxicity of ionic liquids and traditional organic<br />
solvents on microalga Selenastrum capricornutum. Ecotoxicol.<br />
Environ. Saf. 71, 166–171.<br />
Cho, C.-W., Pham, T.P.T., Jeon, Y.-C., Min, J., Jung, H.Y., Lee, D.S.,<br />
Yun, Y.-S., 2008c. Microalgal photosynthetic activity<br />
measurement system for rapid toxicity assessment.<br />
Ecotoxicology 17, 455–463.<br />
Cieniecka-Ros1onkiewicz, A., Pernak, J., Kubis-Feder, J., Ramani, A.,<br />
Robertson, A.J., Seddon, K.R., 2005. Synthesis, anti-microbial<br />
activities and anti-electrostatic properties of phosphoniumbased<br />
ionic liquids. Green Chem. 7, 855–862.<br />
Costello, D.M., Brown, L.M., Lamberti, G.A., 2009. Acute toxic effects<br />
of ionic liquids on zebra mussel (Dreissena polymorpha) surviving<br />
and feeding. Green Chem. 11, 548–553.<br />
Couling, D.J., Bernot, R.J., Docherty, K.M., Dixon, J.K., Maginn, E.J.,<br />
2006. Assessing the factors responsible for ionic liquid toxicity<br />
to aquatic organisms via quantitative structure-property<br />
relationship modeling. Green Chem. 8, 82–90.<br />
Dipeolu, O., Green, E., Stephens, G., 2008. Effects of watermiscible<br />
ionic liquids on cell growth and nitro reduction using<br />
Clostridium sporogenes. Green Chem. 11, 397–401.<br />
Docherty, K.M., Kulpa, C.F., 2005. Toxicity and antimicrobial<br />
activity of imidazolium and pyridinium ionic liquids. Green<br />
Chem. 7, 185–189.<br />
Docherty, K.M., Dixon, J.K., Kulpa, C.F., 2007. Biodegradability of<br />
imidazolium and pyridinium ionic liquids by an activated<br />
sludge microbial community. Biodegradation 18, 481–493.<br />
Earle, M.J., Seddon, K.R., 2000. Ionic liquids. Green solvents for the<br />
future. Pure Appl. Chem. 72, 1391–1398.<br />
Frade, R.F.M., Matias, A., Branco, L.C., Afonso, C.A.M., Duarte, C.<br />
M.M., 2007. Effect of ionic liquids on human colon carcinoma<br />
HT-29 and CaCo-2 cell lines. Green Chem. 9, 873–877.<br />
Fuler, J., Carlin, R.T., Osteryoung, R.A., 1997. The room temperature<br />
ionic liquids 1-ethyl-3-methylimidazolium tetrafluoroborate:
370<br />
electrochemical couples and physical properties. J.<br />
Electrochem. Soc. 144, 3881–3886.<br />
Fulton, M.H., Key, P.B., 2001. Acetylcholinesterase inhibition in<br />
estuarine fish and invertebrates as an indicator of<br />
organophosphorus insecticide exposure and effects. Environ.<br />
Toxicol. Chem. 20, 37–45.<br />
Ganske, F., Bornscheuer, U.T., 2006. Growth of Escherichia coli,<br />
Pichia pastoris and Bacillus cereus in the presence of the ionic<br />
liquids [BMIM] [BF 4] and [BMIM] [PF 6] and organic solvents.<br />
Biotechnol. Lett. 28, 465–469.<br />
Garcia, M.T., Gathergood, N., Scammells, P.J., 2005. Biodegradable<br />
ionic liquids. part II: effect of the anion and toxicology. Green<br />
Chem. 7, 9–14.<br />
García-Lorenzo, A., Tojo, E., Tojo, J., Teijeira, M., Rodríguez-<br />
Berrocal, F.J., González, M.P., Martínez-Zorzano, V.S., 2008.<br />
Cytotoxicity of selected imidazolium-derived ionic liquids in<br />
the human Caco-2 cell line. Sub-structural toxicological<br />
interpretation through a QSAR study. Green Chem. 10, 508–516.<br />
Gathergood, N., Scammells, P.J., 2002. Design and preparation of<br />
room-temperature ionic liquids containing biodegradable side<br />
chains. Aust. J. Chem. 55, 557–560.<br />
Gathergood, N., Garcia, M.T., Scammells, P.J., 2004. Biodegradable<br />
ionic liquids: part I. Concept, preliminary targets and<br />
evaluation. Green Chem. 6, 166–175.<br />
Gathergood, N., Scammells, P.J., Garcia, M.T., 2006. Biodegradable<br />
ionic liquids. Part III: the first readily biodegradable ionic<br />
liquids. Green Chem. 8, 156–160.<br />
Gordon, C.M., 2001. New developments in catalysis using ionic<br />
liquids. Appl. Catal. A 222, 101–117.<br />
Gorman-Lewis, D.J., Fein, J.B., 2004. Experimental study of the<br />
adsorption of an ionic liquid onto bacterial and mineral<br />
surfaces. Environ. Sci. Technol 38, 2491–2495.<br />
Grabinska-Sota, E., Kalka, J., 2004. An assessment of the toxicity<br />
of pyridinium chlorides and their biodegradation<br />
intermediates. Environ. Int 28, 687–690.<br />
Grabinska-Sota, E., Kalka, J., 2006. Toxicity of imidazolium chlorides<br />
to aquatic organisms. Polish J. Environ. Stud. 15, 405–409.<br />
Hagiwara, R., Ito, Y., 2000. Room temperature ionic liquids of<br />
alkylimidazolium cations and fluoroanions. J. Fluorine Chem.<br />
105, 221–227.<br />
Harjani, J.R., Singer, R.D., Garcia, M.T., Scammells, P.J., 2008. The<br />
design and synthesis of biodegradable pyridinium ionic<br />
liquids. Green Chem. 10, 436–438.<br />
Hassoun, E.A., Abraham, M., Kini, V., Al-Ghafri, M., Abushaban, A.,<br />
2002. Cytotoxicity of the ionic liquid, 1- N-butyl-3methylimidazolium<br />
chloride. Res. Commun, Pharmacol.<br />
Toxicol 7, 23–31.<br />
Howard, P.H., Boethling, R.S., Stiteler, W., Meylan, W., Beauman, J.,<br />
1991. Development of a predictive model for biodegradability<br />
based on BIODEG, the evaluated biodegradation data base. Sci.<br />
Total Environ. 110, 635–641.<br />
Huddleston, J.G., Visser, A.E., Reichert, W.M., Willauer, H.D.,<br />
Broker, G.A., Rogers, R.D., 2001. Characterization and<br />
comparison of hydrophilic and hydrophobic room<br />
temperature ionic liquids incorporating the imidazolium<br />
cation. Green Chem. 3, 156–164.<br />
Itakura, T., Hirata, K., Aoki, M., Sasai, R., Yoshida, H., Itoh, H.,<br />
2008. Decompostition and removal of ionic liquid in aqueous<br />
solution by hydrothermal and photocatalytic treatment.<br />
Environ. Chem. Lett. doi:10.1007/s10311-008-0177-7.<br />
Jastorff, B., Störmann, R., Ranke, J., Mölter, K., Stock, F.,<br />
Oberheitmann, B., Hoffmann, W., Hoffmann, J., Nüchter, M.,<br />
Ondruschka, B., Filser, J., 2003. How hazardous are ionic<br />
liquids? Structure-activity relationships and biological testing<br />
as important elements for sustainability evaluation. Green<br />
Chem. 5, 136–142.<br />
Jastorff, B., Mölter, K., Behrend, P., Bottin-Weber, U., Filser, J.,<br />
Heimers, A., Ondruschka, B., Ranke, J., Schaefer, M.,<br />
water research 44 (2010) 352–372<br />
Schröder, H., Stark, A., Stepnowski, P., Stock, F., Störmann, R.,<br />
Stolte, S., Welz-Biermann, U., Ziegert, S., Thöming, J., 2005.<br />
Progress in evaluation of risk potential of ionic liquids-basis<br />
for an eco-design of sustainable products. Green Chem. 7,<br />
362–372.<br />
Kaiser, K.L.E., Palabrica, V.S., 1991. Photobacterium phosphoreum,<br />
toxicity data index. Water Poll. Res. J. Can. 26, 361–431.<br />
Kaiser, J.P., Feng, Y.C., Bollag, J.M., 1996. Microbial metabolism of<br />
pyridine, quinoline, acridine, and their derivatives under<br />
aerobic and anaerobic conditions. Microbiol. Rev. 60, 483–498.<br />
Kazarian, S.G., Briscoe, B.J., Welton, T., 2000. Combining ionic<br />
liquids and supercritical fluids: in situ ATR-IR study of CO 2<br />
dissolved in two ionic liquids at high pressures. Chem.<br />
Commun., 2047–2048.<br />
Kelman, D., Kashman, Y., Rosenberg, E., Ilan, M., Ifrach, I.,<br />
Loya, Y., 2001. Antimicrobial activity of the reef sponge<br />
Amphimedon viridis from the Red Sea: evidence for selective<br />
toxicity. Aquat. Microb. Ecol 24, 9–16.<br />
Kulacki, K.J., Lamberti, G.A., 2008. Toxicity of imidazolium ionic<br />
liquids to freshwater algae. Green Chem. 10, 104–110.<br />
Kumar, R.A., Papaïconomou, N., Lee, J.-M., Salminen, J., Clark, D.S.,<br />
Prausnitz, J.M., 2009. In vitro cytotoxicities of ionic liquids: effect<br />
of cation rings, functional groups, and anions. Environ. Toxicol.<br />
24, 388–395.<br />
Kumar, S., Ruth, W., Sprenger, B., Kragl, U., 2006. On the<br />
biodegradation of ionic liquid 1-butyl-3-methylimidazolium<br />
tetrafluoroborate. Chim. Oggi 24, 24–26.<br />
Lahr, J., 1997. Ecotoxicology of organisms adapted to life in<br />
temporary freshwater ponds in arid and semi-arid regions.<br />
Arch. Environ. Contam. Toxicol 32, 50–57.<br />
Landry, T.D., Brooks, K., Poche, D., Woolhiser, M., 2005. Acute<br />
toxicity profile of 1-butyl-3-methylimidazolium chloride. Bull.<br />
Environ. Contam. Toxicol 74, 559–565.<br />
Larson, J.H., Frost, P.C., Lamberti, G.A., 2008. Variable toxicity of<br />
ionic liquid-forming chemicals to Lemna minor and the<br />
influence of dissolved organic matter. Environ. Toxicol. Chem.<br />
27, 676–681.<br />
Lata1a, A., Stepnowski, P., N˛edzi, M., Mrozik, W., 2005. Marine<br />
toxicity assessment of imidazolium ionic liquids: acute effects<br />
on the Baltic algae Oocystis submarina and Cyclotella<br />
meneghiniana. Aqua. Toxicol 73, 91–98.<br />
Lee, S.-M., Chang, W.-J., Choi, A.-R., Koo, Y.-M., 2005. Influence of<br />
ionic liquids on the growth of Escherichia coli. Korean. J. Chem.<br />
Eng 22, 687–690.<br />
Lewis, M.A., 1995. In: Rand, G.M. (Ed.), Fundamentals of<br />
Aquatic Toxicology: Effects, Environment Fate, and Risk<br />
Assessment, second ed. Taylor and Francis, Washington,<br />
DC, USA, pp. 135–170.<br />
Li, G., Shen, J., Zhu, Y., 1998. Study of pyridinium-type functional<br />
polymers. II. Antibacterial activity of soluble pyridinium-type<br />
polymers. J. Appl. Polym. Sci. 67, 1761–1768.<br />
Li, X., Zhao, J., Li, Q., Wang, L., Tsang, S.C., 2007. Ultrasonic<br />
chemical oxidative degradations of 1,3-dialkylimidazolium<br />
ionic liquids and their mechanistic elucidations. Dalton Trans,<br />
1875–1880.<br />
Li, X.-Y., Zhou, J., Yu, M., Wang, J.-J., Pei, Y.C., 2009. Toxic effects of<br />
1-methyl-3-octylimidazolium bromide on the early embryonic<br />
development of the frog Rana nigromaculata. Ecotox. Environ.<br />
Saf. 72, 552–556.<br />
Luis, P., Ortiz, I., Alcado, R., Irabien, A., 2007. A novel group<br />
contribution method in the development of a QSAR for<br />
predicting the toxicity (Vibrio fischeri EC 50) of ionic liquids.<br />
Ecotoxicol. Environ. Saf 67, 423–429.<br />
Luo, Y.-R., Li, X.-Y., Chen, X.-X., Zhang, B.-J., Sun, Z.-J., Wang, J.-J.,<br />
2008. The developmental toxicity of 1-butyl-3-octylimidazolium<br />
bromide on Daphnia magna. Environ. Toxicol 23, 736–744.<br />
Malich, G., Markovic, B., Winder, C., 1997. The sensitivity and<br />
specificity of the MTS tetrazolium assay for detecting the in
vitro cytotoxicity of 20 chemicals using human cell lines.<br />
Toxicology 124, 179–192.<br />
Mann, R.M., Bidwell, J.R., 2000. Application of the FETAX protocol<br />
to assess the development toxicity of nonylphenol ethoxylate<br />
to Xenopus laevis and two Australian frogs. Aquat. Toxicol 51,<br />
19–29.<br />
Marsh, K.N., Boxall, J.A., Lichtenthaler, R., 2004. Room<br />
temperature ionic liquids and their mixtures–a review. Fluid<br />
Phase Equilibr 219, 93–98.<br />
Massoulié, J., Pezzementi, L., Bon, S., Krejci, E., Vallette, F.-M.,<br />
1993. Molecular and cellular biology of cholinesterases. Prog.<br />
Neurobiol. 41, 31–91.<br />
Matsumoto, M., Mochiduki, K., Fukunishi, K., Kondo, K., 2004a.<br />
Extraction of organic acids using imidazolium-based ionic<br />
liquids and their toxicity to Lactobacillus rhamnosus. Sep.Purif.<br />
Technol. 40, 97–101.<br />
Matsumoto, M., Mochiduki, K., Kondo, K., 2004b. Toxicity of ionic<br />
liquids and organic solvents to lactic acid-producing bacteria.<br />
J. Biosci. Bioeng 98, 344–347.<br />
Matzke, M., Stolte, S., Thiele, K., Juffernholz, T., Arning, J., Ranke, J.,<br />
Welz-Biermann, U., Jastorff, B., 2007. The influence of anion<br />
species on the toxicity of 1-alkyl-3-methylimidazolium ionic<br />
liquids observed in an (eco)toxicological test battery. Green<br />
Chem. 9, 1198–1207.<br />
Matzke, M., Stolte, S., Böschen, A., Filser, J., 2008. Mixture effects<br />
and predictability of combination effects of imidazolium<br />
based ionic liquids as well as imidazolium based ionic liquids<br />
and cadmium on terrestrial plants (Triticum aestivum) and<br />
limnic green algae (Scenedesmus vacuolatus). Green Chem. 10,<br />
784–792.<br />
Matzke, M., Stolte, S., Arning, J., Uebers, U., Filser, J., 2009a. Ionic<br />
liquids in soils: effects of different anion species of<br />
imidazolium based ionic liquids on wheat (Triticum aestivum)<br />
as affected by different clay minerals and clay concentrations.<br />
Ecotoxicology 18, 197–203.<br />
Matzke, M., Thiele, K., Müller, A., Filser, J., 2009b. Sorption and<br />
desorption of imidazolium based ILs in different soil types.<br />
Chemosphere 74, 568–574.<br />
McFarlane, J., Ridenour, W.B., Luo, H., Hunt, R.D., Depaoli, D.W.,<br />
Ren, R.X., 2005. Room temperature ionic liquids for<br />
separating organics from produced water. Sep. Sci. Technol<br />
40, 1245–1265.<br />
McGuinness, D.S., Cavell, K.J., 2000. Donor-functionalized<br />
heterocyclic carbine complexes of palladium(II): efficient<br />
catalysts for C-C coupling reactions. Organometallics 19,<br />
741–748.<br />
McQueen, D.J., Post, J.R., Mills, E.L., 1986. Trophic relationships in<br />
freshwater pelagic ecosystems. Can. J. Fish. Aqua. Sci. 43,<br />
1571–1581.<br />
Morawski, A.W., Janus, M., Goc-Maciejewska, I., Syguda, A.,<br />
Pernak, J., 2005. Decomposition of ionic liquids by<br />
photocatalysis. Pol. J. Chem. 79, 1929–1935.<br />
Myasoedov, B.F., Molochnikova, N.P., Shkinev, V.M., Spivakov, B.Y.,<br />
1995. The Behavior of Actinides in Two-Phase Aqueous Systems<br />
Based on Polyethylene Glycol. Plenum, New York, pp. 91.<br />
Nockemann, P., Thijs, B., Driesen, K., Janssen, C.R., Hecke, K.V.,<br />
Meervelt, L.V., Kossmann, S., Kirchner, B., Binnemans, K.,<br />
2007. Choline saccharinate and choline acesulfamate: ionic<br />
liquids with low toxicities. J. Phys. Chem. B 111, 5254–5263.<br />
Olivier, H., 1999. Recent developments in the use of non-aqueous<br />
ionic liquids for two-phase catalysis. J. Mol. Cat. A: Chem. 146,<br />
285–289.<br />
Osenberg, C.W., Mittelbach, G.G., 1989. Effects of body size on the<br />
predator–prey interaction between pumpkinseed sunfish and<br />
gastropods. Ecol. Monogr. 59, 405–432.<br />
Pernak, J., Czepukowicz, A., 2001. New ionic liquids and their<br />
antielectrostatic properties. Ind. Eng. Chem. Res. 40, 2379–<br />
2383.<br />
water research 44 (2010) 352–372 371<br />
Pernak, J., Rogo _za, J., Mirska, I., 2001a. Synthesis and<br />
antimicrobial activities of new pyridinium and<br />
benzimidazolium chlorides. Eur. J. Med. Chem. 36, 313–320.<br />
Pernak, J., Kalewska, J., Ksycińska, H., Cybulski, J., 2001b.<br />
Synthesis and anti-microbial activities of some pyridinium<br />
salts with alkoxymethyl hydrophobic group. Eur. J. Med.<br />
Chem. 36, 899–907.<br />
Pernak, J., Chwa1a, P., 2003. Synthesis and anti-microbial activities of<br />
choline-like quaternary ammonium chlorides. Eur. J. Med. Chem.<br />
38, 1035–1042.<br />
Pernak, J., Sobaszkiewicz, K., Mirska, I., 2003. Anti-microbial<br />
activities of ionic liquids. Green Chem. 5, 52–56.<br />
Pernak, J., Goc, I., Mirska, I., 2004a. Anti-microbial activities of<br />
protic ionic liquids with lactate anion. Green Chem. 6, 323–329.<br />
Pernak, J., Sobaszkiewicz, K., Foksowicz-Flaczyk, J., 2004b. Ionic<br />
liquids with symmetrical dialkoxymethyl-substituted<br />
imidazolium cations. Chem. Eur. J 10, 3479–3485.<br />
Pham, T.P.T., Cho, C.-W., Vijayaraghavan, K., Min, J., Yun, Y.-S.,<br />
2008a. Effect of imidazolium-based ionic liquids on the<br />
photosynthetic activity and growth rate of Selenastrum<br />
capricornutum. Environ. Toxicol. Chem. 27, 1583–1589.<br />
Pham, T.P.T., Cho, C.-W., Min, J., Yun, Y.-S., 2008b. Alkyl-chain<br />
length effects of imidazolium and pyridinium ionic liquids on<br />
photosynthetic response of Pseudokirchneriella subcapitata.<br />
J. Biosci. Bioeng 105, 425–428.<br />
Pham, T.P.T., Cho, C.-W., Jeon, C.-O., Chung, Y.-J., Lee, M.-W.,<br />
Yun, Y.-S., 2009. Identification of metabolites involved in the<br />
biodegradation of the ionic liquid 1-butyl-3-methylpyridinium<br />
bromide by activated sludge microorganisms. Environ. Sci.<br />
Technol 43, 516–521.<br />
Peraccini, D., Chiappe, C., Intorre, L., Pretti, C., 2007. In: Letcher, T.<br />
M. (Ed.), Thermodynamics, Solubility and Environmental<br />
Issues. Elsevier, UK, pp. 259–278.<br />
Pope, C., Karanth, S., Liu, J., 2005. Pharmacology and toxicology of<br />
cholinesterase inhibitors: uses and misuses of a common<br />
mechanism of action. Environ. Toxicol. Pharmacol 19, 433–446.<br />
Pretti, C., Chiappe, C., Pieraccini, D., Gregori, M., Abramo, F.,<br />
Monni, G., Intorre, L., 2006. Acute toxicity of ionic liquids to<br />
the zebrafish (Danio rerio). Green Chem. 8, 238–240.<br />
Pretti, C., Chiappe, C., Baldetti, I., Brunini, S., Monni, G., Intorre, L.,<br />
2009. Acute toxicity of ionic liquids for three freshwater<br />
organisms: Pseudokirchneriella subcapitata, Daphnia magna and<br />
Danio rerio. Ecotoxicol. Environ. Saf. 72, 1170–1176.<br />
Ranke, J., Mölter, K., Stock, F., Bottin-Weber, U., Poczobutt, J.,<br />
Hoffmann, J., Ondruschka, B., Filser, J., Jastorff, B., 2004.<br />
Biological effects of imidazolium ionic liquids with varying<br />
chain length in acute Vibrio fischeri and WST-1 cell viability<br />
assays. Ecotox. Environ. Saf 58, 396–404.<br />
Ranke, J., Müller, A., Bottin-Weber, U., Stock, F., Stolte, S., Arning, J.,<br />
Störmann, R., Jastorff, B., 2007a. Lipophilicity parameters for<br />
ionic liquid cations and their correlation to in vitro cytotoxicity.<br />
Ecotox. Environ. Saf 67, 430–438.<br />
Ranke, J., Stolte, S., Störmann, R., Arning, J., Jastorff, B., 2007b.<br />
Design of sustainable chemical products – the example of<br />
ionic liquids. Chem. Rev. 107, 2183–2206.<br />
Rogers, R.D., Bond, A.H., Bauer, C.B., Zhang, J., Rein, S.D., Chomko, R.R.,<br />
Roden, D.M., 1995. Partitioning behavior of 99 Tc and, 129 I from<br />
simulated Handford tank wastes using polyethylene-glycol based<br />
aqueous biphasic systems. Solvent Extr. Ion Exch. 13, 689–713.<br />
Romero, A., Santos, A., Tojo, J., Rodríguez, A., 2008. Toxicity and<br />
biodegradability of imidazolium ionic liquids. J. Hazard. Mater<br />
151, 268–273.<br />
Rooney, D.W., Seddon, K.R., 2001. In: Wypych, G. (Ed.), Handbook<br />
of Solvents. ChemTech Publishing, Toronto, pp. 1459–1484.<br />
Salminen, J., Papaiconomou, N., Kumar, R.A., Lee, J.-M., Kerr, J.,<br />
Newman, J., Prausnitz, J.M., 2007. Physicochemical properties<br />
and toxicities of hydrophobic piperidinium and pyrrolidinium<br />
ionic liquids. Fluid Phase Equilibr 261, 421–426.
372<br />
Samorì, C., Pasteris, A., Galletti, P., Tagliavini, E., 2007. Acute<br />
toxicity of oxygenated and nonoxygenated imidazoliumbased<br />
ionic liquids to Daphnia magna and Vibrio fischeri.<br />
Environ. Toxicol. Chem. 26, 2379–2382.<br />
Schmid, A., Kollmer, A., Mathys, R.G., Withot, B., 1998. Development<br />
toward large-scale bacterial bioprocesses in the presence of bulk<br />
amounts of organic solvents. Extremophiles 2, 249–256.<br />
Sheldon, R., 2001. Catalytic reactions in ionic liquids. Chem.<br />
Commun, 2399–2407.<br />
Sheldon, R.A., 2005. Green solvents for sustainable organic<br />
synthesis: state of the art. Green Chem. 7, 267–278.<br />
Siedlecka, E.M., Stepnowski, P., 2009. The effect of alkyl chain<br />
length on the degradation of alkylimidazolium- and<br />
pyridinium-type ionic liquids in a Fenton-like system.<br />
Environ. Sci. Pollut. Res. 16, 453–458.<br />
Siedlecka, E.M., Mrozik, W., Kaczyński, Z., Stepnowski, P.,<br />
2008a. Degradation of 1-butyl-3-methylimidazolium chloride<br />
ionic liquid in a Fenton-like system. J. Hazard. Mater 154,<br />
893–900.<br />
Siedlecka, E.M., Go1˛ebiowski, M., Kumirska, J., Stepnowski, P.,<br />
2008b. Identification of 1-butyl-3-methylimidazolium chloride<br />
degradation products formed in Fe(III)/H2O2 oxidation system.<br />
Chem. Anal. (Warsaw) 53, 943–951.<br />
Sipes, I.G., Knudsen, G.A., Kuester, R.K., 2008. The effects of dose<br />
and route on the toxicokinetics and disposition of 1-butyl-<br />
3-methylimidazolium chloride in male F-344 rats and female<br />
B6C3F1 mice. Drug Metab. Dispos. 36, 284–293.<br />
Sk1adanowski, A.C., Stepnowski, P., Kleszczyński, K.,<br />
Dmochowska, B., 2005. AMP deaminase in vitro inhibition by<br />
xenobiotics. A potential molecular method for risk assessment of<br />
synthetic nitro- and polycyclic musks, imidazolium ionic liquids<br />
and N-glucopyranosylammoniumsalts.Environ.Toxicol.Phar.<br />
19, 291–296.<br />
Stasiewicz, M., Mulkiewicz, E., Tomczak-Wandzel, R.,<br />
Kurmirska, J., Siedlecka, E.M., Go1ebiowski, M., Gajdus, J.,<br />
Czerwicka, M., Stepnowski, P., 2008. Assessing toxicity and<br />
biodegradation of novel, environmentally benign ionic liquids<br />
(1-alkoxymethyl-3-hydroxypyridinium chloride, saccharine<br />
and acesulfamates) on cellular and molecular level. Ecotox.<br />
Environ. Saf 71, 157–165.<br />
Steinberg, S.M., Poziomek, E.J., Engelmann, W.H., Rogers, K.R., 1995.<br />
A review of environmental applications of bioluminescence<br />
measurements. Chemosphere 30, 2155–2197.<br />
Stepnowski, P., Sk1adanowski, A.C., Ludwiczak, A., qaczyńska, E.,<br />
2004. Evaluating the cytotoxicity of ionic liquids using human<br />
cell line HeLa. Hum. Exp. Toxicol 23, 513–517.<br />
Stepnowski, P., 2005. Preliminary assessment of the sorption of<br />
some alkyl imidazolium cations as used in ionic liquids to<br />
soils and sediments. Aust. J. Chem. 58, 170–173.<br />
Stepnowski, P., Zaleska, A., 2005. Comparison of different<br />
advanced oxidation processes for the degradation of room<br />
temperatureionicliquids.J.Photoch.Photobio.A170,45–50.<br />
Stepnowski, P., Mrozik, W., Nichthauser, J., 2007. Adsorption of<br />
alkylimidazolium and alkylpyridinium ionic liquids onto<br />
natural soils. Environ. Sci. Technol 41, 511–516.<br />
Stock, F., Hoffmann, J., Ranke, J., Störmann, R., Ondruschka, B.,<br />
Jastorff, B., 2004. Effects of ionic liquids on the<br />
acetycholinesterase–a structure-activity relationship<br />
consideration. Green Chem. 6, 286–290.<br />
Stolte, S., Arning, J., Bottin-Weber, U.,Matzke,M.,Stock,F.,Thiele,K.,<br />
Uerdingen, M., Welz-Biermann, U., Jastorff, B., Ranke, J., 2006.<br />
Anion effects on the cytotoxicity of ionic liquids. Green Chem. 8,<br />
621–629.<br />
Stolte, S., Matzke, M., Arning, J., Böschen, A., Pitner, W.-R.,<br />
Welz-Biermann, U., Jastorff, B., Ranke, J., 2007a. Effects of<br />
water research 44 (2010) 352–372<br />
different head groups and functionalized side chains on<br />
the aquatic toxicity of ionic liquids. Green Chem. 9, 1170–1179.<br />
Stolte, S., Arning, J., Bottin-Weber, U., Müller, A., Pitner, W.-R.,<br />
Welz-Biermann, U., Jastorff, B., Ranke, J., 2007b. Effects of<br />
different head groups and functionalized side chains on the<br />
cytotoxicity of ionic liquids. Green Chem. 9, 760–767.<br />
Stolte, S., Abdulkarim, S., Arning, J., Blomeyer-Nienstedt, A.-K.,<br />
Bottin-Weber,U.,Matzke,M.,Ranke,J.,Jastorff,B.,Thöming, J.,<br />
2008. Primary biodegradation of ionic liquid cations, identification<br />
of degradation products of 1-methyl-3-octylimidazolium chloride<br />
and electrochemical wastewater treatment of poorly<br />
biodegradable compounds. Green Chem. 10, 214–224.<br />
Studzińska, S., Buszewski, B., 2009. Study of toxicity of imidazolium<br />
ionic liquids to watercress (Lepidium sativum L.). Anal. Bioanal.<br />
Chem. 393, 983–990.<br />
Swatloski, R.P., Holbrey, J.D., Rogers, R.D., 2003. Ionic liquids are<br />
not always green: hydrolysis of 1-butyl-3-methylimidazolium<br />
hexafluorophosphate. Green Chem. 5, 361–363.<br />
Swatloski, R.P., Holbrey, J.D., Memon, S.B., Caldwell, G.A.,<br />
Caldwell, K.A., Rogers, R.D., 2004. Using Caenorhabditis elegans<br />
to probe toxicity of 1-alkyl-3-methylimidazolium chloride<br />
based ionic liquids. Chem. Commun, 668–669.<br />
Torrecilla, J.S., García, J., Rojo, E., Rodríguez, R., 2009. Estimation<br />
of toxicity of ionic liquids in leukemia rat cell line and<br />
acetylcholinesterase enzyme by principal component<br />
analysis, neural networks and multiple linear regressions. J.<br />
Hazard. Mater 164, 182–194.<br />
U.S. Environmental Protection Agency 1998. Fate, transport, and<br />
transformation test guidelines. OPPTS 835.3110. Readily<br />
Biodegradability.<br />
Vijayaraghavan, K., Pham, T.P.T., Cho, C.-W., Won, S.W., Choi, S.<br />
B., Mao, J., Kim, S., Kim, Y.-R., Chung, B.W., Yun, Y.-S., 2009.<br />
An assessment on the interaction of a hydrophilic ionic liquid<br />
with different sorbents. Ind. Eng. Chem. Res. 48, 7283–7288.<br />
Wang, L.-S., Wang, L., Wang, L., Wang, G., Li, Z.-H., Wang, J.-J.,<br />
2009. Effect of 1-butyl-3-methylimidazolium tetrafluoroborate<br />
on the wheat (Triticum aestivum L.) seedlings. Environ. Toxicol<br />
24, 296–303.<br />
Wang, X., Ohlin, C.A., Lu, Q., Fei, Z., Hu, J., Dyson, P.J., 2007.<br />
Cytotoxicity of ionic liquids and precursor compounds<br />
towards human cell line HeLa. Green Chem. 9, 1191–1197.<br />
Wasserscheid, P., Keim, W., 2000. Ionic liquids–New ‘‘solutions’’ for<br />
transition metal catalysis. Angew. Chem. Int. Ed. 39, 3772–3789.<br />
Wells, A.S., Coombe, V.T., 2006. On the freshwater ecotoxicity and<br />
biodegradation properties of some common ionic liquids. Org.<br />
Pro. Res. Dev. 10, 794–798.<br />
Welton, T., 1999. Room-temperature ionic liquids, solvents for<br />
synthesis and catalysis. Chem. Rev. 99, 2071–2083.<br />
Willauer, H.D., Huddleston, J.G., Griffin, S.T., Rogers, R.D., 1999.<br />
Partitioning of aromatic molecules in aqueous biphasic<br />
systems. Sep. Sci. Technol 34, 1069–1090.<br />
Wong, D.S.H., Chen, J.P., Chang, J.M., Chou, C.H., 2002. Phase<br />
equilibria of water and ionic liquids [emim][PF 6] and [bmim][PF 6].<br />
Fluid Phase Equilibr 194–197, 1089–1095.<br />
Yu, M., Li, S.-M., Li, X.-Y., Zhang, B.-J., Wang, J.-J., 2009a. Acute<br />
effects of 1-octyl-3-methylimidazolium bromide ionic liquid<br />
on the antioxidant enzyme system of mouse liver. Ecotox.<br />
Environ. Saf. 71, 903–906.<br />
Yu, M., Wang, S.-H., Luo, Y.-R., Han, Y.-W., Li, X.-Y., Zhang, B.-J.,<br />
Wang, J.-J., 2009b. Effects of the 1-alkyl-3-methylimidazolium<br />
bromide ionic liquids on the antioxidant defense system of<br />
Daphnia magna. Ecotox. Environ. Saf. 72, 1798–1804.<br />
Zhang, C., Malhotra, S.V., 2005. Increased paraoxon detection by<br />
acetylcholinesterase inactivation with ionic liquid additives.<br />
Tatanta 67, 560–563.
A QSAR model for predicting rejection of emerging<br />
contaminants (pharmaceuticals, endocrine disruptors)<br />
by nanofiltration membranes<br />
Victor Yangali-Quintanilla a,b, *, Anwar Sadmani a , Megan McConville a,b ,<br />
Maria Kennedy a , Gary Amy a,b<br />
a<br />
UNESCO-IHE, Institute for Water Education, Westvest 7, 2611AX Delft, The Netherlands<br />
b<br />
Delft University of Technology, Stevinweg 1, 2628CN Delft, The Netherlands<br />
article info<br />
Article history:<br />
Received 25 February 2009<br />
Received in revised form<br />
4 June 2009<br />
Accepted 26 June 2009<br />
Available online 3 July 2009<br />
Keywords:<br />
Pharmaceuticals<br />
Endocrine disruptors<br />
Nanofiltration<br />
Modelling<br />
QSAR<br />
water research 44 (2010) 373–384<br />
Available at www.sciencedirect.com<br />
journal homepage: www.elsevier.com/locate/watres<br />
abstract<br />
A quantitative structure activity relationship (QSAR) model has been produced for predicting<br />
rejection of emerging contaminants (pharmaceuticals, endocrine disruptors,<br />
pesticides and other organic compounds) by polyamide nanofiltration (NF) membranes.<br />
Principal component analysis, partial least square regression and multiple linear regressions<br />
were used to find a general QSAR equation that combines interactions between<br />
membrane characteristics, filtration operating conditions and compound properties for<br />
predicting rejection. Membrane characteristics related to hydrophobicity (contact angle),<br />
salt rejection, and surface charge (zeta potential); compound properties describing<br />
hydrophobicity (log Kow, log D), polarity (dipole moment), and size (molar volume, molecular<br />
length, molecular depth, equivalent width, molecular weight); and operating conditions<br />
namely flux, pressure, cross flow velocity, back diffusion mass transfer coefficient,<br />
hydrodynamic ratio (Jo/k), and recovery were identified as candidate variables for rejection<br />
prediction. An experimental database produced by the authors that accounts for 106<br />
rejection cases of emerging contaminants by NF membranes as result of eight experiments<br />
with clean and fouled membranes (NF-90, NF-200) was used to produce the QSAR model.<br />
Subsequently, using the QSAR model, rejection predictions were made for external<br />
experimental databases. Actual rejections were compared against predicted rejections and<br />
acceptable R 2 correlation coefficients were found (0.75 and 0.84) for the best models.<br />
Additionally, leave-one-out cross-validation of the models achieved a Q 2 of 0.72 for internal<br />
validation. In conclusion, a unified general QSAR equation was able to predict rejections of<br />
emerging contaminants during nanofiltration; moreover the present approach is a basis to<br />
continue investigation using multivariate analysis techniques for understanding<br />
membrane rejection of organic compounds.<br />
ª 2009 Elsevier Ltd. All rights reserved.<br />
* Corresponding author: UNESCO-IHE, Institute for Water Education, Westvest 7, 2611AX Delft, The Netherlands. Tel.: þ31 15 215 1745;<br />
fax: þ31 15 215 2921.<br />
E-mail addresses: v.a.yangaliquintanilla@tudelft.nl, v.yangaliquintanilla@unesco-ihe.org (V. Yangali-Quintanilla).<br />
0043-1354/$ – see front matter ª 2009 Elsevier Ltd. All rights reserved.<br />
doi:10.1016/j.watres.2009.06.054
374<br />
1. Introduction<br />
Nanofiltration (NF) and reverse osmosis (RO) are technologies<br />
that provide medium to high rejections of organic<br />
compounds present as emerging contaminants (micropollutants)<br />
in water, namely endocrine disrupting<br />
compounds (EDCs), pharmaceutically active compounds<br />
(PhACs), and pesticides (Kiso et al., 2001; Schäfer et al., 2003;<br />
Kimura et al., 2003, 2004; Nghiem et al., 2004). The presence<br />
of micropollutants has been identified in surface water<br />
bodies, sewage treatment plant effluents, and stages of<br />
drinking water treatment plants and even at trace-levels in<br />
finished drinking water (Kolpin et al., 2002; Heberer, 2002;<br />
Castiglioni et al., 2006). The possible effects on aquatic<br />
organisms and human health, associated with the<br />
consumption of water containing low concentrations of<br />
single compounds, have been presented in toxicology studies<br />
(Pomati et al., 2006; Escher et al., 2005; Vosges et al., 2008).<br />
The studies demonstrate that researchers do not yet<br />
understand the exact risks from decades of persistent<br />
exposure to random combinations of low levels of pharmaceuticals,<br />
EDCs, and other organic compounds; hence, the<br />
long-term effects of consumption of water containing low<br />
concentrations of micropollutants will remain as an unanswered<br />
question for the foreseeable future. Meanwhile,<br />
water treatment facilities are implementing monitoring<br />
programs, research organizations dealing with water reuse<br />
have published reports, and studies have addressed the topic<br />
(Drewes et al., 2006; Verliefde et al., 2007). An important<br />
aspect to deal with the problem has been the identification<br />
of compound physicochemical properties and membrane<br />
characteristics to explain transport, adsorption and removal<br />
of micropollutants by different mechanisms, explicitly size/<br />
steric exclusion, hydrophobic adsorption and partitioning,<br />
and electrostatic repulsion (Kiso et al., 2001; Schäfer et al.,<br />
2003; Kimura et al., 2003; Nghiem et al., 2004, Ozaki and Li,<br />
2002; Van der Bruggen and Vandecasteele, 2002; Bellona and<br />
Drewes, 2005; Xu et al., 2005). A number of articles have<br />
proposed a mechanistic understanding of the interaction<br />
between membranes and organic compounds; others have<br />
tried to apply fitting parameter models to model rejection<br />
(Cornelissen et al., 2005; Kim et al., 2007; Verliefde et al.,<br />
2008). However, there have been few models to ‘‘predict’’ the<br />
rejection of compounds. To overcome that status, our<br />
objective was to create a general QSAR model to predict<br />
rejections based on an integral approach that considers<br />
membrane characteristics, filtration operating conditions<br />
and physicochemical compound properties. A quantitative<br />
structure activity relationship (QSAR) is a method that<br />
relates an activity of a set of compounds quantitatively to<br />
chemicals descriptors (structure or property) of those<br />
compounds (Sawyer et al., 2003). QSAR has the objective of<br />
prediction but maintaining a relationship to mechanistic<br />
interpretation. Applications of QSAR for the development of<br />
models to find relationships between membranes and<br />
organic compounds have been presented in journals related<br />
to drug discovery and medicinal chemistry for analysis of<br />
permeability of membranes to organic compounds (Ren<br />
et al., 1996; Fujikawa et al., 2007). The study of reverse<br />
water research 44 (2010) 373–384<br />
osmosis membranes has also experienced the application of<br />
QSAR principles. Campbell et al. (1999) performed a QSAR<br />
analysis of surfactants influencing attachment of a mycobacterium<br />
to cellulose and aromatic polyamide reverse<br />
osmosis membranes; their objective was to understand the<br />
relationship between surfactant molecular properties and<br />
activity on the membrane surface to inhibit bacterial<br />
attachment to the membrane in order to reduce biofilm<br />
formation and to increase permeate production. More<br />
recently, Libotean et al. (2008) developed an artificial neural<br />
network model based on quantitative structure-property<br />
relations, the model claim to predict organic solute passage<br />
through reverse osmosis membranes considering simultaneous<br />
correlation of organic solutes (molecular descriptors)<br />
and membrane properties. Alike, our study uses the concept<br />
of QSAR analysis to quantify an activity, compound rejection<br />
by a membrane, in terms of organic compound physicochemical<br />
properties, membrane characteristics (salt rejection,<br />
pure water permeability, molecular weight cut-off,<br />
charge, hydrophobicity) and operating conditions (pressure,<br />
flux, cross flow velocity, back diffusion mass transfer coefficient,<br />
recovery). In this work a QSAR model was constructed<br />
with internal experimental data used for training.<br />
The model was internally validated using measures of<br />
goodness of fit and prediction. Subsequently, after identification<br />
of a relationship in form of an equation, estimation of<br />
rejections for an external dataset for different compounds<br />
and membranes were used to externally validate the model.<br />
Similarly, rejections of more emerging organic contaminants<br />
can be predicted in advance, before nanofiltration or reverse<br />
osmosis applications. Nevertheless, the QSAR model is<br />
applicable in the range of boundary experimental conditions<br />
that will be defined in the experimental section of this<br />
publication (Section 3).<br />
2. Theory<br />
2.1. Principal component analysis<br />
Principal component analysis (PCA) is a method that allows<br />
simplification of many variables into a group of a few variables<br />
that might be measuring the same principles of a system. It<br />
may occur that a system considers an abundance of variables<br />
to explain a process; in this case principal component analysis<br />
reduces the redundancy of information. The general objectives<br />
of PCA are data reduction and interpretation. Although p<br />
variables (components) are required to reproduce the total<br />
system variability, often much of this variability can be<br />
accounted for by a small number k of principal components.<br />
The k is the number of components (reduced) that represent<br />
the initial p variables. In general, PCA is concerned with<br />
whether the covariances or correlations between a set of<br />
observed variables x 1, x 2, ., x p can be explained in terms of<br />
a smaller number of unobservable components, c1, c2, ., ck,<br />
where k < p. Thus, there is as much information in the k<br />
components as there is in the original p variables. Comprehensive<br />
details about the theory of PCA can be found elsewhere<br />
(Jolliffe, 2002; Johnson and Wichern, 2007; Everitt and<br />
Dunn, 2001).
2.2. Multiple linear regression<br />
Multiple linear regression is a method of analysis for assessing<br />
the strength of the relationship between a set of explanatory<br />
variables known as independent variables, and a single<br />
response or dependent variable. Applying multiple regression<br />
analysis to a set of data results in what are known as regression<br />
coefficients, one for each explanatory variable. The<br />
multiple regression model for a response variable, y, with<br />
observed values, y1, y2, ., yn (where n is the sample size) and q<br />
explanatory variables, x1, x2, ., xq with observed values, x1i,<br />
x2i, ., xqi for i ¼ 1, ., n, is<br />
yi ¼ b 0 þ b 1x1i þ b 2x2i þ / þ b qxqi þ 3i<br />
The regression coefficients, b0, b1, ., bq, are generally estimated<br />
by least squares. The term 3i is the residual or error for<br />
individual i and represents the deviation of the observed value<br />
of the response for this individual from that expected by the<br />
model. These error terms are assumed to have a normal<br />
distribution with variance s 2 . The fit of a multiple regression<br />
model can be judged with calculation of the multiple correlation<br />
coefficient, R, defined as the correlation between the<br />
observed values of the response variable and the values predicted<br />
by the model. The squared value of R (R 2 ) gives the<br />
proportion of the variability of the response variable accounted<br />
for by the explanatory variables. Analysis of variance<br />
(ANOVA) will provide an F-test of the null hypothesis that<br />
each of b0, b1, ., bq, is equal to zero, or in other words that R 2 is<br />
zero (Landau and Everitt, 2004).<br />
2.3. Principal component regression and partial least<br />
squares regression<br />
Principal component regression (PCR) is a method in which<br />
the components from the principal component method are<br />
used for regression. Hence, the principal components of<br />
the matrix X are used as regressors of a dependent Y. The<br />
orthogonality of the principal components eliminates the<br />
multicollinearity problem. But, the problem of choosing an<br />
optimum subset of predictors remains. A possible strategy is<br />
to keep only a few of the first components. But they are chosen<br />
to explain X rather than Y, and therefore, nothing guarantees<br />
that the principal components, which ‘‘explain’’ X, are relevant<br />
for Y. Problems may arise, however, if there is a lot of<br />
variation in X. PCR finds, somewhat uncritically, those latent<br />
variables that describe as much as possible of the variation in<br />
X. But sometimes the variable itself gives rise to only small<br />
variations in X, and if the interferences vary a lot, then the<br />
latent variables found by PCR may not be particularly good at<br />
describing Y. In the worst case important information may be<br />
hidden in directions in the X-space that PCR interprets as<br />
disturbance, and therefore leaves out. Partial least squares<br />
regression (PLS) is able to cope better with this problem, by<br />
forming variables that are relevant for describing Y. By<br />
contrast, PLS regression finds components from X that are<br />
also relevant for Y. Specifically, PLS regression searches for<br />
a set of components (called latent vectors) that performs<br />
a simultaneous decomposition of X and Y with the constraint<br />
that these components explain as much as possible of the<br />
water research 44 (2010) 373–384 375<br />
(1)<br />
covariance between X and Y. This step generalizes PCA. The<br />
goal of PLS regression is to predict Y from X and to describe<br />
their common structure (Abdi, 2003; Jørgensen and Goegebeur,<br />
2006).<br />
3. Experimental<br />
3.1. Chemicals, membranes, materials and<br />
experimental conditions<br />
The list of organic compounds representing emerging<br />
contaminants is presented in Table 1. The compounds were<br />
selected considering: their occurrence in surface water and<br />
drinking water, their identification as priority emerging<br />
contaminants, the availability of analytical methods, their<br />
quantitative and qualitative representation of physicochemical<br />
properties. The pharmaceutical compounds (caffeine,<br />
sulfamethoxazole, acetaminophen, phenacetin, phenazone,<br />
carbamazepine, naproxen, ibuprofen, metronidazole), endocrine<br />
disrupting compounds (17b-estradiol, estrone, bisphenol<br />
A, nonylphenol, atrazine) and sodium alginate were<br />
purchased from Sigma–Aldrich (Sigma–Aldrich, Schnelldorf,<br />
Germany). Potassium chloride, sodium hydroxide, hydrochloric<br />
acid and magnesium sulphate anhydrous were<br />
purchased from J.T.Baker (J.T.Baker, Deventer, Netherlands).<br />
Two thin film composite NF membranes were selected for<br />
this study (NF-200 and NF-90, Dow-Filmtec, Dow Chemical<br />
Co., Midland, MI). The experimental setup consisted of two<br />
filtration SEPA CF II (GE Osmonics, Minnetonka, MN) cells and<br />
cell holders in parallel, in order to increase permeate<br />
production and achieve hydrodynamic conditions, two<br />
hydraulic pumps (Power Team, Bega Int. BV, Netherlands),<br />
a 60 litres stainless steel tank (Tummers, Netherlands),<br />
a positive displacement pump (Hydra-Cell pump, Wanner<br />
Eng. Inc., Minneapolis, MN), a frequency converter (VLT<br />
microdrive, Danfoss, SA), a chiller/heater (Julabo, Germany),<br />
control needle valves, pressure gauges, flow meters,<br />
a proportional pressure relief valve and stainless steel<br />
tubings (Swagelok BV, Netherlands), a digital balance<br />
(Sartorius, Germany) and, a computer for flow rate data<br />
acquisition. A piece of membrane was compacted with<br />
deionised water for 6 h at a pressure of 276 kPa before performing<br />
an experiment with the membrane. The experiments<br />
were conducted in a recycle mode in which permeate<br />
and concentrate were recirculated into the feed tank for the<br />
first 72 h (a pre-equilibration period); then, permeate was<br />
collected within the next 24 h. The feed solution of all the<br />
experiments contained a cocktail of 14 compounds (concentration<br />
ranging from 6.5 to 65 mg/L). The main reason for<br />
conducting experiments at concentrations of mg/L was to<br />
accelerate steady state (after membrane adsorption) conditions<br />
in a limited time (3 days). At very low concentrations<br />
more time would be needed to achieve steady state conditions<br />
and at short time tests low concentrations may lead to<br />
over-estimation of rejections. A specific flux decline of 15% of<br />
initial flux was targeted to foul the membranes using a feed<br />
solution containing w10 mg/L DOC of sodium alginate. The<br />
study of Lee et al. (2004) concluded that polysaccharides were<br />
important membrane foulants. Sodium alginate was used as
376<br />
Table 1 – List of compounds and physicochemical properties.<br />
Name Molecular Acid<br />
weight<br />
(g/mol)<br />
pKa<br />
20 C a<br />
b<br />
Log Kow a<br />
Log D Dipole<br />
(pH 7) moment<br />
(debye) c<br />
surrogate of polysaccharides. Alginate is frequently used as<br />
a model for organic matter of algae origin (Henderson et al.,<br />
2008). Experiments were carried out for both clean and fouled<br />
NF-90 and NF-200 membranes. The selection of membranes<br />
was based on a qualitative rejection assessment of emerging<br />
contaminants with molecular weight of more than 150 Da by<br />
membranes with a MWCO between 200 and 300 Da. A total of<br />
eight experiments were performed; at hydrodynamic ratio of<br />
pure water permeation flux to back diffusion mass transfer<br />
coefficient (J0/k) ofw1 and recovery 3% (NF-90 clean, NF-200<br />
clean), at J0/k of w1 and recovery 8% (NF-200 clean), at J0/k of<br />
w2 and recovery 3 (NF-90 clean), and at J 0/k of w2 and<br />
recovery 8 (NF-90 clean and fouled, NF-200 clean and fouled).<br />
The calculation method of k (back diffusion mass transfer<br />
coefficient) was presented in a previous publication (Yangali-<br />
Quintanilla et al., 2009). All the experiments were carried out<br />
at a controlled temperature of 20 C, an ionic strength of<br />
10 mM as KCl and a pH of 7. The transmembrane pressures<br />
were in the range of 276–483 kPa. The fluxes were between<br />
4.3 and 30.2 L/m 2 per day; cross flow velocities were between<br />
0.5 and 4.5 cm/s. The experiments produced a total internal<br />
dataset of 106 rejection cases; the dataset can be accessed as<br />
supplementary data. The boundary experimental conditions<br />
of the internal dataset are presented in Table 2. The internal<br />
dataset was used to develop the model. An external dataset<br />
that gathered three different datasets was used for validation<br />
of the model. The external dataset is presented as supplementary<br />
data. Experimental conditions for the first part of<br />
the external dataset can be obtained from Kim et al. (2007)<br />
and Yangali-Quintanilla et al. (2008). Experimental conditions<br />
for the second and third parts can be found in Verliefde et al.<br />
(2008); the data correspond to filtration experiments using<br />
synthetic water solutions.<br />
Molar<br />
volume d<br />
(cm 3 /<br />
mol)<br />
Molec.<br />
length<br />
(nm) e<br />
3.2. Analytical equipment, analyses of compounds<br />
and membranes<br />
The pharmaceuticals and endocrine disruptors (with the<br />
exception of atrazine) were analyzed by Technologiezentrum<br />
Wasser (TZW, Karlsruhe, Germany). The detection limit was<br />
10 ng/L per compound. The uncertainty of estimates was of<br />
15% according to a validation method of the analysis<br />
protocol and due to high concentrations of the samples (mg/L).<br />
The analyses of pharmaceuticals were performed according to<br />
Table 2 – Data range of membrane characteristics,<br />
operating conditions and rejections.<br />
Variable Unit Min. value Max. value<br />
Molecular weight cut-off<br />
(MWCO)<br />
Pure water permeability<br />
(PWP)<br />
Molec.<br />
width<br />
(nm) e<br />
Da 200 300<br />
L/m 2 per<br />
day/kPa<br />
Molec.<br />
depth<br />
(nm) e<br />
Equiv.<br />
width<br />
(nm) e<br />
Classification f<br />
Acetaminophen 151 10.2 0.46 0.23 4.55 120.90 1.14 0.68 0.42 0.53 HL-neutral<br />
Phenacetine 179 N/A 1.58 1.68 4.05 163.00 1.35 0.69 0.42 0.54 HL-neutral<br />
Caffeine 194 N/A 0.07 0.45 3.71 133.30 0.98 0.87 0.56 0.70 HL-neutral<br />
Metronidazole 171 N/A 0.02 0.27 6.30 117.80 0.93 0.90 0.48 0.66 HL-neutral<br />
Phenazone 188 N/A 0.38 0.54 4.44 162.70 1.17 0.78 0.56 0.66 HL-neutral<br />
Sulfamethoxazole 253 5.7 0.89 0.45 7.34 173.10 1.33 0.71 0.58 0.64 HL-ionic<br />
Naproxen 230 4.3 3.18 0.34 2.55 192.20 1.37 0.78 0.75 0.76 HB-ionic<br />
Ibuprofen 206 4.3 3.97 0.77 4.95 200.30 1.39 0.73 0.55 0.64 HB-ionic<br />
Carbamazepine 236 N/A 2.45 2.58 3.66 186.50 1.20 0.92 0.58 0.73 HB-neutral<br />
Atrazine 216 N/A 2.61 2.52 3.43 169.80 1.26 1.00 0.55 0.74 HB-neutral<br />
17b-estradiol 272 10.3 4.01 3.94 1.56 232.60 1.39 0.85 0.65 0.74 HB-neutral<br />
Estrone 270 10.3 3.13 3.46 3.45 232.10 1.39 0.85 0.67 0.76 HB-neutral<br />
Nonylphenol 220 10.3 5.71 5.88 1.02 236.20 1.79 0.75 0.59 0.66 HB-neutral<br />
Bisphenol A 228 10.3 3.32 3.86 2.13 199.50 1.25 0.83 0.75 0.79 HB-neutral<br />
a ADME/Tox Web Software.<br />
b Experimental database: SRC PhysProp Database.<br />
c Chem3D Ultra 7.0.<br />
d ACD/ChemSketch Properties Batch.<br />
e Molecular Modeling Pro.<br />
f HL ¼ hydrophilic, HB ¼ hydrophobic.<br />
water research 44 (2010) 373–384<br />
0.86 2.23<br />
Salt rejection (SR) a<br />
– 0.96 0.98<br />
Zeta potential (ZP) MV 48.04 10.78<br />
Contact angle (CA) 39.3 58.0<br />
Pressure (P) kPa 276 483<br />
Cross flow velocity (v) cm/s 0.5 4.5<br />
Back diffusion mass<br />
transfer coefficient (k)<br />
cm/s 2.70E-04 5.99E-04<br />
Flux (J ) L/m 2 per day 4.3 30.2<br />
Hydrodynamic ratio<br />
(J0/k)<br />
– 1 2<br />
Recovery (recov) % 3 8<br />
Rejection (rejection) % 17.7 99.0<br />
a 2000 mg/L MgSO4, 25 C, recovery 15%, pressure 1034 kPa, pH 8.
the protocols described by Sacher et al. (2001, 2008). Analyses<br />
of 17b-estradiol, estrone, nonylphenol and bisphenol A were<br />
done by gas chromatography/mass spectrometry (GC/MS)<br />
after automated solid-phase extraction onto a polymeric<br />
material and subsequent silylation of the analytes. First, 10 ml<br />
of a 50 ng/ml solution of 4-n-nonylphenol in acetone which<br />
was used as internal standard for the overall procedure were<br />
added to an aliquot of the water sample (1000 ml). Automated<br />
solid-phase extraction (Tekmar AutoTrace, Germany) was<br />
done on plastic cartridges filled with 200 mg of bondelut<br />
material (Fa. Varian, Darmstadt, Germany). After the enrichment<br />
step the solid-phase material was dried in a gentle<br />
stream of nitrogen. Elution was done with 4 ml of acetone. The<br />
acetone was evaporated to 100 ml in a stream of nitrogen and<br />
to dryness in a drying oven at 80 C. The dry residue was<br />
reconstituted with 100 ml of a silylation reagent mixture<br />
(N-methyl-N-trimethylsilyltrifluoro acetamide (MSTFA)/2%<br />
trimethyliodo silane). After a reaction time of 20 min at 80 C<br />
(drying oven), determination of the derivatives was done by<br />
GC/MS using a 6890 GC/MS system from Agilent Technologies<br />
(Waldbronn, Germany).<br />
Concentrations of atrazine were determined using microplate<br />
enzyme-linked immunoabsorbent assay (ELISA) kits<br />
(Abraxis LLC, Warminster, PA). Atrazine was determined with<br />
a detection limit of 0.04 mg/L, and uncertainty of 15%. To<br />
determine the hydrophobicity of membranes, contact angles<br />
of clean and fouled membrane surfaces were measured with<br />
CAM200 optical contact angle and surface tension meter (KSV<br />
Instruments, Finland) at Delft University of Technology; to<br />
measure contact angle, the sessile drop method was used.<br />
Surface charge, in terms of zeta potential, of clean and fouled<br />
membranes was quantified using ELS-8000 zeta potential<br />
analyzer (Otsuka Electronics, Japan). The zeta potential analyses<br />
were determined using a Milli-Q water solution at pH 7<br />
and ionic strength of 10 mM KCl. The zeta potential was<br />
determined using the electrophoresis method using a cell<br />
consisting of membrane and quartz cells. The zeta potential<br />
was calculated from the electrophoretic mobility using the<br />
Smoluchowski formula, a detailed explanation of calculation<br />
was provided in a previous publication (Shim et al., 2002). The<br />
pH of the solutions was measured using a calibrated Metrohm<br />
691 pH-meter (Metrohm AG, Herisau, Switzerland); the electrical<br />
conductivity and temperature were measured with<br />
a WTW Cond 330i (WTW GmbH, Weilheim, Germany) portable<br />
conductivity meter. Clean and fouled membranes were characterized<br />
to determine magnesium sulphate salt rejection at<br />
standard conditions specified by manufacturers, a pure water<br />
solution containing 2000 mg/L of magnesium sulphate at 25 C<br />
and pH 8 was filtrated at pressure of 1034 kPa and recovery of<br />
15%.<br />
3.3. Characterization and classification of compounds<br />
The acid dissociation constant as log K a (pK a) was used to<br />
determine the speciation of the organic compound in ionic<br />
species at pH 7. For hydrophobicity determination, log Kow and<br />
log D were used; log Kow is the octanol–water partition coefficient<br />
and log D is the ratio of the equilibrium concentrations<br />
of all species (unionized and ionized) of a molecule in octanol<br />
to the same species in the water phase. Values of pKa and log D<br />
water research 44 (2010) 373–384 377<br />
were calculated by ADME/Tox web software. Solubility and<br />
log Kow values were obtained from SRC Physprop experimental<br />
database. The value of the molecular dipole moment<br />
was equal to the vector sum of the individual bond dipole<br />
moments. Dipole moment was calculated by Chem3D Ultra<br />
7.0, Cambridgesoft. Size descriptors included molar volume<br />
(MV), molecular length, molecular width, molecular depth and<br />
equivalent molecular width. The molecular length is defined<br />
as the distance between the two most distant atoms. The<br />
molecular width and molecular depth (width > depth) are<br />
measured by projecting the molecule on the plane perpendicular<br />
to the length axis and the equivalent molecular width<br />
is defined as the geometric mean of width and depth (Santos<br />
et al., 2006). Molar volumes were calculated using the program<br />
ACD/ChemSketch Properties Batch, ACD/Labs; and Molecular<br />
Modeling Pro, ChemSW, was used to compute size descriptors<br />
after optimization geometry of a molecule from the interaction<br />
of conformational analysis and energy minimization with<br />
a semi-empiric method MOPAC-PM3. Based on pKa and log -<br />
Kow values, the compounds were classified as hydrophilic<br />
neutral, hydrophilic ionic, hydrophobic ionic and hydrophobic<br />
neutral (see Table 1). Compounds with log Kow 2 were<br />
referred to as hydrophobic; therefore those with log Kow < 2<br />
were hydrophilic. The classification was based on an early<br />
reference (Connell, 1990). Although the value of 2 may seem<br />
low to consider hydrophobicity, the classification was not<br />
used in constructing the models and therefore the magnitude<br />
of log Kow or log D became more important. Table 1 shows the<br />
calculated values of molecular weight, pKa, log Kow, log D,<br />
dipole moment, molar volume, molecular length, molecular<br />
width, molecular depth and equivalent width.<br />
4. Results and discussion<br />
4.1. QSAR methodology<br />
The procedure to find a general QSAR equation to describe<br />
rejection was performed in four phases. The first phase was<br />
the organization of data from the experimental part. The data<br />
comprised of 106 rejection cases. The database showing the<br />
rejection cases is presented as supplementary data. A total of<br />
21 initial variables were used. The variables considered as<br />
compound descriptors were molecular weight (MW), solubility,<br />
log Kow, log D, dipole moment, molar volume, molecular<br />
length, molecular width, molecular depth and equivalent<br />
width; variables describing membrane characteristics were<br />
molecular weight cut-off (MWCO), pure water permeability<br />
(PWP), magnesium sulphate salt rejection (SR), charge of the<br />
membrane as zeta potential (ZP), and hydrophobicity as<br />
contact angle (CA); variables describing operating conditions<br />
were operating pressure (P), cross flow velocity (v), back<br />
diffusion mass transfer coefficient (k), flux (J ), ratio of pure<br />
water permeation flux J 0 and back diffusion mass transfer<br />
coefficient (J0/k) and recovery. The range of values for<br />
membrane characteristics, operating conditions and rejections<br />
is presented in Table 2. The second phase was dedicated<br />
to the process of variables reduction using a correlation<br />
matrix and factor analysis with principal component analysis.<br />
The third phase corresponded to the regression analysis. In
378<br />
the third phase three methodologies were implemented; the<br />
first was principal component analysis (PCA), with sequential<br />
application of multiple linear regressions (MLR). The second<br />
method was the use of partial least squares (PLS) regression<br />
and MLR; and the third method was the use of MLR only. The<br />
last phase was the validation process. The model was internally<br />
validated using measures of goodness of fit (regression<br />
coefficients) and prediction (leave-one-out cross-validation);<br />
Section 4.5 gives details about the validation process. External<br />
validation of the general QSAR model was implemented by<br />
predicting rejections for an external dataset of experiments<br />
performed with different compounds and membranes, and<br />
with comparable operating conditions. PCA and PLS were<br />
performed using the research and statistical package SPSS<br />
Statistics 16.0. Leave-one-out cross-validations of the models<br />
were performed with MobyDigs (Talete, Milano, Italy).<br />
4.2. Variables reduction with principal component<br />
analysis and QSAR model<br />
The correlation matrix of the initial 21 variables was scrutinized<br />
in order to obtain a not positive definite matrix as<br />
a requisite of PCA; the matrix is accompanied as supplementary<br />
data. A matrix is called not positive definite when there<br />
are both positive and negative eigenvalues. In the case of<br />
symmetric matrices, such as a correlation matrix, positive<br />
definiteness will only hold if the matrix and every principal<br />
submatrix have a positive determinant. A non-positive definite<br />
input matrix may signal a perfect linear dependency of<br />
one variable on another, known as collinearity. This was the<br />
case for MWCO and salt rejection (SR) that were perfectly<br />
linearly correlated. Therefore application of PCA considering<br />
independently one variable or the other will give the same<br />
results of variables reduction and number of components. In<br />
other words, MWCO will not be excluded with PCA, the<br />
variable will be separated in advance and the results obtained<br />
for SR may be replaced by the variable MWCO, or vice versa.<br />
Once an appropriate matrix was defined, the variables were<br />
analyzed in terms of how significant their correlations with<br />
rejection were; those correlations are also shown as an additional<br />
row and column of the 21 21 variables matrix. Rejection<br />
is only a reference variable to evaluate correlation with<br />
the rest of variables. After a sequential implementation of<br />
PCA, three components were extracted; they defined the<br />
initial database of 21 variables with 11 variables describing<br />
three relations namely membrane/operating-conditions<br />
(comp. 1: flux, pure water permeability, salt rejection, zeta<br />
potential, mass transfer coefficient, cross flow velocity),<br />
hydrophobicity/size (comp. 2: length, log Kow, log D) and size<br />
(comp. 3: equivalent width, depth). The final three components<br />
accounted for 89.3% explanation of total variance. It is<br />
important to mention that these results were produced for the<br />
experimental dataset.<br />
The next step was the implementation of multiple linear<br />
regression (MLR) using the new set of variables. The use of<br />
MLR after PCA presents the advantage of a more simplified<br />
modelling approach. Moreover, the analysis of data before<br />
MLR may help to identify variables that are similar in<br />
response, which was the case of SR and MWCO. The dependent<br />
variable for all regression analyses was rejection. Two<br />
water research 44 (2010) 373–384<br />
methods of linear regression were used, the first method is<br />
called enter (forced) method; which performs a regression<br />
with the contribution of all variables entered to model the<br />
dependent variable. The second method is stepwise regression;<br />
which is a more sophisticated method. Each variable is<br />
entered in sequence and its contribution is assessed according<br />
to an F-test. In the present study an F-test with a statistical<br />
significance >0.10 implied removal of the variable, and F-test<br />
with a significance
Elimelech, 2003). Thus, SR is ultimately serving as a comparison<br />
parameter between membranes of the same type<br />
(aromatic polyamide) but with narrow differences in pore size,<br />
and possibly with differences in charge. The QSAR equation<br />
merges information about interaction of membrane characteristics,<br />
filtration operating conditions and organic<br />
compounds properties to predict rejections during nanofiltration.<br />
According to Eq. (2), contact angle and zeta potential<br />
as measurements of hydrophilicity and membrane surface<br />
charge, respectively, were not part of the equation and<br />
therefore did not contribute quantitatively to model rejection.<br />
However, size/steric hindrance effects related to salt rejection,<br />
and hydrophobicity of the solutes were part of the model<br />
equation. In conclusion, rejection increased by size/steric<br />
hindrance effects; but hydrophobicity decrease rejection due<br />
to adsorption and partitioning mechanisms.<br />
The results of PCA and the model cannot be generalized to<br />
experimental conditions outside of the boundary experimental<br />
conditions defined in Section 3. For instance, excessive<br />
changes of pH affect the ionic speciation of charged<br />
compounds, obviously pK a values of solutes and pH of feed<br />
waters will determine boundary conditions for applicability of<br />
the model. Changes of membrane properties such as charge<br />
and pore size due to swelling will also influence the model.<br />
Other considerations for application of the model are the type<br />
of membrane used (aromatic polyamide), fluxes, pressures<br />
and cross flow velocities. Cellulose acetate or even different<br />
membrane chemistry than NF-90 or NF-200 will influence the<br />
PCA and the model. Nonetheless, the approach is valid and<br />
can be generalized under certain conditions in upscale NF<br />
applications.<br />
4.3. QSAR model after partial least squares<br />
regression and MLR<br />
The following variables: MW, solubility, MV, MWCO, ZP, CA, P,<br />
v, J, Jo/k, recovery and width, were removed after partial least<br />
squares (PLS) regression. Therefore, the final PLS model is<br />
defined by variables named log Kow, log D, dipole, length,<br />
depth, eqwidth, PWP, SR and k. The main advantage of PLS is<br />
the ability to handle collinearity among the independent<br />
variables. In contrast, principal component analysis can not<br />
control collinearity. Another advantage of PLS was that the<br />
calculation process was simpler. However, PLS is used more as<br />
a predictive technique and not as an interpretive technique<br />
such as MLR. Therefore, in this exploratory analysis, PLS<br />
regression serves as a variable selection process and as<br />
prelude for implementation of MLR. Once again as occurred<br />
with PCA, a reduced number of variables simplified the<br />
implementation of MLR. After applying stepwise regression to<br />
the reduced number of variables, the model result obtained<br />
was also Eq. (2).<br />
4.4. QSAR model after multiple linear regression<br />
To finalize the model development under various statistical<br />
application scenarios, the implementation of direct multiple<br />
linear regression was performed. The R 2 is calculated for all<br />
possible subset models. Using this technique, the model with<br />
the largest R 2 is declared the best linear model. However, this<br />
water research 44 (2010) 373–384 379<br />
technique has some disadvantages. First, the R 2 increases<br />
with each variable included in the model. Therefore, this<br />
approach encourages including all variables in the best model<br />
although some variables may not significantly contribute to<br />
the model. This approach also contradicts the principal of<br />
parsimony that encourages as few parameters in a model as<br />
possible. Thus, the application of MLR without previous data<br />
analysis is a possibility when the number of variables is<br />
limited. Another disadvantage during the present study was<br />
that MLR was not able to distinguish collinearity between<br />
variables. This was the case for variables MWCO and SR;<br />
MWCO can replace the role of SR in Eq. (2). The new equation<br />
presented an R 2 of 0.75; the F-test was 52.5 with a significance<br />
of w0%. The equation was the following<br />
265:150 eqwidth 117:356 depth þ 81:662 length<br />
rejection ¼<br />
5:229 log D 0:272 MWCO 62:565<br />
(3)<br />
It is important to mention that Eq. (3) may have been equally<br />
defined in Section 4.2 if the variable selected before implementing<br />
PCA was MWCO and not SR as explained before;<br />
therefore it is not surprising that Eqs. (2) and (3) only show<br />
differences in two coefficients. However, considering practical<br />
or operational facts, it may be more difficult to determine<br />
changes of MWCO when fouling occurs; on the other hand,<br />
salt rejections tests are part of monitoring practices. In addition,<br />
the variable of MWCO may be difficult to define when<br />
a range of MWCO exists for a membrane rather than an<br />
approximate single MWCO value if compared to salt rejection,<br />
although that fact is not desirable for a membrane because<br />
will greatly influence rejection of solutes with sizes close to or<br />
in the range of MWCO.<br />
It may be argued that a simple stepwise MLR will produce<br />
the same results of a PCA or PLS followed by MLR, however,<br />
the application of MLR without PCA or PLS has the disadvantage<br />
of removing or adding appropriate variables during the<br />
iterative process of stepwise MLR. The stepwise MLR is only<br />
dependent on the fulfillment of a statistical condition; it even<br />
may happen that different combination of variables defines<br />
a good equation. However, the advantage of PCA or PLS is that<br />
only important variables are considered, and only those will<br />
be part of the final MLR implementation.<br />
4.5. Internal and external validation of the QSAR model<br />
Actual (measured) rejection values (106 rejection cases) versus<br />
modelled (fitted) rejections of the data used to generate the<br />
model are shown in Fig. 1, the dataset is provided as supplementary<br />
data; a 95% confidence interval shows that very few<br />
modelled rejections were out of that interval. Besides of a good<br />
fit of a model, it is necessary an assessment of the predictive<br />
power of the model, i.e. appropriate robustness (Eriksson<br />
et al., 2003). The R 2 is the most widely used measure of the<br />
ability of a QSAR model to reproduce the internal data in the<br />
training (goodness of fit), but does not explain its robustness<br />
and prediction power. One technique to evaluate prediction is<br />
the leave-one-out cross-validation technique, in which one<br />
case at a time is iteratively held-out from the training set and<br />
the rest is used for model development and the excluded case<br />
is predicted by the developed model (Gramatica, 2007).
380<br />
According to Gramatica, the predictive power of a model may<br />
be estimated by the goodness of prediction parameter Q 2<br />
leave-one-out (1-PRESS/TSS, where PRESS is the predictive<br />
error sum of squares and TSS is the total sum of squares). In<br />
general, a Q 2 > 0.5 is regarded as good and Q 2 > 0.9 as excellent<br />
(Eriksson et al., 2003). For the developed QSAR models, the<br />
model with SR (Eq. (2)) presented a Q 2 leave-one-out of 0.72,<br />
and the model with MWCO (Eq. (3)) presented a Q 2 leave-oneout<br />
of 0.72. After internal cross-validation it was demonstrated<br />
that Eqs. (2) and (3) were valid to model rejection,<br />
however, an adjustment must be made to the equation before<br />
using it to compare measured vs. predicted rejections for<br />
external databases. This adjustment was necessary to overcome<br />
the mathematical structure of the equation. Using<br />
a physical interpretation, it was evident that size parameters<br />
referring to variables length and equivalent width may be<br />
large enough to cause rejection predictions over 100%, which<br />
can be explained after observing positive coefficients for<br />
equivalent width and length. This situation may also be<br />
detrimental for rejection predictions of ionic compounds of<br />
medium to large size (0.6–1.2 nm as equivalent width) that<br />
mostly are rejected due to electrostatic repulsion and less<br />
steric hindrance. Therefore Eqs. (2) and (3) can be transformed<br />
to the following conditional equation<br />
rejection ¼<br />
a b<br />
measured rejection (%)<br />
HB-ion<br />
HB-neu<br />
HL-ion<br />
HL-neu<br />
95%<br />
confidence<br />
interval<br />
100 if QSAR model 100<br />
QSAR model<br />
modeled rejection (%)<br />
length, eqwidth,<br />
depth, log D, SR<br />
R² = 0.75<br />
An external dataset (that gathered three different datasets)<br />
was selected for external validation of the QSAR model. The<br />
external dataset is presented as supplementary data. The first<br />
part of the external dataset corresponds to membrane Filmtec<br />
NF-90. The second part corresponds to NF membrane Trisep<br />
TS-80 and the third part corresponds to NF membrane Desal<br />
HL. Desal HL membrane has a main difference with NF-90 and<br />
Trisep TS-80 membranes, viz. the MWCO of Desal HL is in the<br />
range of 150–300 Da, while NF-90 and Trisep TS-80 were<br />
reported to have a MWCO of 200 Da. Therefore an average<br />
MWCO of 225 Da was assumed for Desal HL during the<br />
application of Eq. (3). It is worthwhile to mention that the<br />
water research 44 (2010) 373–384<br />
(4)<br />
measured rejection (%)<br />
HB-ion<br />
HB-neu<br />
HL-ion<br />
HL-neu<br />
95%<br />
confidence<br />
interval<br />
modeled rejection (%)<br />
Fig. 1 – QSAR model of experimental database: (a) Eq. (2); and (b) Eq. (3).<br />
length, eqwidth,<br />
depth, log D,<br />
MWCO<br />
R² = 0.75<br />
second and third parts of the external dataset were generated<br />
using spiral wound membrane elements instead of flat sheet<br />
membranes. Fig. 2a show plotted results of measured rejections<br />
vs. predicted rejections after calculations with Eq. (4), for<br />
QSAR model with Eq. (2) (SR). According to Fig. 2aanR 2 of 0.75<br />
was obtained after considering all compounds of the external<br />
dataset. However, after observing the rejection cases by NF-90<br />
for bromoform (BF) and trichloroethene (TCE), it appeared that<br />
BF and TCE may be considered atypical results because their<br />
rejections did not correspond to their size and hydrophobicity<br />
when compared to other compounds with comparable<br />
molecular descriptors. According to Table 3 BF has approximately<br />
the same hydrophobicity and polarity as CF, but BF is<br />
bigger than CF; therefore, measured rejection for BF was<br />
expected to be higher than rejection of CF (0%) due to size<br />
exclusion. The measured rejection of TCE (3%) was not<br />
comparable to the rejection of perchloroethene (39%)<br />
although they have the same length but a very small difference<br />
in equivalent width. In an experiment conducted by Kim<br />
et al. (2007) rejections of BF and TCE were of 50 and 33%,<br />
respectively, for a low pressure reverse osmosis membrane. It<br />
was also observed that the measured rejection (53%) of<br />
linuron (LNU) may be considered as atypical observation<br />
because a higher rejection was expected. According to Table 3,<br />
monolinuron is close to LNU in size and polarity; thus<br />
measured rejection of monolinuron was of 77%, higher than<br />
53%. Also according to Table 3, carbamazepine is close to LNU<br />
in size and hydrophobicity; thus measured rejection of<br />
carbamazepine was of 94%, higher than 53%. A similar<br />
explanation was given for the low rejection (70%) of N-acetyl-<br />
L-tyrosine (NAT) by Desal HL compared to 94% rejection by<br />
TS-80 as can be seen in Table 3. Other non-expected rejection<br />
was that of 2-(1H)-quinoline (QNL), with a rejection of 22%<br />
when compared to 2-methoxyethanol and perchloroethene<br />
with rejections of 32 and 39%, respectively; even though the<br />
length of QNL is greater than the length of 2-methoxyethanol<br />
and perchloroethene, a lower rejection of QNL was observed.<br />
However, rejections of methacetin (MTC) and NDPA may be<br />
influenced by their small equivalent width as can be seen in
a b<br />
measured rejection (%)<br />
100<br />
80<br />
60<br />
40<br />
ETH<br />
MTBE<br />
R 2 20<br />
QNL<br />
NDPA<br />
0<br />
BF TCE<br />
MTC<br />
= 0.75<br />
0 20 40 60 80 100<br />
predicted rejection (%)<br />
Table 3. The response of the model for rejection of those<br />
particular compounds (with 2*equivalent width < or w length)<br />
was of over prediction, but the model presented better<br />
response for rejections of metribuzin, atrazin and N-acetyl-Ltyrosine<br />
(all with length >2*equivalent width) as observed in<br />
Table 3. Ethanol (ETH) and MTBE with rejections of 38 and 60%,<br />
respectively, were expected to be lower for Desal HL<br />
membrane because it was observed that rejection of ETH was<br />
of 9% for TS-80; and MTBE was expected to have a rejection<br />
compared to that of 2-methoxyethanol (32%) or perchloroethene<br />
(39%) due to proximities in size. After selection<br />
water research 44 (2010) 373–384 381<br />
NDPA<br />
MTC<br />
NAT<br />
LNU<br />
0<br />
0 20 40 60 80 100<br />
predicted rejection (%)<br />
R 2 = 0.84<br />
and justified separation of the mentioned rejection cases,<br />
Fig. 2b presents the predictions of the external dataset with an<br />
R 2 of 0.84. In a similar explanatory scenario, Figs. 3a,b show<br />
measured rejections vs. predicted rejections after calculations<br />
with Eq. (4), for QSAR model with equation 3 (MWCO). The<br />
main difference between Figs. 2b and 3b was that the model<br />
with MWCO (Fig. 3b) showed a lower R 2 (0.80) than the model<br />
with SR (R 2 ¼ 0.84), meaning that the latter had a better<br />
goodness of fit for external prediction response. Moreover, the<br />
characterization of magnesium sulphate salt rejection for<br />
a membrane may be preferred instead of MWCO, particularly<br />
Table 3 – Partial list of rejections and compound properties of external dataset.<br />
Compound Abb. Length Eqwidth Depth Log D Dipole Measured rejection Predicted rejection Membrane<br />
Chloroform 0.53 0.42 0.35 1.97 1.12 0 0 NF-90<br />
Ethanol 0.64 0.52 0.51 0.31 1.55 9 14 TS-80<br />
Ethanol ETH 0.64 0.52 0.51 0.31 1.55 38 0 Desal HL<br />
Carbontetrachloride 0.64 0.6 0.57 2.83 0.30 35 26 NF-90<br />
Bromoform BF 0.69 0.56 0.48 2.40 1.00 0 33 NF-90<br />
MTBE MTBE 0.77 0.63 0.59 0.94 1.37 60 25 Desal HL<br />
Trichloroethene TCE 0.78 0.49 0.36 2.29 0.95 3 36 NF-90<br />
Perchloroethene 0.78 0.59 0.45 3.40 0.11 39 46 NF-90<br />
2-methoxyethanol 0.87 0.52 0.51 0.77 0.25 32 34 TS-80<br />
2-(1H)-quinoline QNL 1.00 0.52 0.36 1.26 3.38 22 53 TS-80<br />
NDPA NDPA 1.16 0.60 0.53 1.36 3.40 45 69 TS-80<br />
NDPA NDPA 1.16 0.60 0.53 1.36 3.40 19 55 Desal HL<br />
Metribuzin 1.17 0.74 0.64 0.47 0.52 97 99 TS-80<br />
Carbamazepine 1.20 0.73 0.58 2.45 3.66 88 94 TS-80<br />
Linuron LNU 1.21 0.69 0.53 3.20 2.11 53 84 TS-80<br />
Monolinuron 1.22 0.69 0.65 2.30 2.02 79 77 TS-80<br />
Atrazin 1.26 0.74 0.55 2.61 3.43 91 100 TS-80<br />
Methacetin MTC 1.28 0.52 0.42 1.03 2.20 38 72 TS-80<br />
Methacetin MTC 1.28 0.52 0.42 1.03 2.20 5 58 Desal HL<br />
N-acetyl-L-tyrosine NAT 1.33 0.71 0.60 2.18 3.45 94 100 TS-80<br />
N-acetyl-L-tyrosine NAT 1.33 0.71 0.60 2.18 3.45 70 100 Desal HL<br />
measured rejection (%)<br />
NF-90 TS-80 HL Linear ( ) NF-90 TS-80 HL Linear ( )<br />
Fig. 2 – Predicted rejections for external dataset using magnesium sulphate SR: (a) all external data; and (b) selected external<br />
data.<br />
100<br />
80<br />
60<br />
40<br />
20
382<br />
measured rejection (%)<br />
100<br />
80<br />
60<br />
40<br />
20<br />
for nanofiltration and low pressure reverse osmosis<br />
membranes; besides, the effect of fouling in membranes can<br />
also be quantified by salt rejection experiments. We can state<br />
that the QSAR model with SR demonstrated to be acceptable<br />
for the external dataset of NF-90, Trisep TS-80 and Desal HL<br />
with an R 2 of 0.75 and 0.84 for the total external dataset and<br />
justified selected external dataset, respectively. Although the<br />
model can be valid with limitations related to boundary<br />
experimental conditions mentioned in Section 3, its applicability<br />
and approach can be of value for the construction of<br />
a model with combined datasets organized in training and<br />
testing groups.<br />
5. Conclusions<br />
0<br />
ETH<br />
BF<br />
MTBE<br />
TCE<br />
QNL<br />
NDPA<br />
MTC<br />
R 2 = 0.74<br />
0 20 40 60 80 100<br />
predicted rejection (%)<br />
– A general QSAR model equation was developed to merge<br />
information about interaction of membrane characteristics,<br />
filtration operating conditions and solute properties<br />
to predict rejections of emerging contaminants during<br />
nanofiltration.<br />
– The QSAR model identified that the most important<br />
variables that influence rejection of organic solutes were<br />
log D, salt rejection, equivalent width, depth and length.<br />
– Rejection increased by size/steric hindrance effects,<br />
solute hydrophobicity decreased rejection due to adsorption<br />
and partitioning mechanisms.<br />
– Salt rejection incorporated steric hindrance and electrostatic<br />
repulsion effects that were related to the membrane<br />
structure and operating conditions.<br />
– The use of MWCO was acceptable for modelling purposes;<br />
however NF membranes with a broad range of MWCO<br />
(pore size and distribution) may difficult estimation of<br />
rejection of contaminants, thus magnesium sulphate salt<br />
rejection may be more appropriate.<br />
water research 44 (2010) 373–384<br />
a b<br />
NDPA<br />
MTC<br />
NAT<br />
LNU<br />
measured rejection (%)<br />
100<br />
80<br />
60<br />
40<br />
20<br />
0<br />
0 20 40 60 80 100<br />
predicted rejection (%)<br />
Acknowledgements<br />
The authors acknowledge Delft Cluster and EU Techneau<br />
Project for funding this project. The authors also acknowledge<br />
Tae-Uk Kim and Arne Verliefde for providing data and details<br />
of their research. The authors thank Filmtec (Dow Chemical<br />
Co.) for donating the membranes, Dr. Jaeweon Cho of GIST<br />
(Korea), Dr. Frank Sacher of TZW (Germany) and Steven<br />
Mookhoek of TU Delft (Netherlands) for contributing with<br />
analytical results and facilities.<br />
Appendix.<br />
Supplementary data<br />
Supplementary data associated with this article can be found,<br />
in the online version, at doi:10.1016/j.watres.2009.06.054<br />
references<br />
R 2 = 0.80<br />
NF-90 TS-80 HL Linear ( ) NF-90 TS-80 HL Linear ( )<br />
Fig. 3 – Predicted rejections for external dataset using MWCO: (a) all external data; and (b) selected external data.<br />
Abdi, H., 2003. Partial least squares (PLS) regression. In: Lewis-<br />
Beck, M., Bryman, A., Futing, T. (Eds.), Encyclopaedia of Social<br />
Sciences Research Methods. Sage, Thousand Oaks, CA.<br />
Bellona, C., Drewes, J.E., 2005. The role of membrane surface<br />
charge and solute physico-chemical properties in the rejection<br />
of organic acids by NF. Journal of Membrane Science 249 (1–2),<br />
227–234.<br />
Campbell, P., Srinivasan, R., Knoell, T., Phipps, D., Ishida, K.,<br />
Safarik, J., Cormack, T., Ridgway, H., 1999. Quantitative<br />
structure-activity relationship (QSAR) analysis of surfactants<br />
influencing attachment of a Mycobacterium sp. to cellulose
acetate and aromatic polyamide reverse osmosis membranes.<br />
Biotechnology and Bioengineering 64 (5), 527–544.<br />
Castiglioni, S., Bagnati, R., Fanelli, R., Pomati, F., Calamari, D.,<br />
Zuccato, E., 2006. Removal of pharmaceuticals in sewage<br />
treatment plants in Italy. Environmental Science &<br />
Technology 40 (1), 357–363.<br />
Connell, D.W., 1990. Bioaccumulation of Xenobiotic Compounds.<br />
CRC Press, Boca Raton, FL, pp. 75-110.<br />
Cornelissen, E.R., Verdouw, J., Gijsbertsen-Abrahamse, A.J.,<br />
Hofman, J.A.M.H., 2005. A nanofiltration retention model for<br />
trace contaminants in drinking water sources. Desalination<br />
178 (1–3), 179–192.<br />
Drewes, J.E., Amy, G., Kim, T.-U., Xu, P., Bellona, C.,<br />
Oedekoven, M., Macalady, D., 2006. Rejection of Wastewater-<br />
Derived Micropollutants in High-Pressure Membrane<br />
Applications Leading to Indirect Potable Reuse Effect of<br />
Membranes and Micropollutant Properties. WateReuse,<br />
Alexandria, VA.<br />
Eriksson, L., Jaworska, J., Worth, A.P., Cronin, M.T.D.,<br />
McDowell, R.M., Gramatica, P., 2003. Methods for reliability<br />
and uncertainty assessment and for applicability evaluations<br />
of classification and regression-based QSARs. Environmental<br />
Health Perspectives 111 (10), 1361–1375.<br />
Escher, B.I., Bramaz, N., Eggen, R.I.L., Richter, M., 2005. In vitro<br />
assessment of modes of toxic action of pharmaceuticals in<br />
aquatic life. Environmental Science & Technology 39 (9),<br />
3090–3100.<br />
Everitt, B.S., Dunn, G., 2001. Applied Multivariate Data Analysis.<br />
Arnold, London.<br />
Fujikawa, M., Nakao, K., Shimizub, R., Akamatsu, M., 2007. QSAR<br />
study on permeability of hydrophobic compounds with<br />
artificial membranes. Bioorganic & Medicinal Chemistry 15<br />
(11), 3753–3767.<br />
Gramatica, P., 2007. Principles of QSAR models validation:<br />
internal and external. QSAR & Combinatorial Science 26 (5),<br />
694–701.<br />
Heberer, T., 2002. Occurrence, fate, and removal of<br />
pharmaceutical residues in the aquatic environment: a review<br />
of recent research data. Toxicology Letters 131 (1–2), 5–17.<br />
Henderson, R.K., Baker, A., Parsons, S.A., Jefferson, B., 2008.<br />
Characterisation of algogenic organic matter extracted from<br />
cyanobacteria, green algae and diatoms. Water Research 42<br />
(13), 3435–3445.<br />
Hoek, E.M.V., Elimelech, M., 2003. Cake-enhanced concentration<br />
polarization: a new fouling mechanism for salt-rejecting<br />
membranes. Environmental Science & Technology 37 (24),<br />
5581–5588.<br />
Johnson, R.A., Wichern, D.W., 2007. Applied Multivariate<br />
Statistical Analysis. Pearson Prentice Hall, Upper Saddle<br />
River, NJ.<br />
Jolliffe, I.T., 2002. Principal Component Analysis. Springer-Verlag,<br />
New York.<br />
Jørgensen, B., Goegebeur, Y., 2006. Multivariate Data Analysis and<br />
Chemometrics. Department of Statistics, University of<br />
Southern Denmark.<br />
Kim, T.-U., Drewes, J.E., Summer, R.S., Amy, G., 2007. Solute<br />
transport model for trace organic neutral and charged<br />
compounds through nanofiltration and reverse osmosis<br />
membranes. Water Research 41 (17), 3977–3988.<br />
Kimura, K., Amy, G., Drewes, J., Heberer, T., Kim, T., Watanabe, Y.,<br />
2003. Rejection of organic micropollutants (disinfection byproducts,<br />
endocrine disrupting compounds, and<br />
pharmaceutically active compounds) by NF/RO membranes.<br />
Journal of Membrane Science 227 (1–2), 113–121.<br />
Kimura, K., Toshima, S., Amy, G., Watanabe, Y., 2004. Rejection of<br />
neutral endocrine disrupting compounds (EDCs) and<br />
pharmaceutical active compounds (PhACs) by RO membranes.<br />
Journal of Membrane Science 245 (1–2), 71–78.<br />
water research 44 (2010) 373–384 383<br />
Kiso, Y., Sugiura, T., Kitao, T., Nishimura, K., 2001. Effects of<br />
hydrophobicity and molecular size on rejection of aromatic<br />
pesticides with nanofiltration membranes. Journal of<br />
Membrane Science 192 (1–2), 1–10.<br />
Kolpin, D.W., Furlong, E.T., Meyer, M.T., Thurman, E.M., Zaugg, S.D.,<br />
Barber, L.B., Buxton, H.T., 2002. Pharmaceuticals, hormones,<br />
and other organic wastewater contaminants in U.S. Streams,<br />
1999–2000: a National Reconnaissance. Environmental Science<br />
& Technology 36 (6), 1202–1211.<br />
Landau, S., Everitt, B., 2004. A Handbook of Statistical Analysis<br />
using SPSS. Chapman & Hall/CRC, Boca Raton, FL.<br />
Lee, N., Amy, G., Croue, J.-P., Buisson, H., 2004. Identification and<br />
understanding of fouling in low-pressure membrane (MF/UF)<br />
filtration by natural organic matter (NOM). Water Research 38<br />
(20), 4511–4523.<br />
Libotean, D., Giralt, J., Rallo, R., Cohen, Y., Giralt, F., Ridgway, H.,<br />
Rodriguez, G., Phipps, D., 2008. Organic compounds passage<br />
through RO membranes. Journal of Membrane Science 313<br />
(1-2), 23–43.<br />
Nghiem, L.D., Schäfer, A.I., Elimelech, M., 2004. Removal of<br />
natural hormones by nanofiltration membranes:<br />
measurement, modelling, and mechanisms. Environmental<br />
Science & Technology 38 (6), 1888–1896.<br />
Nghiem, L.D., Schäfer, A.I., Elimelech, M., 2005. Pharmaceutical<br />
retention mechanisms by nanofiltration membranes.<br />
Environmental Science & Technology 39 (19), 7698–7705.<br />
Ozaki, H., Li, H., 2002. Rejection of organic compounds by ultralow<br />
pressure reverse osmosis membrane. Water Research 36<br />
(1), 123–130.<br />
Pomati, F., Castiglioni, S., Zuccato, E., Fanelli, R., Vigetti, D.,<br />
Rossetti, C., Calamari, C., 2006. Effects of a complex mixture of<br />
therapeutic drugs at environmental levels on human<br />
embryonic cells. Environmental Science & Technology 40 (7),<br />
2442–2447.<br />
Ren, S., Das, A., Lien, E.J., 1996. QSAR analysis of membrane<br />
permeability to organic compounds. Journal of Drug Targeting<br />
4 (2), 103–107.<br />
Sacher, F., Lange, F.T., Brauch, H.-J., Blankenhorn, I., 2001.<br />
Pharmaceuticals in groundwaters: analytical methods and<br />
results of a monitoring program in Baden-Württemberg,<br />
Germany. Journal of Chromatography A 938 (1–2), 199–210.<br />
Sacher, F., Ehmann, M., Gabriel, S., Graf, C., Brauch, H.-J., 2008.<br />
Pharmaceutical residues in the river Rhinedresults of a onedecade<br />
monitoring programme. Journal of Environmental<br />
Monitoring 10 (5), 664–670.<br />
Santos, J.L.C., de Beukelaar, P., Vankelecom, I.F.J., Velizarov, S.,<br />
Crespo, J.G., 2006. Effect of solute geometry and orientation on<br />
the rejection of uncharged compounds by nanofiltration.<br />
Separation and Purification Technology 50 (1), 122–131.<br />
Sawyer, C.N., McCarty, P.L., Parkin, G.F., 2003. Chemistry for<br />
Environmental Engineering and Science. McGraw-Hill, New<br />
York, pp. 479-520.<br />
Schäfer, A.I., Nghiem, L.D., Waite, T.D., 2003. Removal of the<br />
natural hormone estrone from aqueous solutions using<br />
nanofiltration and reverse osmosis. Environmental Science &<br />
Technology 37 (1), 182–188.<br />
Shim, Y., Lee, H.-J., Lee, S., Moon, S.-H., Cho, J., 2002. Effects of<br />
natural organic matter and ionic species on membrane surface<br />
charge. Environmental Science & Technology 36 (17), 3864–3871.<br />
Van der Bruggen, B., Vandecasteele, C., 2002. Modeling of the<br />
retention of uncharged molecules with nanofiltration. Water<br />
Research 36 (5), 1360–1368.<br />
Van der Bruggen, B., Schaep, J., Wilms, D., Vandecasteele, C.,<br />
1999. Influence of molecular size, polarity and charge on the<br />
retention of organic molecules by nanofiltration. Journal of<br />
Membrane Science 156 (1), 29–41.<br />
Verliefde, A., Cornelissen, E., Amy, G., Van der Bruggen, B., van<br />
Dijk, H., 2007. Priority organic micropollutants in water
384<br />
sources in Flanders and the Netherlands and assessment of<br />
removal possibilities with nanofiltration. Environmental<br />
Pollution 146 (1), 281–289.<br />
Verliefde, A.R.D., Cornelissen, E.R., Heijman, S.G.J., Verberk, J.Q.J.C.,<br />
Amy, G.L., Van der Bruggen, B., van Dijk, J.C., 2008. The role of<br />
electrostatic interactions on the rejection of organic solutes in<br />
aqueous solutions with nanofiltration. Journal of Membrane<br />
Science 322 (1), 52–66.<br />
Vosges, M., Braguer, J.-C., Combarnous, Y., 2008. Long-term<br />
exposure of male rats to low-dose ethinylestradiol (EE2)<br />
in drinking water: effects on ponderal growth and on<br />
litter size of their progeny. Reproductive Toxicology 25<br />
(2), 161–168.<br />
water research 44 (2010) 373–384<br />
Xu, P., Drewes, J.E., Bellona, C., Amy, G., Kim, T.-U., Adam, M.,<br />
Heberer, T., 2005. Rejection of emerging organic<br />
micropollutants in nanofiltration-reverse osmosis membrane<br />
applications. Water Environment Research 77 (1), 40–48.<br />
Yangali-Quintanilla, V., Kim, T.-U., Kennedy, M., Amy, G., 2008.<br />
Modeling of RO/NF membrane rejections of PhACs and organic<br />
compounds: a statistical analysis. Drinking Water Engineering<br />
and Science 1 (1), 7–15.<br />
Yangali-Quintanilla, V., Sadmani, A., McConville, M.,<br />
Kennedy, M., Amy, G., 2009. Rejection of pharmaceutically<br />
active compounds and endocrine disrupting compounds by<br />
clean and fouled nanofiltration membranes. Water Research<br />
43 (9), 2349–2362.
Harmful algae and their potential impacts on desalination<br />
operations off southern California<br />
David A. Caron a, *, Marie-Ève Garneau a , Erica Seubert a , Meredith D.A. Howard a,b ,<br />
Lindsay Darjany a , Astrid Schnetzer a , Ivona Cetinić a , Gerry Filteau c , Phil Lauri d ,<br />
Burton Jones a , Shane Trussell e<br />
a<br />
Department of Biological Sciences, University of Southern California, 3616 Trousdale Parkway, Los Angeles, CA 90089-0371, USA<br />
b<br />
Southern California Coastal Water Research Project, 3535 Harbor Blvd., Suite 110, Costa Mesa, CA 92626, USA<br />
c<br />
Separation Processes, Inc., 3156 Lionshead Avenue, Suite 2, Carlsbad, CA 92010, USA<br />
d<br />
West Basin Municipal Water District, 17140 Avalon Blvd., Suite 210, Carson, CA 90746, USA<br />
e<br />
Trussell Technologies, Inc., 6540 Lusk Boulevard, Suite C175, San Diego, CA 92121, USA<br />
article info<br />
Article history:<br />
Received 2 April 2009<br />
Received in revised form<br />
12 June 2009<br />
Accepted 23 June 2009<br />
Available online 30 June 2009<br />
Keywords:<br />
Harmful algal blooms<br />
Desalination<br />
Red tides<br />
Phytoplankton<br />
Phytotoxins<br />
water research 44 (2010) 385–416<br />
Available at www.sciencedirect.com<br />
journal homepage: www.elsevier.com/locate/watres<br />
abstract<br />
* Corresponding author. Tel.: þ1 213 740 0203; fax: þ1 213 740 8123.<br />
E-mail address: dcaron@usc.edu (D.A. Caron).<br />
0043-1354/$ – see front matter ª 2009 Elsevier Ltd. All rights reserved.<br />
doi:10.1016/j.watres.2009.06.051<br />
Seawater desalination by reverse osmosis (RO) is a reliable method for augmenting<br />
drinking water supplies. In recent years, the number and size of these water projects have<br />
increased dramatically. As freshwater resources become limited due to global climate<br />
change, rising demand, and exhausted local water supplies, seawater desalination will play<br />
an important role in the world’s future water supply, reaching far beyond its deep roots in<br />
the Middle East. Emerging contaminants have been widely discussed with respect to<br />
wastewater and freshwater sources, but also must be considered for seawater desalination<br />
facilities to ensure the long-term safety and suitability of this emerging water supply.<br />
Harmful algal blooms, frequently referred to as ‘red tides’ due to their vibrant colors, are<br />
a concern for desalination plants due to the high biomass of microalgae present in ocean<br />
waters during these events, and a variety of substances that some of these algae produce.<br />
These compounds range from noxious substances to powerful neurotoxins that constitute<br />
significant public health risks if they are not effectively and completely removed by the RO<br />
membranes. Algal blooms can cause significant operational issues that result in increased<br />
chemical consumption, increased membrane fouling rates, and in extreme cases, a plant to<br />
be taken off-line. Early algal bloom detection by desalination facilities is essential so that<br />
operational adjustments can be made to ensure that production capacity remains unaffected.<br />
This review identifies the toxic substances, their known producers, and our present<br />
state of knowledge regarding the causes of toxic episodes, with a special focus on the<br />
Southern California Bight.<br />
ª 2009 Elsevier Ltd. All rights reserved.
386<br />
1. Introduction<br />
1.1. General overview of harmful algal blooms:<br />
a growing global concern<br />
Microscopic algae constitute an essential component of all<br />
aquatic food webs. Photosynthetic production of organic<br />
material by this diverse group of species comprises the<br />
primary source of nutrition for all heterotrophic forms of life<br />
in much of the world’s ocean and freshwater ecosystems.<br />
Microalgae can reach high abundances in the plankton during<br />
periods of optimal growth and reduced grazing pressure by<br />
herbivores. Such localized mass proliferations are known as<br />
algal (or phytoplankton) blooms. In addition, a small proportion<br />
of microalgal species are capable of producing a number<br />
of noxious or toxic compounds that cause a variety of adverse<br />
effects on ecosystem structure and function. These<br />
substances pose the potential for ecosystem damage, food<br />
web disruption and marine animal mortality, and present<br />
a significant human health risk through the consumption of<br />
contaminated seafood and, in at least one case, direct exposure<br />
to water or aerosols containing these toxic compounds.<br />
Additionally, the algal biomass and the associated organic<br />
load cause significant desalination operational issues,<br />
impacting the pretreatment system and possibly forcing the<br />
treatment plant to be taken off-line (Petry et al., 2007).<br />
Countless human deaths resulting from the consumption<br />
of seafood contaminated with algal toxins have been avoided<br />
through rigorous monitoring programs, but sea life has not<br />
been so fortunate. Approximately one half of all unusual<br />
marine mammal mortality incidents are now attributable to<br />
the ingestion of food or prey contaminated by harmful algal<br />
blooms (Ramsdell et al., 2005). Losses in revenue due to the<br />
direct contamination of seafood products and indirect effects<br />
on tourism and other uses of coastal areas have been estimated<br />
in the tens of millions of dollars annually in the U.S.<br />
states along the Pacific coast (Trainer et al., 2002).<br />
There is now convincing evidence that harmful algal<br />
bloom (HAB) events are increasing at local, regional and global<br />
scales worldwide (Smayda, 1990; Hallegraeff, 1993, 2003;<br />
Anderson et al., 2002; Glibert et al., 2005a) and along the North<br />
American west coast in particular (Horner et al., 1997; Trainer<br />
et al., 2003). This increased occurrence may be due in part to<br />
better detection of HAB episodes in recent years or the global<br />
dispersal of toxic algal species via the transport of resting<br />
spores in ships’ ballast waters (Hallegraeff and Bolch, 1992;<br />
Burkholder et al., 2007), but another very likely cause is the<br />
increasing impact of anthropogenic activities on coastal<br />
ecosystems (Smayda, 1990; Anderson et al., 2002; Glibert et al.,<br />
2005b, 2006; Howard et al., 2007; Cochlan et al., 2008; Kudela<br />
et al., 2008a). Recent reports reveal extensive and, in some<br />
cases, newly emerging occurrences of HABs along the coasts<br />
of the U.S. (Fig. 1). These incidents engender a variety of<br />
noxious impacts on ecosystems and public health, including<br />
direct effects on organisms due to the production of acutely<br />
toxic substances, and indirect effects such as reduced availability<br />
of dissolved oxygen in the water column resulting from<br />
the decomposition of the extensive amounts of organic<br />
substances usually produced during such blooms. The<br />
water research 44 (2010) 385–416<br />
Fig. 1 – Distribution of some well-known regional HAB<br />
issues along U.S. shores, including (a) Alaska and (b)<br />
Hawaii. Causes and impacts of these poisoning events are<br />
defined in Tables 1–3. Summarized from information<br />
presented on the Harmful Algae webpage (http://www.<br />
whoi.edu/redtide/).<br />
dramatic increases in biomass and organic load that accompany<br />
these events pose a significant threat to seawater desalination<br />
facilities (Gaid and Treal, 2007).<br />
1.2. Regional HAB issues along U.S. coastlines<br />
Harmful algae are present throughout U.S. coastal waters, but<br />
not all species are of equal concern in all regions (Fig. 1). For<br />
example, toxic species of the dinoflagellate genus Alexandrium<br />
are common over vast stretches of the U.S. coastline, but<br />
coastal regions of the northeastern and northwestern U.S.<br />
appear to experience particularly high rates of occurrence of<br />
toxic ‘red tides’ caused by these species. The neurotoxins<br />
produced by Alexandrium, called saxitoxins, cause paralytic<br />
shellfish poisoning (PSP) in humans when ingested through<br />
contaminated seafood (particularly filter-feeding shellfish).<br />
Similarly, several toxic species of the diatom genus Pseudonitzschia<br />
occur along the entire U.S. coastline but significant<br />
concentrations of the neurotoxin, domoic acid, produced by<br />
these species have historically constituted a health threat<br />
primarily in the northeastern and northwestern U.S. (Bates<br />
et al., 1989) where it has been documented as the cause of<br />
amnesic shellfish poisoning (ASP) in humans. However, high<br />
concentrations of domoic acid in the plankton and in diverse<br />
planktivorous organisms have been recently documented<br />
along the entire Pacific coast of the U.S. (Scholin et al., 2000;<br />
Trainer et al., 2002; Schnetzer et al., 2007), as well as in the<br />
Gulf of Mexico (Pan et al., 1998). Domoic acid has been<br />
attributed to numerous marine animal mortalities along the<br />
U.S. west coast. In the Gulf of Mexico, primarily along the<br />
west coast of Florida, extensive and recurrent blooms of the<br />
dinoflagellate Karenia brevis produce a suite of toxins, known<br />
as brevetoxins, that can be aerosolized by breaking waves and
induce neurotoxic shellfish poisoning (NSP) in people inhaling<br />
the aerosols (see review of Kirkpatrick et al., 2004). The<br />
Tampa Bay seawater desalination facility is the only operating<br />
seawater desalination treatment plant of significant<br />
size in the United States. It is located along the west coast of<br />
Florida and is likely to encounter algal blooms that contain<br />
brevetoxin.<br />
Less toxic blooms also take place with regional specificity.<br />
The pelagophyte Aureococcus anophagefferens causes ‘brown<br />
tides’ in coastal waters of Rhode Island, near Long Island (NY)<br />
and southward along the mid-Atlantic coast of the U.S. since<br />
1985. No specific toxins have been identified from A. anophagefferens,<br />
and no human fatalities have been directly attributed<br />
to these blooms. Nevertheless, this species appears to be<br />
unpalatable or inhibitory to many filter-feeding mollusks and<br />
has caused substantial mortality among these populations,<br />
including commercially valuable species (Bricelj and Lonsdale,<br />
1997).<br />
Other microalgal species can disrupt food webs or cause<br />
reductions in water quality without producing acutely toxic<br />
conditions. Among these are the ‘colorful’ red tides of the<br />
dinoflagellate Lingulodinium polyedrum, a yessotoxin producer,<br />
that have occurred periodically throughout several decades<br />
along the south and central Californian coasts (Horner et al.,<br />
1997; Gregorio and Pieper, 2000). These blooms have so far<br />
been found to be relatively innocuous in these waters but<br />
massive accumulations of these cells could have significant<br />
impact on desalination plants because of increased turbidity,<br />
high suspended solids and organic loading of influent water.<br />
Furthermore, accumulations of cells in protected harbors can<br />
cause fish mortality by depleting oxygen dissolved in the<br />
water, further challenging influent screening and pretreatment<br />
systems at desalination plants. Other taxa, such as<br />
species of the prymnesiophyte genus Phaeocystis, produce<br />
substances that can lead to enormous buildups of sea foam<br />
along coasts (Armonies, 1989).<br />
1.3. Desalination, plankton and water quality issues<br />
Large research programs have developed within different<br />
geographic areas throughout the U.S. to address regional HAB<br />
issues. These programs are designed to study the species,<br />
toxins and environmental causes of HAB outbreaks. These<br />
efforts, as well as local, county, state and federal monitoring<br />
programs provide basic information for marine resource use<br />
and have focused almost exclusively on threats to human<br />
health via the consumption of contaminated seafood. Unfortunately,<br />
few if any of these programs provide sufficient<br />
information on appropriate temporal and spatial resolution<br />
for thoroughly assessing the potential impact of HAB events<br />
on reverse osmosis desalination operations. Moreover, toxin<br />
analyses have primarily examined the presence of these<br />
substances in particulate material (plankton or animal tissue,<br />
particularly shellfish and finfish), and therefore may be poor<br />
predictors for the amount of toxins that might occur in<br />
seawater in the dissolved state during algal blooms, which<br />
would be most likely to be loaded onto reverse osmosis<br />
membranes during desalination.<br />
There are two potential impacts that HABs may have on<br />
seawater desalination facilities: (1) algal toxins in ocean water<br />
water research 44 (2010) 385–416 387<br />
pose a significant treatment challenge for the reverse osmosis<br />
system to ensure that these molecules are effectively removed<br />
and (2) increased turbidity, total suspended solids and total<br />
organic content resulting from algal biomass and growth<br />
challenge the entire desalination facility’s treatment train.<br />
The significance of these issues will depend on the specific<br />
algae forming a bloom and the toxin(s) or other substances<br />
that they produce, the magnitude and duration of the bloom,<br />
and the specific desalination process conducted. For example,<br />
multi-effect distillation and multi-flash distillation might be<br />
susceptible to (2) but would be much less affected by toxins in<br />
the water (1). Desalination using reverse osmosis presumably<br />
would be vulnerable to both issues. Therefore, for the latter<br />
desalination approach, a thorough understanding of HAB<br />
episodes in terms of incidence and seasonality, vertical and<br />
horizontal spatial distribution, as well as biological aspects<br />
such as algal composition within a geographical region could<br />
help optimize the design and operational efficiency of desalination<br />
plants employing reverse osmosis.<br />
This paper provides an overview of HABs occurring along<br />
the continental U.S. coastline with special emphasis on the<br />
southwestern U.S., and provides some insight on the potential<br />
impacts that these events may have on the seawater desalination<br />
process. In recent years, this geographical area has<br />
become a focal point of discussions regarding desalination<br />
(Cooley et al., 2006) because of its sizable population and the<br />
particularly tenuous nature of the water supply to this region.<br />
Although numerous issues involving the desalination process<br />
are now being examined (Separation Processes Inc., 2005; Gaid<br />
and Treal, 2007; Pankratz, 2008, 2009), very limited information<br />
exists on the risks that algal blooms pose to seawater<br />
desalination facilities. A review of the major species<br />
producing harmful blooms, the substances they produce, and<br />
information on the spatial and temporal distributions of<br />
blooms are presented along with some conclusions on their<br />
potential impacts. This paper also provides some general<br />
guidelines on how early detection may help prevent or minimize<br />
the impact of HABs on a desalination facility’s production<br />
capacity or its water quality.<br />
2. Toxin producers and toxin concentrations<br />
of the west coast<br />
A variety of toxins including several powerful neurotoxins are<br />
produced by microalgae, and a number of these toxins and<br />
potentially toxic algal species have been detected on the U.S.<br />
west coast (Table 1). The ability to rapidly detect and quantify<br />
toxic algae in natural water samples is problematic at this<br />
time. Many of these species are difficult to identify using light<br />
microscopy. For this reason, new genetic and immunological<br />
methods for species identification and enumeration have<br />
been appearing rapidly in the literature (Miller and Scholin,<br />
1998; Bowers et al., 2000, 2006; Coyne et al., 2001; Caron et al.,<br />
2003; Galluzzi et al., 2004; Anderson et al., 2005; Mikulski et al.,<br />
2005, 2008; Ahn et al., 2006; Handy et al., 2006; Moorthi et al.,<br />
2006; Iwataki et al., 2007, 2008; Demir et al., 2008; Matsuoka<br />
et al., 2008). Moreover, many toxin-producing algal species<br />
exhibit variable toxin production in response to environmental<br />
conditions, and among different strains of the same species
388<br />
Table 1 – Planktonic species occurring along the west coast of the U.S. that are potential concerns for reverse osmosis<br />
operations.<br />
Microalgae Toxin(s) Poisoning Event References<br />
Diatoms<br />
Pseudo-nitzschia spp.<br />
P. australis b<br />
P. cuspidata b<br />
P. delicatissima b<br />
P. fraudulenta b<br />
P. multiseries b<br />
P. pungens b<br />
P. pseudodelicatissima b<br />
P. seriata a<br />
Dinoflagellates<br />
Alexandrium spp.<br />
A. acatenella a<br />
A. catenella b<br />
A. fundyense a<br />
A. hiranoi a<br />
A. ostenfeldii a<br />
A. tamarense a<br />
Dinoflagellates<br />
Lingulodinium polyedrum b<br />
Gonyaulax spinifera a<br />
Protoceratium reticulatum a,c<br />
Dinoflagellates<br />
Dinophysis spp.<br />
D. acuminata a<br />
D. acuta a<br />
D. caudate<br />
D. fortii a<br />
D. norvegica a<br />
D. rotundata a<br />
D. tripos a<br />
Prorocentrum spp.<br />
P. micans<br />
P. minimum a,d<br />
Raphidophytes<br />
Chattonella marina a<br />
Fibrocapsa japonica a<br />
Heterosigma akashiwo a<br />
Domoic acid (DA) Amnesic Shellfish Poisoning (ASP)<br />
even when isolated from the same geographic region (Smith<br />
et al., 2001; Trainer et al., 2001; Kudela et al., 2004).<br />
Laboratory experiments have revealed a wide range of<br />
physico-chemical factors that increase or decrease toxin<br />
Human effects<br />
Gastro-intestinal symptoms<br />
Neurologic symptoms<br />
Death<br />
Ecosystem effects<br />
Marine mammal mortalities<br />
Bird mortalities<br />
Saxitoxins (STXs) Paralytic Shellfish Poisoning (PSP)<br />
Human effects<br />
Gastro-intestinal symptoms<br />
Paralysis<br />
Death<br />
Ecosystem effects<br />
Marine mammal mortalities<br />
Yessotoxins (YTXs) Human and ecosystem<br />
effects<br />
None reported<br />
Okadaic acid (OA)<br />
Dinophysistoxins (DTXs)<br />
Pectenotoxins (PTXs)<br />
water research 44 (2010) 385–416<br />
Diarrhetic Shellfish<br />
Poisoning (DSP)<br />
Human effects<br />
Gastro-intestinal symptoms<br />
Ecosystem effects<br />
None reported<br />
Brevetoxins (PbTxs) Neurotoxic Shellfish<br />
Poisonining (NSP)<br />
Human effects<br />
Gastroenteritis<br />
Neurologic symptoms<br />
Respiratory irritation<br />
and/or failure<br />
Ecosystem effects<br />
Marine mammal<br />
mortalities<br />
Fish mortality events<br />
a Reported to produce toxin.<br />
b Reported to produce toxin on the west coast of the United States.<br />
c Conflicting reports on toxicity of P. reticulatum cultures isolated from California, Washington and Florida.<br />
d Reported to be present on the west coast of Mexico.<br />
Subba Rao et al. (1988), Bates et al. (1989),<br />
Martin et al. (1990), Buck et al. (1992),<br />
Garrison et al. (1992), Rhodes et al. (1996),<br />
Horner et al. (1997), Lundholm et al. (1997),<br />
Rhodes et al. (1998), Trainer et al. (2000, 2001),<br />
Baugh et al. (2006)<br />
Sommer and Meyer (1937),<br />
Gaines and Taylor (1985),<br />
Steidinger (1993),<br />
Scholin et al. (1994),<br />
Taylor and Horner (1994),<br />
Jester (2008)<br />
Holmes et al. (1967),<br />
Satake et al. (1997, 1999),<br />
Draisci et al. (1999a),<br />
Paz et al. (2004, 2007),<br />
Armstrong and Kudela (2006),<br />
Rhodes et al. (2006), Howard et al. (2007)<br />
Holmes et al. (1967),<br />
Yasumoto et al. (1980),<br />
Murata et al. (1982),<br />
Yasumoto et al., (1985), Cembella (1989),<br />
Lee et al. (1989), Horner et al. (1997),<br />
Cembella (2003), Miles et al. (2004),<br />
Shipe et al. (2008), Sutherland (2008)<br />
Loeblich and Fine (1977),<br />
Hershberger et al. (1997),<br />
Gregorio and Connell (2000),<br />
Hard et al. (2000),<br />
Tyrell et al. (2002), O’Halloran et al. (2006)<br />
production by harmful species of algae, and which appear to be<br />
species-specific (see review of Granéli and Flynn, 2006). Reports<br />
of factors inducing toxin production have sometimes been<br />
conflicting, presumably indicating that multiple factors, or
perhaps generally stressful conditions, may stimulate toxin<br />
production. Factors affecting toxin production include: (1)<br />
temperature (Ono et al., 2000); (2) light intensity (Ono et al., 2000);<br />
(3) salinity (Haque and Onoue, 2002a,b); (4) trace metal availability,<br />
especially iron (Ladizinsky and Smith, 2000; Rue and<br />
Bruland, 2001; Maldonado et al., 2002; Wells et al., 2005; Sunda,<br />
2006) but also copper (Maldonado et al., 2002) andselenium<br />
(Mitrovic et al., 2004, 2005); (5) macronutrient availability<br />
including silicate (Pan et al., 1996b; Fehling et al., 2004; Kudela<br />
et al., 2004), phosphate (Pan et al., 1996a, 1998; Fehling et al.,<br />
2004), nitrogen (Bates et al., 1991; Pan et al., 1998; Kudela<br />
et al., 2004) and combinations of nutrient limitation (Anderson<br />
et al., 1990; Flynn et al., 1994; John and Flynn, 2000); (6) cellular<br />
elemental ratios of nutrients and physiological stress (Granéli<br />
and Flynn, 2006; Schnetzer et al., 2007); (7) growth phase<br />
(Anderson et al., 1990; Bates et al., 1991; Flynn et al., 1994;<br />
Johansson et al., 1996; Maldonado et al., 2002; Mitrovic et al.,<br />
2004). The precise combination(s) of environmental factors that<br />
select for population growth of particular algal species within<br />
diverse natural assemblages, and the specific conditions that<br />
induce toxin production, are poorly understood for most<br />
harmful algae. This present state of knowledge makes it difficult<br />
to predict the timing, duration or spatial extent of the vast<br />
majority of HAB events and the toxic events resulting from<br />
them.<br />
Our ability to thoroughly characterize HABs is also<br />
complicated by the complex array of toxins produced by algae.<br />
Marine algal species produce a suite of toxic components<br />
(Yasumoto and Murata, 1993), and unidentified toxins<br />
undoubtedly remain to be described. Additionally, most toxins<br />
are actually composed of families of closely related<br />
compounds. Slightly different forms of a toxin can exhibit very<br />
different levels of toxicity, or may be characterized differently<br />
by some detection methods and analytical approaches<br />
(Garthwaite et al., 2001; Lefebvre et al., 2008). Such complexity<br />
and variability can sometimes yield vague or contradictory<br />
conclusions regarding the exact source of toxicity in a natural<br />
sample (Bates et al., 1978; Paz et al., 2004, 2007).<br />
Finally, characterization of HAB events is complicated by<br />
inherent difficulties associated with linking specific toxins<br />
measured in natural water samples to a specific algal species<br />
in a complex, natural phytoplankton assemblage and, as<br />
noted above, the presence of toxic species in a water sample<br />
does not necessarily indicate the presence of toxins. Despite<br />
these shortcomings, there is considerable knowledge of many<br />
of the major algal toxins and their producers in U.S. coastal<br />
waters that constitute the most important potential concerns<br />
for desalination activities because they are the most likely to<br />
be encountered in ocean water intakes.<br />
2.1. Domoic acid<br />
2.1.1. Toxin description and activity<br />
Domoic acid (Fig. 2; Table 3) is an amino acid derivative<br />
belonging to the kainoid class of compounds containing three<br />
carboxyl groups and one secondary amino group (Wright<br />
et al., 1990; Jeffery et al., 2004). All four groups are charged at<br />
neutral pH, and the carboxyl groups become successively<br />
protonated as pH decreases, yielding five possible protonated<br />
forms of domoic acid (Quilliam, 2003; Jeffery et al., 2004). There<br />
water research 44 (2010) 385–416 389<br />
are currently ten known isomers of domoic acid, including the<br />
isodomoic acids A through H and the domoic acid 5’ diestereomer<br />
(Jeffery et al., 2004).<br />
Domoic acid and other members of the kainoid class are<br />
glutamate analogues that interfere with neurochemical<br />
pathways by binding to glutamate receptors of brain neurons<br />
(Wright et al., 1990; Quilliam, 2003). The resulting effect of<br />
these neuroexcitants, or excitotoxins, is a continuous stimulation<br />
of the neurons, which can lead to rupture and/or<br />
eventual formation of lesions (Wright et al., 1990). Depolarized<br />
neurons result in short-term memory loss (Clayden et al.,<br />
2005), which has led to the common name for the illness<br />
related to the consumption of seafood contaminated with<br />
domoic acid: amnesic shellfish poisoning (ASP). Symptoms of<br />
ASP include gastroenteritis (vomiting, diarrhea, abdominal<br />
cramps) that can be experienced in humans within 24 h after<br />
ingestion, and neurological symptoms of confusion, memory<br />
loss, disorientation, seizures, coma and/or cranial nerve<br />
palsies that are typically experienced within 48 h (Perl et al.,<br />
1990; Wright et al., 1990). The number of human illnesses<br />
resulting from domoic acid poisoning has been few (Horner<br />
et al., 1997), likely due to active monitoring of fisheries.<br />
However, cultured blue mussels (Mytilus edilus) contaminated<br />
with domoic acid poisoned 107 people and killed three during<br />
the first major documented ASP outbreak in 1987 on Prince<br />
Edward Island, Canada (Perl et al., 1990).<br />
ASP poses a serious threat to marine wildlife, and the<br />
deaths of thousands of marine mammals and sea birds have<br />
been attributed to domoic acid intoxication (Bates et al., 1989;<br />
Scholin et al., 2000; Gulland et al., 2002; Caron et al., unpublished<br />
data). The first documented poisoning episode of<br />
marine animals related to domoic acid on the U.S. west coast<br />
was attributed to Pseudo-nitzschia australis and occurred in<br />
September 1991 off central California (Table 2; Buck et al.,<br />
1992; Fritz et al., 1992). High concentrations of domoic acid<br />
were also detected in Washington and Oregon in the 1990s<br />
(Wekell et al., 1994; Adams et al., 2000; Trainer et al., 2002), and<br />
a decade later in coastal waters off southern California<br />
(Schnetzer et al., 2007). The frequency and severity of these<br />
toxic events appears to be increasing (Trainer et al., 2007).<br />
2.1.2. Producers<br />
The production of domoic acid and its isomers is confined to<br />
approximately a dozen chain-forming marine pennate diatom<br />
species within the genus Pseudo-nitzschia (Bates and Trainer,<br />
2006), a genus containing species that form long chains of<br />
cells attached at their ends (Fig. 3a and b). The main toxin<br />
producing species that have been documented on the U.S.<br />
west coast include: P. australis, P. delicatissima, P. fraudulenta, P.<br />
multiseries, P. pungens, P. pseudodelicatissima, P. seriata and<br />
P. cuspidata (Tables 1 and 2). These species are distinguished<br />
based on fine morphological features of their silica frustules<br />
(Fig. 3a and b). These distinctions are subtle and require<br />
careful electron microscopical analysis and elaborate taxonomic<br />
training. As a consequence, historical misidentifications<br />
are not unusual and debates regarding some species<br />
descriptions are still unresolved.<br />
It is surprising that the first reports of ASP on the west coast<br />
of the U.S. were not recorded until the 1990s, even though<br />
Pseudo-nitzschia species have been recorded in surveys of
390<br />
phytoplankton species in the Southern California Bight since<br />
1917 (Allen, 1922, 1924, 1928, 1936, 1940, 1941; Reid et al., 1970,<br />
1985; Lange et al., 1994; Fryxell et al., 1997; Thomas et al., 2001).<br />
Given that these species generally comprise a significant<br />
portion of the total diatom assemblage in these waters, it can<br />
be surmised that either toxin production has increased in these<br />
west coast species, or that poisoning events prior to the 1990s<br />
have occurred but have not been attributed to these diatoms.<br />
Historical accounts of ‘unusual animal mortality events’ along<br />
the U.S. west coast tend to support the latter hypothesis.<br />
There have been increasing numbers of toxic events<br />
recorded along the U.S. west coast (Table 2), notably in Puget<br />
Sound (Trainer et al., 2003, 2007), Monterey Bay (Vigilant and<br />
Silver, 2007; R. Kudela, unpubl. data), Santa Barbara Channel<br />
(Trainer et al., 2000; Anderson et al., 2006; Mengelt, 2006), San<br />
Pedro Channel (Busse et al., 2006; Schnetzer et al., 2007),<br />
Newport Beach (Busse et al., 2006) and San Diego (Lange et al.,<br />
1994; Busse et al., 2006). Most recently, toxic blooms of Pseudonitzschia<br />
in the Long Beach-Los Angeles Harbor and San Pedro<br />
Channel have been particularly toxic, with some of the highest<br />
domoic acid concentrations recorded for the U.S. west<br />
coast (Caron et al., unpublished data). The increased incidence<br />
and severity of these toxic episodes off the western U.S. coast<br />
parallels the increase in frequency and intensity of harmful<br />
water research 44 (2010) 385–416<br />
Fig. 2 – Chemical structures of commonly encountered toxins produced by microalgae in U.S. coastal waters.<br />
algal blooms observed globally (Smayda, 1990; Hallegraeff,<br />
1993, 2003; Anderson et al., 2002; Glibert et al., 2005b).<br />
2.2. Saxitoxins<br />
2.2.1. Toxin description and activity<br />
Saxitoxin is a complex guanidine-based alkaloid that exists as<br />
more than 30 identified analogues in nature (Llewellyn, 2006).<br />
It is the most powerful marine toxin currently known and<br />
among the most dangerous poisons on Earth, except for some<br />
venoms and bacterial toxins (Schantz et al., 1957). Due to its<br />
acute toxicity, saxitoxin is currently listed as a chemical<br />
weapon in Schedule 1 of the Chemical Weapons Convention<br />
(Llewellyn, 2006). Saxitoxins display a rigid tricyclic core<br />
(Fig. 2; Table 3) and are very stable in biological and physiological<br />
solutions (Rogers and Rapoport, 1980). This nitrogenrich<br />
molecule and its chemical relatives are polar and have<br />
a positive charge at pH 7.7 (Shimizu et al., 1981). Consequently,<br />
they are soluble in water and alcohols, and insoluble<br />
in organic solvents (Schantz et al., 1957).<br />
Saxitoxins are known to disrupt the flow of ions through<br />
voltage gate sodium channels (Catterall, 1992; Cestele and<br />
Catterall, 2000). It has also been recently discovered that they<br />
have the ability to bind to calcium (Su et al., 2004) and
Table 2 – Distribution and concentrations of marine toxins in plankton of confirmed toxin producers in U.S. west coast waters.<br />
Toxin(s) Location and year Causative species Particulate mgL 1<br />
(nmol L 1 Cellular<br />
) pg cell 1<br />
Dissolved pg mL 1<br />
(nmol L 1 )<br />
Domoic<br />
acid<br />
Washington coast and Juan de<br />
Fuca Eddy, WA (1997, 1998)<br />
P. pseudodelicatissima b.d.–2.7<br />
Pseudo-nitzschia spp.<br />
3.6–8.7<br />
References<br />
b.d.–4.6 Adams et al. (2000),<br />
Trainer<br />
et al. (2001, 2002)<br />
Penn Cove, WA (1997) P. pungens b.d.–0.8 Trainer et al. (1998)<br />
P. multiseries<br />
P. australis<br />
P. pseudodelicatissima<br />
Washington coast, WA (2001) P. australis b.d.–0.03 Marchetti et al. (2004)<br />
Washington coast, WA (2003) Pseudo-nitzschia spp. (0.4–15) 2 10 4 –0.3 a<br />
(b.d.–4.3) a<br />
0.1–94.4 (1–5)<br />
Baugh et al. (2006)<br />
Puget Sound, WA (2005) P. pseudodelicatissima b.d.–14 Trainer et al. (2007)<br />
Pseudo-nitzschia spp.<br />
Central Oregon coast, OR (1998) P. australis 0.5 35 Trainer et al. (2001)<br />
Pt. Año Nuevo,<br />
San Francisco, CA (1998)<br />
Bolinas Bay,<br />
San Francisco, CA (2003)<br />
P. pungens 0.1–0.7 0.3–6.3 Trainer et al. (2000)<br />
P. multiseries<br />
P. australis 0.15–9.4 Howard et al. (2007)<br />
Monterey Bay, CA (1991, 1998) P. australis b.d.–12.3 3–37 Buck et al. (1992),<br />
0.1–6.7 7.2–75<br />
Garrison<br />
et al. (1992), Walz<br />
et al. (1994), Scholin<br />
et al. (2000)<br />
Monterey Bay, CA (1998) P. pseudodelicatissima 0.1–0.4 0.8–1.2 Trainer et al. (2000,<br />
P. multiseries 0.67 6<br />
2001)<br />
Monterey Bay, CA (2000) Pseudo-nitzschia spp. b.d.–24 b.d.–8491 Bargu et al. (2002, 2008)<br />
P. australis<br />
Monterey Bay, CA (2002–2003) Pseudo-nitzschia spp. 24 Vigilant and Silver<br />
(2007)<br />
Morro Bay, CA (1998) P. australis 1.3–7.4 37–78 Trainer et al. (2000,<br />
2001)<br />
San Luis Obispo, CA (2003–2005) P. australis 1.5–7.6 9–38 Mengelt (2006)<br />
P. multiseries<br />
Point Conception, CA (1998) P. australis 2.2–6.3 15–22 Trainer et al. (2000)<br />
Santa Barbara, CA (1998) P. australis 0.5–1.2 0.1–0.9 Trainer et al. (2000)<br />
P. pungens<br />
P. pseudodelicatissima<br />
(continued on next page)<br />
water research 44 (2010) 385–416 391
Table 2 (continued)<br />
Toxin(s) Location and year Causative species Particulate mgL 1<br />
(nmol L 1 )<br />
Cellular<br />
pg cell 1<br />
Dissolved pg mL 1<br />
(nmol L 1 )<br />
References<br />
Santa Barbara Channel, CA (2003) P. australis 0.03–1.7 0.14–2.1 Anderson et al. (2006)<br />
Santa Barbara (Santa Rosa Island<br />
and north San Miguel) (2004)<br />
Southern California Bight,<br />
CA (2003, 2004)<br />
San Diego and Orange<br />
counties, CA (2004)<br />
P. australis 6–12 b.d.–80 Mengelt (2006)<br />
P. multiseries<br />
Pseudo-nitzschia spp. 5.6–12.7 b.d.–117 Schnetzer et al. (2007)<br />
P. australis<br />
P. cuspidata<br />
P. australis b.d.–2.33 Busse et al. (2006)<br />
P. multiseries<br />
Saxitoxins Sequim Bay, WA (2004–2007) Alexandrium spp. 0.02–0.5 150–800 Lefebvre et al. (2008)<br />
Oregon coast, OR (2004) Alexandrium spp. 0.004–0.028 Jester et al.<br />
(unpublished data)<br />
Humboldt Bay, CA (2004) A. catenella 1.6–19 a<br />
San Mateo County coast, CA (2004) A. catenella 2.1–62.6 a<br />
Monterey Bay, CA (2004) A. catenella 0.6–31.3 a<br />
Jester (2008)<br />
Jester (2008)<br />
Jester (2008)<br />
Monterey Bay, CA (2003–2005) A. catenella b.d.–0.962 Jester et al. (2009b)<br />
Morro Bay, CA (2004) A. catenella 1.4–16.6 a<br />
Yessotoxin La Jolla, CA (1993) Lingulodinium polyedrum 0.002–0.02 a<br />
Brevetoxins Indian Inlet, Bald Eagle Creek<br />
and Torque Canal, DE (2000)<br />
b.d.: Below detection limit.<br />
a Toxin concentration from cells in culture.<br />
0–0.05 a<br />
Jester (2008)<br />
Armstrong and Kudela<br />
(2006)<br />
Howard et al. (2008)<br />
Chattonella cf. verruculosa 0.008–
potassium channels (Wang et al., 2003) and to be a weak<br />
inhibitor of neuronal nitric oxide synthase (reviewed in Llewellyn,<br />
2006). These activities directly affect the nervous<br />
system, and the consumption of seafood containing saxitoxin<br />
can result in serious human illness and death, commonly<br />
referred to as paralytic shellfish poisoning (PSP).<br />
Minor symptoms of PSP, such as burning or tingling sensation<br />
of the lips and face, dizziness, headache, salivation, intense<br />
thirst and perspiration, vomiting, diarrhea and stomach<br />
cramps, can be experienced within 30 min after the consumption<br />
of contaminated seafood (Llewellyn, 2006). The consumption<br />
of a lethal dose can result in death within hours due to<br />
muscular paralysis and respiratory difficulty followed by<br />
complete respiratory arrest. PSP outbreaks result in more than<br />
2000 illnesses worldwide each year, with a 5–10% mortality rate<br />
(Hallegraeff, 2003). PSP toxins also have adverse effects on<br />
marine wildlife that can cause mortalities among fish, marine<br />
mammal and seabird populations (Geraci et al., 1989; Montoya<br />
et al., 1996; Shumway et al., 2003). There are presently no records<br />
of unusual animal mortality events along the Californian coast<br />
that are attributable to saxitoxin poisoning (Jester et al., 2009b),<br />
but occurrence of the toxins in species consumed by humans is<br />
sufficient to warrant year-round monitoring.<br />
2.2.2. Producers<br />
Saxitoxins are biosynthesized by dinoflagellates in marine<br />
ecosystems, most notably species within the genus Alexandrium<br />
(Fig. 3c), as well as Gymnodinium catenatum, Pyrodinium<br />
bahamense var. copressum, and by some cyanobacteria in<br />
freshwater ecosystems (Hallegraeff, 2003). Blooms of toxic and<br />
noxious dinoflagellates are often referred to as ‘red tides’<br />
because of the red discoloration of water created by the<br />
accessory pigments of the cells. However, toxic levels of saxitoxins<br />
can be attained at dinoflagellate abundances that do not<br />
significantly discolor the water because of the exceptionally<br />
high potency of saxitoxins (Burkholder et al., 2006). This situation<br />
exists for Alexandrium in that it does not typically reach<br />
‘bloom’ abundances on the U.S. west coast, and constant toxin<br />
monitoring is necessary to identify toxic conditions (Langlois,<br />
2007; IOC HAB Programme, 2008; Jester et al., 2009a).<br />
Alexandrium species and measurable saxitoxin concentrations<br />
are common along the U.S. west coast (Table 1), although<br />
concentrations reported for this toxin typically have not been<br />
as high as noted along the U.S. east coast (Table 2). Thus, few<br />
west coast studies have contributed to our understanding the<br />
dynamics of Alexandrium abundances and saxitoxin production<br />
while ongoing research in the Gulf of Maine represents<br />
the most comprehensive regional study of this dinoflagellate<br />
(Anderson et al., 2005; McGillicuddy et al., 2005). Combined<br />
field observations, laboratory studies and modeling efforts<br />
have led to a scenario for toxic events along the northeastern<br />
coast of the U.S. that involve an interplay between river<br />
runoff, resuspension of dinoflagellate cysts from coastal<br />
sediments, favorable offshore growth conditions, and winds<br />
that generate onshore flow into coastal shellfish areas. A<br />
monitoring study of PSP in Puget Sound (WA) from 1993 to<br />
2007 underscores that the timing and location of PSP<br />
outbreaks and high Alexandrium abundances are highly variable<br />
and not easily predicted from local or large-scale climate<br />
data (Moore et al., 2009). However, the study points out that<br />
water research 44 (2010) 385–416 393<br />
periods of warm air and low stream flow may favor saxitoxin<br />
accumulation in sentinel mussels (Moore et al., 2009).<br />
2.3. Brevetoxins<br />
2.3.1. Toxin description and activity<br />
Brevetoxins are polyether, non-polar compounds that depolarize<br />
cell membranes by opening voltage gate sodium ion<br />
channels and induce enhanced inward flux of ions into cells<br />
(Lin et al., 1981; Baden, 1983, 1989; Purkerson et al., 1999).<br />
Brevetoxins exist as two structural types and multiple analogs<br />
possessing various levels of potency (Baden, 1989; Cembella,<br />
2003; Kirkpatrick et al., 2004). The types differ in their ladderframe<br />
polycyclic ether structural backbones and are designated<br />
type A and type B (Fig. 2). The brevetoxin derivatives<br />
found in the marine environment (PbTx-2, PbTx-3 and PbTx-9;<br />
Table 3) are produced most commonly by dinoflagellate and<br />
raphidophyte algae and are of the structural type B (Baden,<br />
1989; Baden et al., 2005).<br />
Brevetoxins bind to site 5 of the voltage-sensitive sodium<br />
channel in neurons, causing these channels to remain open<br />
and fire repeatedly (Catterall, 1992; Cestele and Catterall,<br />
2000). Brevetoxin poisoning in humans is referred to as<br />
neurotoxic shellfish poisoning (NSP), and includes gastrointestinal<br />
symptoms of nausea, diarrhea and abdominal pain,<br />
neurologic symptoms of paresthesia, and respiratory irritation<br />
and/or failure (Kirkpatrick et al., 2004).<br />
The effects of brevetoxins on human health are well<br />
documented along the western coast of Florida where severe,<br />
nearly annual red tides caused by the dinoflagellate Karenia<br />
brevis release large amounts of brevetoxins into the air when<br />
the fragile cells are broken in breaking waves at the water’s<br />
edge (Kusek et al., 1999; Kirkpatrick et al., 2004). The aerosolized<br />
toxins constitute a significant health risk when they<br />
are inhaled and, as a result, K. brevis blooms are one of the<br />
most intensively studied and best-understood regional HABs.<br />
Red tides caused by K. brevis have been implicated in marine<br />
mammal fatalities, fish kills and human illnesses. Brevetoxins<br />
have not been reported from the U.S. west coast, and therefore<br />
no known human fatalities or health issues have yet been<br />
attributed to brevetoxins from that region.<br />
2.3.2. Producers<br />
Several dinoflagellate species and a few raphidophyte species<br />
produce a suite of brevetoxins (Baden, 1989). K. brevis, the<br />
most notorious brevetoxin producer within the Gulf of<br />
Mexico, has not been observed on the west coast of the U.S.,<br />
but several species of raphidophytes that are potential brevetoxin<br />
producers have been documented (Tables 1 and 2).<br />
Heterosigma akashiwo (Fig. 3e), Chattonella marina (Fig. 3f) and<br />
Fibrocapsa japonica have been isolated and cultured from<br />
coastal waters off southern California. In general there are few<br />
reports of significant blooms of these species on the west<br />
coast, although blooms of raphidophytes in San Francisco Bay<br />
and Delaware Inland Bays have been observed with cell<br />
abundances in excess of 10 8 cells L 1 (Herndon et al., 2003;<br />
Coyne et al., 2005). In part, this lack of information is<br />
a consequence of the fact that raphidophyte species are<br />
notoriously difficult to identify using traditional microscopical<br />
techniques because they preserve poorly (Hallegraeff and
394<br />
water research 44 (2010) 385–416
Hara, 1995; Throndsen, 1997). Recently developed genetic<br />
approaches for the identification and quantification of some<br />
raphidophytes are beginning to provide much-needed tools<br />
for studying the distributions and ecology of these HAB<br />
species (Handy et al., 2006; Demir et al., 2008). Despite the<br />
difficulties of characterizing these blooms, fish kills have been<br />
attributed to raphidophyte blooms on the west coast of the<br />
U.S. although these studies have not quantified brevetoxins<br />
(Hershberger et al., 1997; Hard et al., 2000).<br />
2.4. Diarrhetic shellfish toxins<br />
2.4.1. Toxin description and activity<br />
Toxins that cause diarrhetic shellfish poisoning (DSP) include<br />
okadaic acid, dinophysistoxins and pectenotoxins (Ramsdell<br />
et al., 2005). Okadaic acid is a monocarboxylic acid named for<br />
the marine sponge Halichondria okadai from which it was first<br />
isolated (Tachibana et al., 1981). Okadaic acid can also be<br />
found in natural water samples in polar and non-polar esteric<br />
forms (Prassopoulou et al., 2009). The first dinophysistoxin<br />
described was isolated from the mussel M. edilus and was<br />
found to be a methyl form of okadaic acid (Murata et al., 1982).<br />
Okadaic acid and dinophysistoxins are linear polyethers<br />
(Fig. 2) and the mode of action is the inhibition of protein<br />
phosphatases (Takai et al., 1987; Bialojan and Takai, 1988;<br />
Haystead et al., 1989), enzymes that play a key role in<br />
dephosphorylation in many biological processes including cell<br />
cycle regulation. The pectenotoxins are lipid soluble and differ<br />
structurally from other diarrhetic toxins in that they possess<br />
a lactone ring (Fig. 2), and not considered to be protein phosphatase<br />
inhibitors, but have a high actin-depolarizing action<br />
(Hori et al., 1999). There is speculation that pectenotoxins may<br />
not produce diarrhetic effects (Cembella, 2003).<br />
DSP toxins (Table 3) were named for the human symptoms<br />
resulting from the ingestion of contaminated shellfish,<br />
including inflammation of the intestinal tract, diarrhea,<br />
abdominal cramps, vomiting and nausea beginning 30 min to<br />
a few hours after ingestion (Hallegraeff, 2003). In addition to the<br />
symptoms listed above, okadaic acid is known to be a strong<br />
tumor promoter (Suganuma et al., 1988), although the potential<br />
health implications of this activity due to the ingestion of<br />
contaminated seafood is unknown. There are presently no<br />
documented cases of DSP resulting from okadaic acid, dinophysistoxins<br />
or pectenotoxins on the U.S. west coast, and thus<br />
these toxins are not regularly monitored in the marine environment.<br />
DSP toxins have been detected in mussels and water<br />
samples from California (Sutherland, 2008), so it is possible<br />
that DSP has occurred on the U.S. west coast but has been<br />
attributed to other sources of contamination.<br />
water research 44 (2010) 385–416 395<br />
2.4.2. Producers<br />
Okadaic acid and dinophysistoxins are produced by a few<br />
species of the dinoflagellate genus Prorocentrum (Cembella,<br />
2003) and most commonly by species of the genus Dinophysis.<br />
Dinophysis species present on the western U.S coast include<br />
D. acuminata, D. acuta, D. caudata, D. fortii, D. norvegica, D.<br />
rotundata, and D. tripos (Table 2). D. acuminata and D. fortii have<br />
been documented in Californian waters for many years<br />
(Bigelow and Leslie, 1930). D. acuminata produces okadaic acid<br />
(Yasumoto et al., 1985), D. fortii produces okadaic acid, dinophysistoxins<br />
and pectenotoxins (Yasumoto et al., 1980; Murata<br />
et al., 1982) while D. rotundata and D. tripos produce<br />
dinophysistoxin-1 (Lee et al., 1989).<br />
Dinophysis species are technically not phytoplankton, but<br />
heterotrophic protists that retain chloroplasts acquired from<br />
their prey. Dinophysis acquires its chloroplasts by preying on<br />
ciliates, which in turn prey on cryptophyte algae. This<br />
complex trophic relationship has made the culture of these<br />
species unsuccessful until recently (Park et al., 2006), and<br />
therefore no information on the environmental conditions<br />
that influence toxin production by these species presently<br />
exists. However, genus members are easily distinguished by<br />
light microscopy because they possess a pronounced ‘keel’<br />
and other unique morphological aspects (Fig. 3d). These<br />
species are often encountered in plankton samples. Abundances<br />
of 10 3 cells L 1 are commonly encountered (Nishitani<br />
et al., 2005), and occasionally they may reach abundances in<br />
excess of 10 5 cells L 1 (Carpenter et al., 1995).<br />
2.5. Yessotoxin<br />
2.5.1. Toxin Description<br />
Yessotoxin (Fig. 2; Table 3) is a chiral molecule of high polarity<br />
due to the presence of two sulfate groups. The molecule<br />
consists of fused polyether rings organized into a ladder<br />
shaped skeleton (Yasumoto and Murata, 1993; Wright and<br />
Cembella, 1998), a structure similar to other ladder-like polyether<br />
toxins such as the ciguatera toxin complex (ciguatoxins<br />
and maitotoxin), gambieric acids and brevetoxins (Yasumoto<br />
and Murata, 1993; Wright and Cembella, 1998). There are<br />
nearly 100 analogs of yessotoxin that have been identified to<br />
date (Satake et al., 1997, 1999; Ciminiello et al., 1998, 2000,<br />
2001; Daiguji et al., 1998; Miles et al., 2004, 2005a,b, 2006; Paz<br />
et al., 2006).<br />
The yessotoxin class was named for the species of<br />
scallop, Patinopecten yessoensis, in which it was initially<br />
detected (Murata et al., 1987). Yessotoxin was originally<br />
classified in the DSP-toxin class because it was detected with<br />
other DSP toxins, but it appears that it does not induce<br />
Fig. 3 – Major algal toxin producers occurring along the U.S. west coast. (a) Pseudo-nitzschia australis, a producer of domoic<br />
acid; (b) Scanning electron micrograph of Pseudo-nitzschia australis; (c) Alexandrium catenella, a producer of saxitoxin; (d)<br />
Dinophysis sp., a producer of okadaic acid; (e) Heterosigma akashiwo, a producer of brevetoxins; (f) Chattonella marina,<br />
a producer of brevetoxins; (g) Cochlodinium sp.; (h) Lingulodinium polyedrum, a producer of yessotoxin; (i) Phaeocystis globosa<br />
colony; (j) foam produced by the prymnesiophyte, Phaeocystis accumulating along the shore; (k) Unconcentrated seawater<br />
from King Harbor, City of Redondo Beach, with significant discoloration due to an algal bloom; (l) Prorocentrum sp., the<br />
dominant organism in (k); and (m) higher magnification of Prorocentrum sp. from (l). Scale bars [ 10 mm. Photo (b) courtesy of<br />
Peter Miller, (c) courtesy of Carmelo Tomas, (j) courtesy of Cindi Heslin.
Table 3 – Summary of toxins that can be present in Southern California waters. MW: molecular weight.<br />
Toxin Properties Formula MW Mode of action References<br />
Domoic acid (DA) Hydrosoluble<br />
At pH 7: DA 3<br />
Saxitoxins (STXs) Hydrosoluble<br />
pH 7: Stable<br />
C 15H 21NO 6 311.14 Binds to glutamate receptors in the brain<br />
disrupting normal neurochemical<br />
transmission<br />
C 10H 17N 7O 4 299.3 Bind to site 1 of voltage-sensitive sodium<br />
channels and block sodium conductance;<br />
bind to calcium and potassium channels<br />
Brevetoxins (PbTxs) Liposoluble Bind to site 5 of voltage-sensitive sodium<br />
Brevetoxin 2 (PbTx 2) C 50H 70O 14 895.1<br />
Brevetoxin 3 (PbTx 3) C50H72O15 897.1<br />
Brevetoxin 9 (PbTx 9) C 50H 74O 14 899.1<br />
Diarrhetic shellfish toxins<br />
channels, shifting activation to more<br />
negative membrane potentials and block<br />
channel activation<br />
Okadaic acid (OA)<br />
Dinophysistoxins (DTXs)<br />
C44H68O13 805 Inhibits protein phosphatases, inhibits<br />
dephosphorylation of proteins<br />
Pectenotoxins (PTXs) Liposoluble High actin-depolarizaing action<br />
Yessotoxins (YTXs) Hydrosoluble C55H80O21S2Na2 1187.3 Activation of phosphodiesterase in<br />
the presence of external Ca 2þ ;<br />
Disruption of the E-cadherin–catenin<br />
system in epithelial cells and potentially<br />
disrupting its tumour suppressive functions<br />
Wright et al. (1990), Quilliam (2003)<br />
Wong et al. (1971), Wang et al. (2003), Su et al. (2004),<br />
Catterall (1992), Cestele and Catterall (2000)<br />
Lin et al. (1981), Baden (1983, 1989),<br />
Purkerson et al. (1999)<br />
Tachibana et al. (1981), Murata et al. (1982),<br />
Yasumoto et al. (1985), Takai et al. (1987),<br />
Bialojan and Takai (1988), Haystead et al. (1989),<br />
Hori et al. (1999)<br />
Murata et al. (1987), Takahashi et al. (1996),<br />
Alfonso et al. (2003), Ronzitti et al. (2004)<br />
396<br />
water research 44 (2010) 385–416
diarrhetic effects (Ogino et al., 1997). Accordingly, the regulation<br />
of the European Commission on marine biotoxins now<br />
considers yessotoxins separately from DSP toxins (European<br />
Commission, 2002). The great number of yessotoxin analogs<br />
complicates toxicity studies, and may explain the sometimes<br />
contradictory reports of its mode of action. Studies have<br />
shown that lysosomes, the immune system and the thymus<br />
(with tumorigenic implications) are the biological targets of<br />
yessotoxin (Franchini et al., 2004; Malagoli et al., 2006), while<br />
other reports have indicated cardiotoxic effects (Terao et al.,<br />
1990; Ogino et al., 1997; Aune et al., 2002). The cardiotoxicity<br />
of yessotoxin might be attributed to phosphodiesterase<br />
activation in the presence of external calcium ions (Alfonso<br />
et al., 2003). Unlike the other marine toxins mentioned<br />
above, there have been no reported human health issues or<br />
marine mammal deaths associated with yessotoxins.<br />
2.5.2. Producers<br />
There are three known yessotoxin-producing dinoflagellates,<br />
Protoceratium reticulatum, Lingulodinium polyedrum, and<br />
Gonyaulax spinifera, and they have all been observed in coastal<br />
waters off the western U.S. (Table 1). According to phylogenetic<br />
analyses of available rRNA gene sequences, the capacity<br />
for yessotoxin production appears to be restricted to the order<br />
Gonyaulacales (Howard et al., 2009). However, toxin production<br />
among strains within each species appears to be highly<br />
variable.<br />
Yessotoxin has been detected in L. polyedrum isolates<br />
cultured from around the globe (Tubaro et al., 1998; Draisci<br />
et al., 1999a; Strom et al., 2003; Paz et al., 2004), including<br />
isolates from Californian coastal waters (Armstrong and<br />
Kudela, 2006). The reported cellular concentrations in the<br />
latter cells ranged from below detection to 1.5 pg cell 1 , indicating<br />
that L. polyedrum is significantly less toxic than<br />
P. reticulatum or G. spinifera. Yessotoxin has been recorded in<br />
blue mussels at low concentrations along the U.S. west coast<br />
(Table 2) during red tides caused by L. polyedrum, as well as<br />
during non-bloom conditions, but yessotoxin production by<br />
this dinoflagellate appears to be less than toxin levels<br />
produced by isolates from other geographical regions (see<br />
Table 2 in Howard et al., 2008). Expansive and dense blooms of<br />
L. polyedrum (Fig. 3 h) have been reported in California since<br />
1901, but there have only been anecdotal reports of health<br />
problems associated with the red tides caused by L. polyedrum<br />
(Kudela and Cochlan, 2000).<br />
Yessotoxin production by isolates of P. reticulatum has<br />
been confirmed (Satake et al., 1997; Boni et al., 2002; Miles<br />
et al., 2002; Riobo et al., 2002; Stobo et al., 2003; Samdal et al.,<br />
2004; Eiki et al., 2005), but concentrations ranging from<br />
below detection to 79 pg cell 1 have been reported for<br />
isolates from Washington, California and Florida (Paz et al.,<br />
2004, 2007). Isolates of G. spinifera appear to be the most<br />
prolific yessotoxin producers. Concentrations in New Zealand<br />
isolates ranged from below detection to 200 pg cell 1<br />
(Rhodes et al., 2006), more than 20-fold higher than the<br />
per-cell toxicity of P. reticulatum and 600-fold higher than<br />
L. polyedrum. G. spinifera does not generally bloom in high<br />
densities on the U.S. west coast, but it has been frequently<br />
observed at low abundances (Howard et al., 2008; M. Silver,<br />
pers. comm.) and has reached bloom concentrations in<br />
water research 44 (2010) 385–416 397<br />
Tomales Bay, north of San Francisco (G. Langlois pers.<br />
comm.). Yessotoxins are monitored in New Zealand, Europe<br />
and Japan but they are not routinely measured on the U.S.<br />
west coast. Howard et al. (2008) was the first study to confirm<br />
yessotoxins in California and Washington coastal waters,<br />
albeit at very low concentrations.<br />
2.6. Toxin detection and quantification<br />
A wide variety of methodologies and technologies exist to<br />
detect, characterize and quantify the major toxins produced<br />
by microalgae. These approaches can be broadly divided into<br />
those used to characterize biological activity (toxicity assays)<br />
and those used to identify specific chemical structure(s)<br />
(immunological, various analytical techniques). Because of<br />
the highly variable approaches employed, and in most<br />
cases the highly diverse set of compounds comprising a toxin<br />
class, the methods provide somewhat different estimates of<br />
absolute toxin concentrations or presumed toxicity. It is also<br />
noteworthy that the majority of the protocols used to measure<br />
algal toxins have focused on the analysis of tissue samples<br />
that may be a source of human contamination (e.g. shellfish or<br />
finfish tissue) or plankton material filtered from seawater<br />
samples. Relatively few studies have examined the concentrations<br />
of toxins dissolved in seawater (Table 2). For this<br />
reason, limits of detection for toxins in seawater are poorly<br />
known for most approaches. However, based on analytical<br />
approaches presently available, a practical limit of detection<br />
for a few of the major concerns (domoic acid and saxitoxins) is<br />
in the range of 0.1 mg per liter of seawater for immunological<br />
approaches, but a detection limit of approximately 0.01 mg per<br />
liter for dissolved domoic acid in seawater using highperformance<br />
liquid chromatography has been reported<br />
(Pocklington et al., 1990). Detection limits for toxins in<br />
particulate material in the water can be significantly lower<br />
because particles can be concentrated by filtration prior to<br />
extraction. Knowledge of the concentrations of dissolved<br />
toxins would be preferable from the perspective of reverse<br />
osmosis desalination operations because dissolved toxins<br />
(rather than cell-bound toxins) would most likely impact<br />
these membranes.<br />
Domoic acid has been detected and quantified in seawater,<br />
plankton, shellfish extract and homogenate, as well as sea bird<br />
and mammalian body fluid (e.g. blood, urine, amniotic<br />
fluid, cerebral spinal fluid). Analytical approaches for these<br />
measurements include commercially available immunological<br />
techniques (enzyme-linked immunosorbent assay; ELISA)<br />
(Garthwaite et al., 1998, 2001), high pressure liquid chromatography<br />
(HPLC) with ultraviolet (UV) diode array detection<br />
(DAD) (Quilliam et al., 1989; Quilliam, 2003), receptor binding<br />
assay (RBA) (Van Dolah et al., 1994), mouse bioassay (MBA) or<br />
liquid chromatography–mass spectrometry (LC–MS). The<br />
results obtained by these various approaches are not yet<br />
completely compatible or comparable, and therefore comparisons<br />
across studies using different analytical methods can<br />
be problematic. In general, the choice of an approach is<br />
a compromise between cost of analysis (or access to costly<br />
equipment), sample throughput, sensitivity and analytical goal<br />
(e.g. thorough chemical characterization versus overall
398<br />
toxicity, or human health risk versus scientific understanding<br />
of bloom dynamics).<br />
Saxitoxins can be rapidly detected and quantified with<br />
commercial ELISA kits, but the high specificity of these tests<br />
precludes the recognition of certain members of the saxitoxin<br />
family, especially the neo-saxitoxin (Garthwaite et al., 2001).<br />
Saxitoxin detection and quantification is often accomplished<br />
by HPLC, RBA, LC–MS, and MBA. Regulatory programs for<br />
seafood consumption are still based on ‘lethal mouse dosage’.<br />
Similarly, detection and quantification of brevetoxin and its<br />
derivatives in seawater, shellfish, and mammalian body fluids<br />
can be accomplished using commercially available ELISA kits<br />
(Naar et al., 2002), by HPLC, RBA (Van Dolah et al., 1994), LC–MS<br />
and MBA. A comparative study also quantified brevetoxins by<br />
radioimmunoassay (RIA) and a neuroblastoma (N2A) cytotoxicity<br />
assay (Twiner et al., 2007).<br />
Quantification of the DSP toxin suite can be underestimated<br />
by ELISA because commercial ELISA assays are<br />
usually optimized to detect okadaic acid and not the dinophysistoxins<br />
(Garthwaite et al., 2001). HPLC with fluorimetric<br />
detection (HPLC-FLD) has been commonly used to detect and<br />
quantify okadaic acid, its polar and non-polar esters, as well<br />
as dinophysistoxins (Lee et al., 1987). The MBA method has<br />
also been routinely used for the detection of DSP toxins.<br />
The detection and quantification of yessotoxin is problematic<br />
because of the extensive suite of derivatives that may<br />
exist. HPLC-FLD analysis (Yasumoto and Takizawa, 1997),<br />
MBA, LC-MS (Draisci et al., 1999b; Paz et al., 2006, 2008) and<br />
ELISA (Samdal et al., 2004, 2005) have been employed.<br />
2.7. Other potentially toxic, noxious<br />
and nuisance organisms<br />
Reports of newly occurring HAB species, or recognition of<br />
extant issues that have gone previously undocumented, are<br />
water research 44 (2010) 385–416<br />
increasing our awareness of other potentially harmful bloomforming<br />
algae along the U.S. west coast. For example, blooms<br />
of an emerging potentially toxic organism, Cochlodinium sp.<br />
(Fig. 3g) off central California have recently been reported<br />
(Curtiss et al., 2008; Iwataki et al., 2008; Kudela et al., 2008b).<br />
This species is difficult to identify using light microscopy, and<br />
therefore researchers have begun to use gene sequences to<br />
provide accurate identification (Iwataki et al., 2007, 2008;<br />
Matsuoka et al., 2008). While this organism has only recently<br />
reached sufficient abundances to discolor Californian waters,<br />
there has already been one reported abalone loss in central<br />
California that appears to be associated with a bloom of<br />
Cochlodinium (R. Kudela pers. comm).<br />
Species of the dinoflagellate Prorocentrum (Fig. 3k–m) occasionally<br />
bloom in Californian coastal waters where they can<br />
attain very high abundances periodically, causing discolorations<br />
of the water and nuisance accumulations of algae<br />
(Holmes et al., 1967; Shipe et al., 2008). The Prorocentrum<br />
species known to occur in Californian waters have not<br />
yet demonstrated DSP toxin production, but species from<br />
elsewhere in the world are known to produce theses toxins<br />
(Table 1). Similarly, massive blooms of the dinoflagellate<br />
Akashiwo sanguinea are common in coastal waters of southern<br />
and central California and have recently been the cause of<br />
seabird mortality due to surfactant-like proteins (Jessup et al.,<br />
2009). Although they may not be overtly toxic, these blooms<br />
can cause animal mortalities, deplete oxygen, and result in an<br />
increased organic and biomass loading to a seawater desalination<br />
facility.<br />
The prymnesiophyte Phaeocystis globosa infrequently<br />
attains high abundances off the Californian coast (Armonies,<br />
1989). This species produces single cells that are
Phaeocystis are consumed by many zooplankton species but<br />
the colonies are typically a poor food source. Selective feeding<br />
on single cells appears to favor colony formation and the<br />
accumulation of colonies in the water column (Netjstgaard<br />
et al., 2007). When released via colony destruction or algal life<br />
cycle events, the colony matrix material is easily worked into<br />
a ‘sea foam’ that can form layers many centimeters (even<br />
meters) thick at the ocean surface or along the coastline over<br />
fairly extensive regions (Fig. 3j).<br />
3. Spatial and temporal patterns<br />
of harmful algae<br />
A fundamental aspect of the biology of harmful algal blooms,<br />
and of vital importance for desalination operations, is the<br />
tendency for rapid and dramatic changes in the spatial and<br />
temporal distributions of these species. These changes occur<br />
rapidly across a wide range of scales, and pose challenges for<br />
documenting and predicting these distributions. Numerous<br />
approaches and instruments have been developed to characterize<br />
the dynamics of phytoplankton communities. These<br />
approaches presently have significant limitations on their<br />
abilities to identify species composition of a bloom, but they<br />
provide crucial information on the emergence and longevity<br />
of bloom events as well as their vertical and geographical<br />
extent. This information can help seawater desalination<br />
facilities adjust operations to ensure reliable production for<br />
the duration of the bloom.<br />
3.1. Temporal variability<br />
Significant temporal variations in the abundances of phytoplankton<br />
take place on time scales ranging from hours to<br />
decades. Short-term temporal variability (hours to a few weeks)<br />
can be a consequence of rapid population growth or consumption<br />
by herbivores, sinking of senescent populations, diel<br />
vertical migration, tidal movement, and aggregation or<br />
dispersal by physical processes such as water mass convergences<br />
or divergences.<br />
Diel vertical migration of several dinoflagellates has been<br />
attributed to geotaxis, phototaxis and nutrient status (Eppley<br />
et al., 1968; Blasco, 1978; Cullen and Horrigan, 1981; Levandowsky<br />
and Kaneta, 1987). Classic responses involve nighttime<br />
sinking out of surface waters to deeper water where<br />
nutrient concentrations are greater, and rising into surface<br />
waters for photosynthesis during daytime. Shifts in nutrient<br />
cell quotas that accompany these migrations may have<br />
significant implications for toxin production because cell<br />
toxicity can be related to nutrient status of the cells (Anderson<br />
et al., 1990; Flynn et al., 1994; John and Flynn, 2000; Flynn,<br />
2002).<br />
Diel-to-weekly variations in phytoplankton abundance can<br />
be characterized using self-contained or wirelessly networked<br />
sensor packages. Data collected in King Harbor of the City of<br />
Redondo Beach, CA (Fig. 4) demonstrate the efficacy of these<br />
instruments for providing high temporal resolution of chlorophyll<br />
a fluorescence (which approximates phytoplankton<br />
biomass) and pertinent environmental factors (e.g. dissolved<br />
oxygen and temperature) that provide insight into the factors<br />
water research 44 (2010) 385–416 399<br />
controlling the pattern. These data reveal a 4-fold variation in<br />
phytoplankton standing stock over a two-day period. In<br />
addition, changes in water quality criteria were easily and<br />
rapidly identified (e.g. decrease in dissolved oxygen concentration<br />
near noontime on September 13). Daily variations in<br />
the latter parameter can be extreme at night during algal<br />
blooms. High resolution, short-term monitoring approaches<br />
allow rapid detection of sentinel parameters, and in turn<br />
provide information for the development of predictive models<br />
of HAB events.<br />
Chlorophyll a fluorescence sensors provide valuable<br />
information on the short-term temporal patterns of total algal<br />
biomass, but these instruments cannot identify noxious algal<br />
species within a phytoplankton assemblage. More sophisticated<br />
instruments now coming online offer that possibility.<br />
For example, the Environmental Sample Processor (http://<br />
www.mbari.org/ESP) isanin situ instrument capable of performing<br />
real-time identifications of HAB and other microbial<br />
species, as well as toxin analyses such as domoic acid using<br />
on-board molecular analyses (Greenfield et al., 2006). Similarly,<br />
handheld devices now exist for the detection of some<br />
HAB species and toxins in the field (Casper et al., 2007). These<br />
rapid and highly specific analyses are becoming valuable tools<br />
for quick determinations of toxin presence resulting from<br />
algal blooms. These more costly instruments can be used to<br />
improve the information available on bloom composition<br />
once the cheaper and more readily available sensors identify<br />
an emerging bloom event.<br />
Seasonal variability of HAB species and their toxins along<br />
the Californian coast can be gleaned from the records of the<br />
Marine Biotoxin Monitoring Program (MBMP) of the California<br />
Department of Public Health (CDPH), a program started after<br />
a major domoic acid outbreak in the fall of 1991. At present,<br />
the annual effort involves the analysis of approximately 300<br />
shellfish samples for domoic acid and >1000 samples for PSP<br />
toxins from all fifteen Californian coastal counties (CDPH,<br />
2007). Shellfish toxin information provides a reasonable<br />
representation of toxins in the upper water column over<br />
seasonal and annual scales such as demonstrated in studies<br />
on toxic dinoflagellates (Montojo et al., 2006; Jester et al.,<br />
2009a; Moore et al., 2009). Detailed information on the<br />
sampling effort is provided in the MBMP annual reports<br />
(Langlois, 2007).<br />
Distributions of domoic acid and PSP toxins along the<br />
Californian coast during the period 2002–2007 based on the<br />
MBMP data reveal both seasonal and geographical trends (Figs.<br />
5 and 6). Monthly averages for the 6-year period (histograms)<br />
and maximal concentrations (triangles and lines) showed<br />
detectable concentrations of PSP toxins and domoic acid in<br />
shellfish in nearly all months for all Californian coastal<br />
counties (Fig. 5). Domoic acid concentrations showed<br />
pronounced seasonality, with very high peaks during spring<br />
and much smaller peaks during fall. This temporal trend is<br />
concordant with previous observations along the southern<br />
Californian coast (Lange et al., 1994; Walz et al., 1994; Schnetzer<br />
et al., 2007; Shipe et al., 2008). Fall domoic acid peaks correspond<br />
to minor blooms of toxic Pseudo-nitzschia occasionally<br />
noted on the west coast of the U.S. (Bolin and Abbott, 1960;<br />
Buck et al., 1992; Walz et al., 1994; Trainer et al., 2002;<br />
Schnetzer et al., 2007).
400<br />
Domoic acid<br />
(µg/100g)<br />
PSP toxins<br />
(µg/100g)<br />
70000<br />
60000<br />
50000<br />
40000<br />
30000<br />
20000<br />
15000<br />
10000<br />
5000<br />
Seasonal peaks in domoic acid along the U.S. west coast<br />
differ along a latitudinal gradient. Highest domoic acid<br />
concentrations in Washington have been observed during the<br />
fall (Trainer, 2002; Office of Shellfish and Water Protection, 2008;<br />
Moore et al., 2009) compared to spring blooms that are common<br />
in southern California (Schnetzer et al., 2007). This north-south<br />
trend in seasonality is also evident on a smaller latitudinal scale.<br />
Blooms along the Californian coast have been more frequently<br />
observed later in the year in northern counties (Humboldt<br />
versus Santa Barbara counties in Fig. 6). The time lag between<br />
bloom periods observed from south to north may be related to<br />
the timing of the California Current System (CCS) upwelling<br />
maximum, which brings nutrients into surface waters and<br />
promotes phytoplankton growth. The CCS upwelling occurs in<br />
early spring in southern California, in June off Washington,<br />
0<br />
1600<br />
1400<br />
1200<br />
1000<br />
800<br />
600<br />
400<br />
300<br />
200<br />
100<br />
0<br />
water research 44 (2010) 385–416<br />
Mean of monthly averages<br />
Absolute maximum<br />
Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec<br />
Fig. 5 – Seasonal variability of domoic acid (upper panel) and PSP toxin (lower panel) concentrations along the Californian<br />
coast. Monthly averages of data collected from 2002 and 2007 summarized for fifteen Californian coastal counties are shown<br />
as histograms. Also shown are the maximal values recorded during each month over the entire study period (triangles and<br />
solid lines). Toxin concentrations were derived from shellfish tissue. Data from CDPH (Langlois, 2007).<br />
and throughout summer in northern California and Oregon<br />
(Reid et al., 1956; Landry, 1989).<br />
Monthly averages as well as maximal PSP toxin concentrations<br />
showed less pronounced seasonal variability than<br />
domoic acid during the period 2002–2007, although the highest<br />
concentrations were recorded from July to September (Fig. 5).<br />
This seasonal pattern of maximal PSP is in agreement with the<br />
last 25 years of monitoring results (Langlois, 2007). It is also<br />
consistent with the 1927–1989 observations on the Californian<br />
coast indicating that most significant concentrations of the<br />
toxin take place between May and October (Price et al., 1991).<br />
The highest PSP toxin concentrations in shellfish have also<br />
been observed in the summer and the fall periods off the<br />
Washington and Oregon coasts (Trainer et al., 2002; Determan,<br />
2003).
North<br />
South<br />
Domoic acid<br />
(µg/100 g)<br />
Domoic acid<br />
(µg/100 g)<br />
14000<br />
12000<br />
10000<br />
8000<br />
6000<br />
4000<br />
2000<br />
0<br />
14000<br />
12000<br />
10000<br />
8000<br />
6000<br />
4000<br />
2000<br />
0<br />
Domoic acid<br />
Santa Barbara County<br />
Diel to seasonal temporal patterns in ASP and PSP<br />
outbreaks off the U.S. west coast are augmented by large<br />
interannual variability in the intensity and frequency of<br />
HABs. Interannual variations in HABs presumably are<br />
related, at least in part, to changes in atmospheric and<br />
hydrographic features modulated by the 3–7 year cycles of El<br />
Niño-Southern Oscillations (ENSOs) (Price et al., 1991; Horner<br />
et al., 1997), but the exact relationship between HABs and<br />
these climatic events is not clear because there are still<br />
relatively few observations spanning these temporal<br />
regimes. Warming off California during El Niño episodes<br />
reduces seasonal upwelling, enhances physical stratification<br />
in the CCS and lowers the nutricline in the water column<br />
(the depth at which nutrient concentrations increase<br />
rapidly). It is clear that these changes in water stability and<br />
nutrient availability have significant impacts on plankton<br />
productivity and community structure, but the specific<br />
responses of the phytoplankton communities vis-à-vis HAB<br />
events are not yet predictable (Barber and Chavez, 1983).<br />
ASP outbreaks occurred during El Niño episodes of 1991 and<br />
1997–98 along the coasts of Oregon, Washington and California<br />
(Table 2; Moore et al., 2008), but the specific factors contributing<br />
to toxic Pseudo-nitzschia blooms during these events<br />
could not be clearly identified (Horner and Postel, 1993; Trainer<br />
et al., 2000). Interannual variability in ASP and PSP concentrations<br />
in coastal waters along the Californian coast was<br />
evident and substantial in the MBMP dataset during the period<br />
2002–2007 (Fig. 6). ASP and PSP outbreaks were frequent, but<br />
water research 44 (2010) 385–416 401<br />
Humboldt County<br />
**** * ** ** ** **<br />
April May June<br />
** * * **<br />
April May June<br />
PSP toxins<br />
(µg/100 g)<br />
PSP toxins<br />
(µg/100 g)<br />
140<br />
120<br />
100<br />
80<br />
60<br />
40<br />
20<br />
0<br />
140<br />
120<br />
100<br />
80<br />
60<br />
40<br />
20<br />
0<br />
PSP toxins<br />
Marin County<br />
July August September<br />
San Luis Obispo County<br />
2002<br />
2003<br />
2004<br />
2005<br />
2006<br />
2007<br />
* not detected<br />
** * * * * * * ** *<br />
July August September<br />
Fig. 6 – Interannual variation in toxin concentration in four Californian coastal counties from 2002 to 2007 during three<br />
months exhibiting high concentrations of domoic acid (April, May, June) and saxitoxin (July, August, September). Humboldt<br />
and Marin counties are located north of Santa Barbara and San Luis Obispo counties. Toxin concentrations were measured<br />
from shellfish tissues. Data from CDPH (Langlois, 2007).<br />
they varied in intensity and the timing of peak concentrations<br />
between the years within a single geographical location, and<br />
between southern and northern Californian counties in the<br />
same year and season. Notably, the magnitude of this variability<br />
is on the same order of magnitude as the variability<br />
observed on short-term (daily) or seasonal time scales.<br />
Little is known regarding the longer time scale fluctuations<br />
in HABs along the U.S. west coast. Multi-decadal fluctuations<br />
in ocean temperature are known to provoke shifts in the<br />
biological regime (Chavez et al., 2003), and it is anticipated<br />
that climatic shifts would affect the timing, intensity or<br />
frequency of phytoplankton blooms. Long-term regime shifts<br />
may affect the occurrence of blooms of L. polyedrum,<br />
a producer of yessotoxin, along the coast of southern<br />
California (Tables 1 and 2). The Pacific Ocean was cooler in the<br />
years preceding 1976, and red tides dominated by L. polyedrum<br />
commonly developed along the Southern California Bight<br />
during fall (Gregorio and Pieper, 2000). During a recent warm<br />
regime (1976-mid 1990s), red tides occurred during winter and<br />
spring and persisted until summer in the region of the Los<br />
Angeles River mouth (Gregorio and Pieper, 2000). Recent<br />
massive blooms of L. polyedrum during fall may indicate<br />
a return to the pre-1976 conditions (Moorthi et al., 2006).<br />
The generality surmised from data depicting short- to longterm<br />
temporal variability in phytotoxin dynamics is that<br />
variability can be high at all scales. Given our present state of<br />
understanding regarding the specific combination of forcing<br />
factors that give rise to this high variability, it is difficult to
402<br />
Fig. 7 – Two-dimensional presentations of chemical and physical data collected by an autonomous vehicle (Webb Slocum<br />
glider, Teledyne Webb Research, East Falmouth, MA) along a nearshore–offshore transect at Newport Beach, CA shown in (f)<br />
(indicated on map by the blue line). Contour plots are shown for (a) temperature, (b) chlorophyll a fluorescence, (c) salinity,<br />
(d) backscattered light and (e) water density. The Webb Slocum glider is an autonomous vehicle commonly employed in<br />
coastal ecosystems. This buoyancy-driven underwater vehicle generates horizontal motion by ascending and descending<br />
with pitched wings (Schofield et al., 2007). A rudder directs heading while buoyancy is controlled by pumping seawater into<br />
and out of the nose of the vehicle. This long-lived, low-power glider achieves horizontal velocities of approximately 25–<br />
30 cm s L1 with vertical velocities of 10–15 cm s L1 .<br />
accurately predict the timing and magnitude of toxic blooms.<br />
For these reasons, monitoring at multiple temporal scales is<br />
necessary to adequately characterize plankton dynamics.<br />
3.2. Spatial variability<br />
Spatial variability of HABs is considerable at multiple scales,<br />
analogous (and strongly related) to the temporal variability<br />
described above. Blooms can be highly localized (10s of<br />
meters) or expansive (100s of kilometers), and distributions<br />
vertically within the water column are heterogeneous over<br />
scales of centimeters to meters. The geographical extent and<br />
heterogeneous nature of the U.S. west coast results from<br />
differences in local hydrography that are manifested in smalland<br />
large-scale differences in spatial patterns of toxic blooms.<br />
Regional-scale variations in HAB distributions are illustrated<br />
by MBMP data during 2002–2007 for ASP and PSP<br />
concentrations observed in counties along northern and<br />
central California (Fig. 6). ASP events (frequency and level of<br />
water research 44 (2010) 385–416<br />
toxicity) were generally lower in northerly Humboldt County<br />
during this period relative to Santa Barbara County nearly<br />
1000 km to the south. More recently, high domoic acid<br />
concentrations have been observed within the Southern California<br />
Bight, presumably indicating a continuing southward<br />
movement of the Pseudo-nitzschia spring blooms (Langlois,<br />
2007; Schnetzer et al., 2007). On the other hand, Marin County<br />
(north of San Francisco) exhibited higher monthly PSP values<br />
during most months than San Luis Obispo County located<br />
nearly 500 km to the south. This general latitudinal trend in<br />
PSP events is consistent with findings that the three southernmost<br />
counties (Los Angeles, Orange and San Diego)<br />
generally experience low concentrations of PSP relative to<br />
northern California (Price et al., 1991; Langlois, 2007).<br />
Regional-scale and geographical differences in ASP and PSP<br />
events have also been reported along the coastline of Oregon<br />
(Trainer et al., 2002).<br />
Small-scale spatial variability (horizontally and vertically)<br />
can also be dramatic. HAB events can be highly restricted
geographically (e.g. relegated to a protected embayment).<br />
Even within spatially extensive blooms, phytoplankton<br />
biomass is often highly discontinuous over very small spatial<br />
scales because of differences in local circulation, and wind or<br />
wave forcing. Considerable spatial variability in the abundance<br />
of P. australis within Monterey Bay has been observed,<br />
a phenomenon that has been attributed to advective forces<br />
(Buck et al., 1992). Discontinuities within a water column, such<br />
as thermoclines, haloclines, nutriclines and light absorption<br />
often leads to the establishment of subsurface microlayers<br />
where phytoplankton biomass can be many-fold elevated<br />
relative to algal standing stocks only centimeters above or<br />
below (Dekshenieks et al., 2001; Rines et al., 2002).<br />
Characterization of phytoplankton spatial distributions<br />
include approaches ranging from shore-based sampling, to<br />
the use of oceanographic ships, to remote sensing of largescale<br />
patterns using satellite imagery. Vertical profiling of<br />
phytoplankton assemblages can be accomplished using<br />
over-the-side, ship-based instrument packages and more<br />
recently autonomous vehicles equipped with a variety of<br />
sensor packages. The use of autonomous vehicles to provide<br />
synoptic measurements of phytoplankton biomass from<br />
chlorophyll a fluorescence and pertinent chemical/physical<br />
parameters is state-of-the-art for obtaining two-dimensional<br />
cross-sections or three-dimensional patterns in the water<br />
column (Fig. 7).<br />
Autonomous vehicles provide time- and depth-stamped<br />
measurements of a variety of parameters that can be optimized<br />
for a specific mission. Such an instrument deployed off Newport<br />
Beach, CA (blue line in Fig. 7f) during May–June 2008 yielded<br />
detailed patterns of chemical/biological parameters that<br />
provided information on the extent and vertical distribution of<br />
phytoplankton biomass (Fig. 7a–e). Evidence of upwelling in this<br />
118°10’W<br />
Particulate domoic acid<br />
(µg L-1)<br />
118°00’W<br />
0.01-0.50<br />
0.51-1.00<br />
1.01-1.50<br />
1.51-2.00<br />
2.01-2.50<br />
2.51-3.00<br />
3.01-3.50<br />
3.51-4.00<br />
4.01-4.50<br />
4.51-5.00<br />
33°45’N<br />
33°40’N<br />
33°35’N<br />
Fig. 8 – Spatial variability in domoic acid concentrations<br />
contained in the total particulate material (algal cells, other<br />
microbes and detritus) in surface waters within the San<br />
Pedro Bay area including the Long Beach-Los Angeles<br />
harbor area. Data were collected at 20 sampling stations<br />
(locations indicated by filled circles).<br />
water research 44 (2010) 385–416 403<br />
spring deployment was apparent from the upward-pointing<br />
isotherms (temperature) and isopycnals (density) on the<br />
shoreward end of the transect (left sides of Fig. 7a, e). This was<br />
particularly evident in the temperature plot 5–10 km from shore<br />
between the surface and 20 m (Fig. 7a, red circle). Nutrients (not<br />
shown) were generally depleted in surface waters, and<br />
increased with decreasing temperature. This physico-chemical<br />
structure is concordant with a significant subsurface maximum<br />
in chlorophyll a concentration, indicative of a response of the<br />
phytoplankton assemblage to elevated nutrient concentrations<br />
at this depth (red circle on left in Fig. 7b), as previously observed<br />
(Jones et al., 2002). The cross-sectional picture provided by the<br />
autonomous vehicle indicated that the phytoplankton<br />
community had a patchy structure on the scale of meters<br />
(vertically) and kilometers (horizontally). Two major horizontal<br />
patches were observed at 6–12 km and at 15–21 km from shore<br />
(red circles in Fig. 7b).<br />
Measurements in addition to chlorophyll a fluorescence can<br />
add information on the observed general patterns such as<br />
particle size distributions derived from the optical backscatter<br />
spectrum. The backscatter profile obtained at a wavelength of<br />
532 nm indicated a patch of particles that were not of algal<br />
nature (Fig. 7d, red circle). The wavelength-dependent slope of<br />
the backscatter, which is dependent on the particle size distribution,<br />
can also be mapped to indicate the size-class of phytoplankton<br />
particles that dominate the chlorophyll a maxima.<br />
This information is particularly important for a seawater<br />
desalination facility, where the incoming particle size distribution<br />
is known to impact the source water filterability.<br />
Autonomous vehicles allow nearly synoptic measurements<br />
of the spatial distribution of phytoplankton and ancillary<br />
parameters. Many of these instruments operate for<br />
significant periods of time (weeks) and thereby supply<br />
temporal as well as spatial coverage. Knowledge of the<br />
temporal evolution and spatial organization of coastal marine<br />
systems enables a better understanding of the linkages<br />
between physical processes and the biological responses that<br />
contribute to the formation of algal blooms. Moreover, data<br />
from these instruments can be telemetered to the laboratory<br />
in near-real time and used to direct costly efforts such as<br />
shipboard sampling, or plan operations for land-based activities<br />
and measurements.<br />
Broad-scale, horizontal distributions can also be acquired<br />
via shipboard sampling programs (Fig. 8). Shipboard work is<br />
time and labor intensive, but onboard sample processing<br />
enables more sophisticated analyses than autonomous vehicles<br />
are presently capable of providing. Moreover, ships and<br />
other manned platforms permit time series studies at a single<br />
study site. Shipboard sampling conducted in April 2008 in the<br />
Long Beach-Los Angeles harbor area and the adjacent San<br />
Pedro Channel demonstrated considerable spatial variability<br />
in the distribution of HAB species and their toxins within this<br />
relatively small area (approximately 500 km 2 ). Results indicated<br />
a patchy distribution of domoic acid in particulate<br />
material (i.e. within phytoplankton cells) collected near the<br />
surface with highest concentrations in the vicinity of the<br />
harbor breakwater and at several offshore locations (Fig. 8).<br />
Intermediate regions exhibited toxin concentrations that<br />
were more than an order of magnitude less than these<br />
maxima.
404<br />
3.3. Environmental driving factors<br />
Characterizing the factors that lead to the stimulation of<br />
harmful algae and the production of toxins by these algae has<br />
been an area of very active research for decades. These studies<br />
involve field observations to document the spatiotemporal<br />
extent of blooms and toxin concentrations in plankton and<br />
marine life, and laboratory experiments aimed at understanding<br />
the key environmental factors leading to HAB events<br />
and toxin production. The overall results gleaned from many<br />
years of work group into three basic categories: (1) factors and<br />
conditions leading to phytoplankton blooms in general, (2)<br />
factors leading specifically to the growth of HAB species, and<br />
(3) factors leading to toxin production.<br />
Numerous factors have been implicated as contributors to<br />
the observed global expansion of HABs (Smayda, 1990; Hallegraeff,<br />
1993, 2003; Anderson et al., 2002; Glibert et al., 2005b).<br />
Phytoplankton blooms occur naturally as a consequence of<br />
the vertical mixing of deep, nutrient-rich waters into lighted<br />
surface waters. This process occurs seasonally in temperate<br />
environments due to winter storm events, and due to coastal<br />
upwelling events caused by appropriate regional wind conditions.<br />
There is no a priori reason why these ‘natural’ sources<br />
of nutrients cannot lead to HAB events, but the global increase<br />
in frequency and severity of HABs implies that human activities<br />
may be an underlying reason for this escalation.<br />
Eutrophication of coastal ecosystems is a growing global<br />
concern that has clear consequences for blooms of nearshore<br />
algal populations (Anderson et al., 2008; Heisler et al., 2008;<br />
Howarth, 2008). Nutrient enrichment has been implicated in<br />
harmful blooms occurring in some protected bays, but the<br />
linkage between nutrient discharges mediated by human<br />
activities and many HAB events is still unconfirmed. For<br />
example, field studies have shown that coastal upwelling of<br />
nitrate-rich waters can be a driving factor leading to toxigenic<br />
Pseudo-nitzschia blooms along the U.S. west coast (Horner<br />
et al., 2000; Scholin et al., 2000; Anderson et al., 2006) but the<br />
specific role of river/coastal runoff in domoic acid production<br />
is unclear (Scholin et al., 2000; Schnetzer et al., 2007). The<br />
importance of nutrient discharge into coastal waters is, of<br />
course, dependent on the amount of nutrients available for<br />
phytoplankton growth from natural sources but the latter<br />
term is poorly defined in most situations. Constructing<br />
nutrient budgets for coastal ecosystems is an area ripe for<br />
future work. In the meantime, it has been speculated that<br />
anthropogenic nutrient sources, such as elevated nutrient<br />
concentrations in river discharge, coastal runoff from agricultural<br />
land, and sewage discharge may significantly<br />
increase the total amounts of nutrients available for the<br />
growth of coastal phytoplankton (Scholin et al., 2000; Glibert<br />
et al., 2005a,b, 2006; Howard et al., 2007; Kudela et al., 2008a).<br />
There now exists a basic understanding of the general<br />
conditions that favor the growth of phytoplankton per se<br />
(Allen et al., 2008). Despite this basic understanding, there is<br />
still only limited information on the specific conditions that<br />
selectively stimulate the growth of harmful algal bloomforming<br />
species of phytoplankton. As a result, mathematical<br />
models that attempt to predict HAB events tend to be more<br />
correlative than deterministic (i.e., they identify the<br />
water research 44 (2010) 385–416<br />
conditions that may promote a HAB, rather than the conditions<br />
that will promote a bloom). One generality is that rarely<br />
can one identify a ‘silver bullet’, a single parameter or set of<br />
circumstances that provide an accurate prediction of the<br />
occurrence of a particular HAB species. The environmental<br />
circumstances leading to the dominance of a HAB population<br />
over all other species of algae in a given locale are composed of<br />
a complex set of physical, chemical and biological conditions<br />
with poorly known variances, and these conditions appear to<br />
be species-specific.<br />
Biological factors contributing to the success or demise of<br />
individual HAB taxa include allelopathy among competing<br />
phytoplankton, mixotrophy by HAB species, and the deterrence<br />
of potential consumers via the production of noxious or<br />
toxic compounds (Strom et al., 2003; Burkholder et al., 2008;<br />
Buskey, 2008; Flynn, 2008; Smayda, 2008). These biological<br />
interactions presuppose a period of stable environmental<br />
conditions in order that the scenarios of allelopathy, grazer<br />
deterrence or phagotrophic activity of HAB species can play<br />
themselves out. This requirement may explain why the<br />
formation of a stable water mass appears to play a role in the<br />
development of some HAB events (Scholin et al., 2000).<br />
Models predicting the population growth of potentially<br />
toxic algae are necessary for understanding bloom dynamics,<br />
but these models must also integrate information on toxin<br />
induction. Many toxins do not appear to be constitutively<br />
produced by algae, but are induced by a variety of specific<br />
environmental conditions that are not completely understood.<br />
Silica, phosphorus, nitrogen and trace metal limitations,<br />
and nutrient or elemental ratios (in addition to or<br />
instead of absolute concentrations) have all been implicated<br />
in toxin induction (Pan et al., 1996a, 1998; Rue and Bruland,<br />
2001; Fehling et al., 2004; Wells et al., 2005; Granéli and Flynn,<br />
2006; Schnetzer et al., 2007). Again, species-specific differences<br />
(and perhaps strain-specific differences) may exist in<br />
the factors promoting toxin production. Physical aspects such<br />
as temperature and light intensity may stimulate toxin<br />
production by some harmful algae (Ono et al., 2000).<br />
Relative fluorescence<br />
7<br />
45<br />
6<br />
Relative fluorescence<br />
TMP<br />
40<br />
35<br />
5<br />
30<br />
4<br />
25<br />
3<br />
20<br />
2<br />
15<br />
10<br />
1<br />
5<br />
0<br />
0<br />
0 5 10 15 20 25 30 35 40 45<br />
Days<br />
Transmembrane pressure<br />
(TMP, psi)<br />
Fig. 9 – Correlation between chlorophyll a fluorescence<br />
(open squares, in relative fluorescence units) and<br />
transmembrane pressure (filled circles) in a pilot-scale<br />
reverse osmosis desalination system. Increased loading of<br />
phytoplankton biomass resulted in greater<br />
transmembrane pressure.
Pseudo-nitzschia spp.<br />
(cells L-1 )<br />
Particulate domoic acid<br />
(µg L-1 )<br />
Dissolved domoic acid<br />
(µg L-1 )<br />
2.0x10 5<br />
1.5x10 5<br />
1.0x10 5<br />
5.0x10 4<br />
4. Desalination operations and HAB events<br />
A thorough understanding of the specific factors and conditions<br />
giving rise to harmful algal blooms and toxin production<br />
in coastal waters will require a great deal of additional<br />
research before accurate models for predicting these toxic<br />
events will be readily available. Until then, appropriate<br />
monitoring strategies to detect imminent bloom events and<br />
the ability to track the evolution of an active bloom, coupled<br />
with an understanding of the potential toxins being produced,<br />
0<br />
4.0<br />
3.0<br />
2.0<br />
1.0<br />
0.5<br />
0.0<br />
3.0<br />
2.5<br />
2.0<br />
1.5<br />
1.0<br />
0.5<br />
0.0<br />
01-Mar-05<br />
01-May-05<br />
water research 44 (2010) 385–416 405<br />
01-Jul-05<br />
01-Sep-05<br />
01-Nov-05<br />
Fig. 10 – Time series of abundances of Pseudo-nitzschia spp. cells (top), domoic acid concentrations in particulate material<br />
(middle) and dissolved in seawater (bottom) at a coastal monitoring site in El Segundo, CA.<br />
01-Jan-06<br />
01-Mar-06<br />
01-May-06<br />
01-Jul-06<br />
01-Sep-06<br />
01-Nov-06<br />
the toxin chemistry, and their rejection by seawater reverse<br />
osmosis membranes, provide a seawater desalination facility<br />
with the best strategy for making operational adjustments to<br />
ensure that the treatment plant capacity or product water<br />
quality remains unaffected.<br />
4.1. Concern with harmful algal blooms and their<br />
toxin production<br />
As seawater desalination has continued to become more costeffective<br />
and less energy intensive (Al-Sahlawi, 1999), many
406<br />
communities are planning or implementing seawater desalination<br />
facilities (Al-Sahlawi, 1999; Burbano et al., 2007). The<br />
selected pretreatment procedures and the process engineering<br />
that determines the ultimate facility design is entirely<br />
dependent upon the source water quality (e.g. suspended<br />
solids, turbidity, organic material content, algal cell content,<br />
etc.) and its variations, particularly for facilities incorporating<br />
open intakes (Bonnelye et al., 2004b; Gaid and Treal, 2007).<br />
When the seawater desalination process is performed by<br />
reverse osmosis membranes, the selection and proper operation<br />
of a pretreatment system is paramount to the success of<br />
the downstream desalination process (Tenzer et al., 1999;<br />
Bonnelye et al., 2004b; Separation Processes Inc., 2005; Gaid<br />
and Treal, 2007; Petry et al., 2007).<br />
Algal blooms are known to have significant negative<br />
impacts on reverse osmosis desalination facilities. A variety of<br />
pretreatment trains have been considered to address the<br />
difficult source water quality associated with algal blooms,<br />
where the organic and biomass load increase dramatically<br />
(Adin and Klein-Banay, 1986; Al Arrayedhy, 1987; Hasan Al-<br />
Sheikh, 1997; Watson, 1997; Abdul Azis et al., 2000; Bonnelye<br />
et al., 2004a,b; Burbano et al., 2007; Gaid and Treal, 2007; Petry<br />
et al., 2007; Peleka and Matis, 2008). Recently, microfiltration<br />
and ultrafiltration membrane pretreatment has been identified<br />
as a component of a preferred pretreatment train due to the<br />
consistent, high quality water produced by membrane filtration,<br />
especially when compared to conventional processes<br />
(Wilf and Schierach, 2001). However, one significant drawback<br />
to implementation of these modern pretreatment technologies<br />
is that they are as susceptible, or possibly more so, to significant<br />
algal blooms (Bonnelye et al., 2004a).<br />
An early warning system can provide information to<br />
a seawater desalination facility so that functional changes can<br />
be made to efficiently maintain operations even as source<br />
water quality deteriorates. Turbidity sensors offer a rapid<br />
measurement of the total amount of suspended particles in<br />
the intake water for making these decisions. On in situ<br />
instruments such as the Slocum glider, backscatter<br />
measurements yield this type of information (Fig. 7d).<br />
Measurements of chlorophyll a fluorescence can augment this<br />
surveillance approach by estimating the degree to which algal<br />
bimoass contributes to the total load of suspended particles.<br />
Responses to deteriorating water quality may include chemical<br />
additions, use of additional pretreatment equipment, or<br />
additional staff preparations (e.g. maintenance activities,<br />
guaranteeing all membranes and filters are clean in preparation<br />
for an event) to continuously deliver a high quality<br />
feedwater to the reverse osmosis system that will produce the<br />
desalinated drinking water.<br />
4.2. Addressing spatiotemporal variability in HAB<br />
abundance and early detection<br />
As detailed above, it is essential that a desalination facility<br />
incorporate a means of rapid algal bloom detection so that,<br />
when necessary, proper process changes can be made to<br />
maintain the production capacity. Sensors for detecting an<br />
eminent algal bloom can be located at the desalination facility<br />
to inform personnel regarding changes in water quality that<br />
are directly observed on the source water. Fig. 9 presents the<br />
water research 44 (2010) 385–416<br />
transmembrane pressure (TMP) of a microfiltration system<br />
that serves as pretreatment to a pilot-scale reverse osmosis<br />
desalination system along with the levels of chlorophyll<br />
a fluorescence observed in the feedwater. It is clear from this<br />
figure that increased membrane fouling rates (e.g. faster daily<br />
rise in the TMP) were associated with increasing chlorophyll<br />
a fluorescence (i.e. increased algal biomass) in the source<br />
water.<br />
It is well known that higher concentrations of algae cause<br />
increased membrane fouling rates in microfiltration systems<br />
that are frequently incorporated, or considered, in today’s<br />
desalination facilities (Gijsbertsen-Abrahamse et al., 2006; Lee<br />
and Walker, 2006; Reiss et al., 2006). A more complete approach<br />
might include a monitoring system located offshore that<br />
measures some of the primary factors influencing algal<br />
blooms, such as nutrient monitoring in near-real time using<br />
new in situ sensor technology (Glibert et al., 2008). Such information<br />
would be useful to both the desalination facility and<br />
HAB researchers who are continually improving their understanding<br />
of the causative factors that produce HABs and their<br />
associated toxins. Using the information provided by offshore<br />
sensors, the desalination facility personnel could note trends<br />
and shifts in driving factors that generate algal blooms and<br />
make any chemical orders or perform maintenance procedures<br />
that have significant lead times. The same offshore sensor<br />
might also incorporate real-time monitors of sentinel parameters<br />
for changes in algal biomass, such as turbidity and chlorophyll<br />
a, allowing the facility to prepare for changes in<br />
chemical additions and redundant equipment service.<br />
Monitoring of basic chemical parameters of seawater (e.g.<br />
chlorophyll a concentrations) will provide valuable information<br />
for facility operations, but this activity is not sufficient to<br />
fully assess potentially toxic conditions that might arise from<br />
algae that do not require high standing stocks to constitute<br />
a significant toxic threat, such as Alexandrium species.<br />
Species-specific approaches, such as automated in situ<br />
instruments or laboratory-based methods, as well as chemical/immunological<br />
analyses that identify and quantify<br />
specific algal toxins are necessary to more thoroughly characterize<br />
the potential hazards posed by HAB species. The<br />
consistent removal of these potentially toxic substances<br />
through the reverse osmosis process is both a function of<br />
size (e.g. molecular weight) and charge (e.g. zeta potential)<br />
(Amy et al., 2005). Depending on the size and charge of<br />
the contaminant of concern, the rejection, or removal, by the<br />
reverse osmosis process will differ. It is important that we<br />
continue to broaden our knowledge on potentially toxic<br />
substances excreted by algal stock and their associated<br />
blooms.<br />
The approach for obtaining this information would be best<br />
complemented with knowledge of the species that are present<br />
regionally, the potential problems they pose (e.g. specific toxins<br />
and the amounts of soluble microbial products and extracellular<br />
polymeric substances excreted), the spatial extent of HAB<br />
episodes, and their seasonality. The seasonality of Pseudo-nitzschia<br />
spp. and domoic acid near the intake of a pilot desalination<br />
plant in El Segundo, CA, exemplifies the usefulness of routine<br />
monitoring for identifying potentially toxic conditions in coastal<br />
waters adjacent to a plant (Fig. 10). Abundances of Pseudo-nitzschia<br />
spp. and concentrations of domoic acid contained in the
algal cells or dissolved in the intake water exhibited a springtime<br />
peak. Knowledge of the seasonality of this toxic bloom-forming<br />
species allows intensive sampling of coastal waters during<br />
spring when toxic events are more common, improving the<br />
overall effectiveness of the monitoring effort and making it<br />
more cost effective.<br />
Historical and real-time information on the spatial distribution<br />
of HABs can provide information vital for optimizing<br />
design and performance of desalination operations. Local/<br />
regional hydrography, and resultant algal blooms can differ<br />
dramatically. When constructing a new intake pipeline, the<br />
selection of its location (e.g. depth and distance from shore) can<br />
be greatly enhanced through the use of offshore monitoring<br />
devices and efforts to take into account the presence of any<br />
local accumulations of algal biomass due to currents, water<br />
mass convergences/divergences or internal waves, and also<br />
subsurface maxima in algal abundance. Properly locating<br />
offshore monitoring can provide significant information that<br />
will allow optimal location of a new intake pipeline or identification<br />
of issues that might affect an existing one, thereby<br />
significantly reducing the organic and suspended solid loads<br />
present in the feedwater during algal bloom events. These<br />
considerations will ease pretreatment operations, reduce the<br />
cost of water production, and help improve the facility’s<br />
longevity.<br />
5. Conclusions<br />
5.1. Potential impacts, unresolved issues<br />
and research prospectus<br />
The presence of harmful algae in coastal waters that might be<br />
employed in reverse osmosis desalination pose potential<br />
problems for these operations that have been known to even<br />
cause desalination facilities to temporarily cease production<br />
(Tenzer et al., 1999; Pankratz, 2008). As the number of seawater<br />
desalination facilities continues to grow with lower costs and<br />
increasing demand, it is essential that these operating facilities<br />
develop the tools necessary to allow process changes and<br />
ensure capacity objectives continue to be met. Regardless of<br />
the pretreatment configuration, changes in source water<br />
quality require adjustments and these changes need to carefully<br />
coordinate to ensure that the reverse osmosis membranes<br />
are not irreversibly fouled or damaged in the process.<br />
Benchmark work is required to establish the effectiveness<br />
of the seawater reverse osmosis process in dealing with HAB<br />
toxins and other phytoplankton-derived substances. Even if<br />
advanced pretreatment technologies such as microfiltration<br />
are implemented upstream of the reverse osmosis process,<br />
passage of transparent extracellular material produced by the<br />
algal bloom (Alldredge et al., 1993; Hong et al., 1997) may affect<br />
reverse osmosis membrane performance. Additionally, the<br />
physical durability of phytoplankton varies greatly and the<br />
pretreatment process might disrupt cells and create significantly<br />
higher concentrations of dissolved organic substances,<br />
including toxins, than were originally present in the source<br />
water. For example, dissolved domoic acid has been observed<br />
in the seawater passing through the prefiltration process<br />
(Fig. 10, bottom panel) but it is unclear if these values are<br />
water research 44 (2010) 385–416 407<br />
higher due to cell breakage as a result of the prefiltration<br />
process. Therefore, it is important that the international<br />
desalination community carefully characterize these potential<br />
contaminants and their removal to improve treatment<br />
approaches in seawater desalination.<br />
To our knowledge, there are no published reports on the<br />
effectiveness of reverse osmosis for removing dissolved algal<br />
toxins from seawater. Some of these toxin molecules (e.g.<br />
domoic acid) are near the theoretical molecular size of molecules<br />
rejected by reverse osmosis membranes, but experimental<br />
studies are required to validate the effective of this<br />
process on toxin removal. In addition, more information will<br />
be needed to understand the potential impact of discharged<br />
brine and pretreatment backwash water resulting from the<br />
reverse osmosis desalination process on the ecology of coastal<br />
ecosystems. The use of ferric sulfate or ferric chloride as<br />
a pretreatment coagulant would concentrate toxic algae and<br />
their associated toxins if they are present in the intake water.<br />
Similarly, the discharge of brine resulting from the reverse<br />
osmosis process would contain elevated concentrations of<br />
dissolved algal toxins relative to unfiltered seawater. The<br />
degree of concentration of these toxins would not be expected<br />
to be large, but the significance of these processes will depend<br />
on the starting concentrations in the raw intake or prefiltered<br />
water and the degree of concentration due to treatment. There<br />
is presently no information on algal toxins in these<br />
discharges.<br />
HABs on the U.S. west coast exhibit significant generalities<br />
across geographical and temporal scales (e.g. many of the<br />
same species occur throughout the region), but the details of<br />
bloom dynamics differ with geographic location, depth and<br />
season (and perhaps on interannual and decadal scales). The<br />
high degree of variability associated with these events makes<br />
constant monitoring of HABs in intake water for desalination<br />
a vital issue. Regional HAB programs and regulatory agencies<br />
along the U.S. west coast presently provide useful information<br />
for some known potential problems (e.g. ASP and PSP toxins)<br />
for end users that need information on coastal water quality.<br />
Awareness (and augmentation) of this information could<br />
improve planning and safe operation of desalination facilities.<br />
Monitoring of newly emerging HAB concerns (e.g. Cochlodinium<br />
spp.), or HABs and toxins that are presently poorly characterized<br />
(e.g. NSP, DSP, and yessotoxin poisoning) should also<br />
be implemented in the future to allow evaluation of their<br />
potential impacts on desalination processes.<br />
New technologies for toxin detection and quantification,<br />
and in situ monitoring of biological and chemical parameters<br />
are rapidly improving our ability to monitor coastal ecosystems<br />
and identify potentially problematic situations involving<br />
HABs. Advances in in situ observing technologies (sensor<br />
networks, autonomous sensor-equipped vehicles) provide the<br />
capability for obtaining unprecedented resolution in the<br />
spatial and temporal distributions of chemical and physical<br />
parameters, and some biologically important features<br />
(Sukhatme et al., 2007). New approaches and instruments for<br />
toxin detection help identify contaminated seafood products,<br />
and constitute sentinels for the threats of HABs to marine<br />
animal populations. Future uses of coastal waters for desalination<br />
will also benefit from, and contribute to, these<br />
activities.
408<br />
Acknowlegements<br />
The authors are grateful to Mr. Mark Donovan at Separation<br />
Processes, Inc. for providing the data for Fig. 9. The preparation<br />
of this manuscript was supported in part by funding from<br />
a contract between the West Basin Municipal Water District,<br />
Department of Water Resources and the University of Southern<br />
California, National Oceanic and Atmospheric Administration<br />
grants NA05NOS4781228 and NA07OAR4170008, Sea Grant<br />
NA07OAR4170008, National Science Foundation grants CCR-<br />
0120778 (Center for Embedded Networked Sensing; CENS),<br />
DDDAS-0540420, MCB-0703159, and a NASA Earth and Space<br />
Science Fellowship Grant NNX06AF88H. M.-È. Garneau was<br />
supported by a fellowship from the Fonds québécois de<br />
recherche sur la nature et les technologies (FQRNT).<br />
references<br />
Abdul Azis, P.K., Al-Tisan, I., Al-Daili, M., Green, T.N., Dalvi, A.G.I.,<br />
Javeed, M.A., 2000. Effects of environment on source water for<br />
desalination plants on the eastern coast of Saudi Arabia.<br />
Desalination 132 (1–3), 29–40.<br />
Adams, N.A., Lesoing, M., Trainer, L., 2000. Environmental<br />
influences on domoic acid accumulation in razor clams on the<br />
Washington coast. Journal of Shellfish Research 19 (2),<br />
1007–1015.<br />
Adin, A., Klein-Banay, C., 1986. Pretreatment of seawater by<br />
flocculation and settling for particulates removal.<br />
Desalination 58 (3), 227–242.<br />
Ahn, S., Kulis, D.M., Erdner, D.L., Anderson, D.M., Walt, D.R., 2006.<br />
Fiber-optic microarray for simultaneous detection of multiple<br />
harmful algal bloom species. Applied and Environmental<br />
Microbiology 72 (9), 5742–5749.<br />
Al Arrayedhy, M., 1987. Pre- and post-treatment at the RO plant at<br />
RA’s Abu Jarjur, Bahrain. Desalination 63 (C), 81–94.<br />
Al-Sahlawi, M.A., 1999. Seawater desalination in Saudi Arabia:<br />
economic review and demand projections. Desalination 123<br />
(2–3), 143–147.<br />
Alfonso, A., de la Rosa, L., Vieytes, M.R., Yasumoto, T., Botana, L.M.,<br />
2003. Yessotoxin, a novel phycotoxin, activates<br />
phosphodiesterase activity: effect of yessotoxin on cAMP<br />
levels in human lymphocytes. Biochemical Pharmacology 65<br />
(2), 193–208.<br />
Alldredge, A.L., Passow, U., Logan, B.E., 1993. The abundance and<br />
significance of a class of large, transparent organic particles in<br />
the ocean. Deep-Sea Research Part I Oceanographic Research<br />
Papers 40 (6), 1131–1140.<br />
Allen, W.E., 1922. Observations on surface distribution of marine<br />
diatoms between San Diego and Seattle. Ecology 3 (2), 140–145.<br />
Allen, W.E., 1924. Surface catches of marine diatoms and<br />
dinoflagellates made by U.S.S. Pioneer between San Diego and<br />
Seattle in 1923. In: University of California Publications in<br />
Zoology, vol. 26, pp. 243–248.<br />
Allen, W.E., 1928. Review of five years of studies on phytoplankton at<br />
Southern California piers, 1920–1924, inclusive. Bulletin of the<br />
Scripps Institution of Oceanography Technical Services 1,<br />
357–401.<br />
Allen, W.E., 1936. Occurrence of marine plankton diatoms in a tenyear<br />
series of daily catches in Southern California. American<br />
Journal of Botany 23 (2), 60–63.<br />
Allen, W.E., 1940. Summary of results of twenty years of<br />
researches on marine phytoplankton. Proceedings of the 6th<br />
Pacific Scientific Congress 3, 576–583.<br />
water research 44 (2010) 385–416<br />
Allen, W.E., 1941. Twenty years’ statistical studies of marine<br />
plankton dinoflagellates of Southern California. American<br />
Midland Naturalist 26, 603–635.<br />
Allen, J.I., Smyth, T.J., Siddom, J.R., Holt, M., 2008. How well can<br />
we forecast high biomass algal biomass events in a eutrophic<br />
coastal sea? Harmful Algae 8 (1), 70–76.<br />
Amy, G., Kim, T.U., Yoon, J., Bellona, C., Drewes, J., Pellegrino, J.,<br />
Heberer, T., 2005. Removal of micropollutants by NF/RO<br />
membranes. Water Science and Technology: Water Supply 5 (5),<br />
25–33.<br />
Anderson, C.R., Brzezinski, M.A., Washburn, L., Kudela, R., 2006.<br />
Circulation and environmental conditions during a toxigenic<br />
Pseudo-nitzschia australis bloom in the Santa Barbara Channel,<br />
California. Marine Ecology Progress Series 327, 119–133.<br />
Anderson, D.M., Kulis, D.M., Sullivan, J.J., Hall, S., Lee, C., 1990.<br />
Dynamics and physiology of saxitoxin production by the<br />
dinoflagellates Alexandrium spp. Marine Biology 104 (3), 511–524.<br />
Anderson, D.M., Glibert, P.M., Burkholder, J.M., 2002. Harmful algal<br />
blooms and eutrophication: nutrient sources, composition, and<br />
consequences. Estuaries 25 (4B), 704–726.<br />
Anderson, D.M., Kulis, D.M., Keafer, B.A., Gribble, K.E., Marin, R.,<br />
Scholin, C.A., 2005. Identification and enumeration of<br />
Alexandrium spp. from the Gulf of Maine using molecular<br />
probes. Deep-Sea Research Part II 52 (19–21), 2467–2490.<br />
Anderson, D.M., Burkholder, J.M., Cochlan, W.P., Glibert, P.M.,<br />
Gobler, C.J., Heil, C.A., Kudela, R.M., Parsons, M.L., Rensel, J.E.J.,<br />
Townsend, D.W., Trainer, V.L., Vargo, G.A., 2008. Harmful algal<br />
blooms and eutrophication: examining linkages from selected<br />
coastal regions of the United States. Harmful Algae 8 (1), 39–53.<br />
Armonies, W., 1989. Occurrence of meiofauna in Phaeocystis<br />
seafoam. Marine Ecology Progress Series 53, 305–309.<br />
Armstrong, M., Kudela, R., 2006. Evaluation of California isolates<br />
of Lingulodinium polyedrum for the production of yessotoxin.<br />
African Journal of Marine Science 28 (2), 399–401.<br />
Aune, T., Sørby, R., Yasumoto, T., Ramstad, H., Landsverk, T., 2002.<br />
Comparison of oral and intraperitoneal toxicity of yessotoxin<br />
towards mice. Toxicon 40 (1), 77–82.<br />
Baden, D.G., 1983. Marine food-borne dinoflagellate toxins.<br />
International Review of Cytology 83, 99–150.<br />
Baden, D.G., 1989. Brevetoxins: unique polyether dinoflagellate<br />
toxins. The FASEB Journal 3 (7), 1807–1817.<br />
Baden, D.G., Bourdelais, A.J., Jacocks, H., Michelliza, S., Naar, J.,<br />
2005. Natural and derivative brevetoxins: historical<br />
background, multiplicity and effects. Environmental Health<br />
Perspectives 113 (5), 621–625.<br />
Barber, R.T., Chavez, F.P., 1983. Biological consequences of El<br />
Niño. Science 222 (4629), 1203–1210.<br />
Bargu, S., Powell, C.L., Coale, S.L., Busman, M., Doucette, G.J.,<br />
Silver, M.W., 2002. Krill: a potential vector for domoic acid in<br />
marine food webs. Marine Ecology Progres Series 237, 209–216.<br />
Bargu, S., Powell, C.L., Wang, Z., Doucette, G.J., Silver, M.W., 2008.<br />
Note on the occurrence of Pseudo-nitzschia australis and<br />
domoic acid in squid from Monterey Bay, CA (USA). Harmful<br />
Algae 7 (1), 45–51.<br />
Bates, H.A., Kostriken, R., Rapoport, A., 1978. The occurrence of<br />
saxitoxin and other toxins in various dinoflagellates. Toxicon<br />
16 (6), 595–601.<br />
Bates, S.S., Trainer, V.L., 2006. The ecology of harmful diatoms. In:<br />
Granéli, E., Turner, J.T. (Eds.), Ecology of Harmful Algae.<br />
Springer, Berlin, pp. 81–93.<br />
Bates, S.S., Bird, C.J., Defreitas, A.S.W., Foxall, R., Gilgan, M.,<br />
Hanic, L.A., Johnson, G.R., McCulloch, A.W., Dodense, P.,<br />
Pocklington, R., Quilliam, M.A., Sim, P.G., Smith, J.C., Rao, D.V.<br />
S., Todd, C.D., Walter, J.A., Wrigh, J.L.C., 1989. Pennate diatom<br />
Nitzschia pungens as the primary source of domoic acid, a toxin<br />
in shellfish from eastern Prince Edwards Island, Canada.<br />
Canadian Journal of Fisheries and Aquatic Sciences 46 (7),<br />
1203–1215.
Bates, S.S., DeFreitas, A.S.W., Milley, J.E., Pocklington, R.,<br />
Quilliam, M.A., Smith, J.C., Worms, J., 1991. Controls on<br />
domoic acid production by the diatom Nitzschia pungens f.<br />
multiseries in cultures: nutrients and irradiance. Canadian<br />
Journal of Fisheries and Aquatic Sciences 48 (7), 1136–1144.<br />
Baugh, K.A., Bush, J.M., Bill, B.D., Lefebvre, K.A., Trainer, V.L.,<br />
2006. Estimates of specific toxicity in several Pseudo-nitzschia<br />
species from the Washington coast, based on culture and field<br />
studies. African Journal of Marine Science 28 (2), 403–407.<br />
Bialojan, C., Takai, A., 1988. Inhibitory effect of a marine-sponge<br />
toxin, okadaic acid, on protein phosphatases. Specificity and<br />
kinetics. Biochemical Journal 256 (1), 283–290.<br />
Bigelow, H.B., Leslie, M., 1930. Reconnaissance of the waters and<br />
plankton of Monterey Bay July 1928. Bulletin of the Museum of<br />
Comparative Zoology 70 (5), 430–581.<br />
Blasco, D., 1978. Observations on the diel migration of marine<br />
dinoflagellates off the Baja Californian coast. Marine Biology<br />
46 (1), 41–47.<br />
Bolin, R.L., Abbott, D.P., 1960. Studies on the marine climate and<br />
phytoplankton of the central coastal area of California,<br />
1954–1960. California Cooperative Oceanic Fisheries<br />
Investigations 9, 23–45.<br />
Boni, L., Ceredi, A., Guerrini, F., Milandri, A., Pistocchi, R., Poletti,<br />
R., Pompei, M., 2002. Toxic Gonyaulax Grindley Reinecke in the<br />
North-Western Adriatic Sea (Italy). In: Tenth International<br />
Conference on Harmful Algal Blooms, St. Petersburg, FL<br />
(abstract).<br />
Bonnelye, V., Snaz, M.A., Comte, C., Colas, F., Plasse, L., Gueguen, F.,<br />
2004a. Ultrafiltration versus conventional pre-treatment<br />
upstream RO: advantages and limits, Seoul, Korea.<br />
Bonnelye, V., Sanz, M.A., Durand, J.P., Plasse, L., Gueguen, F.,<br />
Mazounie, P., 2004b. Reverse osmosis on open intake<br />
seawater: pre-treatment strategy. Desalination 167 (1–3),<br />
191–200.<br />
Bourdelais, A.J., Tomas, C.R., Naar, J., Kubanek, J., Baden, D.G., 2002.<br />
New fish-killing alga in coastal Delaware produces neurotoxins.<br />
Environmental Health Perspectives 110 (5), 465–470.<br />
Bowers, H.A., Tengs, T., Glasgow Jr., H.B., Burkholder, J.M.,<br />
Rublee, P.A., Oldach, D.W., 2000. Development of real-time<br />
PCR assays for rapid detection of Pfiesteria piscicida and related<br />
dinoflagellates. Applied and Environmental Microbiology 66<br />
(11), 4641–4648.<br />
Bowers, H.A., Tomas, C., Tengs, T., Kempton, J.W., Lewitus, A.J.,<br />
Oldach, D.W., 2006. Raphidophyceae (Chadefaud ex Silva)<br />
systematics and rapid identification: sequence analyses and<br />
real-time PCR assays. Journal of Phycology 42 (6), 1333–1348.<br />
Bricelj, V.M., Lonsdale, D.J., 1997. Aureococcus anophagefferens:<br />
causes and ecological consequences of brown tides in U.S.<br />
mid-Atlantic coastal waters. Limnology and Oceanography 42<br />
(5), 1023–1038.<br />
Buck, K.R., Uttal-Cooke, L., Pilskaln, C.H., Roelke, D.L., Villac, M.C.,<br />
Fryxell, G.A., Cifuentes, L., Chavez, F.P., 1992. Autecology of<br />
the diatom Pseudo-nitzschia australis, a domoic acid producer,<br />
from Monterey Bay, California. Marine Ecology Progress Series<br />
84 (3), 293–302.<br />
Burbano, A.A., Adham, S.S., Pearce, W.R., 2007. The state of fullscale<br />
RO/NF desalination - Results from a worldwide survey.<br />
Journal of American Water Works Association 99 (4), 116–127.<br />
Burkholder, J.M., Azanza, R.V., Sako, Y., 2006. The ecology of<br />
harmful dinoflagellates. In: Granéli, E., Turner, J.T. (Eds.),<br />
Ecology of Harmful Algae. Springer, Berlin, pp. 53–66.<br />
Burkholder, J.M., Hallegraeff, G.M., Melia, G., Cohen, A.,<br />
Bowers, H.A., Oldach, D.W., Parrow, M.W., Sullivan, M.J.,<br />
Zimba, P.V., Allen, E.H., Kinder, C.A., Mallin, M.A., 2007.<br />
Phytoplankton and bacterial assemblages in ballast water of<br />
U.S. military ships as a function of port of origin, voyage time,<br />
and ocean exchange practices. Harmful Algae 6 (4), 486–518.<br />
water research 44 (2010) 385–416 409<br />
Burkholder, J.M., Glibert, P.M., Skelton, H.M., 2008. Mixotrophy,<br />
a major mode of nutrition for harmful algal species in eutrophic<br />
waters. Harmful Algae 8 (1), 77–93.<br />
Buskey, E.J., 2008. How does eutrophication affect the role of<br />
grazers in harmful algal bloom dynamics? Harmful Algae 8 (1),<br />
152–157.<br />
Busse, L.B., Venrick, E.L., Antrobus, R., Miller, P.E., Vigilant, V.,<br />
Silver, M.W., Mengelt, C., Mydlarz, L., Prezelin, B.B., 2006.<br />
Domoic acid in phytoplankton and fish in San Diego, CA, USA.<br />
Harmful Algae 5 (1), 91–101.<br />
Caron, D.A., Dennett, M.R., Moran, D.M., Schaffner, R.A., Lonsdale, D.<br />
J., Gobler, C.J., Nuzzi, R., McLean, T.I., 2003. Development and<br />
application of a monoclonal-antibody technique for counting<br />
Aureococcus anophagefferens, an alga causing recurrent brown<br />
tides in the Mid-Atlantic United States. Applied and<br />
Environmental Microbiology 69 (9), 5492–5502.<br />
Carpenter, E.J., Janson, S., Boje, R., Pollehne, F., Chang, J., 1995.<br />
The dinoflagellate Dinophysis norvegica: biological and<br />
ecological observations in the Baltic Sea. European Journal of<br />
Phycology 109.<br />
Casper, E.T., Patterson, S.S., Bhanushali, P., Farmer, A., Smith, M.,<br />
Fries, D.P., Paul, J.H., 2007. A handheld NASBA analyzer for the<br />
field detection and quantification of Karenia brevis. Harmful<br />
Algae 6 (1), 112–118.<br />
Catterall, W.A., 1992. Cellular and molecular biology of voltagegated<br />
sodium channels. Physiological Reviews 72 (Suppl. 4.),<br />
S15–S48.<br />
CDPH, 2007. Preharvest Shellfish Protection and Marine Biotoxin<br />
Monitoring Program. California Department of Public Health.<br />
Cembella, A.D., 1989. Occurrence of okadaic acid, a major<br />
diarrheic shellfish toxin, in natural populations of Dinophysis<br />
spp. from the eastern coast of North America. Journal of<br />
Applied Phycology 1 (4), 307–310.<br />
Cembella, A.D., 2003. Chemical ecology of eukaryotic microalgae<br />
in marine ecosystems. Phycologica 42 (4), 420–447.<br />
Cestele, S., Catterall, W.A., 2000. Molecular mechanisms of<br />
neurotoxin action on voltage-gated sodium channels.<br />
Biochimie 82 (9–10), 883–892.<br />
Chavez, F.P., Ryan, J., Lluch-Cota, S.E., Niquen, M., 2003. From<br />
anchovies to sardines and back: multidecadal change in the<br />
Pacific ocean. Science 299 (5604), 217–221.<br />
Ciminiello, P., Fattorusso, E., Forino, M., Poletti, R., Viviani, R., 2000.<br />
Structure determination of carboxyhomoyessotoxin, a new<br />
yessotoxin analogue isolated from Adriatic mussels. Chemistry<br />
Research in Toxicology 13 (8), 770–774.<br />
Ciminiello, P., Fattorusso, E., Forino, M., Magno, S., Poletti, R., 2001.<br />
42,43,44,45,46,47,55-Heptanor-41-oxohomoyessotoxin, a new<br />
biotoxin from mussels of the Northern Adriatic Sea. Chemical<br />
Research in Toxicology 14 (5), 596–599.<br />
Ciminiello, P., Fattorusso, E., Forino, M., Magno, S., Poletti, R.,<br />
Viviani, R., 1998. Isolation of adriatoxin a new analogue of<br />
yessotoxin from mussels of the Adriatic Sea. Tetrahedron Letters<br />
39 (48), 8897–8900.<br />
Clayden, J., Read, B., Hebditch, K., 2005. Chemistry of domoic acid,<br />
isodomoic acids, and their analogues. Tetrahedron 61 (24),<br />
5713–5724.<br />
Cochlan, W.P., Herndon, J., Kudela, R.M., 2008. Inorganic and organic<br />
nitrogen uptake by the toxigenic diatom Pseudo-nitzschia australis<br />
(Bacillariophyceae). Harmful Algae 8 (1), 111–118.<br />
Cooley, H., Gleick, P.H., Wolff, G., 2006. Desalination, with a grain<br />
of salt: a California perspective. In: Hart, I. (Ed.). Pacific<br />
Institute for Studies in Development, Environment, and<br />
Security, Hayward, CA, p. 88.<br />
Coyne, K.J., Hutchins, D.A., Hare, C.E., Cary, S.C., 2001. Assessing<br />
temporal and spatial variability in Pfiesteria piscicida<br />
distributions using molecular probing techniques. Aquatic<br />
Microbial Ecology 24 (3), 275–285.
410<br />
Coyne, K.J., Handy, S.M., Demir, E., Whereat, E.B., Hutchins, D.A.,<br />
Portune, K.J., Doblin, M.A., Cary, S.C., 2005. Improved<br />
quantitative real-time PCR assays for enumeration of harmful<br />
algal species in field samples using an exogenous DNA reference<br />
standard. Limnology and Oceanography Methods 3, 381–391.<br />
Cullen, J.J., Horrigan, S.G., 1981. Effects of nitrate on the diurnal<br />
vertical migration, carbon to nitrogen ratio, and the<br />
photosynthetic capacity of the dinoflagellate Gymnodinium<br />
splendens. Marine Biology 62 (2–3), 81–89.<br />
Curtiss, C.C., Langlois, G.W., Busse, L.B., Mazzillo, F., Silve, M.W.,<br />
2008. The emergence of Cochlodinium along the California<br />
Coast (USA). Harmful Algae 7 (3), 337–346.<br />
Daiguji, M., Satake, M., Ramstad, H., Aune, T., Naoki, H.,<br />
Yasumoto, T., 1998. Structure and fluorometric HPLC<br />
determination of 1-desulfoyessotoxin, a new yessotoxin<br />
analog isolated from mussels from Norway. Natural Toxins 6<br />
(6), 235–239.<br />
Dekshenieks, M.M., Donaghay, P.L., Sullivan, J.M., Rines, J.E.B.,<br />
Osborn, T.R., Twardowski, M.S., 2001. Temporal and spatial<br />
occurrence of thin phytoplankton layers in relation to physical<br />
processes. Marine Ecology Progress Series 223, 61–71.<br />
Demir, E., Coyne, K.J., Doblin, M.A., Handy, S.M., Hutchins, D.A.,<br />
2008. Assessment of microzooplankton grazing on Heterosigma<br />
akashiwo using a species-specific approach combining<br />
quantitative real-time PCR (QPCR) and dilution methods.<br />
Microbial Ecology 55 (4), 583–594.<br />
Determan, T., 2003. Paralytic shellfish poisoning (PSP) patterns in<br />
Puget Sound shellfish in 2001-A report for the Puget Sound<br />
Ambient Monitoring Program Office of Food Safety and<br />
Shellfish Programs. Washington State Department of Health,<br />
Olympia.<br />
Draisci, R., Ferretti, E., Palleschi, L., Marchiafava, C., Poletti, R.,<br />
Milandri, A., Ceredi, A., Pompei, M., 1999a. High levels of<br />
yessotoxin in mussels and presence of yessotoxin and<br />
homoyessotoxin in dinoflagellates of the Adriatic Sea. Toxicon<br />
37, 1187–1193.<br />
Draisci, R., Palleschi, L., Giannetti, L., Lucentini, L., James, K.J.,<br />
Bishop, A.G., Satake, M., Yasumoto, T., 1999b. New approach to<br />
the direct detection of known and new diarrhetic shellfish toxins<br />
in mussels and phytoplankton by liquid chromatography-mass<br />
spectrometry. Journal of Chromatrography 847 (1–2), 213–221.<br />
Eiki, K., Satake, M., Koike, K., Ogata, T., Mitsuya, T., Oshima, Y.,<br />
2005. Confirmation of yessotoxin production by the<br />
dinoflagellate Protoceratium reticulatum in Mutsu Bay. Fisheries<br />
Science 71 (3), 633–638.<br />
Eppley, R.W., Holm-Harisen, O., Strickland, J.D.H., 1968. Some<br />
observations on the vertical migration of dinoflagellates.<br />
Journal of Phycology 4 (4), 333–340.<br />
European Commission, 2002. Directive Commission Decision, No.<br />
C2002 1001. Official Journal of the European Communities<br />
March 15, 2002, 1.75/63, 16.3.<br />
Fehling, J., Davidson, K., Bolch, C.J., Bates, S.S., 2004. Growth and<br />
domoic acid production by Pseudo-nitzschia seriata<br />
(Bacillariophyceae) under phosphate and silicate limitation.<br />
Journal of Phycology 40 (4), 674–683.<br />
Flynn, K.J., 2002. Toxin production in migrating dinoflagellates:<br />
a modelling study of PSP producing Alexandrium. Harmful<br />
Algae 1 (2), 147–155.<br />
Flynn, K.J., 2008. Attack is not the best form of defense: lessons from<br />
harmful algal bloom dynamics. Harmful Algae 8 (1), 129–139.<br />
Flynn, K., Franco, J.M., Fernandez, P., Reguera, B., Zapata, M.,<br />
Wood, G., Flynn, K.J., 1994. Changes in toxin content, biomass<br />
and pigments of the dinoflagellate Alexandrium minutum during<br />
nitrogen refeeding and growth into nitrogen or phosphorus<br />
stress. Marine Ecology Progres Series 111, 99–109.<br />
Franchini, A., Marchesini, E., Poletti, R., Ottaviani, E., 2004. Lethal<br />
and sub-lethal yessotoxin dose-induced morpho-functional<br />
water research 44 (2010) 385–416<br />
alterations in intraperitoneal injected Swiss CD1 mice. Toxicon<br />
44 (1), 83–90.<br />
Fritz, L., Quilliam, M.A., Wright, J.L.C., Beale, A.M., Work, T.M.,<br />
1992. An outbreak of domoic acid poisoning attributed to the<br />
pennate diatom Pseudo-nitzschia australis. Journal of Phycology<br />
28 (4), 439–442.<br />
Fryxell, G.A., Villac, M.C., Shapiro, L.P., 1997. The occurrence of<br />
the toxic diatom genus Pseudo-nitzschia (Bacillariophyceae) on<br />
the West Coast of the USA, 1920–1996: a review. Phycologia 36<br />
(6), 419–437.<br />
Gaid, K., Treal, Y., 2007. Le dessalement des eaux par osmose<br />
inverse: l’expÈrience de VÈolia Water. Desalination 203 (1–3),<br />
1–14.<br />
Gaines, G., Taylor, F.J.R., 1985. An exploratory analysis of PSP<br />
patterns in British Columbia: 1942–1984. In: Anderson, D.M.,<br />
White, A.W., Baden, D.G. (Eds.), Toxic dinoflagellates. Elsevier<br />
Science, New York, pp. 439–444.<br />
Galluzzi, L., Penna, A., Bertozzini, E., Vila, M., Garces, E.,<br />
Magnani, M., 2004. Development of a real-time PCR assay for<br />
rapid detection and quantification of Alexandrium minutum<br />
(a dinoflagellate). Applied and Environmental Microbiology 70<br />
(2), 1199–1206.<br />
Garrison, D.L., Conrad, S.M., Eilers, P.P., Waldron, E.M., 1992.<br />
Confirmation of domoic acid production by Pseudo-nitzschia<br />
australis (Bacillariophyceae) cultures. Journal of Phycology 28<br />
(5), 604–607.<br />
Garthwaite, I., Ross, K.M., Miles, C.O., Hansen, R.P., Foster, D.,<br />
Wilkins, A.L., Towers, N.R., 1998. Polyclonal antibodies to<br />
domoic acid, and their use in immunoassays for domoic acid<br />
in sea water and shellfish. Natural Toxins 6 (3–4), 93–104.<br />
Garthwaite, I., Ross, K.M., Miles, C.O., Briggs, L.R., Towers, N.R.,<br />
Borrell, T., Busby, P., 2001. Integrated Enzyme-Linked<br />
Immunosorbent Assay screening system for amnesic,<br />
neurotoxic, diarrhetic, and paralytic shellfish poisoning toxins<br />
found in New Zealand. Journal of AOAC International 84 (5),<br />
1643–1648.<br />
Geraci, J.R., Anderson, D.M., Timperi, R.J., St. Aubin, D.J., Early, G.A.,<br />
Prescott, J.H., Mayo, C., 1989. Humpback whales (Megaptera<br />
novaeangliae) fatally poisoned by dinoflagellate toxin. Canadian<br />
Journal of Fisheries and Aquatic Sciences 46 (11), 1895–1898.<br />
Gijsbertsen-Abrahamse, A.J., Schmidt, W., Chorus, I., Heijman, S.G.J.,<br />
2006. Removal of cyanotoxins by ultrafiltration and<br />
nanofiltration. Journal of Membrane Science 276 (1–2), 252–259.<br />
Glibert, P.M., Anderson, D.M., Gentien, P., Granéli, E., Sellner, K.G.,<br />
2005a. The global, complex phenomena of harmful algal<br />
blooms. Oceanography 18 (2), 137–147.<br />
Glibert, P.M., Seitzinger, S., Heil, C.A., Burkholder, J.M., Parrow, M.W.,<br />
Codispoti, L.A., Kelly, V., 2005b. The role of eutrophication in<br />
the global proliferation of harmful algal blooms. Oceanography<br />
18 (2), 198–209.<br />
Glibert, P.M., Harrison, J., Heil, C., Seitzinger, S., 2006. Escalating<br />
worldwide use of urea – a global change contributing to<br />
coastal eutrophication. Biogeochemistry 77 (3), 441–463.<br />
Glibert, P.M., Kelly, V., Alexander, J., Cosdispoti, L.A., Boicourt, W.C.,<br />
Trice, T.M., Michael, B., 2008. In situ nutrient monitoring:<br />
a tool for capturing nutrient variability and the antecedent<br />
conditions that support algal blooms. Harmful Algae 8 (1),<br />
175–181.<br />
Granéli, E., Flynn, K., 2006. Chemical and physical factors<br />
influencing toxin content. In: Granéli, E., Turner, J.T. (Eds.),<br />
Ecology of Harmful Algae. Springer, Berlin, pp. 229–241.<br />
Greenfield, D.I., Marin, R., Jensen, S., Massion, E., Roman, B.,<br />
Feldman, J., Scholin, C., 2006. Application of the Environmental<br />
Sample Processor (ESP) methodology for quantifying Pseudonitzschia<br />
australis using ribosomal RNA-targeted probes in<br />
sandwich and fluorescent in situ hybridization. Limnology and<br />
Oceanography Methods 4, 426–435.
Gregorio, D.E., Connell, L., 2000. Range of Heterosigma akashiwo<br />
(Raphidophyceae) expanded to include California, USA. In:<br />
Ninth International Conference on Harmful Algal Blooms,<br />
Tasmania, Australia (abstract).<br />
Gregorio, D.E., Pieper, R.E., 2000. Investigations of red tides along<br />
the southern California. Bulletin of Southern California<br />
Academy of Sciences 99 (8), 147–160.<br />
Gulland, F.M.D., Haulena, M., Fauquier, D., Langlois, G., Lander, M.E.,<br />
Zabka, T., Duerr, R., 2002. Domoic acid toxicity in California sea<br />
lions (Zalophus californianus): clinical signs, treatment and<br />
survival. Veterinary Record 150 (15), 475–480.<br />
Hallegraeff, G., Hara, Y., 1995. Taxonomy of harmful marine<br />
raphidophytes. In: Hallegraef, G., Anderson, D.M., Cembella, A.D.<br />
(Eds.), Manual on Harmful Marine Microalgae. IOC UNCESCO,<br />
Paris, pp. 365–371.<br />
Hallegraeff, G.M., 1993. A review of harmful algal blooms and<br />
their apparent global increase. Phycologia 32 (2), 79–99.<br />
Hallegraeff, G.M., 2003. Harmful algal blooms: a global overview.<br />
In: Hallegraeff, G.M., Anderson, D.M., Cembella, A.D. (Eds.),<br />
Manual on Harmful Marine Microalgae. Monographs on<br />
Oceanographic Methodology. UNESCO Publishing, Paris<br />
pp. 25–49.<br />
Hallegraeff, G.M., Bolch, C.J., 1992. Transport of diatom and<br />
dinoflagellate resting spores in ships’ ballast water:<br />
implications for plankton biogeography and aquaculture.<br />
Journal of Plankton Research 14 (8), 1067–1084.<br />
Handy, S.M., Hutchins, D.A., Cary, S.C., Coyne, K.J., 2006.<br />
Simultaneous enumeration of multiple raphidophyte species<br />
by multiprobing and mutiplexing: capabilities and limitations.<br />
Limnology and Oceanography: Methods 4, 193–204.<br />
Haque, S.M., Onoue, Y., 2002a. Effects of salinity on growth and<br />
toxin production of a noxious phytoflagellate, Heterosigma<br />
akashiwo (Raphidophyceae). Botanica Marina 45 (4), 356–363.<br />
Haque, S.M., Onoue, Y., 2002b. Variation in toxin compositions of<br />
two harmful raphidophytes, Chattonella antiqua and Chattonella<br />
marina, at different salinities. Environmental Toxicology 17 (2),<br />
113–118.<br />
Hard, J.J., Connell, L., Hershberger, W.K., Harrell, L.W., 2000.<br />
Genetic variation in mortality of chinook salmon during<br />
a bloom of the marine alga Heterosigma akashiwo. Journal of<br />
Fish Biology 56 (6), 1387–1397.<br />
Hasan Al-Sheikh, A.H., 1997. Seawater reverse osmosis<br />
pretreatment with an emphasis on the Jeddah Plant operation<br />
experience. Desalination 110 (1–2), 183–192.<br />
Haystead, T.A.J., Sim, A.T.R., Carling, D., Honnor, R.C., Tsukitani, Y.,<br />
Cohen, P., Hardie, D.G., 1989. Effects of the tumour promoter<br />
okadaic acid on intracellular protein phosphorylation and<br />
metabolism. Nature 337, 78–81.<br />
Heisler, J., Glibert, P.M., Burkholder, J.M., Anderson, D.M.,<br />
Cochlan, W., Dennison, W.C., Dortch, Q., Gobler, C.J., Heil, C.A.,<br />
Humphries, E., Lewitus, A., Magnien, R., Marshall, H.G.,<br />
Sellner, K., Stockwell, D.A., Stoecker, D.K., Suddleson, M., 2008.<br />
Eutrophication and harmful algal blooms: a scientific<br />
consensus. Harmful Algae 8 (1), 3–13.<br />
Herndon, J., Cochlan, W.P., Horner, R., 2003. Heterosigma akashiwo<br />
blooms in San Francisco Bay. Interagency Ecological Program<br />
for the San Francisco Estuary Newsletter 16, 46–48.<br />
Hershberger, P.K., Rensel, J.E., Postel, J.R., Taub, F.B., 1997.<br />
Heterosigma bloom and associated fish kill. Harmful Algae<br />
News 16 (1), 4.<br />
Holmes, R.W., Williams, P.M., Eppley, R.W., 1967. Red water in<br />
La Jolla, 1964–1966. Limnology and Oceanography 12 (3),<br />
503–512.<br />
Hong, Y., Smith Jr., W.O., White, A.M., 1997. Studies on<br />
transparent exopolymer particles (TEP) produced in the Ross<br />
Sea (Antarctica) and by Phaeocystis antarctica<br />
(Prymnesiophyceae). Journal Of Phycology 33 (3), 368–376.<br />
water research 44 (2010) 385–416 411<br />
Hori, M., Matsuura, Y., Yoshimoto, R., Ozaki, H., Yasumoto, T.,<br />
Karaki, H., 1999. Actin depolymerizing action by marine toxin,<br />
pectenotoxin-2. Folia Pharmacologica Japonica 114 (Suppl. 1),<br />
225–229.<br />
Horner, R.A., Postel, J.R., 1993. Toxic diatoms in western<br />
Washington waters (U.S. west coast). Hydrobiologia 269/270<br />
(1), 197–205.<br />
Horner, R.A., Garrison, D.L., Plumley, F.G., 1997. Harmful algal<br />
blooms and red tide problems on the U.S. west coast.<br />
Limnology and Oceanography 42 (2), 1076–1088.<br />
Horner, R.A., Hickey, B.M., Postel, J.R., 2000. Pseudo-nitzschia<br />
blooms and physical oceanography off Washington state.<br />
South African Journal of Marine Science 22 (1), 299–308.<br />
Howard, M.D.A., Cochlan, W.P., Ladizinsky, N., Kudela, R.M., 2007.<br />
Nitrogenous preference of toxigenic Pseudo-nitzschia australis<br />
(Bacillariophyceae) from field and laboratory experiments.<br />
Harmful Algae 6 (2), 206–217.<br />
Howard, M.D.A., Silvera, M., Kudela, R.M., 2008. Yessotoxin<br />
detected in mussel (Mytilus californicus) and phytoplankton<br />
samples from the U.S. west coast. Harmful Algae 7 (5),<br />
646–652.<br />
Howard, M.D., Smith, G.J., Kudela, R.M., 2009. Phylogenetic<br />
relationships of yessotoxin-producing dinoflagellates, based<br />
on the large subunit and internal transcribed spacer ribosomal<br />
DNA domains. Applied and Environmental Microbiology 75 (1),<br />
54–63.<br />
Howarth, R.W., 2008. Coastal nitrogen pollution: a review of<br />
sources and trends globally and regionally. Harmful Algae 8<br />
(1), 14–20.<br />
IOC HAB Programme, 2008. IOC Harmful Algal Bloom Website.<br />
Intergovernmental Oceanic Commission of UNESCO.<br />
Iwataki, M., Kawami, H., Matsuoka, K., 2007. Cochlodinium<br />
fulvescens sp nov (Gymnodiniales, Dinophyceae), a new chainforming<br />
unarmored dinoflagellate from Asian coasts.<br />
Phycological Research 55 (3), 231–239.<br />
Iwataki, M., Kawami, H., Mizushima, K., Mikulski, C.M.,<br />
Doucette, G.J., Relox, J.R., Anton, A., Fukuyo, Y., Matsuoka, K.,<br />
2008. Phylogenetic relationships in the harmful dinoflagellate<br />
Cochlodinium polykrikoides (Gymnodiniales, Dinophyceae)<br />
inferred from LSU rDNA sequences. Harmful Algae 7 (3),<br />
271–277.<br />
Jeffery, B., Barlow, T., Moizer, K., Paul, S., Boyle, C., 2004. Amnesic<br />
shellfish poison. Food and Chemical Toxicology 42 (4),<br />
545–557.<br />
Jessup, D.A., Miller, M.A., Ryan, J.P., Nevins, H.M., Kerkering, H.A.,<br />
Mekebri, A., Crane, D.B., Johnson, T.A., Kudela, R.M., 2009.<br />
Mass stranding of marine birds caused by a surfactantproducing<br />
red tide. PLOS One 4, E4550.<br />
Jester, R., 2008. An investigation into the prevalence of<br />
Alexandrium derived toxins in marine food webs. Ph.D.,<br />
University of California, Santa Cruz, Santa Cruz, pp. 116.<br />
Jester, R., Lefebvre, K.A., Langlois, G., Vigilant, V., Baugh, K.A.,<br />
Silver, M.W., 2009a. A shift in the dominant toxin-producing<br />
algal species in central California alters phycotoxins in food<br />
webs. Harmful Algae 8 (2), 291–298.<br />
Jester, R.J., Baugh, K.A., Lefebvre, K.A., 2009b. Presence of<br />
Alexandrium catenella and paralytic shellfish toxins in finfish,<br />
shellfish and rock crabs in Monterey Bay, California, USA.<br />
Marine Biology 156 (3), 493–504.<br />
Johansson, N., Granéli, E., Yasumoto, T., Carlsson, P., Legrand, C.,<br />
1996. Toxin production by Dinophysis acuminata and D. acuta<br />
cells grown under nutrient sufficient and deficient conditions.<br />
In: Yasumoto, T., Oshima, Y., Fukuyo, Y. (Eds.), Harmful and<br />
Toxic Algal Blooms. Intergovernmental Oceanographic<br />
Commission of UNESCO, Paris, pp. 277–279.<br />
John, E.H., Flynn, K.J., 2000. Growth dynamics and toxicity of<br />
Alexandrium fundyense (Dinophyceae): the effect of changing N:
412<br />
P supply ratios on internal toxin and nutrient levels. European<br />
Journal of Phycology 35 (1), 11–23.<br />
Jones, B.H., Noble, M.A., Dickey, T.D., 2002. Hydrographic and<br />
particle distributions over the Palos Verdes continental shelf:<br />
spatial, seasonal and daily variability. Continental Shelf<br />
Research 22 (6–7), 945–965.<br />
Kirkpatrick, B., Fleming, L.E., Squicciarini, D., Backer, L.C.,<br />
Clark, R., Abraham, W., Benson, J., Cheng, Y.S., Johnson, D.,<br />
Pierce, R., Zaias, J., Bossart, G.D., Baden, D.G., 2004. Literature<br />
review of Florida red tide: implications for human health<br />
effects. Harmful Algae 3 (2), 99–115.<br />
Kudela, R.M., Cochlan, W.P., 2000. Nitrogen and carbon uptake<br />
kinetics and the influence of irradiance for a red tide bloom off<br />
southern California. Aquatic Microbial Ecology 21 (1), 31–47.<br />
Kudela, R.M., Roberts, A., Armstrong, M., 2004. Laboratory<br />
analyses of nutrient stress and toxin production in Pseudonitzschia<br />
spp. from Monterey Bay, California. In: Steidinger, K.,<br />
Landsberg, J., Tomas, C., Vargo, G. (Eds.). Florida<br />
Environmental Research Institute and UNESCO, St. Petersberg,<br />
pp. 136–138.<br />
Kudela, R.M., Lane, J.Q., Cochlan, W.P., 2008a. The potential role<br />
of anthropogenically derived nitrogen in the growth<br />
of harmful algae in California, USA. Harmful Algae 8 (1),<br />
103–110.<br />
Kudela, R.M., Ryan, J.P., Blakely, M.D., Lane, J.Q., Peterson, T.D.,<br />
2008b. Linking the physiology and ecology of Cochlodinium to<br />
better understand harmful algal bloom events: a comparative<br />
approach. Harmful Algae 7 (3), 278–292.<br />
Kusek, K.M., Vargo, G., Steidinger, K., 1999. Gymnodinium breve in<br />
the field in the lab and in the newspaper - scientific and<br />
journalistic analysis of Florida red tides. Contributions in<br />
Marine Science 34, 1–229.<br />
Ladizinsky, N., Smith, G.J., 2000. Accumulation of domoic acid by<br />
the coastal diatom Pseudo-nitzschia multiseries: a possible<br />
copper complexation strategy. Journal of Phycology 36, 41.<br />
Landry, M.R., 1989. Broad-scale distributional patterns of<br />
hydrographic variables on the Washington/Oregon shelf. In:<br />
Landry, M.R., Hickey, B.M. (Eds.), Coastal Oceanography of<br />
Washington and Oregon. Elsevier, New York, pp. 1–40.<br />
Lange, C.B., Reid, F.M.H., Vernet, M., 1994. Temporal distribution<br />
of the potentially toxic diatom Pseudo-nitzschia australis at<br />
a coastal site in Southern California. Marine Ecology Progress<br />
Series 104 (3), 309–312.<br />
Langlois, G., 2007. Marine Biotoxin Monitoring Program Annual<br />
Report. California Department of Public Health for California<br />
Department of Fish and Game, Sacramento.<br />
Lee, J., Walker, H.W., 2006. Effect of process variables and natural<br />
organic matter on removal of microcystin-LR by PAC-UF.<br />
Environmental Science and Technology 40 (23), 7336–7342.<br />
Lee, J.S., Yanagi, T., Kenma, R., Yasumoto, T., 1987. Fluorometric<br />
determination of diarrhetic shellfish toxins by highperformance<br />
liquid chromatography. Agricultural and<br />
Biological Chemistry 51 (3), 877–881.<br />
Lee, J.S., Igarashi, T., Fraga, S., Dah, E., Hovgaard, P., Yasumoto, T.,<br />
1989. Determination of diarrhetic toxins in various<br />
dinoflagellate species. Journal of Applied Phycology 1 (2),<br />
147–152.<br />
Lefebvre, K.A., Bill, B.D., Erickson, A., Baugh, K.A., O’Rourke, L.,<br />
Costa, P.R., Nance, S., Trainer, V.L., 2008. Characterization of<br />
intracellular and extracellular saxitoxin levels in both field<br />
and cultured Alexandruim spp. samples from Sequim Bay,<br />
Washington. Marine Drugs 6, 103–116.<br />
Levandowsky, M., Kaneta, P.J., 1987. Behaviour in dinoflagellates.<br />
In: Taylor, F.J.R. (Ed.), The Biology of Dinoflagellates. Blackwell,<br />
Oxford, pp. 360–397.<br />
Lin, Y.Y., Risk, M., Ray, S.M., Van Engen, D., Clardy, J., Golik, J.,<br />
James, J.C., Nakanishi, K., 1981. Isolation and structure of<br />
brevetoxin B from the ‘red tide’ dinoflagellate Ptychodiscus<br />
water research 44 (2010) 385–416<br />
brevis (Gymnodinium breve). Journal of American Chemical<br />
Society 103 (22), 6773–6775.<br />
Llewellyn, L.E., 2006. Saxitoxin, a toxic marine natural product<br />
that targets a multitude of receptors. Natural Product Reports<br />
23 (2), 200–222.<br />
Loeblich III, A.R., Fine, K.E., 1977. Marine chloromonads: More<br />
widely distributed in neritic environments than previously<br />
thought. Proceedings of the Biological Society of Washington<br />
90 (2), 388–399.<br />
Lundholm, N., Skov, J., Pocklington, R., Moestrup, Ã., 1997. Studies<br />
on the marine planktonic diatom Pseudo-nitzschia. 2.<br />
Autecology of P. pseudodelicatissima based on isolates from<br />
Danish coastal waters. Phycologia 36, 381–388.<br />
Malagoli, D., Marchesini, E., Ottaviani, E., 2006. Lysosomes as the<br />
target of yessotoxin invertebrate and vertebrate cell lines.<br />
Toxicology Letters 167 (1), 75–83.<br />
Maldonado, M.T., Hughes, M.P., Rue, E.L., Wells, M.L., 2002. The<br />
effect of Fe and Cu on growth and domoic acid production by<br />
Pseudo-nitzschia multiseries and Pseudo-nitzschia australis.<br />
Limnology and Oceanography 47 (2), 515–526.<br />
Marchetti, A., Trainer, V.L., Harrison, P.J., 2004. Environmental<br />
conditions and phytoplankton dynamics associated<br />
with Pseudo-nitzschia abundance and domoic acid in the<br />
Juan de Fuca eddy. Marine Ecology Progress Series 281,<br />
1–12.<br />
Martin, J.L., Haya, K., Burridge, L.E., Wildish, D.J., 1990. Nitzschia<br />
pseudodelicatissima - a source of domoic acid in the Bay of<br />
Fundy, eastern Canada. Marine Ecology Progress Series 67 (2),<br />
177–182.<br />
Matsuoka, K., Iwataki, M., Kawami, H., 2008. Morphology and<br />
taxonomy of chain-forming species of the genus Cochlodinium<br />
(Dinophyceae). Harmful Algae 7 (3), 261–270.<br />
McGillicuddy Jr., D.J., Signell, R.P., Stock, C.A., Keafer, B.A.,<br />
Keller, M.D., Ehtland, R.D., Anderson, D.M., 2005. A<br />
mechanisms for offshore initiations of harmful algal blooms<br />
in the coastal Gulf of Maine. Journal of Plankton Research 25,<br />
1131–1138.<br />
Mengelt, C., 2006. Ultraviolet Photoecology, Dark Survival, and<br />
Seasonal Abundance of Pseudo-nitzschia australis and P.<br />
multiseries in Coastal Waters of Central California.<br />
Dissertation, University of California, Santa Barbara.<br />
Mikulski, C.M., Morton, S.L., Doucette, G.J., 2005. Development<br />
and application of LSU rRNA probes for Karenia brevis in the<br />
Gulf of Mexico, USA. Harmful Algae 4 (1), 49–60.<br />
Mikulski, C.M., Park, Y.T., Jones, K.L., Lee, C.K., Lim, W.A., Lee, Y.,<br />
Scholin, C.A., Doucette, G.J., 2008. Development and field<br />
application of rRNA-targeted probes for the detection of<br />
Cochlodinium polykrikoides Margalef in Korean coastal waters<br />
using whole cell and sandwich hybridization formats.<br />
Harmful Algae 7 (3), 347–359.<br />
Miles, C.O., Aasen, J., Dahl, E., Samdal, I., Aune, T., Briggs, L.R.,<br />
Olseng, C.D., Edvardsen, B., Naustvoll, L.J., 2002. Yessotoxin in<br />
Norwegian Blue Mussels caused by Protoceratium reticulatum.<br />
In: Tenth International Conference on Harmful Algal Blooms,<br />
St. Petersburg, FL (abstract).<br />
Miles, C.O., Wilkins, A.L., Samdal, I.A., Sandvik, M., Petersen, D.,<br />
Quilliam, M.A., Nausvoll, L.J., Rundberget, T., Torgersen, T.,<br />
Hovgaard, P., Jensen, D.J., Cooney, J.M., 2004. A novel<br />
pectenotoxin, PTX-12, in Dinophysis spp. and shellfish from<br />
Norway. Chemical Research and Toxicology 17 (11),<br />
1423–1433.<br />
Miles, C.O., Wilkins, A.L., Hawkes, A.D., Selwood, A.I., Jensen, D.J.,<br />
Munday, R., Cooney, J.M., Beuzenberg, V., 2005a.<br />
Polyhydroxylated amide analogs of yessotoxin from<br />
Protoceratium reticulatum. Toxicon 45 (1), 61–71.<br />
Miles, C.O., Samdal, I.A., Aasen, J.A.G., Jensen, D.J., Quilliam, M.A.,<br />
Petersen, D., Briggs, L.M., Wilkins, A.L., Rise, F., Cooney, J.M.,<br />
MacKenzie, A.L., 2005b. Evidence for numerous analogs of
yessotoxin in Protoceratium reticulatum. Harmful Algae 4 (6),<br />
1075–1091.<br />
Miles, C.O., Wilkins, A.L., Hawkes, A.D., Selwood, A.I., Jensen, D.J.,<br />
Cooney, J.M., Beuzenberg, V., MacKenzie, L., 2006.<br />
Identification of 45-hydroxy-46,47-dinoryessotoxin, 44-oxo-<br />
45,46,47-trinoryessotoxin, and 9-methyl-42,43,44,45,46,47,55hepta-nor-38-en-41-oxoyessotoxin,<br />
and partial<br />
characterization of some minor yessotoxins, from<br />
Protoceratium reticulatum. Toxicon 47 (2), 229–240.<br />
Miller, P.E., Scholin, C.A., 1998. Identification and enumeration of<br />
cultured and wild Pseudo-nitzschia (Bacillariophyceae) using<br />
species-specific LSU rRNA-targeted fluorescent probes and<br />
filter-based whole cell hybridization. Journal of Phycology 34<br />
(2), 371–382.<br />
Mitrovic, S.M., Fernández Amandi, M., MacKenzie, L., Furey, A.,<br />
James, K.J., 2004. Effects of selenium, iron and cobalt<br />
addition to growth and yessotoxin production of the toxic<br />
marine dinoflagellate Protoceratium reticulatum in culture.<br />
Journal of Experimental Marine Biology and Ecology 313 (2),<br />
337–351.<br />
Mitrovic, S.M., Hamilton, B., MacKenzie, L., Furey, A., James, K.J.,<br />
2005. Persistence of yessotoxin under light and dark<br />
conditions. Marine Environmental Research 60 (3), 397–401.<br />
Montojo, U.M., Sakamoto, S., Cayme, M.F., Gatdula, N.C., Furio, E.<br />
F., Relox, J.R.J., Sato, S., Fukuyo, Y., Kodam, M., 2006.<br />
Remarkable difference in accumulation of paralytic shellfish<br />
poisoning toxins among bivalve species exposed to Pyrodinium<br />
bahamense var. compressum bloom in Masinloc bay. Philippines<br />
Toxicon 48 (1), 85–92.<br />
Montoya, N.G., Akselman, R., Franco, J.M., Carreto, J.I., 1996.<br />
Paralytic shellfish toxins and mackerel (Scomber japonicus)<br />
mortality in the Argentine sea. In: Yasumoto, T., Oshima, Y.,<br />
Fukuyo, Y. (Eds.), Harmful and Toxic Algal Blooms.<br />
Intergovernmental Oceanographic Community, UNESCO,<br />
Paris, pp. 417–420.<br />
Moore, S.K., Mantua, N.J., Kellogg, J.P., Newton, J.A., 2008. Local<br />
and large-scale climate forcing of Puget Sound oceanographic<br />
properties on seasonal to interdecadal timescales. Limnology<br />
and Oceanography 53 (5), 1746–1758.<br />
Moore, S.K., Mantua, N.J., Hickey, B.M., Trainer, V.L., 2009. Recent<br />
trends in paralytic shellfish toxins in Puget Sound,<br />
relationships to climate, and capacity for prediction of toxic<br />
events. Harmful Algae 8 (3), 463–477.<br />
Moorthi, S.D., Countway, P.D., Stauffer, B.A., Caron, D.A., 2006.<br />
Use of quantitative real-time PCR to investigate the dynamics<br />
of the red tide dinoflagellate Lingulodinium polyedrum.<br />
Microbial Ecology 52 (1), 136–150.<br />
Murata, M., Shimatani, M., Sugitani, Y., Oshima, Y.,<br />
Yasumoto, T., 1982. Isolation and structural elucidation of<br />
the causative toxin of the diarrhetic shellfish poisoning.<br />
Bulletin of the Japanese Society of Scientific Fisheries 48 (4),<br />
549–552.<br />
Murata, M., Kumagai, M., Lee, J.S., Yasumoto, T., 1987. Isolation<br />
and structure of Yessotoxin, a novel polyether compound<br />
implicated in diarrhetic shellfish poisoning. Tetrahedron<br />
Letters 28 (47), 5869–5872.<br />
Naar, J., Bourdelais, A., Tomas, C., Kubanek, J., Whitney, P.L.,<br />
Flewelling, L., Steidinger, K., Lancaster, J., Baden, D.G., 2002.<br />
A competitive ELISA to detect brevetoxins from Karenia brevis<br />
(formerly Gymnodinium breve) in seawater, shellfish, and<br />
mammalian body fluid. Environmental Health Perspectives<br />
110 (2), 179–185.<br />
Netjstgaard, J.C., Tang, K.W., Steinke, M., Dutz, J., Koski, M.,<br />
Antajan, E., Long, J.D., 2007. Zooplankton grazing on<br />
Phaeocystis: a quantitative review and future challenges.<br />
Biogeochemistry 83 (1–3), 147–172.<br />
Nishitani, G., Yamaguchi, M., Ishikawa, A., Yanagiya, S.,<br />
Mitsuya, T., Imai, M., 2005. Relationships between occurrences<br />
water research 44 (2010) 385–416 413<br />
of toxic Dinophysis species (Dinophyceae) and small<br />
phytoplanktons in Japanese coastal waters. Harmful Algae 4<br />
(4), 755–762.<br />
O’Halloran, C., Silver, M., Holman, T., Scholin, C., 2006.<br />
Heterosigma akashiwo in central California waters. Harmful<br />
Algae 5 (2), 124–132.<br />
Office of Shellfish and Water Protection, 2008. 2007 Annual<br />
Report: Commercial and Recreational Shellfish Areas in<br />
Washington State Olympia. Office of Shellfish and Water<br />
Protection.<br />
Ogino, H., Kumagai, M., Yasumoto, T., 1997. Toxicologic<br />
evaluation of Yessotoxin. Natural Toxins 5 (6), 255–259.<br />
Ono, K., Khan, S., Onoue, Y., 2000. Effects of temperature and light<br />
intensity on the growth and toxicity of Heterosigma akashiwo<br />
(Raphidophyceae). Aquaculture Research 31 (5), 427–433.<br />
Pan, Y., Subba Rao, D.V., Mann, K.H., 1996a. Changes in domoic<br />
acid production and cellular chemical composition of the<br />
toxigenic diatom Pseudo-nitzschia muliseries under<br />
phosphate limitation. Journal of Phycology 32 (3), 371–381.<br />
Pan, Y., Subba Rao, D.V., Mann, K.H., Li, K.W., Harrison, W.G.,<br />
1996b. Effects of silicate limitation on production of domoic<br />
acid, a neurotoxin, by the diatom Pseudo-nitzschia multiseries. I.<br />
Batch culture studies. Marine Ecology Progress Series 131 (1–3),<br />
225–233.<br />
Pan, Y., Bates, S.S., Cembella, A.D., 1998. Environmental stress<br />
and domoic acid production by Pseudo-nitzschia:<br />
a physiological perspective. Natural Toxins 6 (3–4), 127–135.<br />
Pankratz, T., 2008. Red tides close desal plants. Water<br />
Desalination Report 44, 1.<br />
Pankratz, T., 2009. Speaking of red tides. Water Desalination<br />
Report 45, 2.<br />
Park, M.G., Kim, S., Kim, H.S., Myung, G., Kang, Y.G., Yih, W., 2006.<br />
First successful culture of the marine dinoflagellate Dinophysis.<br />
Aquatic Microbial Ecology 45 (2), 101–106.<br />
Paz, B., Riobo, P., Fernandez, A.L., Fraga, S., Franco, J.M., 2004.<br />
Production and release of yessotoxins by the dinoflagellates<br />
Protoceratium reticulatum and Lingulodinium polyedrum in<br />
culture. Toxicon 44 (3), 251–258.<br />
Paz, B., Riobo, P., Souto, M.L., Gil, L.V., Norte, M., Fernandez, J.J.,<br />
Franco, J.M., 2006. Detection and identification of<br />
glycoyessotoxin A in a culture of the dinoflagellate<br />
Protoceratium reticulatum. Toxicon 48 (6), 611–619.<br />
Paz, B., Daranas, A.H., Cruz, P.G., Franco, J.M., Pizarro, G.,<br />
Souto, M.L., Norte, M., Fernández, J.J., 2007. Characterization of<br />
okadaic acid related toxins by liquid chromatography coupled<br />
with mass spectrometry. Toxicon 50 (2), 225–235.<br />
Paz, B., Daranas, A.H., Norte, M., Riobo, P., Franco, J.M.,<br />
Fernández, J.J., 2008. Yessotoxins, a group of marine polyether<br />
toxins: an overview. Marine Drugs 6 (6), 73–102.<br />
Peleka, E.N., Matis, K.A., 2008. Application of flotation as<br />
a pretreatment process during desalination. Desalination 222<br />
(1–3), 1–8.<br />
Perl, T.M., Bedard, L., Kosatsky, T., Hockin, J.C., Todd, E.C.D.,<br />
McNutt, L.A., Remis, R.S., 1990. Amnesic shellfish poisoning:<br />
a new clinical syndrome due to domoic acid. Canadian<br />
Diseases Weekly Report 16 (Suppl 1E), 7–8.<br />
Petry, M., Sanz, M.A., Langlais, C., Bonnelye, V., Durand, J.P.,<br />
Guevara, D., Nardes, W.M., Saemi, C.H., 2007. The El Coloso<br />
(Chile) reverse osmosis plant. Desalination 203 (1–3),<br />
141–152.<br />
Pocklington, R., Milley, J.E., Bates, S.S., Bird, C.J., deFreitas, A.S.W.,<br />
Quilliam, M.A., 1990. Trace determination of domoic acid and<br />
phytoplankton by high-performance liquid chromatography<br />
of the fluorenylmethoxycarbonyl (FMOC) derivative.<br />
International Journal of Environmental Analytical Chemistry<br />
38, 351–358.<br />
Prassopoulou, E., Katikou, P., Georgantelis, D., Kyritsakis, A.,<br />
2009. Detection of okadaic acid and related esters in mussels
414<br />
during diarrhetic shellfish poisoning (DSP) episodes in<br />
Greece using the mouse bioassay, the PP2A inhibition assay<br />
and HPLC with fluorimetric detection. Toxicon 53 (2),<br />
214–227.<br />
Price, D.W., Kizer, K.W., Hansgen, K.H., 1991. California’s paralytic<br />
shellfish poisoning prevention program, 1927–89. Journal of<br />
Shellfish Research 10 (11), 119–145.<br />
Purkerson, S.L., Baden, D.G., Fieber, L.A., 1999. Brevetoxin<br />
modulates neuronal sodium channels in two cell lines derived<br />
from rat brain. Neurotoxicology 20 (6), 909–920.<br />
Quilliam, M., 2003. Chemical methods for domoic acid, the<br />
amnesic shellfish poisoning (ASP) toxin. In: Hallegraeff, G.,<br />
Anderson, D., Cembella, A. (Eds.), Manual on Harmful Marine<br />
Microalgae, Monographs on Oceanographic Methodology.<br />
Intergovernmental Oceanographic Commission (UNESCO),<br />
Paris, pp. 247–266.<br />
Quilliam, M.A., Sim, P.G., McCulloch, A.W., McInnes, A.G., 1989.<br />
High-performance liquid chromatography of domoic acid,<br />
a marine neurotoxin, with application to shellfish and<br />
plankton. International Journal of Environmental Analytical<br />
Chemistry 36 (3), 139–154.<br />
Ramsdell, J.S., Anderson, D.M., Glibert, P.M., 2005. HARRNESS,<br />
Harmful Algal Research and Response: a National<br />
Environmental Science Strategy 2005–2015. Ecological Society<br />
of America, Washington, DC, 96 pp.<br />
Reid, J.L., Roden, G.I., Wyllie, J.G., 1958. Studies of the California<br />
Current system. California Cooperative Oceanic Fisheries<br />
Investigations Progress Report, 1 July 1956–1 January 1958,<br />
pp. 27–56.<br />
Reid, F.M.H., Fuglister, E., Jordan, J.B., 1970. The ecology of the<br />
plankton off La Jolla, California, in the period April through<br />
September, 1967. Part 5: phytoplankton taxonomy and<br />
standing crop. Bulletin of the Scripps Institute of<br />
Oceanography Techical Series 17, 51–66.<br />
Reid, F.M.H., Lange, C.B., White, M.M., 1985. Microplankton<br />
species assemblages at Scripps Pier from March to November<br />
1983 during the 1982–1983 El Niño Event. Botanica Marina 28<br />
(10), 443–452.<br />
Reiss, C.R., Robert, C., Owen, C., Taylor, J.S., 2006. Control of MIB,<br />
geosmin and TON by membrane systems. Journal of Water<br />
Supply: Research and Technology – AQUA 55 (2), 95–108.<br />
Rhodes, L., White, D., Syhre, M., Atkinson, M., 1996. Pseudonitzschia<br />
species isolated from New Zealand coastal waters:<br />
domoic acid production in vitro and links with shellfish<br />
toxicity. In: Yasumoto, T., Oshima, Y., Fukuyo, Y. (Eds.),<br />
Harmful and Toxic Algal Blooms. IOC of UNESCO, Paris<br />
pp. 155–158.<br />
Rhodes, L., Scholin, C., Garthwaite, I., Haywood, A., Thomas, A.,<br />
1998. Domoic acid producing Pseudo-nitzschia species deduced<br />
with whole cell DNA probe-based and immunochemical<br />
assays. In: Reguera, B., Blanco, J., Fernández, M.L., Wyatt, T.<br />
(Eds.), Harmful Algae. Xunta de Galicia and<br />
Intergovernmental Oceanographie Commission of UNESCO,<br />
pp. 274–277.<br />
Rhodes, L., McNabb, P., de Salas, M., Briggs, L., Beuzenberg, V.,<br />
Gladstone, M., 2006. Yessotoxin production by Gonyaulax<br />
spinifera. Harmful Algae 5 (2), 148–155.<br />
Rines, J.E.B., Donaghay, P.L., Dekshenieks, M.M., Sullivan, J.M.,<br />
Twardowski, M.S., 2002. Thin layers and camouflage: hidden<br />
Pseudo-nitzschia spp. (Bacillariophyceae) populations in a fjord<br />
in the San Juan Island, Washington, USA. Marine Ecology<br />
Progress Series 225, 123–137.<br />
Riobo, P., Paz, B., Fernandez, M., Fraga, S., Franco, J.M., 2002.<br />
Lipophylic toxins of different strains of Ostreopsidaceae and<br />
Gonyaulacaceae. In: Tenth International Conference on<br />
Harmful Algal Blooms, St. Petersburg, FL (abstract).<br />
Rogers, R.S., Rapoport, H., 1980. The pKas of saxitoxin. Journal of<br />
American Chemical Society 102 (24), 7335–7339.<br />
water research 44 (2010) 385–416<br />
Ronzitti, G., Callegari, F., Malaguti, C., Rossini, G.P., 2004. Selective<br />
disruption of the E-cadherin–catenin system by an algal toxin.<br />
British Journal of Cancer 90 (5), 1100–1107.<br />
Rue, E.L., Bruland, K.W., 2001. Domoic acid binds iron and copper:<br />
a possible role for the toxin produced by the marine diatom<br />
Pseudo-nitzschia. Marine Chemistry 76 (1–2), 127–134.<br />
Samdal, I.A., Naustvoll, L.J., Olseng, C.D., Briggs, L.R., Miles, C.O.,<br />
2004. Use of ELISA to identify Protoceratium reticulatum as<br />
a source of yessotoxin in Norway. Toxicon 44 (1),<br />
75–82.<br />
Samdal, I.A., Aesen, J.A.G., Briggs, L.R., Dahl, E., Miles, C.O., 2005.<br />
Comparison of ELISA and LC–MS analyses for yessotoxins in<br />
blue mussels (Mytilus edulis). Toxicon 46, 7–15.<br />
Satake, M., Mackenzie, L., Yasumoto, T., 1997. Identification of<br />
Protoceratium reticulatum as the biogenetic origin of yessotoxin.<br />
Natural Toxins 5 (4), 164–167.<br />
Satake, M., Ichimura, T., Sekiguchi, K., Yoshimatsu, S., Oshima, Y.,<br />
1999. Confirmation of yessotoxin and 45, 46, 47trinoryessotoxin<br />
production by Protoceratium reticulatum<br />
collected in Japan. Natural Toxins 7 (4), 147–150.<br />
Schantz, E.J., Mold, J.D., Stanger, D.W., Shavel, J., Riel, F.J.,<br />
Bowden, J.P., Lynch, J.M., Wyler, R.S., Riegel, B., Sommer, H.,<br />
1957. Paralytic shellfish poison. VI. A Procedure for the<br />
isolation and purification of the poison from toxic clam and<br />
mussel tissues. Journal of American Chemical Society 79 (19),<br />
5230–5235.<br />
Schnetzer, A., Miller, P.E., Schaffner, R.A., Stauffer, B.A., Jones, B.H.,<br />
Weisberg, S.B., DiGiacomo, P.M., Berelson, W.M., Caron, D.A.,<br />
2007. Blooms of Pseudo-nitzschia and domoic acid in the San<br />
Pedro Channel and Los Angeles harbor areas of the Southern<br />
California Bight, 2003-2004. Harmful Algae 6 (3), 372–387.<br />
Schofield, O., Kohut, J., Aragon, D., Creed, L., Graver, J.,<br />
Haldeman, C., Kerfoot, J., Roarty, H., Jones, C., Webb, D.,<br />
Glenn, S., 2007. Slocum gliders: robust and ready. Journal of<br />
Field Robotics 24 (6), 473–485.<br />
Scholin, C.A., Herzog, M., Sogin, M., Anderson, D.M., 1994.<br />
Identification of group-and strain-specific genetic markers for<br />
globally distributed Alexandrium (Dinophyceae). Part II.<br />
Sequence analysis of a fragment of the LSU rRNA gene. Journal<br />
of Phycology 30 (6), 999–1011.<br />
Scholin, C.A., Gulland, F., Doucette, G.J., Benson, S., Busman, M.,<br />
Chavez, F.P., Cordaro, J., DeLong, R., De Vogelaere, A.,<br />
Harvey, J., Haulena, M., Lefebvre, K.A., Lipscomb, T.,<br />
Loscutoff, S., Lowenstine, L.J., Marin 3rd, R., Miller, P.E.,<br />
McLellan, W.A., Moeller, P.D., Powell, C.L., Rowles, T.,<br />
Silvagni, P., Silver, M., Spraker, T., Trainer, V.L., Van<br />
Dolah, F.M., 2000. Mortality of sea lions along the central<br />
California coast linked to a toxic diatom bloom. Nature 403<br />
(6765), 80–84.<br />
Separation Processes Inc., 2005. Investigation of Microfiltration<br />
and Reverse Osmosis Seawater Desalination. National Water<br />
Research Institute.<br />
Shimizu, Y., Hsu, C.P., Genenah, A.A., 1981. Structure of saxitoxin<br />
in solutions and stereochemistry of dihydrosaxitoxins. Journal<br />
of American Chemical Society 103, 605–609.<br />
Shipe, R.F., Leinweber, A., Gruber, N., 2008. Abiotic controls of<br />
potentially harmful algal blooms in Santa Monica Bay,<br />
California. Continental Shelf Research 28 (18), 2584–2593.<br />
Shumway, S.E., Allen, S.M., Boersma, P.D., 2003. Marine birds and<br />
harmful algal blooms: sporadic victims or under-reported<br />
events? Harmful Algae 2 (1), 1–17.<br />
Smayda, T.J., 1990. Novel and nuisance phytoplankton blooms in<br />
the sea: evidence for a global epidemic. In: Granéli, E.,<br />
Gundström, B., Edler, L., Anderson, D.M. (Eds.), Toxic Marine<br />
Phytoplankton. Elsevier, New York, pp. 29–40.<br />
Smayda, T.J., 2008. Complexity in the eutrophication–harmful<br />
algal bloom relationship, with comment on the importance of<br />
grazing. Harmful Algae 8 (1), 140–151.
Smith, G., Ladizinsky, N., Miller, P., 2001. Amino acid profiles in<br />
species and strains of Pseudo-nitzschia from Monterey Bay,<br />
California: insights into the metabolic role(s) of domoic acid.<br />
In: Hallegraef, G., Blackburn, S., Bolch, C., Lewis, R. (Eds.). IOC,<br />
UNESCO, Hobart, Tasmania, pp. 324–327.<br />
Sommer, H., Meyer, K.F., 1937. Paralytic shellfish poisoning.<br />
Archives of Pathology 24, 560–598.<br />
Steidinger, K.A., 1993. Some taxonomic and biologic aspects of<br />
toxic dinoflagellates. In: Falconer, I. (Ed.), Algal Toxins in<br />
Seafood and Drinking Water. Academic Press, London, pp. 1–28.<br />
Stobo, L.A., Lewis, J., Quilliam, M.A., Hardstaff, W.R., Gallacher, S.,<br />
Webster, L., Smith, E.A., McKenzie, M., 2003. Detection of<br />
yessotoxin in UK and Canadian isolates of phytoplankton and<br />
optimization and validation of LC–MS methods. Canadian<br />
Technical Report of Fisheries and Aquatic Sciences 2498, 8–14.<br />
Strom, S., Wolfe, G.V., Slajer, A., Lambert, S., Clough, J., 2003.<br />
Chemical defenses in the microplankton II: inhibition of<br />
protist feeding by B-dimethylsulfoniopropionate (DMSP).<br />
Limnology and Oceanography 48 (1), 230–237.<br />
Su, Z., Sheets, M., Ishida, H., Li, F., Barry, W.H., 2004. Saxitoxin<br />
blocks L-type ICa. Journal of Pharmacology and Experimental<br />
Therapeutics 308 (1), 324–329.<br />
Subba Rao, D.V., Quilliam, M.A., Pocklington, R., 1988. Domoic<br />
acid – a neurotoxic amino acid produced by the marine diatom<br />
Nitzschia pungens in culture. Canadian Journal of Fisheries and<br />
Aquatic Sciences 45 (12), 2076–2079.<br />
Suganuma, M., Fujiki, H., Suguri, H., Yoshizawa, S., Hirota, M.,<br />
Nakayasu, M., Ojika, M., Wakamatsu, K., Yamada, K.,<br />
Sugimura, T., 1988. Okadaic acid: an additional non-phorbol-<br />
12-tetradecanoate-13-acetate-type tumor promoter.<br />
Proceedings of the National Academy of Sciences, United<br />
States of America 85 (6), 1768–1771.<br />
Sukhatme, G.A.D., Zhang, B., Oberg, C., Stauffer, B.A., Caron, D.A.,<br />
2007. The design and development of a wireless robotic<br />
networked aquatic microbial observing system.<br />
Environmental Engineering Science 24, 205–215.<br />
Sunda, W., 2006. Trace metals and harmful algal blooms. In:<br />
Granéli, E., Turner, J.T. (Eds.), Ecology of Harmful Algae.<br />
Springer, Berlin, pp. 203–214.<br />
Sutherland, C.M., 2008. Diarrhetic Shellfish Toxins Linked to Local<br />
Dinophysis Populations in the California Coastal Waters of<br />
Monterey Bay. Dissertation, University of California, Santa<br />
Cruz, Santa Cruz.<br />
Tachibana, K., Scheuer, P.J., Tsukitani, Y., Kikuchi, H., Engen, D.V.,<br />
Clardy, J., Gopichand, Y., Schmitz, F.J., 1981. Okadaic acid,<br />
a cytotoxic polyether from two marine sponges of the genus<br />
Halichondria. Journal of American Chemical Society 103 (9),<br />
2469–2471.<br />
Takahashi, H., Kusumi, T., Kan, Y., Satake, M., Yasumoto, T., 1996.<br />
Determination of the absolute configuration of yessotoxin,<br />
a polyether compound implicated in diarrhetic shellfish<br />
poisoning, by NMR spectroscopic method using a chiral<br />
anisotropic reagent, methoxy-(2-naphthyl) acetic acid.<br />
Tetrahedron Letters 37 (39), 7087–7090.<br />
Takai, A., Bialojan, C., Troschka, M., Rüegg, J.C., 1987. Smooth<br />
muscle myosin phosphatase inhibition and force<br />
enhancement by black sponge toxin. FEBS Letters 217 (1),<br />
81–84.<br />
Taylor, F.J.R., Horner, R.A., 1994. Red tides and other problems<br />
with harmful algal blooms in Pacific Northwest coastal waters.<br />
In: Review of the Marine Environment and Biota of Strait of<br />
Georgia, Puget Sound and Juan de Fuca Strait. Canadian<br />
Fisheries and Aquatic Science Technical Report 1948,<br />
pp. 175–186.<br />
Tenzer, B., Adin, A., Priel, M., 1999. Seawater filtration for fouling<br />
prevention under stormy conditions. Desalination 125 (1–3),<br />
77–88.<br />
water research 44 (2010) 385–416 415<br />
Terao, K., Ito, E., Oarada, M., Murata, M., Yasumoto, T., 1990.<br />
Histopathological studies on experimental marine toxin<br />
poisoning: the effects in mice of yessotoxin isolated from<br />
Patinopecten yessoensis and of a desulfated derivative. Toxicon<br />
28 (9), 1095–1104.<br />
Thomas, W.H., Zavoico, T., Hewes, C., 2001. Historical<br />
Phytoplankton Species Time-Series Data (1917–1939) from the<br />
North American Pacific Coast. Scripps Institution of<br />
Oceanography. Ref. no. 01-12.<br />
Throndsen, J., 1997. The plankton marine flagellates. In:<br />
Tomas, C.R. (Ed.), Identifying marine phytoplankton.<br />
Academic Press, San Diego, pp. 591–729.<br />
Trainer, V.L., 2002. Harmful algal blooms on the U.S. west coast.<br />
In: Taylor, F.J., Trainer, V.L. (Eds.), Harmful Algal Blooms in the<br />
PICES Region of the North Pacific, pp. 89–118. PICES Scientific<br />
Report 23.<br />
Trainer, V.L., Adams, N.G., Bill, B.D., Anulacion, F., Wekell, J.C.,<br />
1998. Concentration and dispersal of a Pseudo-nitzschia bloom<br />
in Penn Cove, Washington, USA. Natural Toxins 6 (3–4),<br />
113–126.<br />
Trainer, V.L., Adams, N.G., Bill, B.D., Stehr, C.M., Wekell, J.C.,<br />
Moeller, P., Busman, M., Woodruff, D., 2000. Domoic acid<br />
production near California coastal upwelling zones, June 1998.<br />
Limnology and Oceanography 45 (8), 1818–1833.<br />
Trainer, V.L., Adams, N.G., Wekell, J.C., 2001. Domoic acid<br />
producing Pseudo-nitzschia species off the U.S. west coast<br />
associated with toxification events. In: Hallegraeff, G.M.,<br />
Blackburn, S.I., Bolch, C.J., Lewis, R.J. (Eds.), Harmful Algal<br />
Blooms 2000. Intergovernmental Oceanographic Commission<br />
of UNESCO, Paris, pp. 46–49.<br />
Trainer, V.L., Hickey, B.M., Horner, R.A., 2002. Biological and<br />
physical dynamics of domoic acid production off the<br />
Washington Coast. Limnology and Oceanography 47 (5),<br />
1438–1446.<br />
Trainer, V.L., Eberhart, B.T.L., Wekell, J.C., Adams, N.G.,<br />
Hanson, L., Cox, F., Dowell, J., 2003. Paralytic shellfish toxins in<br />
Puget Sound, Washington State. Journal of Shellfish Research<br />
22 (1), 213–223.<br />
Trainer, V.L., Cochlan, W.P., Erickson, A., Bill, B.D., Cox, F.H.,<br />
Borchert, J.A., Lefebvre, K., 2007. Recent domoic acid closures<br />
of shellfish harvest areas in Washington State inland<br />
waterways. Harmful Algae 6 (3), 449–459.<br />
Tubaro, A., Sidari, L., Della Loggia, R., Yasumoto, T., 1998.<br />
Occurrence of yessotoxin in phytoplankton and mussels from<br />
Northern Adriatic Sea. In: Proceedings of the VIII International<br />
Conference on Harmful Algae, Vigo, Spain (abstract).<br />
Twiner, M.J., Dechraoui, M.Y.B., Wang, Z., Mikulski, C.M., Henry, M.<br />
S., Pierce, R.H., Doucette, G.J., 2007. Extraction and analysis of<br />
lipophilic brevetoxins from the red tide dinoflagellate Karenia<br />
brevis. Analytical Biochemistry 369 (1), 128–135.<br />
Tyrell, J.V., Connell, L.B., Scholin, C.A., 2002. Monitoring for<br />
Heterosigma akashiwo using a sandwich hybridization assay.<br />
Harmful Algae 1 (2), 205–214.<br />
Van Dolah, F.M., Finley, E.L., Haynes, B.L., Doucette, G.J.,<br />
Moeller, P.D., Ramsdell, J.S., 1994. Development of rapid and<br />
sensitive high throughput pharmacologic assays for marine<br />
phycotoxins. Natural Toxins 2 (4), 189–196.<br />
Vigilant, V.L., Silver, M., 2007. Domoic acid in benthic flatfish on<br />
the continental shelf of Monterey Bay, California, USA. Marine<br />
Biology 151 (6), 2053–2062.<br />
Walz, P.M., Garrison, D.L., Graham, W.M., Cattey, M.A.,<br />
Tjeerdema, R.S., Silver, M.W., 1994. Domoic acid-producing<br />
diatom blooms in Monterey Bay, California: 1991–1993.<br />
Natural Toxins 2 (5), 271–279.<br />
Wang, J., Salata, J.J., Bennett, P.B., 2003. Saxitoxin is a gating<br />
modifier of hERG Kþ channels. Journal of General Physiology<br />
121 (6), 583–598.
416<br />
Watson, M., 1997. Control of Membrane Fouling Due to Algae.<br />
Applied Research Department. Technology Transfer Note.<br />
No.12.<br />
Wekell, J.C., Gauglitz, E.J.J., Barnett, H.J., Hatfield, C.L., Simons, D.,<br />
Ayres, D., 1994. Occurrence of domoic acid in Washington<br />
state razor clams (Siliqua patula) during 1991–1993. Natural<br />
Toxins 2 (4), 197–205.<br />
Wells, M.L., Trick, C.G., Cochlan, W.P., Hughes, M.P., Trainer, V.L.,<br />
2005. Domoic acid: the synergy of iron, copper, and the<br />
toxicity of diatoms. Limnology and Oceanography 50 (6),<br />
1908–1917.<br />
Wilf, M., Schierach, M.K., 2001. Improved performance and cost<br />
reduction of RO seawater systems using UF pretreatment.<br />
Desalination 135 (1–3), 61–68.<br />
Wong, J.L., Oesterlin, R., Rapoport, H., 1971. Struture of<br />
saxitoxin. Journal of American Chemical Society 93 (26),<br />
7344–7345.<br />
Wright, J.L.C., Cembella, A.D., 1998. Ecophysiology and<br />
biosynthesis of polyether marine biotoxins. In: Anderson, D.M.,<br />
Cembella, A.D., Hallegraeff, G.M. (Eds.), The Physiological<br />
water research 44 (2010) 385–416<br />
Ecology of Harmful Algal Blooms. Springer-Verlag, Heidelberg,<br />
pp. 427–451.<br />
Wright, J.L.C., Bird, C., de Freitas, A., Hampson, D., McDonald, J.,<br />
Quilliam, M., 1990. Chemistry, biology, and toxicology of<br />
domoic acid and its isomers. Canadian Diseases Weekly<br />
Report 16 (Suppl. 1E), 21–26.<br />
Yasumoto, T., Murata, M., 1993. Marine toxins. Chemical Reviews<br />
93, 1867–1909.<br />
Yasumoto, T., Takizawa, A., 1997. Fluorometric measurement of<br />
yessotoxins in shellfish by high-pressure liquid<br />
chromatography. Bioscience, Biotechnology and Biochemistry<br />
61 (10), 1775–1777.<br />
Yasumoto, T., Oshima, Y., Sugawara, W., Fukuyo, Y., Oguri, H.,<br />
Igarashi, T., Fujita, N., 1980. Identification of Dinophysis fortii as<br />
the causative organism of diarrhetic shellfish poisoning.<br />
Bulletin of the Japanese Society of Scientific Fisheries 46 (11),<br />
1405–1411.<br />
Yasumoto, T., Murata, M., Oshima, Y., Sano, M., Matsumoto, G.K.,<br />
Clardy, J., 1985. Diarrhetic shellfish toxins. Tetrahedron 41 (6),<br />
1019–1025.
Occurrence and removal of pharmaceuticals, caffeine and<br />
DEET in wastewater treatment plants of Beijing, China<br />
Qian Sui, Jun Huang, Shubo Deng, Gang Yu*, Qing Fan<br />
POPs Research Centre, Department of Environmental Science & Engineering, Tsinghua University, Beijing 10084, China<br />
article info<br />
Article history:<br />
Received 14 January 2009<br />
Received in revised form<br />
5 July 2009<br />
Accepted 8 July 2009<br />
Available online 15 July 2009<br />
Keywords:<br />
Pharmaceuticals<br />
Wastewater<br />
Removal efficiency<br />
Advanced treatment<br />
Risk assessment<br />
China<br />
1. Introduction<br />
abstract<br />
With the progress of sensitive analytical techniques, the<br />
frequent detection of various pharmaceuticals in the aquatic<br />
environment has received global concerns of both the<br />
academic community and the public (Daughton and Ternes,<br />
1999; Jones et al., 2005). After intake by humans or animals,<br />
the pharmaceuticals will be partially converted to metabolites,<br />
however, partially excreted unchanged or as conjugates,<br />
and finally delivered to the wastewater treatment plants<br />
(WWTPs). As there is no unit specifically designed to remove<br />
these compounds, the elimination by most WWTPs seems to<br />
be inefficient (Ternes, 1998; Castiglioni et al., 2006; Lishman<br />
et al., 2006; Nakada et al., 2006; Santos et al., 2007; Vieno et al.,<br />
2007b; Xu et al., 2007; Gulkowska et al., 2008; Paxeus, 2004).<br />
Together with treated wastewater, these compounds are<br />
water research 44 (2010) 417–426<br />
Available at www.sciencedirect.com<br />
journal homepage: www.elsevier.com/locate/watres<br />
* Corresponding author. Tel.: þ86 10 62787137; fax: þ86 10 62794006.<br />
E-mail address: yg-den@tsinghua.edu.cn (G. Yu).<br />
0043-1354/$ – see front matter ª 2009 Elsevier Ltd. All rights reserved.<br />
doi:10.1016/j.watres.2009.07.010<br />
The occurrence and removal of 13 pharmaceuticals and 2 consumer products, including<br />
antibiotic, antilipidemic, anti-inflammatory, anti-hypertensive, anticonvulsant, stimulant,<br />
insect repellent and antipsychotic, were investigated in four wastewater treatment plants<br />
(WWTPs) of Beijing, China. The compounds were extracted from wastewater samples by<br />
solid-phase extraction (SPE) and analyzed by ultra-performance liquid chromatography<br />
coupled with tandem mass spectrometry (UPLC–MS/MS). Most of the target compounds were<br />
detected, with the concentrations of 4.4 ng L 1 –6.6 mgL 1 and 2.2–320 ng L 1 in the influents<br />
and secondary effluents, respectively. These concentrations were consistent with their<br />
consumptions in China, and much lower than those reported in the USA and Europe. Most<br />
compounds were hardly removed in the primary treatment, while their removal rates ranging<br />
from 12% to 100% were achieved during the secondary treatment. In the tertiary treatment,<br />
different processes showed discrepant performances. The target compounds could not be<br />
eliminated by sand filtration, but the ozonation and microfiltration/reverse osmosis (MF/RO)<br />
processes employed in two WWTPs were very effective to remove them, showing their main<br />
contributions to the removal of such micro-pollutants in wastewater treatment.<br />
ª 2009 Elsevier Ltd. All rights reserved.<br />
released to the aquatic environment, and consequently found<br />
to contaminate the receiving water bodies (Lindqvist et al.,<br />
2005; Kasprzyk-Hordern et al., 2009), or even raw water sources<br />
of drinking water treatment plant (Ternes et al., 2002;<br />
Vieno et al., 2007a; Radjenovic et al., 2008). Meanwhile, results<br />
of toxicology studies have revealed that some pharmaceuticals<br />
are suspected to have direct toxicity to certain aquatic<br />
organisms (Ferrari et al., 2003; Jjemba, 2006; Grung et al., 2008;<br />
Quinn et al., 2008). Besides, their continual but undetectable<br />
effects could accumulate slowly, and finally lead to irreversible<br />
change on wildlife and human beings (Daughton and<br />
Ternes, 1999). Therefore, the occurrence and behavior of<br />
pharmaceuticals in the WWTPs, which are both the sink and<br />
source of the compounds, should be focused on. So far,<br />
concentrations of pharmaceuticals from various therapeutic<br />
classes in the WWTPs have been well documented in the
418<br />
North America (Thomas and Foster, 2005; Lishman et al.,<br />
2006), Japan (Nakada et al., 2006) and some European countries<br />
(Ternes, 1998; Castiglioni et al., 2006; Santos et al., 2007; Vieno<br />
et al., 2007b; Jones et al., 2007; Paxeus, 2004). Reported species<br />
and concentrations of pharmaceuticals varied from country to<br />
country, and plant to plant, owing to the different usage<br />
patterns. Meanwhile, the removal efficiencies of pharmaceuticals<br />
also varied much (Nakada et al., 2006; Gulkowska et al.,<br />
2008), indicating that the removal could be affected by both<br />
the compound-specific properties, and the factors concerning<br />
specific WWTPs, such as types of treatment processes, solids<br />
retention time (SRT), hydraulic retention time (HRT),<br />
temperature, etc. In recent years, very few studies about the<br />
situation in China have been reported. Only one specific<br />
therapeutic class, antibiotics, has been investigated by limited<br />
previous studies (Xu et al., 2007; Gulkowska et al., 2008; Chen<br />
et al., 2008). Therefore, it is necessary and important to<br />
investigate the occurrence and removal of pharmaceuticals<br />
from different therapeutic classes in the WWTPs of China.<br />
Due to the low efficiency of conventional wastewater<br />
treatment processes, some advanced treatment technologies<br />
have been evaluated. Ozonation was found to be effective to<br />
remove pharmaceuticals in real municipal WWTPs of Japan<br />
(Nakada et al., 2007; Okuda et al., 2008) and Germany (Ternes<br />
et al., 2003). Nanofiltration (NF) and reverse osmosis (RO)<br />
membrane filtration, the well-proven technologies to remove<br />
pharmaceuticals from different kinds of waters, have also been<br />
applied at bench, pilot and full scale (Khan et al., 2004; Nghiem<br />
et al., 2005; Drewes et al., 2005; Al-Rifai et al., 2007; Watkinson<br />
et al., 2007; Comerton et al., 2008; Radjenovic et al., 2008).<br />
Retention behavior of pharmaceuticals during the processes<br />
associated with physicochemical properties of pharmaceuticals,<br />
membranes as well as the solution chemistry, and<br />
mechanisms of pharmaceutical rejection have been discussed<br />
in Kimura et al. (2004), Nghiem et al. (2005), Nghiem and<br />
Coleman (2008) and Comerton et al. (2008). Recently, considering<br />
the requirement of reclaimed water, several advanced<br />
treatment facilities have been installed in the WWTPs of<br />
Beijing. However, the removal efficiency of micro-pollutants,<br />
such as pharmaceuticals, has not been evaluated yet.<br />
In the present study, we investigated the contamination<br />
levels of 13 pharmaceuticals and 2 consumer products from 8<br />
classes (i.e. antibiotic, antilipidemic, anti-inflammatory, antihypertensive,<br />
anticonvulsant, stimulant, insect repellent and<br />
antipsychotic) in four WWTPs of Beijing, China, which have<br />
different advanced treatment units, and evaluated the elimination<br />
efficiencies of the target pharmaceuticals. To the best<br />
of our knowledge, this is the first report on the occurrence and<br />
removal of pharmaceuticals and consumer products from<br />
multiple classes in the WWTPs of China, especially for the<br />
situation during the advanced treatment processes.<br />
2. Materials and methods<br />
2.1. Chemicals<br />
All the standards including chloramphenicol (CP), nalidixic<br />
acid (NA), trimethoprim (TP), bezafibrate (BF), clofibric acid<br />
(CA), gemfibrozil (GF), diclofenac (DF), indometacin (IM),<br />
water research 44 (2010) 417–426<br />
ketoprofen (KP), mefenamic acid (MA), metoprolol (MTP),<br />
carbamazepine (CBZ), caffeine (CF), N,N-diethyl-meta-toluamide<br />
(DEET) and sulpiride (SP) (Appendix) were of analytical<br />
grade (>90%), and purchased from Sigma–Aldrich (Steinheim,<br />
Germany). Isotopically labeled compounds, used as internal<br />
standards, were 13 C-phenacetin obtained from Sigma–<br />
Aldrich, and 3 D-mecoprop from Dr. Ehrenstorfer (Augsburg,<br />
Germany). HPLC grade methanol, acetone, dichloromethane,<br />
hexane, as well as formic acid were provided by Dikma (USA),<br />
and ultra-pure water was produced by a Milli-Q unit (Millipore,<br />
USA). Stock solutions of individual compound were prepared in<br />
methanol and mixture standards with different concentrations<br />
were prepared by diluting the stock solutions before each<br />
analytical run. All the solutions were stored at 4 C in the dark.<br />
2.2. Sample collection<br />
Four full-scale municipal WWTPs, referred as A, B, C and D,<br />
were selected in our study. These WWTPs employ similar<br />
conventional treatment processes: primary treatment to<br />
remove particles coupled with secondary biological treatment.<br />
For the secondary biological treatment processes,<br />
WWTPs A and D employ anaerobic/anoxic/oxic (A 2 /O) activated<br />
sludge process, anoxic/oxic (A/O) activated sludge<br />
process is adopted in WWTP B, and WWTP C employs oxidation<br />
ditch (OD). Other detailed information on each WWTP,<br />
such as inhabitants served, daily flow, HRT and SRT are shown<br />
in Table 1. Part of the secondary effluents was further treated<br />
in WWTPs A, B and D, by the processes of ultrafiltration<br />
(UF)/ozone, sand filtration (SF) and microfiltration/reverse<br />
osmosis (MF/RO), respectively. In WWTP A, a dead-end<br />
ultrafiltration system (Zenon GE) is used. The whole system<br />
has 6 trains of Zee-Weed 1000 membrane. Each train contains<br />
9 cassettes of 57–60 modules per cassette. The membrane,<br />
with the pore size of 0.02 mm, is made by PVDF. The module is<br />
operated in an outside/in configuration at a constant flow of<br />
23 L (m 2 h) 1 and the total treatment capacity reaches<br />
80,000 m 3 d 1 . The membrane is hydraulically backwashed at<br />
a constant flow rate of 34 (m 2 h) 1 , and 29 times per day. The<br />
backwash phase lasts for 1 min. Maintenance cleaning is<br />
conducted once per day. Membranes are soaked in the sodium<br />
hypochlorite solution (50 mg L 1 ) for 25 min. For the ozonation<br />
process, gaseous ozone is generated from an ozone generator<br />
(Mitsubishi Electric). The ozone dosage and contact time in the<br />
reaction tank is 5 mg L 1 and 15 min, respectively. The pH of<br />
the wastewater before ozonation ranges 6.5–8.0 and shows no<br />
significant change after ozonation. As the heart of the<br />
advanced treatment in WWTP D, a spiral-wound crossflow<br />
module is employed for the reverse osmosis (RO) membrane<br />
filtration. The RO membrane (Filmtec, DOW) is made from<br />
a thin-film composite polyamide material. Each module is<br />
designed to operate at a water flux of 1.3 m 3 h 1 , and a product<br />
water recovery of 75–80%. The trans-membrane pressure is<br />
between 0.04 and 0.06 MPa, and the salt rejection remained at<br />
the level of 99%. Every 3–6 months, normally when the transmembrane<br />
pressure reaches above 0.06 MPa, the membrane is<br />
cleaned with 0.1%(w) sodium hydroxide solution (for organic<br />
foulants), 2%(w) citric acid (for inorganic foulants) and 0.5%(w)<br />
formaldehyde (as biocide). Schematic diagram of treatment<br />
processes in the four WWTPs is shown in Fig. 1.
The samples were collected once from the four WWTPs<br />
during June and July 2008, with no compensation for HRT. All<br />
of them were collected as grab samples in duplicate (500 mL<br />
for influents and 1000 mL for the others) in prewashed amber<br />
glass bottles, kept in the cooler and transported to the laboratory.<br />
Immediately after delivery to the laboratory, they were<br />
filtered through prebaked (400 C, >4 h) glass microfiber filters<br />
(GF/F, Whatman) to remove particles and stored at 4 C before<br />
extraction.<br />
2.3. Sample extraction and analysis<br />
The method for the extraction and analysis of pharmaceuticals<br />
and consumer products is presented elsewhere (Sui et al.,<br />
in press) and briefly described here. After the solid-phase<br />
extraction (SPE) cartridges (Oasis, HLB, 200 mg, 6 mL) were<br />
a WWTP A<br />
Influent<br />
Screen<br />
b WWTP B<br />
Influent<br />
Screen<br />
c WWTP C<br />
Influent<br />
Screen<br />
d WWTP D<br />
Influent<br />
Screen<br />
Grit<br />
Removal<br />
Grit<br />
Removal<br />
Grit<br />
Removal<br />
Grit<br />
Removal<br />
water research 44 (2010) 417–426 419<br />
Table 1 – Information of the WWTPs investigated.<br />
WWTP Inhabitants<br />
served 10 3<br />
Daily flow<br />
( 10 3 m 3 HRT (h) SRT (d) Secondary treatment Tertiary treatment<br />
)<br />
A 814 400 11 12–15 A 2 /O UF/ozone<br />
B 2400 1000 11 20 A/O SF<br />
C 480 200 15 12–16 OD –<br />
D 2415 600 10.67 15 A 2 /O MF/RO<br />
A/O<br />
treatment<br />
Primary<br />
Clarifier<br />
Primary<br />
Clarifier<br />
Oxidation<br />
Ditch<br />
conditioned, wastewater samples, added with internal standards<br />
and adjusted to pH ¼ 7, were introduced to the cartridge<br />
via a PTFE tube, at a flow rate of 5–10 mL min 1 . After washing<br />
by 5 mL of 5.0% methanol solution, the cartridge was dried<br />
under vacuum for 2 h and eluted with 5 mL of methanol. The<br />
extract was then concentrated to 0.4 mL under a gentle<br />
nitrogen stream and stored at 4 C for analysis. Concentrations<br />
of the target compounds were analyzed using ultraperformance<br />
liquid chromatography coupled with tandem<br />
mass spectrometry (UPLC–MS/MS). Analytes were separated<br />
using Waters Acquity UPLC system (Waters Corporation, USA)<br />
equipped with Acquity UPLC BEH C18 column (50 2.1 mm,<br />
particle size of 1.7 mm), and detected by Quattro Premier XE<br />
tandem quadrupole mass spectrometry (Waters Corp., USA)<br />
equipped with an electrospray ionization source. The analysis<br />
was carried out in multiple reaction monitoring (MRM) mode,<br />
Secondary<br />
Clarifier<br />
A 2 /O<br />
treatment<br />
A 2 /O<br />
treatment<br />
Secondary<br />
Clarifier<br />
Secondary<br />
Clarifier<br />
Ultrafiltration<br />
Secondary Effluent<br />
Secondary<br />
Clarifier<br />
Secondary<br />
Effluent<br />
Ozonation<br />
Sand Filtration<br />
Secondary Effluent<br />
MF<br />
`<br />
RO<br />
Secondary Effluent<br />
Tertiary<br />
Effluent<br />
Tertiary<br />
Effluent<br />
Tertiary<br />
Effluent<br />
Fig. 1 – Schematic diagram of the treatment processes in the four WWTPs and sampling site location (C).
420<br />
and in general, two precursor ion/product ion transitions were<br />
monitored for one compound with the purpose of quantification<br />
and confirmation.<br />
2.4. Quality control<br />
For each sampling, 500 mL Milli-Q water in an amber glass<br />
bottle as a field blank was brought to the WWTPs, exposed to<br />
the environment where the samples were taken from, and<br />
then delivered back to the laboratory with samples. For each<br />
set of samples (normally 10 samples), at least one procedural<br />
blank was prepared from ultra-pure water in the laboratory.<br />
Both the field blanks and procedural blanks were run identically<br />
to the wastewater samples, and the concentrations of<br />
target compounds were below the limit of quantification<br />
(LOQ). The absolute recoveries, calculated by comparing the<br />
concentrations of target compounds in spiked and unspiked<br />
wastewaters, were proved to be 73–102% and 50–95% in the<br />
effluent and influent for most compounds, respectively. While<br />
for several compounds (i.e. sulpiride, gemfibrozil, mefenamic<br />
acid), the absolute recoveries were not satisfactory. However,<br />
13 C-phenacetin and 3 D-mecoprop, the surrogate standards<br />
used for positive and negative ion mode respectively were able<br />
to compensate for the loss of most analytes, and relative<br />
recoveries were 67–130% for all the analytes in the effluent<br />
and 79–140% in the influents except mefenamic acid (251%)<br />
and nalidixic acid (178%). Therefore, the concentrations of<br />
these two compounds in the wastewater influents were not<br />
quantitatively determined and reported. The LOQs were<br />
0.3–5.5 ng L 1 and 0.7–20 ng L 1 in the effluent and influent,<br />
respectively. Detailed information about the calibration,<br />
recoveries, LOQ, matrix effects, etc. were described in Sui et al.<br />
(in press), and briefly listed in Table 2. As duplicate samples<br />
were collected at each sampling site, mean concentrations<br />
were adopted. In most cases, deviations of duplicate samples<br />
were less than 20%. For some tertiary effluent samples, low<br />
concentrations of some target compounds (i.e. caffeine, DEET,<br />
carbamazepine) resulted in slightly higher deviations.<br />
3. Result and discussion<br />
3.1. Influents<br />
As shown in Fig. 2, 12 target compounds were detected in all<br />
the influent samples from the four WWTPs, while ketoprofen<br />
was below LOQ in all wastewater samples. The most<br />
abundant compounds detected were the consumer products,<br />
caffeine (3.4–6.6 mgL 1 ) and N,N-diethyl-meta-toluamide<br />
(0.6–1.2 mgL 1 ), probably due to the large consumption of<br />
drinks containing caffeine (i.e. coffee, tea, etc.) and wide<br />
application of insect repellent during the summer time when<br />
we sampled. Diclofenac, trimethoprim, sulpiride, carbamazepine,<br />
indometacin and metoprolol showed relatively high<br />
concentrations (Fig. 2). A similar composition distribution<br />
was observed among all the influents of the four WWTPs.<br />
The concentrations of target pharmaceuticals except<br />
diclofenac and trimethoprim, were much lower than those<br />
reported in the European and North American countries<br />
(Thomas and Foster, 2005; Lishman et al., 2006; Vanderford<br />
and Snyder, 2006; Santos et al., 2007; Gomez et al., 2007; Vieno<br />
et al., 2007b; Huerta-Fontela et al., 2008). For instance, the<br />
concentrations of ketoprofen in the wastewater influents<br />
were recorded to be 2.0 0.6 mgL 1 in Finland (Lindqvist et al.,<br />
2005), 200 ng L 1 in Australia (Al-Rifai et al., 2007), and<br />
300–1360 ng L 1 in Spain (Santos et al., 2007), while in the<br />
influents of four WWTPs in Beijing, it could not be detected.<br />
Concerning gemfibrozil, which is used to lower cholesterol<br />
and triglyceride levels in the blood, the contamination level<br />
found in the present study was 24–140 ng L 1 , even 1 or 2 order<br />
of magnitude lower than those in the USA (4770 ng L 1 ,<br />
Vanderford and Snyder, 2006) and Canada (418 ng L 1 ,<br />
Table 2 – Instrumental quantification limit (IQL), limit of quantification (LOQ), absolute recovery (AR), relative recovery (RR)<br />
and matrix suppression of target compounds.<br />
Compounds IQL (pg) LOQ (ng L 1 ) AR (n¼6, %) RR (n ¼ 6, %) Matrix effect (%)<br />
Effluent Influent Effluent a<br />
Influent a<br />
Effluent a<br />
Influent a<br />
BF 0.5 0.3 0.7 74 (3) 58 (5) 94 (4) 94 (10) 27<br />
CA 10 5.4 16 74 (4) 50 (4) 94 (6) 82 (8) 30<br />
CBZ 2.5 1.0 2.8 100 (4) 71 (12) 130 (6) 133 (23) 6<br />
CF 2.5 1.7 3.3 60 (3) 61 (37) 77 (4) 114 (70) 49<br />
CP 2.5 1.0 2.3 98 (7) 86 (9) 124 (10) 140 (17) 39<br />
DEET 2.5 1.3 3.2 76 (3) 63 (6) 99 (5) 118 (14) 24<br />
DF 10 4.7 9.4 86 (12) 85 (19) 109 (16) 139 (32) 4<br />
GF 10 4.2 20 95 (12) 40 (11) 120 (16) 65 (18) 28<br />
IM 2.5 1.3 2.8 80 (12) 73 (13) 101 (16) 119 (22) 1<br />
KP 10 5.5 18 69 (9) 44 (3) 87 (12) 72 (7) 46<br />
MA 10 5.5 5.2 73 (11) 154 (24) 92 (15) 251 (41) 13<br />
MTP 2.5 1.1 3.3 88 (3) 60 (3) 115 (5) 113 (9) 15<br />
NA 2.5 1.1 2.1 91 (9) 95 (9) 118 (12) 178 (20) 3<br />
SP 0.5 0.4 1.0 51 (5) 42 (3) 67 (7) 79 (7) 63<br />
TP 2.5 1.0 2.7 102 (7) 74 (16) 132 (9) 139 (31) 5<br />
a Value in the brackets refers to the deviation of the recovery.<br />
water research 44 (2010) 417–426
a<br />
Concentration (ng L -1 )<br />
b<br />
Concentration (ng L -1 )<br />
10000<br />
1000<br />
100<br />
10<br />
1<br />
10000<br />
1000<br />
100<br />
10<br />
1<br />
DEET<br />
DEET<br />
CF<br />
CF<br />
CP<br />
TP<br />
BF<br />
CA<br />
Lishman et al., 2006). The low levels of target pharmaceuticals<br />
were probably due to the lower per capita consumption in<br />
China than in the countries with higher socioeconomic<br />
statuses, where medical care is more prevalent (Thomas and<br />
Foster, 2005). The per capita consumption rate of gemfibrozil<br />
in China is estimated to be 0.036 mg person 1 d 1 (Table 3),<br />
lower than those in Germany (0.2 mg person 1 d 1 , Ternes,<br />
1998) and Canada (0.2 mg person 1 d 1 , Lishman et al., 2006).<br />
Since the levels of target pharmaceuticals were somewhat<br />
different from those of European and North American<br />
countries, we theoretically calculate the concentration of<br />
pharmaceuticals in the wastewater influent by the following<br />
equation (Lindqvist et al., 2005; Nakada et al., 2006)<br />
T e% I 1012<br />
Cpred ¼<br />
365 P Q<br />
where C pred is the predicted concentration of the pharmaceutical<br />
in wastewater influent (ng L 1 ); T is the total production<br />
of a pharmaceutical both for human and animal use in<br />
China per year (ton year 1 ), P is the population of China, e% is<br />
the amount of the pharmaceutical excreted unchanged, I is<br />
the number of inhabitants served and Q is the influent flow<br />
(m 3 d 1 ). The predicted concentrations of gemfibrozil, diclofenac,<br />
indometacin, ketoprofen, carbamazepine, and sulpiride<br />
were comparable to those measured in the influents<br />
(Table 3). Much lower measured concentration than predicted<br />
concentration of chloramphenicol was probably because it<br />
had been forbidden for use in food and aquaculture in China<br />
since 2005, and the available data about the production of<br />
pharmaceuticals were based on the year of 2004. It should be<br />
noticed that since there is no available data on the total<br />
consumption of any pharmaceutical, we used figures for total<br />
production of individual pharmaceutical instead. Therefore,<br />
the differences between the amounts actually produced and<br />
applied as well as the amount used in human and veterinary<br />
medicine could not be distinguished, which might result in<br />
overestimation of the theoretical concentration. Nevertheless,<br />
the comparability between the predicted concentrations<br />
and measured concentrations illustrates the overall reasonability<br />
of the approach.<br />
Table 3 – Outputs, per capita consumption, predicted concentrations (PECs) and measured concentrations (MECs) of some<br />
pharmaceuticals in the wastewater influents of WWTPs investigated.<br />
Compound Output a<br />
(tons year 1 )<br />
GF<br />
DF<br />
IM<br />
MTP<br />
Pharmaceuticals & Consumer Products<br />
CP<br />
TP<br />
BF<br />
CA<br />
GF<br />
DF<br />
IM<br />
MA<br />
MTP<br />
Influents<br />
CBZ<br />
CBZ<br />
SP<br />
Secondary effluents<br />
Pharmaceuticals & Consumer Products<br />
Fig. 2 – Concentrations of target pharmaceuticals in<br />
wastewater influents (a) and secondary effluents (b) of four<br />
WWTPs in Beijing.<br />
water research 44 (2010) 417–426 421<br />
SP<br />
Per capita<br />
consumption<br />
(mg person 1 d 1 )<br />
Excreted<br />
unchanged b (%)<br />
PEC<br />
(ng L 1 )<br />
(1)<br />
MEC<br />
(mean, ng L 1 )<br />
CP 1929 4.051 5–10 844 31<br />
TP 2352 4.939 45 6173 400<br />
GF 17 0.036 76 76 60<br />
DF 328 0.689 15 287 318<br />
IM 277 0.582 10–20 242 129<br />
KP 92 0.193 2.7 15 n.d. c<br />
CBZ 395 0.830 2–3 58 113<br />
SP 98 0.206 15–70 243 157<br />
a From CMEIN (2005).<br />
b From Bolton and Null (1981), Ternes (1998), Khan and Ongerth (2004), Niwa et al. (2005), Nakada et al. (2006), Jjemba (2006).<br />
c n.d. ¼ Not detected.
422<br />
3.2. Secondary effluent<br />
Similar to the influent samples, ketoprofen was below the LOQ<br />
in all the secondary effluent samples. Nalidixic acid and<br />
chloramphenicol were detected only in one WWTP, with the<br />
concentration of 8.1 and 19 ng L 1 , respectively. The mean<br />
concentrations of the other 12 compounds ranged from 5 to<br />
200 ng L 1 (Fig. 2). Diclofenac, N,N-diethyl-meta-toluamide,<br />
trimethoprim, sulpiride and indometacin showed high<br />
concentrations in the secondary effluents. Carbamazepine<br />
and metoprolol followed, with the concentrations ranging<br />
from 69 to 120 ng L 1 , and 60 to 108 ng L 1 , respectively. Other<br />
compounds, such as caffeine, gemfibrozil and mefenamic<br />
acid, occurred at the lowest levels. Despite of a wide variation<br />
of trimethoprim from different WWTPs, the composition<br />
profiles of target pharmaceuticals in secondary effluents from<br />
the four WWTPs were quite similar (Fig. 3).<br />
The concentration levels of most pharmaceuticals and<br />
consumer products detected in the secondary effluent were<br />
also lower than those reported in the Europe. They were over<br />
100 ng L 1 , in some cases even up to 500 ng L 1 in the wastewater<br />
effluents of the European countries (Santos et al., 2007;<br />
Gomez et al., 2007; Ternes, 1998; Vieno et al., 2007b). While in<br />
the present study, 10 out of 15 compounds were less than<br />
100 ng L 1 , and none of them exceeded 400 ng L 1 in any<br />
effluent samples (Fig. 2). Our results were in agreement with<br />
those in Japan (Nakada et al., 2006), Korea (Kim et al., 2007) and<br />
some other cities of China (Xu et al., 2007; Gulkowska et al.,<br />
2008; Chen et al., 2008). For instance, the concentrations of<br />
chloramphenicol in the effluents of 4 WWTPs in Guangzhou<br />
were
et al., 2008; Gomez et al., 2007). The removal rate of bezafibrate<br />
was found to be 87% in six Italian WWTPs in summer time<br />
(Castiglioni et al., 2006), very similar to that observed in our<br />
study. The concentrations of DEET were decreased by more<br />
than 80% during the biological treatment in the WWTP of<br />
Japan (Okuda et al., 2008), slightly better than our results. A<br />
second group of pharmaceuticals, including three antiinflammatory<br />
drugs, clofibric acid, gembrozil, metoprolol and<br />
sulpiride, had lower removal rates with large variation in<br />
different WWTPs studied. For instance, 28–53% of diclofenac,<br />
a representative of the anti-inflammatory drugs, was removed<br />
by secondary treatment in the WWTPs, which was between<br />
26% in Finland (Lindqvist et al., 2005) and 69% in Germany<br />
(Ternes, 1998). The elimination of these compounds may be<br />
highly dependent on the configurations and operation conditions<br />
of individual WWTP as well as wastewater characteristics,<br />
and thus no definitive conclusion could be reached.<br />
Higher load of carbamazepine was found in the secondary<br />
effluent than in the primary effluent, indicating negative<br />
removal efficiency during the secondary treatment. Some<br />
carbamazepine was found to be excreted as the form of<br />
conjugates (Vieno et al., 2007b), which was biodegraded to<br />
carbamazepine by enzymatic processes during the secondary<br />
treatment, resulting in additional amounts of carbamazepine<br />
in the secondary effluent. However, as the calculations of all<br />
the removal efficiencies were based on grab samples that were<br />
not sampled with a hydraulic lag in the present study, some<br />
error might be brought in due to diurnal variation of the<br />
concentration. Therefore, the present study only provided<br />
a snapshot of the removal of pharmaceuticals and consumer<br />
products in the WWTPs of Beijing. To better illustrate that,<br />
24-h composite samples that are lagged by HRT should be<br />
collected and analyzed in further studies.<br />
It has been reported that high HRT (>12 h) and SRT (>10 d)<br />
may contribute to an increased removal rate of pharmaceuticals<br />
(Jones et al., 2007; Vieno et al., 2007b). In the present<br />
study, the WWTP C, in which the HRT was higher than the<br />
others, was the best in removing these compounds, due to<br />
increased contact time of target compounds and the microorganisms.<br />
On the other hand, the different SRTs did not have<br />
significant effects on the removal efficiency, probably because<br />
the SRTs in all the four WWTPs were relatively high (>10 d),<br />
and without large differences. In addition, it is noteworthy<br />
that the WWTP C employed oxidation ditches, which showed<br />
better removal of natural estrogens and estrogenic activity<br />
than A/O (Hashimoto et al., 2007). It also could be the reason<br />
for the higher removal efficiencies in the WWTP C. Further<br />
investigation for different types of WWTPs is necessary to<br />
confirm the results mentioned above.<br />
3.4. Removal efficiency in advanced treatment processes<br />
The removal efficiencies of the pharmaceuticals during the SF,<br />
UF/ozonation, as well as MF/RO treatment in three corresponding<br />
WWTPs are listed in Table 5.<br />
Generally, sand filtration was not effective for these<br />
compounds. Only trimethoprim, DEET and gemfibrozil were<br />
removed slightly during this treatment process. It should be<br />
noticed that these compounds were efficiently removed in the<br />
secondary treatment, indicating that the biodegradation on<br />
water research 44 (2010) 417–426 423<br />
Table 5 – Removal efficiencies (%) of target<br />
pharmaceuticals and consumer products by advanced<br />
treatment processes in studied WWTPs.<br />
Compound WWTP A WWTP B WWTP D<br />
UF Ozone SF MF/RO<br />
DEET 0–50 50–80 0–50 >90<br />
CF 90<br />
BF 0–50 0–50 0–50 >90<br />
CA 90 90<br />
IM 0–50 >90 90<br />
MA 0–50 80–90 90<br />
SP 0–50 >90 0–50 >90<br />
the biofilm present on the sand particle, rather than the<br />
removal with particles, may be the main reason for their<br />
elimination (Gobel et al., 2007).<br />
The results showed that ozonation is effective in removing<br />
most of the target compounds, probably due to the operation<br />
conditions employed in WWTP A (ozone dosage: 5 mg L 1 ,<br />
contact time: 15 min). Carbamazepine, diclofenac, indomethacin,<br />
sulpiride and trimethoprim were significantly<br />
eliminated, with the removal rates of above 95%. The double<br />
bond in the azepine ring of carbamazepine and pyrrole ring of<br />
indomethacin, and the non-protonated amine of diclofenac<br />
and trimethoprim were susceptible to ozone attack (Vieno<br />
et al., 2007a; Nakada et al., 2007; Westerhoff et al., 2005). The<br />
removal efficiencies of DEET and metoprolol were modest.<br />
The amide group, which is not reactive with ozone, could be<br />
the reason for the modest removal of DEET (Nakada et al., 2007).<br />
Low removal efficiencies were found for bezafibrate, clofibric<br />
acid, as well as caffeine. Only 14% of bezafibrate disappeared<br />
in the ozone process, consistent with its low rate constants<br />
with ozone (590 50 M 1 S 1 , Huber et al., 2003). The reaction<br />
site of bezafibrate is the R-oxysubstituent (–O–C(CH 3) 2COOH)<br />
on one of the aromatic rings. However, as the pK a of bezafibrate<br />
is 3.6, the R-oxysubstituent cannot be deprotonated and<br />
consequently the overall rate constant at pH > 4 is much lower<br />
(Huber et al., 2003). It should be noticed that during the<br />
ozonation, most of the pharmaceuticals were not mineralized<br />
but transformed to the oxidation products. For instance, three<br />
oxidation products containing quinazoline-based functional<br />
groups were identified during the ozonation of CBZ (Mcdowell<br />
et al., 2005).<br />
The good performance of ozonation in the present study<br />
was consistent with Ternes et al. (2003), Huber et al. (2005)<br />
and Okuda et al. (2008). When 5 mg L 1 ozone was applied to<br />
the effluent of a municipal WWTP in Germany (contact time:<br />
18 min), target compounds, such as trimethoprim, carbamazepine,<br />
indomethacin, clofibric acid, were removed by<br />
more than 50% (Ternes et al., 2003). Huber et al. (2005)<br />
conducted a pilot study on the oxidation of pharmaceuticals<br />
during ozonation of conventional activated sludge (CAS) and<br />
membrane bio-reactor (MBR) effluents with various ozone<br />
dosages, and found that macrolide and sulfonamide antibiotics,<br />
estrogens, and acidic pharmaceuticals diclofenac,
424<br />
naproxen and indomethacin were oxidized by more than<br />
90–99% for ozone doses 2mgL 1 in all effluents.<br />
The elimination by ultrafiltration in the WWTP A was low<br />
for all the investigated compounds. The molecular weight cutoff<br />
(MWCO) of UF membranes was much higher than 1000 Da,<br />
thus UF membranes showed poor retention of all the investigated<br />
pharmaceuticals, of which the molecular weight are<br />
less than 400 Da. The removal of individual target compound<br />
was less than 50%, and might be due to the adsorption onto<br />
the membrane. It has been also demonstrated that UF<br />
membrane typically had less than 40% retention of 27 PPCPs,<br />
and the mass balances calculated based on the concentration<br />
of each compound in feed, permeate and retentate showed<br />
the observed retention was significantly governed by adsorption<br />
(Yoon et al., 2006).<br />
In contrast, MF/RO employed in WWTP D was very effective.<br />
In the effluent of MF/RO, all the target compounds except<br />
caffeine were not detected. Generally, one or combination of<br />
three basic mechanisms could be involved during the rejection<br />
of solute by NF/RO membrane: steric effect, charge<br />
exclusion and adsorption (Radjenovic et al., 2008). For most<br />
pharmaceuticals, the rejections were considered to be dominated<br />
by steric interaction in ‘‘tight’’ NF or RO membrane<br />
filtration (Nghiem et al., 2005; Radjenovic et al., 2008). As most<br />
investigated compounds have molecular weights about<br />
200–400 Da, smaller than MWCO of RO membrane applied,<br />
excellent rejection of most pharmaceuticals by RO membrane<br />
was observed in this study as well as in previous studies<br />
(Kimura et al., 2004; Al-Rifai et al., 2007; Radjenovic et al.,<br />
2008). Besides, membrane fouling and the presence of organic<br />
matter in the wastewater effluents likely contributed to higher<br />
rejections of pharmaceuticals, especially for some hydrophobic<br />
ionogenic compound (Nghiem and Coleman, 2008;<br />
Comerton et al., 2008).<br />
Nevertheless, the rejections of two compounds, caffeine<br />
and mefenamic acid were slightly lower (i.e. 50–80% and<br />
0–50%, respectively). The concentration of mefenamic acid in<br />
feed wastewaters of MF/RO membrane process was very low,<br />
only a bit higher than its LOQ in the wastewater effluent,<br />
which could be the reason for the low rejection rate. The low<br />
retention of caffeine in the present study was inaccordance<br />
with Drewes et al. (2005). They found that in two full-scale RO<br />
facilities, target EDCs and PPCPs were efficiently rejected to<br />
below detection limit except for caffeine, still detected in the<br />
permeates. The physiochemical properties might explain the<br />
low rejection rate of caffeine. As a representative of hydrophilic<br />
and non-ionic compounds, the rejection driven by<br />
charge exclusion and adsorption is negligible, and steric<br />
exclusion is solely responsible for the retention of caffeine<br />
(Nghiem et al., 2005). However, the molecular weight of<br />
caffeine is 195 Da, smaller than other target compounds, and<br />
might result in the decreased removal efficiency during the RO<br />
membrane filtration process.<br />
Compared to the other two, the WWTPs employing ozone<br />
and RO membrane filtration as advanced treatment were<br />
more efficient in removing pharmaceuticals. For these<br />
WWTPs, the advanced treatment made a significant contribution<br />
to the total elimination of most pharmaceuticals<br />
(Fig. 5). Therefore, the utility of efficient advanced treatment<br />
could be considered as a tool to reduce pharmaceuticals in the<br />
water research 44 (2010) 417–426<br />
a<br />
Removal Contribution (%)<br />
b<br />
Contribution to removal efficiency (%)<br />
120<br />
100<br />
80<br />
60<br />
40<br />
20<br />
0<br />
-20<br />
180<br />
160<br />
140<br />
120<br />
100<br />
80<br />
60<br />
40<br />
20<br />
0<br />
-20<br />
-40<br />
-60<br />
DEET<br />
DEET<br />
CF<br />
CP<br />
municipal wastewater treatment plants. However, the problems<br />
of membrane fouling and further treatment or disposal<br />
of retentate challenge the application of RO membrane<br />
filtration (Van der Bruggen et al., 2008). For ozonation, as most<br />
of the pharmaceuticals could not be mineralized, and oxidation<br />
products are formed from parent pharmaceutical<br />
compounds (Mcdowell et al., 2005), more research is required<br />
to identify the oxidation products and their potential toxicity<br />
during the partial oxidation process (Nakada et al., 2007).<br />
Besides, economic feasibility should be evaluated by estimating<br />
the energy consumption and investment and<br />
operation costs for both advanced treatment processes (Joss<br />
et al., 2008).<br />
4. Conclusion<br />
Tertiary treatment<br />
Conventional treatment<br />
CF<br />
TP<br />
TP<br />
BF<br />
BF<br />
13 out of 15 pharmaceuticals and consumer products from<br />
eight classes were detected at four WWTPs in Beijing, China.<br />
The concentrations of most compounds in the influent and<br />
secondary effluent were lower than those reported in the USA<br />
and Europe, but consistent with the production profile of the<br />
CA<br />
GF<br />
DF<br />
Pharmaceutical<br />
CA<br />
GF<br />
DF<br />
Pharmaceutical<br />
IM<br />
IM<br />
MTP<br />
MTP<br />
CBZ<br />
CBZ<br />
SP<br />
SP<br />
Tertiary treatment<br />
Secondary treatment<br />
Primary treatment<br />
Fig. 5 – Contributions of primary, secondary (or<br />
conventional treatment) and tertiary treatment to the total<br />
elimination of selected pharmaceuticals in WWTP A (a)<br />
and WWTP D (b).
pharmaceuticals in China. According to the result of risk<br />
assessment for the secondary effluent, only diclofenac might<br />
pose a risk to the aquatic environment. The removal efficiencies<br />
by the conventional treatment varied for different<br />
compounds, depending on their chemical structures, physiochemical<br />
properties, as well as the specific treatment<br />
processes utilized at each WWTP. Further removal could be<br />
achieved by adopting some advanced treatment processes,<br />
such as ozonation and MF/RO. However, others, such as<br />
sand filtration, showed low efficiency in removing these<br />
compounds from secondary effluent.<br />
Acknowledgement<br />
This study was supported by the National Science Fund for<br />
Distinguished Young Scholars (No. 50625823).<br />
Appendix.<br />
Supplementary data<br />
Supplementary information related to this article can be<br />
found at doi:10.1016/j.watres.2009.07.010.<br />
references<br />
Al-Rifai, J.H., Gabelish, C.L., Schafer, A.I., 2007. Occurrence of<br />
pharmaceutically active and non-steroidal estrogenic<br />
compounds in three different wastewater recycling schemes<br />
in Australia. Chemosphere 69 (5), 803–815.<br />
Bolton, S., Null, G., 1981. Caffeine: psychological effects, use and<br />
abuse. Orthomolecular Psychiatry 10 (3), 202–211.<br />
Castiglioni, S., Bagnati, R., Fanelli, R., Pomati, F., Calamari, D.,<br />
Zuccato, E., 2006. Removal of pharmaceuticals in sewage<br />
treatment plants in Italy. Environmental Science and<br />
Technology 40 (1), 357–363.<br />
Chen, H.C., Wang, P.L., Ding, W.H., 2008. Using liquid<br />
chromatography–ion trap mass spectrometry to determine<br />
pharmaceutical residues in Taiwanese rivers and<br />
wastewaters. Chemosphere 72 (6), 863–869.<br />
China Medicine Economic Information Net (CMEIN), 2005.<br />
Chinese Medical Statistical Yearbook. Products section. China<br />
Medicine Economic Information Net (CMEIN), Beijing, China.<br />
Comerton, A.M., Andrews, R.C., Bagley, D.M., Hao, C.Y., 2008. The<br />
rejection of endocrine disrupting and pharmaceutically active<br />
compounds by NF and RO membranes as a function of<br />
compound and water matrix properties. Journal of Membrane<br />
Science 313 (1–2), 323–335.<br />
Daughton, C.G., Ternes, T.A., 1999. Pharmaceuticals and personal<br />
care products in the environment: agents of subtle change?<br />
Environmental Health Perspectives 107 (S6), 907–938.<br />
Drewes, J.E., Bellona, C., Oekekoven, M., Xu, P., Kim, T.U., Amy, G.,<br />
2005. Rejection of wastewater-derived micropollutants in<br />
high-pressure membrane applications leading to indirect<br />
potable reuse. Environmental Progress 24 (4), 400–409.<br />
European Environment Agency, 1998. Environmental Risk<br />
Assessment – Approaches, Experiences and Information<br />
Source. European Environment Agency, London, pp. 68–86.<br />
Ferrari, B., Paxeus, N., Giudice, R.L., Pollio, A., Grric, J., 2003.<br />
Ecotoxicological impact of pharmaceuticals found in treated<br />
water research 44 (2010) 417–426 425<br />
wastewaters: study of carbamazepine, clofibric acid, and<br />
diclofenac. Ecotoxicology and Environmental Safety 55 (3),<br />
359–370.<br />
Gobel, A., Mcardell, C.S., Joss, A., Siegrist, H., Giger, W., 2007. Fate<br />
of sulfonamides, macrolides, and trimethoprim in different<br />
wastewater treatment technologies. Science of the Total<br />
Environment 372 (2–3), 361–371.<br />
Gomez, M.J., Bueno, M.J.M., Lacorte, S., Fernandez-Alba, A.R.,<br />
Aguera, A., 2007. Pilot survey monitoring pharmaceuticals and<br />
related compounds in a sewage treatment plant located on the<br />
Mediterranean coast. Chemosphere 66 (6), 993–1002.<br />
Grung, M., Kallqvist, T., Sakshaug, S., Skurtveit, S., Thomas, K.V.,<br />
2008. Environmental assessment of Norwegian priority<br />
pharmaceuticals based on the EMEA guideline. Ecotoxicology<br />
and Environmental Safety 71 (2), 328–340.<br />
Gulkowska, A., Leung, H.W., So, M.K., Taniyasu, S., Yamashita, N.,<br />
Yeung, L.W.Y., Richardson, B.J., Lei, A.P., Giesy, J.P., Lam, P.K.S.,<br />
2008. Removal of antibiotics from wastewater by sewage<br />
treatment facilities in Hongkong and Shenzhen, China. Water<br />
Research 42 (1–2), 395–403.<br />
Hashimoto, T., Onda, K., Nakamura, Y., Tada, K., Miya, A.,<br />
Murakami, T., 2007. Comparison of natural estrogen removal<br />
efficiency in the conventional activated sludge process and<br />
the oxidation ditch process. Water Research 41 (10),<br />
2117–2126.<br />
Huber, M.M., Canonica, S., Park, G.Y., Gunten, U.V., 2003.<br />
Oxidation of pharmaceuticals during ozonation and advanced<br />
oxidation processes. Environmental Science and Technology<br />
37 (5), 1016–1024.<br />
Huber, M.M., Gobel, A., Joss, A., Hermann, N., Loffler, D.,<br />
Mcardell, C.S., Ried, A., Siegrist, H., Ternes, T.A., Gunten, U.V.,<br />
2005. Oxidation of pharmaceuticals during ozonation of<br />
municipal wastewater effluents: a pilot study. Environmental<br />
Science and Technology 39 (11), 4290–4299.<br />
Huerta-Fontela, M., Galceran, M.T., Martin-Alonso, J., Ventura, F.,<br />
2008. Occurrence of psychoactive stimulatory drugs in<br />
wastewaters in north-eastern Spain. Science of the Total<br />
Environment 397 (1–3), 31–40.<br />
Huschek, G., Hansen, P.D., Maurer, H.H., Krengel, D., Kayser, A.,<br />
2004. Environmental risk assessment of medicinal products<br />
for human use according to European Commission<br />
recommendations. Environmental Toxicology 19 (3), 226–240.<br />
Jjemba, P.K., 2006. Excretion and ecotoxicity of pharmaceutical<br />
and personal care products in the environment. Ecotoxicology<br />
and Environmental Safety 63 (1), 113–130.<br />
Jones, O.A.H., Voulvoulis, N., Lester, J.N., 2005. Human<br />
pharmaceuticals in wastewater treatment processes. Critical<br />
Reviews in Environmental Science and Technology 35 (4),<br />
401–427.<br />
Jones, O.A.H., Voulvoulis, N., Lester, J.N., 2007. The occurrence<br />
and removal of selected pharmaceutical compounds in<br />
a sewage treatment works utilising activated sludge<br />
treatment. Environmental Pollution 145 (3), 738–744.<br />
Joss, A., Siegrist, H., Ternes, T.A., 2008. Are we about to upgrade<br />
wastewater treatment for removing organic micropollutants.<br />
Water Science and Technology 57 (2), 251–255.<br />
Kasprzyk-Hordern, B., Dinsdale, R.M., Guwy, A.J., 2009. The<br />
removal of pharmaceuticals, personal care products,<br />
endocrine disruptors and illicit drugs during wastewater<br />
treatment and its impact on the quality of receiving waters.<br />
Water Research 43 (2), 363–380.<br />
Khan, S.J., Wintgens, T., Sherman, P., Zaricky, J., Schafer, A.I.,<br />
2004. Removal of hormones and pharmaceuticals in the<br />
advanced water recycling demonstration plant in Queensland,<br />
Australia. Water Science and Technology 50 (5), 15–22.<br />
Khan, S.J., Ongerth, J.E., 2004. Modelling of pharmaceutical<br />
residues in Australian sewage by quantities of use and<br />
fugacity calculations. Chemosphere 54 (3), 355–367.
426<br />
Kim, S.D., Cho, J., Kim, I.S., Vanderford, B.J., Snyder, S.A., 2007.<br />
Occurrence and removal of pharmaceuticals and endocrine<br />
disruptors in South Korean surface, drinking, and waste<br />
waters. Water Research 41 (5), 1013–1021.<br />
Kimura, K., Toshima, S., Amy, G., Watanabe, Y., 2004. Rejection of<br />
neutral endocrine disrupting compounds (EDCs) and<br />
pharmaceutical active compounds (PhACs) by RO membranes.<br />
Journal of Membrane Science 245 (1–2), 71–78.<br />
Lishman, L., Smyth, S.A., Sarafin, K., Kleywegt, S., Toito, J.,<br />
Peart, T., Lee, B., Servos, M., Beland, M., Seto, P., 2006.<br />
Occurrence and reductions of pharmaceuticals and personal<br />
care products and estrogens by municipal wastewater<br />
treatment plants in Ontario, Canada. Science of the Total<br />
Environment 367 (2–3), 544–558.<br />
Lindqvist, N., Tuhkanen, T., Kronberg, L., 2005. Occurrence of<br />
acidic pharmaceuticals in raw and treated sewages and in<br />
receiving waters. Water Research 39 (11), 2219–2228.<br />
Mcdowell, D.C., Huber, M.M., Wagner, M., Gunten, U.V.,<br />
Ternes, T.A., 2005. Ozonation of carbamazepine in drinking<br />
water: identification and kinetic study of major oxidation<br />
products. Environmental Science and Technology 39 (20),<br />
8014–8022.<br />
Nakada, N., Tanishima, T., Shinohara, H., Kiri, K., Takada, H.,<br />
2006. Pharmaceutical chemicals and endocrine disrupters in<br />
municipal wastewater in Tokyo and their removal during<br />
activated sludge treatment. Water Research 40 (17), 3297–3303.<br />
Nakada, N., Shinohara, H., Murata, A., Kiri, K., Managaki, S.,<br />
Sato, N., Takada, H., 2007. Removal of selected<br />
pharmaceuticals and personal care products (PPCPs) and<br />
endocrine-disrupting chemicals (EDCs) during sand filtration<br />
and ozonation at a municipal sewage treatment plant. Water<br />
Research 41 (19), 4372–4382.<br />
Nghiem, L.D., Schafer, A.I., Elimelech, M., 2005. Pharmaceutical<br />
retention mechanisms by nanofiltration membrane.<br />
Environmental Science and Technology 39 (19), 7698–7705.<br />
Nghiem, L.D., Coleman, P.J., 2008. NF/RO filtration of the<br />
hydrophobic ionogenic compound triclosan: transport<br />
mechanisms and the influence of membrane fouling.<br />
Separation and Purification Technology 62 (3), 709–716.<br />
Niwa, T., Inoue, S., Shiraga, T., Takagi, A., 2005. No inhibition of<br />
cytochrome P450 activities in human liver microsomes by<br />
sulpiride, an antipsychotic drug. Biological and<br />
Pharmaceutical Bulletin 28 (1), 188–191.<br />
Okuda, T., Kobayashi, Y., Nagao, R., Yamashita, N., Tanaka, H.,<br />
Tanaka, S., Fujii, S., Konishi, C., Houwa, I., 2008. Removal<br />
efficiency of 66 pharmaceuticals during wastewater treatment<br />
process in Japan. Water Science and Technology 57 (1), 65–71.<br />
Paxeus, N., 2004. Removal of selected non-steroidal antiinflammatory<br />
drugs (NSAIDs), gemfibrozil, carbamazepine,<br />
beta-blockers, trimethoprim and triclosan in conventional<br />
wastewater treatment plants in five EU countries and their<br />
discharge to the aquatic environment. Water Science and<br />
Technology 50 (5), 253–260.<br />
Quinn, B., Gagne, F., Blaise, C., 2008. An investigation into the acute<br />
and chronic toxicity of eleven pharmaceuticals (and their<br />
solvents) found in wastewater effluent on the cnidarian, Hydra<br />
attenuata. Science of the Total Environment 389 (2–3), 306–314.<br />
Radjenovic, J., Petrovic, M., Ventura, F., Barcelo, D., 2008. Rejection<br />
of pharmaceuticals in nanofiltration and reverse osmosis<br />
water research 44 (2010) 417–426<br />
membrane drinking water treatment. Water Research 42 (14),<br />
3601–3610.<br />
Santos, J.L., Aparicio, I., Alonso, E., 2007. Occurrence and risk<br />
assessment of pharmaceutically active compounds in<br />
wastewater treatment plants. A case study: Seville city<br />
(Spain). Environment International 33 (4), 596–601.<br />
Sui, Q., Huang, J., Deng, S. B., Yu, G. Rapid determination of<br />
pharmaceuticals from multiple therapeutic classes in<br />
wastewater by solid-phase extraction and ultra-performance<br />
liquid chromatography tandem mass spectrometry. Chinese<br />
Science Bulletin, in press.<br />
Ternes, T.A., 1998. Occurrence of drugs in German sewage<br />
treatment plants and rivers. Water Research 32 (11),<br />
3245–3260.<br />
Ternes, T.A., Meisenheimer, M., Mcdowell, D., Sacher, F.,<br />
Brauch, H.J., Haist-gulde, B., Preuss, G., Wilme, U.,<br />
Zulei-seibert, N., 2002. Removal of pharmaceuticals during<br />
drinking water treatment. Environmental Science and<br />
Technology 36 (17), 3855–3863.<br />
Ternes, T.A., Stuber, J., Herrmann, N., Mcdowell, D., Ried, A.,<br />
Kampmann, M., Teiser, B., 2003. Ozonation: a tool for removal<br />
of pharmaceuticals, contrast media and musk fragrances from<br />
wastewater? Water Research 37 (8), 1976–1982.<br />
Thomas, P.M., Foster, G.D., 2005. Tracking acidic<br />
pharmaceuticals, caffeine, and triclosan through the<br />
wastewater treatment process. Environmental Toxicology and<br />
Chemistry 24 (1), 25–30.<br />
Vanderford, B.J., Snyder, S.A., 2006. Analysis of pharmaceuticals<br />
in water by isotope dilution liquid chromatography/tandem<br />
mass spectrometry. Environmental Science and Technology<br />
40 (23), 7312–7320.<br />
Van der Bruggen, B., Manttari, M., Nystrom, M., 2008.<br />
Drawbacks of applying nanofiltration and how to avoid them:<br />
a review. Separation and Purification Technology 63 (2),<br />
251–263.<br />
Vieno, N.M., Harkki, H., Tuhkanen, T., Kronberg, L., 2007a.<br />
Occurrence of pharmaceuticals in river water and their<br />
elimination in a pilot-scale drinking water treatment plant.<br />
Environmental Science and Technology 41 (14), 5077–5084.<br />
Vieno, N., Tuhkanen, T., Kronberg, L., 2007b. Elimination of<br />
pharmaceuticals in sewage treatment plants in Finland. Water<br />
Research 41 (5), 1001–1012.<br />
Watkinson, A.J., Murby, E.J., Costanzo, S.D., 2007. Removal of<br />
antibiotics in conventional and advanced wastewater<br />
treatment: implications for environmental discharge and<br />
wastewater recycling. Water Research 41 (18), 4164–4176.<br />
Westerhoff, P., Yoon, Y., Snyder, S., Wert, E., 2005. Fate of<br />
endocrine-disruptor, pharmaceutical, and personal care<br />
product chemicals during simulated drinking water treatment<br />
processes. Environmental Science and Technology 39 (17),<br />
6649–6663.<br />
Xu, W.H., Zhang, G., Li, X.D., Zou, S.C., Li, P., Hu, Z.H., Li, J., 2007.<br />
Occurrence and elimination of antibiotics at four sewage<br />
treatment plants in the Pearl River Delta, South China. Water<br />
Research 41 (19), 4526–4534.<br />
Yoon, Y., Westerhoff, P., Snyder, S.A., Wert, E.C., 2006.<br />
Nanofiltration and ultrafiltration of endocrine disrupting<br />
compounds, pharmaceuticals and personal care products.<br />
Journal of Membrane Science 270 (1–2), 88–100.
Application of the combination index (CI)-isobologram<br />
equation to study the toxicological interactions of lipid<br />
regulators in two aquatic bioluminescent organisms<br />
Ismael Rodea-Palomares a,1 , Alice L. Petre b,1 , Karina Boltes b , Francisco Leganés a ,<br />
José Antonio Perdigón-Melón b , Roberto Rosal b , Francisca Fernández-Piñas a, *<br />
a<br />
Departamento de Biología, Facultad de Ciencias, Universidad Autónoma de Madrid, 2 Darwin Street, Cantoblanco, 28049 Madrid, Spain<br />
b<br />
Departamento de Ingeniería Química, Universidad de Alcalá, Alcalá de Henares, E-28871 Madrid, Spain<br />
article info<br />
Article history:<br />
Received 26 March 2009<br />
Received in revised form<br />
13 July 2009<br />
Accepted 18 July 2009<br />
Available online 25 July 2009<br />
Keywords:<br />
Antagonism<br />
Combination index-isobologram<br />
equation<br />
Cyanobacterium<br />
Fibrates<br />
Synergism<br />
Vibrio fischeri<br />
1. Introduction<br />
abstract<br />
Fibrates and statins (HMG-CoA reductase inhibitors) are the<br />
main lipid-lowering drugs prescribed either alone or in<br />
combination therapy in order to decrease plasma cholesterol<br />
levels and reduce the incidence of coronary heart disease.<br />
Although partially displaced by statins, the total number of<br />
fibrate prescriptions is in constant increase in the United<br />
States (Holoshitz et al., 2008). Fibric acids are the active forms<br />
water research 44 (2010) 427–438<br />
Available at www.sciencedirect.com<br />
journal homepage: www.elsevier.com/locate/watres<br />
* Corresponding author. Tel.: þ34 9 1497 8176; fax: þ34 9 1497 8344.<br />
E-mail address: francisca.pina@uam.es (F. Fernández-Piñas).<br />
1<br />
Both authors contributed equally to this work.<br />
0043-1354/$ – see front matter ª 2009 Elsevier Ltd. All rights reserved.<br />
doi:10.1016/j.watres.2009.07.026<br />
Pharmaceuticals in the aquatic environment do not appear singly and usually occur as<br />
complex mixtures, whose combined effect may exhibit toxicity to the aquatic biota. We<br />
report an environmental application of the combination index (CI)-isobologram equation,<br />
a method widely used in pharmacology to study drug interactions, to determine the nature<br />
of toxicological interactions of three fibrates toward two aquatic bioluminescent organisms,<br />
Vibrio fischeri and the self-luminescent cyanobacterial recombinant strain Anabaena<br />
CPB4337. The combination index-isobologram equation method allows computerized<br />
quantitation of synergism, additive effect and antagonism. In the Vibrio test, the fibrate<br />
combinations showed antagonism at low effect levels that turned into an additive effect or<br />
synergism at higher effect levels; by contrast, in the Anabaena test, the fibrate combinations<br />
showed a strong synergism at the lowest effect levels and a very strong antagonism at high<br />
effect levels. We also evaluated the nature of the interactions of the three fibrates with a real<br />
wastewater sample in the cyanobacterial test. We propose that the combination indexisobologram<br />
equation method can serve as a useful tool in ecotoxicological assessment.<br />
ª 2009 Elsevier Ltd. All rights reserved.<br />
of fibrates and belong to the nuclear receptor superfamily of<br />
ligand-activated transcription factors. Gemfibrozil and fenofibrate<br />
are the fibrates currently marketed in the US, whereas<br />
bezafibrate is also available in Europe and other developed<br />
countries (Lambropoulou et al., 2008). Fenofibric acid, 2-[4-(4chlorobenzoyl)phenoxy]-2-methylpropanoic<br />
acid, is the active<br />
metabolite of fenofibrate, the inactive prodrug marketed and<br />
dispensed. Gemfibrozil, 5-(2,5-dimethylphenoxy)-2,2-dimethylpentanoic<br />
acid and bezafibrate, p-[4-[chlorobenzoylamino-
428<br />
ethyl]-phenoxy]-b-methylpropionic acid, are also fibric acid<br />
derivatives with similar pharmacokinetic behaviour (Miller and<br />
Spence, 1998).<br />
The occurrence of lipid regulators in the discharge of<br />
treated urban and municipal wastewater has been relatively<br />
well documented. Bezafibrate has been detected in effluents<br />
of two British STP with averages up to 230 ng/L (Kasprzyk-<br />
Hordern et al., 2009). Metcalfe et al. (2003) found around 1 mg/L<br />
of gemfibrozil in effluents of Canadian STP, whereas fenofibrate<br />
has been reported in concentrations up to 0.5 mg/L in the<br />
influent of several Brazilian STP (Stumpf et al., 1999).<br />
Andreozzi et al. (2003) found lipid regulators in the effluent of<br />
several European STP at concentrations up to 4.76 mg/L (gemfibrozil),<br />
1.07 mg/L (bezafibrate) and 0.16 mg/L (fenofibrate).<br />
Rosal et al. (2008), reported the occurrence of bezafibrate and<br />
gemfibrozil at levels of 139 and 608 ng/L respectively in the<br />
effluent of a Spanish STP. In the same plant Rodríguez et al.<br />
(2008) found 165 ng/L of fenofibric acid, 61 ng/L of bezafibrate<br />
and 143 ng/L of gemfibrozil.<br />
It is also significant that removal efficiencies observed in<br />
current STP are not always high. Fent et al. (2006) reported<br />
maximum removal rates of 50–75% for fenofibric acid and<br />
gemfibrozil and somewhat higher for bezafibrate, although for<br />
the later, efficiencies below 15% have also been reported.<br />
Stumpf et al. (1999) reported a 45% removal of fenofibric acid<br />
by an activated sludge conventional treatment. Kasprzyk-<br />
Hordern et al. (2009) encountered an average degradation of<br />
bezafibrate not higher than 67%. On the other hand, Castiglioni<br />
et al. (2006) reported that the removal efficiency of<br />
bezafibrate during an activated sludge treatment greatly<br />
varied from 15% in winter to 87% in summer.<br />
At measured environmental concentrations as those<br />
reported above (mostly in the ng/L and mg/L range), many<br />
studies have shown that the risk of acute toxicity is unlikely<br />
(Fent et al., 2006; Han et al., 2006; Borgmann et al., 2007);<br />
however, there is a lack of data on chronic toxicity effects.<br />
Moreover, pharmaceuticals in the aquatic environments occur<br />
as complex mixtures from different classes, not as single<br />
contaminants (Gros et al., 2007); thus, although the concentration<br />
of individual pharmaceuticals is low, their mixture<br />
could prove ecotoxicologically significant (Brain et al., 2004).<br />
Current methods of risk assessment usually focus on the<br />
assessment of single chemicals, which may underestimate<br />
the risk associated with toxic action of mixtures; probably for<br />
this reason, in the last years there is an increasing number of<br />
studies dealing with complex mixtures of pharmaceuticals<br />
(Cleuvers, 2003, 2004; Crane et al., 2006; Han et al., 2006;<br />
Borgmann et al., 2007; Christensen et al., 2007; Pomati et al.,<br />
2008; Quinn et al., 2009). However, assessment of combined<br />
toxicities is not an easy issue. Basically, two different models<br />
are in use for the prediction of mixture toxicity, i.e., concentration<br />
addition, when pharmaceuticals have a similar mode<br />
of toxic action, and response addition or independent action,<br />
when pharmaceuticals have different modes of toxic action<br />
(Cleuvers, 2003; Teuschler, 2007). However, toxicological<br />
interactions, synergisms or antagonisms, between the pharmaceuticals<br />
and their effects can occur independently of<br />
mode of action; moreover, in most cases, the pharmacological<br />
mechanisms of action is known but the toxic mode of action<br />
may remain unknown (Cleuvers, 2003; Chou, 2006). In an effort<br />
water research 44 (2010) 427–438<br />
to overcome this limitation, we report an environmental<br />
application of a method widely used in pharmacology to<br />
interpret drug interactions; this method, termed as the<br />
median-effect/combination index (CI)-isobologram equation<br />
(Chou, 2006) allows quantitative determinations of chemical<br />
interactions where CI 1 indicate synergism,<br />
additive effect and antagonism, respectively. One important<br />
property of the method is that previous knowledge of the<br />
mechanisms of action of each chemical is not required.<br />
Besides, the method takes into account both the potency and<br />
the shapes of the dose-effect curve of each chemical. The<br />
method has been computerized allowing an automated<br />
simulation of synergism and antagonism at different<br />
concentrations and at different effect levels of the chemicals<br />
in a mixture.<br />
The aim of our study was to assess the nature of the<br />
toxicological interactions of three fibrates, gemfibrozil, bezafibrate<br />
and fenofibric acid, by the method of combination<br />
index (CI)-isobologram equation. The three pharmaceuticals<br />
were used singly or in two- and three-drug combinations. As<br />
toxicity endpoint we have chosen the bioluminescent<br />
response of two prokaryotes, the naturally luminescent Vibrio<br />
fischeri and the recombinant bioluminescent cyanobacterium<br />
Anabaena sp. PCC 7120 CPB4337 (hereinafter, Anabaena<br />
CPB4337), both bioluminescent organisms have proved very<br />
useful in evaluating the toxicity of individual fibrates in<br />
a previous study (Rosal et al., 2009). For Anabaena CPB4337, we<br />
also evaluated the nature of the interactions of the three<br />
fibrates with a wastewater sample from a local STP, which<br />
already proved very toxic to the cyanobacterium (Rosal et al.,<br />
2009).<br />
2. Materials and methods<br />
2.1. Materials<br />
Gemfibrozil (þ99%) and bezafibrate (þ98%) were purchased<br />
from Sigma–Aldrich. Fenofibric acid was produced from<br />
fenofibrate (Sigma–Aldrich, þ99% purity) by hydrolysis.<br />
A suspension of fenofibrate in isopropanol (30 wt.%, 400 mL)<br />
was refluxed during 4 h with an aqueous sodium hydroxide<br />
solution (2.0 M, 200 mL). After cooling to less than 70 C,<br />
a solution of hydrochloric acid (1.0 M, 325 mL) was slowly<br />
added while keeping the temperature over 60 C. The product<br />
crystallized after cooling and keeping at room temperature<br />
during 4 or more h. The product was filtered and rinsed with<br />
water and dried overnight at 60 C under nitrogen. The purity<br />
of the product was over 97% checked by HPLC. Solubility of<br />
acidic drugs in water is strongly pH dependent with few data<br />
considering this variable. Comerton et al. (2007) reported<br />
a solubility of 10.9 mg/L of gemfibrozil in water, but we could<br />
solve over 125 mg/L in 2 mM MOPS (3-[N-morpholino] propanesulfonic<br />
acid) at pH 6 and higher quantities for the pH at<br />
which V. fischeri bioassays were performed. In all cases, we<br />
avoided the use of solvents and the upper limit for the<br />
concentrations of the studied compounds was their solubility<br />
in pure water or wastewater at the pH of the bioassay.<br />
Wastewater samples were collected from the secondary<br />
clarifier of a STP located in Alcalá de Henares (Madrid) that
eceives domestic wastewater with a minor contribution of<br />
industrial effluents from facilities located near the city. This<br />
STP used a conventional activated sludge treatment and has<br />
been designed for a total capacity of 375,000 equivalent<br />
inhabitants with a maximum flow rate of 3000 m 3 /h. In<br />
a recent previous study (Rosal et al., 2009), we found that this<br />
wastewater was very toxic to Anabaena cells with a wastewater<br />
dilution as low as 0.11 causing 50% luminescence<br />
inhibition (wastewater EC 50).<br />
2.2. Toxicity tests<br />
Bioassays with the photo-luminescent bacteria Vibrio fischeri<br />
were carried out according to ISO 11348-3 standard protocol<br />
(ISO, 2007). This bioassay measures, during the prescribed<br />
incubation period, the decrease in bioluminescence induced<br />
in the cell metabolism due to the presence of a toxic<br />
substance. The bacterial assay used the commercially available<br />
Biofix Lumi test (Macherey-Nagel, Germany). The bacterial<br />
reagent is supplied freeze-dried (Vibrio fischeri NRRL-B<br />
11177) and was reconstituted and incubated at 3 C for 5 min<br />
before use. The desired pH was set by using NaOH or HCl. The<br />
analysis media was 0.34 M NaCl (2% w/v) and tests were<br />
performed at 15 C and the measurements of light were made<br />
using a luminometer (Optocomp I). The effect of toxicants or<br />
toxicant mixtures (i.e., fibrates or fibrate combinations) was<br />
measured as percent inhibition with respect to the light<br />
emitted under test conditions in the absence of any toxic<br />
influence. Toxicity values were routinely obtained after<br />
30 min exposure. Phenol and ZnSO 4 7H 2O have been used as<br />
toxicity standards and all tests have been replicated to ensure<br />
reproducibility.<br />
The bioassays using the recombinant bioluminescent<br />
cyanobacterium Anabaena CPB4337 were based on the inhibition<br />
of constitutive luminescence caused by the presence of<br />
any toxic substance (Rodea-Palomares et al., 2009; Rosal et al.,<br />
2009). Anabaena CPB4337 was routinely grown at 28 C in the<br />
light, ca. 65 mmol photons m 2 s 1 on a rotary shaker in 50 mL<br />
AA/8 (Allen and Arnon, 1955) supplemented with nitrate<br />
(5 mM) in 125 ml Erlenmeyer flasks and 10 mg/mL of neomycin<br />
sulphate (Nm). Luminescence inhibition-based toxicity assays<br />
were performed as follows: 160 mL from five to seven serial<br />
dilutions of each tested toxicant or toxicant mixture (i.e.;<br />
fibrates or fibrate combinations) plus a control (ddH 2O buffered<br />
with MOPS at pH 5.8) were disposed in an opaque white<br />
96-well microtiter plates. 40 mL cells, grown as described, were<br />
washed twice and resuspended in ddH2O buffered with MOPS<br />
at pH 5.8 and were added to the microtiter plate wells to reach<br />
a final cell density at OD750 nm of 0.5. The luminescence of<br />
each sample was recorded every 5 min for up to 1 h in the<br />
Centro LB 960 luminometer. Three independent experiments<br />
with duplicate samples were carried out for all Anabaena<br />
toxicity assays. CuSO 4 has been used as toxicity standard and<br />
all tests have been replicated to ensure reproducibility.<br />
2.3. Experimental design of fibrate combinations<br />
Solutions of gemfibrozil (Gm), bezafibrate (Bz) and fenofibric<br />
acid (Fn) prepared as described above were used singly and in<br />
two (Bz þ Gm; Fn þ Gm; Fn þ Bz) and three (Fn þ Gm þ Bz)<br />
water research 44 (2010) 427–438 429<br />
combinations. Anabaena and Vibrio fischeri cells were treated<br />
with serial dilutions of each fibrate individually and with<br />
a fixed constant ratio (1:1), based on the individual EC50 values,<br />
in their binary and ternary combinations. Five dilutions (serial<br />
dilution factor ¼ 2) of each fibrate and combination plus<br />
a control were tested in three independent experiments with<br />
replicate samples.<br />
For evaluating the nature of the interaction of fibrates with<br />
wastewater, binary combinations of each fibrate plus wastewater<br />
(Fn þ WW; Gm þ WW; Bz þ WW) and a quaternary<br />
combination of the three fibrates plus wastewater (Fn þ Gm þ<br />
Bz þ WW) were also prepared and tested for Anabaena<br />
CPB4337. Anabaena cells were treated with serial dilutions of<br />
each fibrate and wastewater individually and with a fixed<br />
constant ratio (1:1), based on the individual EC50 values, in<br />
their binary and quaternary combinations. Five dilutions<br />
(serial dilution factor ¼ 2) of each fibrate and wastewater and<br />
their combinations plus a control were tested in three independent<br />
experiments with replicate samples. The experimental<br />
design is shown in Table 1.<br />
All individual fibrate, wastewater and their combination<br />
assays were carried out at the same time as recommended by<br />
Chou (2006) to maximize computational analysis of data.<br />
2.4. Median-effect and combination index (CI)isobologram<br />
equations for determining combined<br />
fibrate interactions<br />
The results were analyzed using the median-effect/combination<br />
index (CI)-isobologram equation by Chou (2006) and Chou<br />
and Talalay (1984) which is based on the median-effect principle<br />
(mass-action law) (Chou, 1976) that demonstrates that<br />
there is an univocal relationship between dose and effect<br />
independently of the number of substrates or products and of<br />
the mechanism of action or inhibition. This method involved<br />
plotting the dose-effect curves for each compound and their<br />
combinations in multiple diluted concentrations by using the<br />
median-effect equation:<br />
fa D<br />
¼<br />
fu Dm<br />
m<br />
Where D is the dose, Dm is the dose for 50% effect (e.g., 50%<br />
inhibition of bioluminescence or EC 50), fa is the fraction<br />
affected by dose D (e.g., 0.75 if cell bioluminescence is inhibited<br />
by 75%), fu is the unaffected fraction (therefore, fa ¼ 1 fu),<br />
and m is the coefficient of the sigmoidicity of the dose-effect<br />
curve: m ¼ 1, m > 1, and m < 1 indicate hyperbolic, sigmoidal,<br />
and negative sigmoidal dose-effect curve, respectively.<br />
Therefore, the method takes into account both the potency<br />
(Dm) and shape (m) parameters. If Eq. (1) is rearranged, then:<br />
D ¼ Dm½fa=ð1 faÞŠ 1=m<br />
The Dm and m values for each fibrate are easily determined by<br />
the median-effect plot: x ¼ log (D) versus y ¼ log ( fa/fu) which<br />
is based on the logarithmic form of Eq. (1). In the medianeffect<br />
plot, m is the slope and log (Dm) is the x-intercept. The<br />
conformity of the data to the median-effect principle can be<br />
ready manifested by the linear correlation coefficient (r) of the<br />
data to the logarithmic form of Eq. (1) (Chou, 2006).<br />
(1)<br />
(2)
Table 1 – Experimental design for determining toxicological interactions of fenofibric acid [Fn (D)1], gemfibrozil [Gm (D)2], bezafibrate [Bz (D)3] and their binary and ternary<br />
combinations for Vibrio fischeri and Anabaena CPB4337 bioluminescence tests.<br />
Pure fibrate experiments Fibrates plus wastewater experiments<br />
Vibrio fischeri Anabaena CPB4337 Anabaena CPB4337<br />
Dilutions Single toxicant Single toxicant Single toxicant<br />
Fn Gm Bz Fn Gm Bz Fn Gm Bz WW**<br />
(D) 1 (D) 2 (D) 3 (D) 1 (D) 2 (D) 3 (D) 1 (D) 2 (D) 3 (D) 4<br />
1<br />
⁄4 (EC50) 0.4 8.75 37.5 2.5 2.5 12.5<br />
1<br />
⁄4 (EC50) 2.5 2.5 12.5 0.025<br />
1<br />
⁄2 (EC50) 0.8 17.5 75 5 5 25<br />
1<br />
⁄2 (EC50) 5 5 25 0.05<br />
1 (EC50) 1.6 35 150 10 10 50 1 (EC50) 10 10 50 0.1<br />
2 (EC50) 3.2 70 300 20 20 100 2 (EC50) 20 20 100 0.2<br />
4 (EC50) 6.4 140 600 40 40 200 4 (EC50) 40 40 200 0.4<br />
Two toxicant combo Two toxicant combo Two toxicant combo<br />
(D) 1 þ (D) 2 (1.6:35) (D) 1 þ (D) 2 (1:1) (D) 1 þ (D) 4 (1:0.01)<br />
1<br />
⁄4 (EC50) 0.4 8.75 2.5 2.5<br />
1<br />
⁄4 (EC50) 2.5 0.025<br />
1<br />
⁄2 (EC50) 0.8 17.5 5 5<br />
1<br />
⁄2 (EC50) 5 0.05<br />
1 (EC50) 1.6 35 10 10 1 (EC50) 10 0.1<br />
2 (EC50) 3.2 70 20 20 2 (EC50) 20 0.2<br />
4 (EC50) 6.4 140 25* 25* 4 (EC50) 40 0.4<br />
(D) 1 þ (D) 3 (1.6:150) (D) 1 þ (D) 3 (1:5) (D) 2 þ (D) 4 (1:0.01)<br />
1<br />
⁄4 (EC50) 0.4 37.5 2.5 12.5<br />
1<br />
⁄4 (EC50) 2.5 0.025<br />
1<br />
⁄2 (EC50) 0.8 75 5 25<br />
1<br />
⁄2 (EC50) 5 0.05<br />
1 (EC50) 1.6 150 10 50 1 (EC50) 10 0.1<br />
2 (EC50) 3.2 300 20 100 2 (EC50) 20 0.2<br />
4 (EC50) 6.4 600 30* 150* 4 (EC50) 40 0.4<br />
(D)2 þ (D)3 (35:150) (D)2 þ (D)3 (1:5) (D)3 þ (D)4 (1:0.002)<br />
1<br />
⁄4 (EC50) 8.75 37.5 2.5 12.5<br />
1<br />
⁄4 (EC50) 12.5 0.025<br />
1<br />
⁄2 (EC50) 17.5 75 5 25<br />
1<br />
⁄2 (EC50) 25 0.05<br />
1 (EC50) 35 150 10 50 1 (EC50) 50 0.1<br />
2 (EC50) 70 300 20 100 2 (EC50) 100 0.2<br />
4 (EC50) 140 600 40 200 4 (EC50) 200 0.4<br />
Three toxicant combo Three toxicant combo Four toxicant combo<br />
(D)1 þ (D)2 þ (D)3 (1.6:35:150) (D)1 þ (D)2 þ (D)3 (1:1:5) (D)1 þ (D)2 þ (D)3 þ (D)4 (1:1:5:0.01)<br />
1<br />
⁄4 (EC50) 0.4 8.75 37.5 2.5 2.5 12.5 1/8 (EC50) 1.25 1.25 6.25 0.0125<br />
1<br />
⁄2 (EC50) 0.8 17.5 75 5 5 25<br />
1<br />
⁄4 (EC50) 2.5 2.5 12.5 0.025<br />
1 (EC50) 1.6 35 150 10 10 50<br />
1<br />
⁄2 (EC50) 5 5 25 0.05<br />
2 (EC50) 3.2 70 300 20 20 100 1 (EC50) 10 10 50 0.1<br />
4 (EC50) 6.4 140 600 40 40 200 2 (EC50) 20 20 100 0.2<br />
For the Anabaena test, the design for the experiment with the wastewater [WW (D4)] sample is also included. The experimental design is based on EC50 ratios as proposed by Chou and Talalay (1984).<br />
EC 50 is the effective concentration of a toxicant which caused a 50% bioluminescence inhibition. The combination ratio was approximately equal to the EC 50 ratio of the combination components (i.e.,<br />
close to their equipotency ratio). *Upper maximal possible dose due to the solubility limit of fibrates in pure water. **EC 50 for wastewater is the dilution which caused 50% luminescence inhibition. (D) 1,<br />
(D)2 and (D)3 in mg/L, (D)4 is the dilution of wastewater in ddH2O.<br />
430<br />
water research 44 (2010) 427–438
These parameters were then used to calculate doses of the<br />
fibrates and their combinations required to produce various<br />
effect levels according to Eq. (1); for each effect level, combination<br />
index (CI) values were then calculated according to the<br />
general combination index equation for n chemical combination<br />
at x% inhibition (Chou, 2006):<br />
n ðCIÞx ¼ Xn<br />
j¼1<br />
ðDÞj ðDxÞj ¼ Xn<br />
j¼1<br />
ðDmÞ j<br />
ðDxÞ1 n ½DŠj = Pn 1 ½DŠ<br />
n h<br />
fax = 1 j<br />
fax j<br />
io 1=mj<br />
where n (CI) x is the combination index for n chemicals (e.g.,<br />
fibrates) at x% inhibition (e.g., bioluminescence inhibition);<br />
(Dx) 1 n is the sum of the dose of n chemicals that exerts x%<br />
inhibition in combination, {[Dj]/ Pn 1 ½DŠ} is the proportionality<br />
of the dose of each of n chemicals that exerts x% inhibition in<br />
combination; and (Dm)j {( fax)j/[1 ( fax)j]} 1/mj is the dose of each<br />
drug alone that exerts x% inhibition. From Eq. (3), CI1 indicates synergism, additive effect and antagonism,<br />
respectively.<br />
2.5. Analysis of results<br />
Computer program CompuSyn (Chou and Martin, 2005,<br />
Compusyn Inc, USA) was used for calculation of dose-effect<br />
curve parameters, CI values, fa-CI plot (plot representing CI<br />
(3)<br />
versus fa, the fraction affected by a particular dose; see Eq. (1))<br />
and polygonograms (a polygonal graphic representation<br />
depicting synergism, additive effect and antagonism for three<br />
or more drug combinations). Linear regression analyses were<br />
computed using MINITAB Release 14 for Windows (Minitab<br />
Inc; USA).<br />
3. Results<br />
3.1. Toxicological interactions of fibrate combinations in<br />
Vibrio fischeri and Anabaena CPB4337 bioluminescence<br />
tests<br />
Applying the combination index-isobologram method, we<br />
evaluated the nature of gemfibrozil (Gm), fenofibric acid (Fn)<br />
and bezafibrate (Bz) interactions both in Vibrio fischeri and<br />
Anabaena CPB4337 bioluminescence tests. Table 2 shows the<br />
dose-effect curve parameters (Dm, m and r) of the three<br />
fibrates singly and their binary and ternary combinations, as<br />
well as mean combination index (CI) values of fibrate combinations.<br />
Dm was the dose required to produce the medianeffect<br />
(analogous to the EC 50); Dm values for Fn were the<br />
lowest both, in Vibrio and Anabaena tests, Dm values for Gm<br />
were in the same range for both Vibrio and Anabaena while Bz<br />
Table 2 – Dose-effect relationship parameters and mean combination index (CI) values (as a function of fractional inhibition<br />
of luminescence) of gemfibrozil (Gm), fenofibric acid (Fn), and bezafibrate (Bz) individually and of their binary and ternary<br />
combinations on Vibrio fischeri and Anabaena CPB4337 bioluminescence tests.<br />
Drug combo Vibrio fischeri<br />
Dose-effect parameters CI values<br />
Dm m r EC10 EC50 EC90<br />
mg/L (mM)<br />
Fn 1.45 (4.01) 0.78 0.989 – – –<br />
Gm 20.58 (82.11) 1.53 0.966 – – –<br />
Bz 252.07 (696.46) 1.15 0.975 – – –<br />
Gm þ Bz 78.20 (234.20) 1.54 0.991 1.13 0.13 Add 0.97 0.04 Add 0.86 0.05 Syn<br />
Fn þ Bz 153.79 (424.93) 1.09 0.981 2.98 0.15 Ant 1.71 0.03 Ant 1.17 0.06 Ant<br />
Fn þ Gm 9.84 (38.74) 1.15 0.973 0.99 0.17 Add 0.75 0.05 Syn 0.86 0.08 Syn<br />
Fn þ Gm þ Bz 55.69 (166.69) 1.23 0.993 1.46 0.06 Ant 1.01 0.02 Add 0.99 0.03 Add<br />
Anabaena CPB4337<br />
Dose-effect parameters CI values<br />
Dm m r EC 10 EC 50 EC 90<br />
mg/L (mM)<br />
water research 44 (2010) 427–438 431<br />
Fn 8.53 (23.62) 0.96 0.971 – – –<br />
Gm 10.69 (42.67) 0.81 0.959 – – –<br />
Bz 12.56 (34.70) 1.08 0.990 – – –<br />
Gm þ Bz 19.17 (56.88) 0.84 0.972 1.06 0.15 Add 1.57 0.06 Ant 2.5 0.22 Ant<br />
Fn þ Bz 13.92 (38.49) 0.76 0.965 0.55 0.06 Syn 1.19 0.04 Ant 2.59 0.14 Ant<br />
Fn þ Gm 12.26 (41.45) 0.46 0.955 0.13 0.02 Syn 1.29 0.05 Ant 12.9 2.33 Ant<br />
Fn þ Gm þ Bz 6.62 (19.45) 0.53 0.960 0.09 0.01 Syn 0.57 0.02 Syn 3.92 0.19 Ant<br />
The parameters m, Dm and r are the antilog of x-intercept, the slope and the linear correlation coefficient of the median-effect plot, which<br />
signifies the shape of the dose-effect curve, the potency (EC50), and conformity of the data to the mass-action law, respectively (Chou, 1976; Chou<br />
and Talalay, 1984; Chou, 2006). Dm and m values are used for calculating the CI values (Eq. (3)); CI 1 indicate synergism (Syn),<br />
additive effect (Add), and antagonism (Ant), respectively. EC 10,EC 50 and EC 90, are the doses required to inhibit bioluminescence 10, 50 and 90%,<br />
respectively. Computer software CompuSyn was used for automated calculation and simulation.
432<br />
Dm values were an order of magnitude higher in the Vibrio test<br />
(Rosal et al., 2009); m was the Hill coefficient used to determine<br />
the shape of the dose-response curve, hyperbolic (m ¼ 1),<br />
sigmoidal (m > 1) or negative sigmoidal (m < 1); also shown in<br />
the table, linear regression correlation coefficients (r-values)<br />
of the median-effect plots were >0.95 in all cases, indicating<br />
the conformity of the data to the median-effect principle<br />
which qualifies for further studies using this method.<br />
The Dm and m values for single fibrates and for their<br />
combination mixtures were used for calculating synergism or<br />
antagonism based on the CI Eq. (3) (Chou, 2006). Fig. 1 shows<br />
the fa-CI plot of fibrate interactions both for Vibrio (Fig. 1a) and<br />
Anabaena tests (Fig. 1b); the fa-CI plot depicts the CI value<br />
a<br />
Combination index, CI<br />
b<br />
Combination index, CI<br />
3.0<br />
2.5<br />
2.0<br />
1.5<br />
1.0<br />
0.5<br />
0.0<br />
0.0 0.2 0.4 0.6 0.8 1.0<br />
3.0<br />
2.5<br />
2.0<br />
1.5<br />
1.0<br />
0.5<br />
Fraction affected, fa<br />
0.0<br />
0.0 0.2 0.4 0.6 0.8 1.0<br />
Fraction affected, fa<br />
Fig. 1 – Combination index plot ( fa-CI plot) for a set of three<br />
fibrate combinations: Fn D Bz (–6–), Bz D Gm (–B–),<br />
Fn D Gm (–,–) and Fn D Gm D Bz (–;–) for Vibrio fischeri<br />
test (a) and Anabaena CPB4337 test (b). CI values are plotted<br />
as a function of the fractional inhibition of<br />
bioluminescence ( fa) by computer simulation (CompuSyn)<br />
from fa [ 0.10 to 0.95. CI 1 indicates<br />
synergism, additive effect and antagonism, respectively.<br />
At least three independent experiments with two<br />
replicates were used. The vertical bars indicate 95%<br />
confidence intervals for CI values based on sequential<br />
deletion analysis (SDA) (Chou and Martin, 2005).<br />
Fn [ fenofibric acid, Bz [ bezafibrate and<br />
Gm [ gemfibrozil.<br />
water research 44 (2010) 427–438<br />
versus fa (effect level or fraction of luminescence inhibited by<br />
a fibrate singly or in combination with respect to the control)<br />
for two (Fn þ Bz; Fn þ Gm and Bz þ Gm) and three fibrate<br />
(Fn þ Gm þ Bz) combinations. The fa-CI plot is an effectoriented<br />
plot that shows the evolution of the kind of interaction<br />
(synergism, antagonism, additive effect) as a function of<br />
the level of the effect ( fa) of a particular toxicant on the<br />
reference organism ( fa, where ECa ¼ fa 100; i.e.,<br />
EC 10 ¼ f10 100). In the Vibrio test (Fig. 1a), the Bz þ Gm and<br />
Fn þ Gm binary combination showed a slight antagonism at<br />
very low fa values and slight synergism (Fn þ Gm) or nearly<br />
additive effects (Bz þ Gm) at the highest fa values, the Fn þ Bz<br />
combination showed a strong antagonism at low effect levels<br />
but the antagonism decreased and approached an additive<br />
kind of interaction at the highest fa levels; the ternary<br />
combination (Fn þ Gm þ Bz) showed a moderate antagonism<br />
at low fa values that also turned into a nearly additive effect at<br />
fa values above 0.4. Correlation analyses were made between<br />
CI values of the fibrate ternary combination and CI values of<br />
each of the fibrate binary combinations to determine which<br />
binary combination interaction was predominant in the<br />
ternary mixture (Table 3); the highest correlation coefficient<br />
was found for the Fn þ Bz combination (r ¼ 0.91), suggesting<br />
that this combination interaction predominated in the three<br />
fibrate mixture. The fa-CI plot of the Anabaena test (Fig. 1b)<br />
showed the opposite pattern of interactions as the three<br />
binary and the ternary combinations showed from slight to<br />
strong synergism at the lowest fa values that turned into a very<br />
strong antagonism at fa values over 0.5; the ternary combination<br />
(Fn þ Gm þ Bz) closely followed the interaction pattern<br />
of the binary Fn þ Gm combination, this is confirmed by the<br />
highest correlation coefficient found between the CI values of<br />
the ternary combination and the CI values of the Fn þ Gm<br />
combination (r ¼ 0.996) which suggests that in the Anabaena<br />
test, this particular combination seemed to be the predominant<br />
in the ternary toxicological interaction. Selected average<br />
CI values for both Vibrio fischeri and Anabaena CPB4337 tests at<br />
three representative dose levels (EC10, EC50 and EC90) and the<br />
combined effects are summarized in Table 2.<br />
3.2. Toxicological interactions of wastewater and<br />
fibrate combinations in the Anabaena CPB4337<br />
bioluminescence test<br />
In a recent previous study (Rosal et al., 2009), we found that<br />
a wastewater sample collected from a local STP was very toxic<br />
to Anabaena cells with a wastewater dilution of 0.11 causing<br />
50% luminescence inhibition (wastewater EC50). The observed<br />
toxicity was attributed to the combined toxicities of over thirty<br />
micropollutants, which included fibrates as well as other<br />
pharmaceuticals (Rosal et al., 2008). We sought to investigate<br />
the nature of the interaction between the wastewater (WW)<br />
and the three fibrates in binary (Fn þ WW; Bz þ WW and<br />
Gm þ WW) and quaternary (Fn þ Gm þ Bz þ WW) combinations;<br />
for these experiments, the wastewater itself was<br />
regarded as a toxicant; the experimental design was analogous<br />
to the one for the three fibrate interactions and is also shown<br />
in Table 1. The r-values of the median-effect plots were >0.95<br />
in all cases, indicating that the data conformed to the medianeffect<br />
principle (not shown). Fig. 2 shows the fa-CI plot for each
Table 3 – Correlation analyses between CI values of fibrate ternary and fibrate D wastewater quaternary combinations ( y)<br />
and their binary combinations (x) for Vibrio fischeri and Anabaena CPB4337 tests.<br />
Test organism Combinations Regression parameters<br />
of the binary fibrate-wastewater combination and the<br />
quaternary combination; as can be observed, in a broad range<br />
of fa values, the binary combinations showed a strong synergism;<br />
however, at fa values above 0.8, the binary Fn þ WW and<br />
Bz þ WW combinations approached an additive effect and at<br />
fa values above 0.95, these two combinations yielded antagonism;<br />
by contrast, the Gm þ WW combination became even<br />
more synergistic. The quaternary combination interaction<br />
showed a strong synergism through a broad range of fa values<br />
but also turned into slight antagonism at fa values above 0.95,<br />
Combination index, CI<br />
closely resembling the pattern of the Fn þ WW and Bz þ WW<br />
interactions which is confirmed by the highest r value<br />
(r ¼ 0.999) in the correlation analyses (Table 3), which suggests<br />
a predominant effect of Fn and Bz in the quaternary<br />
interaction.<br />
The computer software CompuSyn (Chou and Martin, 2005)<br />
displays a type of graphic termed polygonogram, which is<br />
a semiquantitative method of representing interactions<br />
between three or more compounds at a determined fa value.<br />
This graphic allows a simplified visual presentation of the<br />
overall results. Fig. 3 shows the polygonogram for the three<br />
fibrates and the wastewater at four fa values; synergism is<br />
indicated by solid lines and antagonism by broken ones; the<br />
thickness of the lines indicates the strength of the interaction.<br />
The polygonogram clearly shows the synergistic interaction of<br />
wastewater in combination with each of the three fibrates at<br />
low fa values and the antagonistic interaction that appeared at<br />
the highest fa value, 0.99, for the Fn þ WW and the Bz þ WW<br />
combinations.<br />
The same wastewater sample collected from a local STP<br />
was proved as responsible of stimulation of the bioluminescence<br />
activity of Vibrio fischeri to 110–120% of that of the<br />
control. Moreover, the EC50 values for the fibrates in the<br />
wastewater were higher than those for fibrates in pure water<br />
(Rosal et al., 2009). The same trend was observed comparing<br />
the dose-effect curve parameters (Dm, m and r) for the ternary<br />
combination (Fn þ Gm þ Bz) of fibrates in ddH 2O and wastewater.<br />
The dose required to produce the median-effect (Dm)in<br />
Vibrio fischeri test when (Fn þ Gm þ Bz) were solved in wastewater<br />
was 131.936 compared to 55.6951 mg/L required when<br />
ddH2O was employed. CI values could not be calculated for<br />
Vibrio fischeri due to the fact that the wastewater itself was not<br />
toxic to this bacterium; synergism or antagonism could not be<br />
properly estimated (Chou, 2006).<br />
4. Discussion<br />
xo m r<br />
V. fischeri Fn þ Gm þ Bz versus Gm þ Bz 0.614 1.77 0.83<br />
Fn þ Bz 0.594 0.281 0.91<br />
Fn þ Gm 0.067 1.40 0.81<br />
Anabaena CPB4337 Fn þ Gm þ Bz versus Gm þ Bz 5.876 4.39 0.91<br />
Fn þ Bz 2.716 3.00 0.941<br />
Fn þ Gm 0.282 0.247 0.996<br />
WW þ Fn þ Gm þ Bz versus Gm þ Bz 0.079 0.372 0.999<br />
Fn þ Bz 0.199 0.246 0.998<br />
Fn þ Gm 0.464 0.017 0.897<br />
Fn þ WW 0.253 1.31 0.999<br />
Gm þ WW 2.131 2.41 0.89<br />
Bz þ WW 0.003 0.865 0.999<br />
Fn ¼ fenofibric acid, Bz ¼ bezafibrate, Gm ¼ gemfibrozil, WW ¼ wastewater. The parameters of linear regression equations: x0 (value of y when<br />
x ¼ 0); m (slope) and r (correlation coefficient) with all p-values of 0.001. Analyses were computed using MINITAB Release 14 for Windows.<br />
2.0<br />
1.5<br />
1.0<br />
0.5<br />
0.0<br />
0.0 0.2 0.4 0.6 0.8 1.0<br />
Fractions affected, fa<br />
Fig. 2 – Combination index plot ( fa-CI plot) for a set of three<br />
fibrates and toxic wastewater sample in their binary and<br />
quaternary combinations: Gn D WW (–6–), Fn D WW (–B–),<br />
Bz D WW (–,–) and Fn D Gm D Bz D WW (–;–) for the<br />
Anabaena CPB4337 test. CI values are plotted as a function of<br />
the fractional inhibition of bioluminescence ( fa) by<br />
computer simulation (CompuSyn) from fa [ 0.10 to 0.95.<br />
CI 1 indicates synergism, additive effect and<br />
antagonism, respectively. At least three independent<br />
experiments with two replicates were used. The vertical<br />
bars indicate 95% confidence intervals for CI values based on<br />
sequential deletion analysis (SDA) (Chou and Martin, 2005).<br />
Fn [ fenofibric acid, Bz [ bezafibrate, Gm [ gemfibrozil,<br />
WW [ wastewater.<br />
water research 44 (2010) 427–438 433<br />
The three fibrates that we have used in our study are lipid<br />
modifying agents that are effective in lowering elevated
434<br />
Fig. 3 – Polygonograms showing the toxicological interactions of three fibrates and a toxic wastewater sample in their<br />
binary combinations (Fn D Bz, Bz D Gm, Fn D Gm, Gm D WW, Fn D WW, Bz D WW) as calculated by CompuSyn (Chou and<br />
Martin, 2005) for Anabaena CPB4337 test at four effect levels: fa [ 0.1 (a), fa [ 0.5 (b), fa [ 0.9 (c) and fa [ 0.99 (d). Solid lines<br />
indicate synergism, broken lines indicate antagonism. The thickness of the line represents the strength of synergism or<br />
antagonism. Figure generated by CompuSyn (Chou and Martin, 2005).<br />
plasma triglycerides and cholesterol in humans (Staels et al.,<br />
1998). These pharmaceuticals are highly used, ubiquitous and<br />
persistent (Daughton and Ternes, 1999), they are found at ng/L<br />
to mg/L levels in many STP effluents, surface waters, estuaries<br />
of rivers and even in sea water (for a review, see Hernando<br />
et al., 2007). Although non-target organisms; the continuous<br />
release of these substances into the environment may cause<br />
acute or chronic toxicity to the aquatic biota. Regarding<br />
fibrates, in the recent literature there are many reports dealing<br />
with individual toxicity of different fibrates in a range of<br />
aquatic organisms from primary producers to consumers;<br />
a great variability has been found in the sensitivity of the<br />
different test organisms toward these pharmaceuticals<br />
(Hernando et al., 2007). However, pharmaceuticals such as<br />
fibrates do not occur singly in a polluted environment and are<br />
usually found as mixtures, therefore, for risk assessment<br />
strategies it is important to know the combined effects of<br />
pharmaceuticals in non-target organisms (Teuschler, 2007).<br />
There are two concepts widely used for the prediction of<br />
mixture toxicity: concentration addition (CA) and independent<br />
action (IA) (Backhaus et al., 2003; Vighi et al., 2003;<br />
Backhaus et al., 2004; Junghans et al., 2006). CA is used for<br />
mixtures whose components act in a similar mode of action<br />
while IA is based on the idea of dissimilar action, meaning<br />
that the compounds have different mechanisms of action;<br />
however, as discussed by Cleuvers (2003) the terms similar/<br />
water research 44 (2010) 427–438<br />
dissimilar action may be misleading. Pharmaceuticals such as<br />
fibrates may have the same pharmacological mechanism of<br />
action [i.e., interaction with the binding peroxisome proliferator-activated<br />
receptor a (PPARa)] in their target organism,<br />
humans; however, if fibrates released in the aquatic environments<br />
prove toxic to different non-target organisms, the<br />
exact mechanism of toxicity (probably different to the pharmacological<br />
mode of action) should be investigated in depth<br />
before choosing which approach, CA or IA, to use. In fact, only<br />
if toxicity is regarded as non-specific at all, the concept of CA<br />
may be used although it may also have limitations. Cleuvers<br />
(2003) found that two totally different pharmaceuticals,<br />
a fibrate and an anti-epileptic drug, followed the concept of CA<br />
in the Daphnia toxicity test and the concept of IA in an algal<br />
test; both pharmaceuticals apparently shared the same nonspecific<br />
toxic mode of action for both organisms; so it<br />
appeared that the concept of CA or IA did not depend on<br />
a similar/dissimilar mode of action but on the tested<br />
organism. The author also discussed that by definition, when<br />
using CA, substances applied below their individual noneffect<br />
concentration (NOEC) will contribute to the total effect<br />
of the mixture while when using IA, substances applied below<br />
their NOEC will not contribute to the total effect of the<br />
mixture, meaning that any combination effect will probably<br />
be higher if the substances follow the concept of CA and this<br />
may be misleading when considering the terms synergism or
antagonism because as also discussed by Chou (2006), synergism<br />
or antagonism may occur independently of a similar or<br />
dissimilar mode of action. In this context, Fent et al. (2006)<br />
tested mixtures of different kinds of pharmaceuticals<br />
(including fibrates) that might have estrogenic activity in<br />
a yeast reporter system; they applied the CA model and found<br />
that it had severe limitations when the dose-response curves<br />
of the individual pharmaceuticals were not identical or at low<br />
effect concentrations. As pharmaceuticals released in the<br />
environment may have such diverse dose-effect relationships,<br />
the lack of appropriate prediction suggests limitation of<br />
the CA mixtures concept.<br />
To study the nature of the combined fibrate interactions<br />
(synergism, additive effect, antagonism) for the Vibrio fischeri<br />
and Anabaena CPB4337 bioluminescence tests, we have followed<br />
the combination index (CI)-isobologram equation<br />
method of Chou (2006) and Chou and Talalay (1984); a method<br />
widely used to study drug interactions in pharmacology. This<br />
method may be considered a fractional analysis technique for<br />
drug interactions (Berenbaum, 1981; Bovill, 1998) that is<br />
independent of the mode of action and considers both the<br />
potency (EC 50, Dm) and the shape (m) of the dose-effect curve<br />
for each drug. The method allows prediction of synergism/<br />
antagonism at all effect levels ( fa) for a combination of n<br />
drugs; in contrast with the classical graphical isobologram<br />
method (Berenbaum, 1981; Bovill, 1998) that cannot be used<br />
for more than three compounds and have also graphical<br />
limitations to show all effect levels. By using this method, we<br />
have been able to determine the nature of interactions for<br />
a wide range of effect levels of three fibrates in binary and<br />
ternary combinations in two different bioluminescent organisms.<br />
However, the nature of these interactions was not<br />
uniform along the fa levels range in any of the two organisms.<br />
In Vibrio fischeri, antagonism predominated at low and intermediate<br />
fa levels but at the highest effect levels, interactions<br />
became additive or slightly synergistic. In Anabaena, a dual<br />
synergistic/antagonistic behaviour was observed with synergism<br />
predominating at fa levels below 0.4–0.5 and strong<br />
antagonism above these fa values. It is difficult to give an<br />
explanation to this phenomenon because the combination<br />
index method only allows quantitative determination of<br />
synergism or antagonism and the elucidation of the mechanism<br />
by which synergism or antagonism occurs is a separate<br />
issue that needs a different kind of approach. However,<br />
tentatively, antagonism, which could be considered the<br />
predominant interaction in Vibrio fischeri and Anabaena, might<br />
be explained by the structural similarity of fibrates which are<br />
related pharmaceuticals that share a common structural<br />
motif, a cyclic head and a hydrophobic tail (Rosal et al., 2009);<br />
at the fa levels where antagonism is found in both organisms,<br />
fibrates may compete with one another for the same target/<br />
receptor sites. The slight synergism found at very high levels<br />
in Vibrio fischeri could perhaps be explained by the fact that at<br />
very high concentrations, fibrates may somehow combine to<br />
increase toxicity by an unspecific way of action that is probably<br />
not related to their pharmacological mechanism.<br />
Perhaps, the most puzzling interaction is the observed high<br />
synergism at very low fa levels in Anabaena; the mechanism of<br />
such synergistic interaction is not readily apparent. One could<br />
speculate that these fibrates at very low concentrations could<br />
water research 44 (2010) 427–438 435<br />
involve what Jia et al. (2009) in their extensive review of<br />
mechanisms of drug combinations call ‘‘facilitating actions’’<br />
that means that secondary actions of one drug enhances the<br />
activity or level of another drug in the mixture or alternatively<br />
‘‘complementary actions’’ when drugs act at the same target<br />
at different sites, at overlapping sites or at different targets of<br />
the same pathway. However, in the literature there are very<br />
few reports on possible targets of fibrates on the prokaryotic<br />
cell; English et al. (1994) reported that peroxisome proliferators<br />
such as fibrates have been shown to induce cytochrome<br />
P450BM-3 which catalyzes the hydroxylation of fatty<br />
acids, in Bacillus megaterium. Garbe (2004) reported that<br />
fibrates induced methyltransferase Rv0560c with a function in<br />
the biosynthesis of isoprenoid compounds in Mycobacterium<br />
tuberculosis; Garbe (2004) suggested that both effects may act<br />
on the plasma membrane, modulating its properties. In<br />
mitochondria, which have significant features that resemble<br />
those of prokaryotes, fibrates have been found to inhibit<br />
respiratory complex I (NDH-1 complex) and to interfere with<br />
mitochondrial fatty acid oxidation (Scatena et al., 2007).<br />
Whether fibrates may exert similar effects in Vibrio fischeri and<br />
Anabaena to those observed in Bacillus or mitochondria needs<br />
further research. In this context, we have found that, as the fa-<br />
CI plots show, fibrate interactions do not follow the same<br />
pattern in both bacteria, this may be due to the different origin<br />
and position in the food web of Vibrio fischeri, a heterotrophic<br />
marine prokaryote and Anabaena CPB4337, a recombinant<br />
strain of an obligate phototrophic freshwater prokaryote; in<br />
fact, Anabaena presents intracellular photosynthetic<br />
membranes called thylakoids where several functionally<br />
distinct NDH-1 complexes have been found with roles both in<br />
respiration and photosynthesis (Battchikova and Aro, 2007). If<br />
fibrates are also affecting NDH-1 complexes in Anabaena, their<br />
effects might be very different to those in Vibrio fischeri; so,<br />
although we have measured the same toxicity endpoint in<br />
both bacteria, i.e., luminescence inhibition, the combined<br />
effects of fibrates seem to depend on the test organism.<br />
Ince et al. (1999) assessed toxic interactions of heavy<br />
metals in binary mixtures on Vibrio fischeri and the freshwater<br />
aquatic plant Lemna minor and found that most binary metal<br />
mixtures exhibited only antagonistic interactions in the plant<br />
opposed to fewer antagonistic and some synergistic interactions<br />
in the heterotrophic bacterium. These authors also<br />
found that in the bacterium, the nature of the interaction<br />
(synergism or antagonism) also changed with the effect level<br />
of the binary metal combinations, although the authors did<br />
not provide a mechanistic explanation for this variability.<br />
Cheng and Lu (2002) made a comparison of joint interactions<br />
of organic toxicants in binary mixtures in Escherichia coli and<br />
Vibrio fischeri and found that toxicants with the same mechanisms<br />
of toxicity displayed mostly additive or antagonistic<br />
interactions in E. coli and Vibrio fischeri; however a synergistic<br />
interaction was found between glutardialdehyde and butyraldehyde<br />
in Vibrio. Synergistic effects in both bacteria were<br />
mostly associated with toxicants with different mechanisms<br />
of toxicity, although antagonism clearly predominated. They<br />
also found that for a total of 44 organic binary mixtures, only<br />
six mixtures resulted in identical type of interaction in both<br />
bacteria. From our results and those of other authors’ (Ince<br />
et al., 1999; Cheng and Lu, 2002; Cleuvers, 2003) one may
436<br />
conclude that previous knowledge of the mechanism of toxic<br />
action of a compound is not useful enough to predict which<br />
kind of interactions it will display when combined with other<br />
toxicants with the same or different toxic mechanism; also, as<br />
we have shown, the nature of the interaction may depend on<br />
the effect level of the mixture. In addition, different types of<br />
organisms will show completely different responses to<br />
mixtures of potential toxicants.<br />
We previously found that a local wastewater was very toxic<br />
for the Anabaena CPB4337 test but non-toxic at all for the Vibrio<br />
fischeri or Daphnia magna tests. This wastewater is a mixture of<br />
over thirty micropollutants, mostly pharmaceuticals of<br />
different therapeutics groups that, besides the fibrates used in<br />
this study, included antibiotics, analgesics/anti-inflammatories,<br />
b-blockers, antidepressants, anti-epileptics/psychiatrics,<br />
ulcer healing compounds, diuretics and bronchodilators;<br />
personal care products and some priority organic pollutants<br />
are also present (Rosal et al., 2008). The method of Chou allows<br />
to combine one drug mixture with another drug mixture and<br />
determine their interactions; therefore, we studied the nature<br />
of the interaction of fibrates and wastewater in the Anabaena<br />
bioluminescence test; interestingly, we found that in a wide<br />
range of effect levels, the interaction of wastewater and the<br />
three fibrate combination was synergistic; particularly, at very<br />
low fa values which means that fibrates are at low concentrations<br />
and the wastewater is diluted several-fold, the method<br />
predicted a strong synergism; this may be due, as discussed<br />
above, to the observed synergistic interactions of fibrates with<br />
one another as well as interactions with some of the detected<br />
micropollutants when present at very low concentrations. This<br />
observed synergism may be environmentally relevant since<br />
most pharmaceuticals such as fibrates do not usually show<br />
acute toxicity on non-target organisms when tested at real<br />
environmental concentrations (Hernando et al., 2007) but in<br />
a mixture, if they act synergistically, they could prove toxic for<br />
a test organism even at low concentrations; these results agree<br />
with those found by Hernando et al. (2004) who reported<br />
synergistic toxic effects for Daphnia magna test when wastewater<br />
was spiked with environmental concentrations of<br />
several pharmaceuticals including fibrates. By contrast, our<br />
results show that at high fa values ( fa > 0.8), the combined<br />
interaction of the quaternary fibrates þ wastewater combination,<br />
the binary Fn þ WW and Bz þ WW combinations<br />
approached an additive effect and eventually became antagonistic;<br />
in our previous study, the wastewater itself decreased<br />
Anabaena bioluminescence by 84% with a lower confidence<br />
limit of 76% and an upper confidence limit of 91%; when the<br />
wastewater was spiked with increasing concentrations of each<br />
fibrate we found that, with the exception of gemfibrozil, the<br />
EC50 values for the fibrates in the wastewater were higher than<br />
those for fibrates in pure water; this was attributed either to<br />
reduced bioavailability or to antagonistic effects of fibrates<br />
with other chemicals present in the wastewater; although we<br />
did not use the method of Chou, we obtained similar results to<br />
the ones we report in this study; that is, at high effect levels<br />
(>84% luminescence inhibition) the interaction of fibrates with<br />
wastewater, except the Gm þ WW combination, showed<br />
antagonism.<br />
Based on our results, we propose that the combination<br />
index (CI)-isobologram equation, a method widely used in<br />
water research 44 (2010) 427–438<br />
pharmacology both for in vitro and in vivo bioassays, may also<br />
be applied in environmental toxicology as a general method to<br />
define interactions of potential toxicants in mixtures in any<br />
test organism and/or toxicological endpoint of interest and<br />
could be especially useful for risk assessment strategies that<br />
take into account the toxicological interactions of substances<br />
in a mixture.<br />
5. Conclusions<br />
We report an environmental application of the combination<br />
index (CI)-isobologram equation to study the nature of the<br />
interactions of fibrate combinations in two bioluminescent<br />
aquatic organisms. The method allowed calculating synergism<br />
or antagonism of binary and ternary fibrate combinations at all<br />
effect levels simultaneously; we could also test the method<br />
with a real wastewater sample in binary and quaternary<br />
combination with the fibrates, finding that at very low effect<br />
levels, the fibrates acted synergistically with the wastewater in<br />
the Anabaena test. The proposed method may be used with<br />
other test organisms and/or toxicological endpoints and could<br />
be particularly useful for risk assessment approaches to<br />
toxicity of complex mixtures.<br />
Acknowledgements<br />
The research was funded by the Spanish Ministry of Education<br />
through grants CTM2005-03080/TECNO and CSD2006-00044<br />
and the Comunidad de Madrid, grants 0505/AMB-0395 and<br />
0505/MB/0321.<br />
references<br />
Allen, M.B., Arnon, D.I., 1955. Studies on nitrogen-fixing blue<br />
green algae. I Growth and nitrogen fixation by Anabaena<br />
cylindrica Lemm. Plant Physiol 30 (4), 366–372.<br />
Andreozzi, R., Raffaele, M., Nicklas, P., 2003. Pharmaceuticals in<br />
STP effluents and their solar photodegradation in aquatic<br />
environment. Chemosphere 50 (10), 1319–1330.<br />
Backhaus, T., Altenburger, R., Arrhenius, A˚ ., Blanck, H., Faust, M.,<br />
Finizio, A., Gramatica, P., Grote, M., Junghans, M., Meyer, W.,<br />
Pavan, M., Porsbring, T., Scholze, M., Todeschini, R., Vighi, M.,<br />
Walter, H., Horst Grimme, L., 2003. The BEAM-project:<br />
prediction and assessment of mixture toxicities in the aquatic<br />
environment. Continental Shelf Res. 23 (17–19), 1757–1769.<br />
Backhaus, T., Arrhenius, A., Blanck, H., 2004. Toxicity of a mixture<br />
of dissimilarly acting substances to natural algal<br />
communities: predictive power and limitations of<br />
independent action and concentration addition. Environ. Sci.<br />
Technol 38 (23), 6363–6370.<br />
Battchikova, N., Aro, E.-M., 2007. Cyanobacterial NDH-1<br />
complexes: multiplicity in function and subunit composition.<br />
Physiol. Plant 131 (1), 22–32.<br />
Berenbaum, M.C., 1981. Criteria for analyzing interactions<br />
between biologically active agents. Adv. Cancer Res. 35,<br />
269–335.<br />
Borgmann, U., Bennie, D.T., Ball, A.L., Palabrica, V., 2007. Effect of<br />
a mixture of seven pharmaceuticals on Hyalella azteca over<br />
multiple generations. Chemosphere 66 (7), 1278–1283.
Bovill, J.G., 1998. Analysis of drug interactions. Baillière’s Clin.<br />
Anaesthesiol 12 (2), 135–168.<br />
Brain, R.A., Johnson, D.J., Richards, S.M., Hanson, M.L.,<br />
Sanderson, H., Lam, M.W., Young, C., Mabury, S.A., Sibley, P.K.,<br />
Solomon, K.R., 2004. Microcosm evaluation of the effects of an<br />
eight pharmaceutical mixture to the aquatic macrophytes Lemna<br />
gibba and Myriophyllum sibiricum. Aquat. Toxicol 70 (1), 23–40.<br />
Castiglioni, S., Bagnati, R., Fanelli, R., Pomati, F., Calamari, D.,<br />
Zuccato, E., 2006. Removal of pharmaceuticals in sewage<br />
treatment plants in Italy. Environ. Sci. Technol 40 (1),<br />
357–363.<br />
Chen, C.-Y., Lu, C.-L., 2002. An analysis of the combined effects of<br />
organic toxicants. Sci. Total Environ 289 (1–3), 123–132.<br />
Chou, T.C., 1976. Derivation and properties of Michaelis–Menten<br />
type and Hill type equations for reference ligands. J. Theor.<br />
Biol. 59 (2), 253–276.<br />
Chou, T.C., 2006. Theoretical basis, experimental design, and<br />
computerized simulation of synergism and antagonism in<br />
drug combination studies. Pharmacol. Rev. 58 (3), 621–681.<br />
Chou, T.C., Martin, N., 2005. CompuSyn for Drug Combinations:<br />
PC Software and User’s Guide: A Computer Program for<br />
Quantification of Synergism and Antagonism in Drug<br />
Combinations and the Determination of IC 50 and ED 50 and<br />
LD 50 Values. ComboSyn, Inc., Paramus, NJ.<br />
Chou, T.C., Talalay, P., 1984. Quantitative analysis of dose-effect<br />
relationships: the combined effects of multiple drugs or<br />
enzyme inhibitors. Adv. Enzyme Regul 22, 27–55.<br />
Christensen, A.M., Faaborg-Andersen, S., Ingerslev, F., Baun, A.,<br />
2007. Mixture and single-substance toxicity of selective<br />
serotonin reuptake inhibitors toward algae and crustaceans.<br />
Environ. Toxicol. Chem. 26 (1), 85–91.<br />
Cleuvers, M., 2003. Aquatic ecotoxicity of pharmaceuticals<br />
including the assessment of combination effects. Toxicol. Lett.<br />
142 (3), 185–194.<br />
Cleuvers, M., 2004. Mixture toxicity of the anti-inflammatory<br />
drugs diclofenac, ibuprofen, naproxen, and acetylsalicylic<br />
acid. Ecotoxicol. Environ. Saf 59 (3), 309–315.<br />
Comerton, A.M., Andrews, R.C., Bagley, D.M., Yang, P., 2007.<br />
Membrane adsorption of endocrine disrupting compounds<br />
and pharmaceutically active compounds. J. Memb. Sci. 303<br />
(1–2), 267–277.<br />
Crane, M., Watts, C., Boucard, T., 2006. Chronic aquatic<br />
environmental risks from exposure to human<br />
pharmaceuticals. Sci. Total Environ 367 (1), 23–41.<br />
Daughton, C.G., Ternes, T.A., 1999. Pharmaceuticals and personal<br />
care products in the environment: agents of subtle change?<br />
Environ. Health Perspect 107 (6), 907–938.<br />
English, N., Hughes, V., Wolf, C.R., 1994. Common pathways of<br />
cytochrome P450 gene regulation by peroxisome proliferators<br />
and barbiturates in Bacillus megaterium ATCC14581. J. Biol.<br />
Chem. 269 (43), 26836–26841.<br />
Fent, K., Escher, C., Caminada, D., 2006. Estrogenic activity of<br />
pharmaceuticals and pharmaceutical mixtures in a yeast<br />
reporter gene system. Reprod. Toxicol 22 (2), 175–185.<br />
Garbe, T.R., 2004. Co-induction of methyltransferase Rv0560c by<br />
naphthoquinones and fibric acids suggests attenuation of<br />
isoprenoid quinone action in Mycobacterium tuberculosis. Can. J.<br />
Microbiol. 50 (10), 771–778.<br />
Gros, M., Petrovic, M., Barcelo, D., 2007. Wastewater treatment<br />
plants as a pathway for aquatic contamination by<br />
pharmaceuticals in the Ebro river basin (northeast Spain).<br />
Environ. Toxicol. Chem. 26 (8), 1553–1562.<br />
Han, G.H., Hur, H.G., Kim, S.D., 2006. Ecotoxicological risk of<br />
pharmaceuticals from wastewater treatment plants in Korea:<br />
occurrence and toxicity to Daphnia magna. Environ. Toxicol.<br />
Chem. 25 (1), 265–271.<br />
Hernando, M.D., Petrovic, M., Fernández-Alba, A.R., Barceló, D.,<br />
2004. Analysis by liquid chromatography-electrospray<br />
water research 44 (2010) 427–438 437<br />
ionization tandem mass spectrometry and acute toxicity<br />
evaluation for beta-blockers and lipid-regulating agents in<br />
wastewater samples. J. Chromatogr. A 1046 (1–2), 133–140.<br />
Hernando, M.D., Agüera, A., Fernández-Alba, A.R., 2007. LC-MS<br />
analysis and environmental risk of lipid regulators. Anal.<br />
Bioanal. Chem. 387 (4), 1269–1285.<br />
Holoshitz, N., Alsheikh-Ali, A.A., Karas, R.H., 2008. Relative safety<br />
of gemfibrozil and fenofibrate in the absence of concomitant<br />
cerivastatin use. Am. J. Cardiol 101 (1), 95–97.<br />
Ince, N.H., Dirilgen, N., Apikyan, I.G., Tezcanli, G., Üstün, B., 1999.<br />
Assessment of toxic interactions of heavy metals in binary<br />
mixtures: a statistical approach. Arch. Environ. Contam.<br />
Toxicol 36, 365–372.<br />
International Organization for Standardization, 2007. Water<br />
quality - Determination of the inhibitory effect of water<br />
samples on the light emission of Vibrio fischeri (Luminescent<br />
bacteria test), ISO 11348-3 Revised version, Geneva,<br />
Switzerland.<br />
Jia, J., Zhu, F., Ma, X., Cao, Z.W., Li, Y.X., Chen, Y.Z., 2009.<br />
Mechanisms of drug combinations: interaction and network<br />
perspectives. Nat. Rev. Drug Discov 8 (2), 111–128.<br />
Junghans, M., Backhaus, T., Faust, M., Scholze, M., Grimme, L.H.,<br />
2006. Application and validation of approaches for the<br />
predictive hazard assessment of realistic pesticide mixtures.<br />
Aquat. Toxicol 76 (2), 93–110.<br />
Kasprzyk-Hordern, B., Dinsdale, R.M., Guwy, A.J., 2009. The<br />
removal of pharmaceuticals, personal care products,<br />
endocrine disruptors and illicit drugs during wastewater<br />
treatment and its impact on the quality of receiving waters.<br />
Water Res. 43 (2), 363–380.<br />
Lambropoulou, D.A., Hernando, M.D., Konstantinou, I.K.,<br />
Thurman, E.M., Ferrer, I., Albanis, T.A., Fernandez-Alba, A.R.,<br />
2008. Identification of photocatalytic degradation products of<br />
bezafibrate in TiO 2 aqueous suspensions by liquid and gas<br />
chromatography. J. Chromatogr. A 1183 (1–2), 38–48.<br />
Metcalfe, C.D., Miao, X.S., Koenig, B.G., Struger, J., 2003.<br />
Distribution of acidic and neutral drugs in surface waters near<br />
sewage treatment plants in the lower Great Lakes, Canada.<br />
Environ. Toxicol. Chem. 22 (12), 2881–2889.<br />
Miller, D.B., Spence, J.D., 1998. Clinical pharmacokinetics of fibric<br />
acid derivatives (fibrates). Clin. Pharmacokinet 34 (2), 155–162.<br />
Pomati, F., Orlandi, C., Clerici, M., Luciani, F., Zuccato, E., 2008.<br />
Effects and interactions in an environmentally relevant<br />
mixture of pharmaceuticals. Toxicol. Sci. 102 (1), 129–137.<br />
Quinn, B., Gagne, F., Blaise, C., 2009. Evaluation of the acute,<br />
chronic and teratogenic effects of a mixture of eleven<br />
pharmaceuticals on the cnidarian, Hydra attenuata. Sci. Total<br />
Environ 407 (3), 1072–1079.<br />
Rodea-Palomares, I., González-García, C., Leganés, F., Fernández-<br />
Piñas, F., 2009. Effect of pH, EDTA, and anions on heavy<br />
metal toxicity toward a bioluminescent cyanobacterial<br />
bioreporter. Arch. Environ. Contam. Toxicol. doi:10.1007/<br />
s00244-008-9280-9.<br />
Rodríguez, A., Rosal, R., Perdigón, J.A., Mezcua, M., Agüera, A.,<br />
Hernando, M.D., Letón, P., Fernández-Alba, A.R., García-<br />
Calvo, E., 2008. In: Barceló, D., Petrovic, M. (Eds.), Emerging<br />
Contaminants from Industrial and Municipal Waste. Springer-<br />
Verlag, Berlin, pp. 127–175.<br />
Rosal, R., Rodríguez, A., Perdigón-Melón, J.A., Mezcua, M.,<br />
Hernando, M.D., Letón, P., García-Calvo, E., Agüera, A.,<br />
Fernández-Alba, A.R., 2008. Removal of pharmaceuticals and<br />
kinetics of mineralization by O 3/H 2O 2 in a biotreated<br />
municipal wastewater. Water Res. 42 (14), 3719–3728.<br />
Rosal, R., Rodea-Palomares, I., Boltes, K., Fernández-Piñas, F.,<br />
Leganés, F., Gonzalo, S., Petre, A., 2009. Ecotoxicity assessment<br />
of lipid regulators in water and biologically treated<br />
wastewater using three aquatic organism. Environ. Sci. Pollut.<br />
Res. doi:10.1007/s11356-009-0137-1.
438<br />
Scatena, R., Bottoni, P., Botta, G., Martorana, G.E., Giardina, B.,<br />
2007. The role of mitochondria in pharmacotoxicology:<br />
a reevaluation of an old, newly emerging topic. Am. J. Physiol.<br />
Cell Physiol 293 (1), C12–C21.<br />
Staels, B., Dallongeville, J., Auwerx, J., Schoonjans, K., Leitersdorf, E.,<br />
Fruchart, J.C., 1998. Mechanism of action of fibrates on lipid and<br />
lipoprotein metabolism. Circulation 98 (19), 2088–2093.<br />
Stumpf, M., Ternes, T.A., Wilken, R.D., Rodrigues, S.V., Baumann, W.,<br />
1999. Polar drug residues in sewage and natural waters in the<br />
state of Rio de Janeiro. Brazil. Sci. Total Environ 225 (1–2), 135–141.<br />
water research 44 (2010) 427–438<br />
Teuschler, L.K., 2007. Deciding which chemical mixtures risk<br />
assessment methods work best for what mixtures. Toxicol.<br />
Appl. Pharmacol 223 (2), 139–147.<br />
Vighi, M., Altenburger, R., Arrhenius, A., Backhaus, T.,<br />
Bodeker,W.,Blanck,H.,Consolaro,F.,Faust,M.,<br />
Finizio, A., Froehner, K., Gramatica, P., Grimme, L.H.,<br />
Gronvall, F., Hamer, V., Scholze, M., Walter, H., 2003.<br />
Water quality objectives for mixtures of toxic chemicals:<br />
problems and perspectives. Ecotoxicol. Environ. Saf 54 (2),<br />
139–150.
Toxication or detoxication? In vivo toxicity assessment of<br />
ozonation as advanced wastewater treatment with the<br />
rainbow trout<br />
Daniel Stalter a, *, Axel Magdeburg a , Mirco Weil b , Thomas Knacker b ,Jörg Oehlmann a<br />
a<br />
Goethe University Frankfurt am Main, Biological Sciences Division, Department of Aquatic Ecotoxicology, Siesmayerstrasse 70, 60054<br />
Frankfurt, Hessen, Germany<br />
b<br />
ECT Oekotoxikologie GmbH, Böttgerstrasse 2–4, 65439 Flörsheim, Germany<br />
article info<br />
Article history:<br />
Received 23 March 2009<br />
Received in revised form<br />
16 July 2009<br />
Accepted 18 July 2009<br />
Available online 25 July 2009<br />
Keywords:<br />
Sewage treatment<br />
Pharmaceuticals<br />
Oxidation byproducts<br />
Vitellogenin<br />
Emerging contaminants<br />
Advanced oxidation process<br />
Rainbow trout<br />
Fish early life stage toxicity test<br />
1. Introduction<br />
abstract<br />
Wastewater (WW) is one of the major sources of micropollutants<br />
introduced into the aquatic environment<br />
(Schwarzenbach et al., 2006). The large spectrum of pharmaceuticals<br />
and personal care products (PPCPs) occurring in<br />
water research 44 (2010) 439–448<br />
Available at www.sciencedirect.com<br />
journal homepage: www.elsevier.com/locate/watres<br />
* Corresponding author. Tel.: þ49 69 7982 4882; fax: þ49 69 7982 4748.<br />
E-mail address: stalter@bio.uni-frankfurt.de (D. Stalter).<br />
0043-1354/$ – see front matter ª 2009 Elsevier Ltd. All rights reserved.<br />
doi:10.1016/j.watres.2009.07.025<br />
Ozonation as advanced wastewater treatment method is an effective technique for<br />
micropollutant removal. However, the application of this method carries the inherent<br />
danger to produce toxic oxidation byproducts. For an ecotoxicological assessment<br />
conventionally treated wastewater, wastewater after ozonation and ozonated wastewater<br />
after sand filtration were evaluated in parallel at an operating treatment plant via the fish<br />
early life stage toxicity test (FELST) using rainbow trout (Oncorhynchus mykiss).<br />
The FELST revealed a considerable developmental retardation of test organisms<br />
exposed to ozonated WW. This was accompanied by a significant decrease in body weight<br />
and length compared to reference water, to the conventionally treated WW and to the<br />
ozonated water after sand filtration. Hence sand filtration obviously prevents from adverse<br />
ecotoxicological effects of ozonation.<br />
An additional test with yolk-sac larvae resulted in a significant reduction of vitellogenin<br />
levels in fish exposed to ozonated wastewater compared to fish reared in conventionally<br />
treated wastewater. This demonstrates the effective removal of estrogenic activity by<br />
ozonation.<br />
Adverse ozonation effects may have been a result of the conversion of chemicals into<br />
more toxic metabolites. However, sand filtration reduced toxication effects indicating that<br />
these oxidation byproducts are readily degradable or adsorbable. The results indicate that<br />
in any case ozonation should not be applied without subsequent post treatment appropriate<br />
for oxidation byproducts removal (e.g. sand filtration).<br />
ª 2009 Elsevier Ltd. All rights reserved.<br />
surface waters (Daughton and Ternes, 1999), particularly with<br />
regard to mixture toxicity, may pose a potential threat to<br />
aquatic wildlife. But also single substances may have the<br />
capability to endanger the ecosystem as for example diclofenac<br />
can lead to an impairment of the general health condition<br />
of rainbow trout at environmentally relevant concentrations
440<br />
Nomenclature<br />
AOC assimilable organic carbon<br />
C control<br />
COD chemical oxygen demand<br />
DOC dissolved organic carbon<br />
EE2 17a-ethinylestradiol<br />
ELISA enzyme linked immunosorbent assay<br />
EQS environmental quality standard<br />
FELST fish early life stage toxicity test<br />
FS wastewater after final sedimentation<br />
GJIC gap junctional intercellular communication<br />
MS222 tricaine methanesulfonate<br />
NH4–N ammonium nitrogen<br />
(Schwaiger et al., 2004; Triebskorn et al., 2004). 17a-ethinylestradiol<br />
(EE2) is considered to be the most potent estrogenic<br />
active compound for fish with the lowest observed effect<br />
concentration of 0.1 ng/L for vitellogenesis in rainbow trout<br />
(Purdom et al., 1994). Moreover it leads to a reduced egg fertilisation<br />
success and skewed sex ratio toward females of fat<br />
head minnows at a LOEC of 0.32 ng/L (Parrott and Blunt, 2005).<br />
Furthermore, industrial surfactants like alkylphenolic ethoxylates<br />
are likely to be responsible for feminization effects in<br />
rainbow trout (Jobling et al., 1996) and complex mixtures of<br />
estrogenic chemicals present in the environment have been<br />
shown to act additively (Brian et al., 2005). The concentrations<br />
of such compounds in WW often exceed effect concentrations<br />
(Ternes et al., 1999; Ying et al., 2002; Tixier et al., 2003) and<br />
thus even treated WW contains high amounts of estrogenic<br />
substances that impair the endocrine system of exposed fish<br />
(Larsson et al., 1999; Rodgers-Gray et al., 2001) resulting in<br />
feminization effects of males and consequently reduced<br />
fertility of field populations (Jobling et al., 2002).<br />
Besides, in the European Union the reduction of the<br />
contamination of surface waters with hazardous substances<br />
is defined by the Water Framework Directive requiring a ‘‘good<br />
status’’ for all coastal and inland waters until 2015 (European<br />
Commission, 2000). One tool is the implementation of environmental<br />
quality standards (EQSs) for mono substance<br />
pollutants exhibiting a significant risk to the aquatic<br />
ecosystem. The discharge of such compounds, classified as<br />
priority substances, is envisaged to be progressively reduced<br />
by 2025 or 5 years after inclusion in the list for priority<br />
substances, respectively.<br />
However, till now so called ‘‘emerging contaminants’’ are<br />
only marginally addressed and PPCPs are not included in this<br />
list. But the latter has to be reviewed at least every 4 years and<br />
provisions have been made to add several hazardous<br />
emerging contaminants. Amongst others the inclusion of<br />
diclofenac, EE2 and carbamazepine is discussed and additionally<br />
recommended by the European Parliament since 2007<br />
(Jensen, 2007). Moreover EQSs meanwhile have been proposed<br />
for several PPCPs based on reliable ecotoxicity data (e.g.<br />
diclofenac: 0.1 mg/L, EE2: 0.03 ng/L, carbamazepine: 0.5 mg/L;<br />
Jahnel et al., 2006; Moltmann et al., 2007). Unfortunately, some<br />
PPCPs exceed the EQSs in surface waters of some urbanised<br />
water research 44 (2010) 439–448<br />
NO 3–N nitrate nitrogen<br />
O wastewater after ozonation<br />
OECD Organisation for Economic Cooperation and<br />
Development<br />
OS wastewater after ozonation and sand filtration<br />
P phosphate<br />
PAH polycyclic aromatic hydrocarbons<br />
PAC powdered activated carbon<br />
PPCPs pharmaceuticals and personal care products<br />
rcf relative centrifugal force<br />
SPM suspended particulate matter<br />
VTG vitellogenin<br />
WW wastewater<br />
WWTP wastewater treatment plant<br />
regions. That is, inter alia, a result of the fact that conventional<br />
WW treatment plants are not designed for appropriate<br />
removal of PPCPs (Daughton and Ternes, 1999). Nevertheless<br />
in many countries surface waters serve as raw water resources<br />
for drinking water supply. Due to the public’s demand for safe<br />
drinking water and ecosystem health advanced WW treatment<br />
methods are required to allow for sufficient removal of<br />
micropollutants from sewage treatment effluents. Therefore,<br />
end of pipe techniques like activated carbon filtration and<br />
ozonation of WW might be crucial to achieve regulative<br />
requirements in a medium-term perspective.<br />
Huber et al. (2005) demonstrated that ozonation of<br />
secondary effluent is an effective tool for the removal of<br />
a wide range of pharmaceuticals among them diclofenac,<br />
carbamazepine and the estrogens estradiol, estrone and EE2,<br />
with degradation rates of more than 90% for ozone doses of<br />
2 mg/L. Merely iodinated X-ray contrast media and a few<br />
acidic pharmaceuticals were oxidized only partially. Nakada<br />
et al. (2007) confirmed the effective elimination of 22 pharmaceuticals<br />
during ozonation in an operating treatment<br />
plant. A further advantage of the ozonation is the sanitizing<br />
property of the method (Tyrrell et al., 1995). The effective<br />
removal of micropollutants and pathogens indicates that<br />
ozonation is a suitable technique for advanced WW treatment<br />
to reduce the contamination of the aquatic environment.<br />
Actually a reduced toxicity of a mixture of six pharmaceuticals<br />
treated by ozonation towards algae and rotifer was<br />
demonstrated by Andreozzi et al. (2004). Reduced estrogenic<br />
activity of an EE2-containing WW after ozonation was shown<br />
by Huber et al. (2004) and ozone treatment of WW significantly<br />
decreased the toxic potency of urban pollutants to the freshwater<br />
mussel Elliptio complanata (Gagne et al., 2007). However,<br />
ozone treatment typically transforms chemical compounds<br />
but does not mineralize them entirely. Consequently, not only<br />
the concentrations of mother compounds but also the inclusion<br />
of transformation products and their mixtures should be<br />
assessed for toxicity, too.<br />
For a risk-benefit analysis an extensive ecotoxicological<br />
evaluation of ozonated WW is essential. In vitro bioassays<br />
covering different modes of toxic action (e.g. estrogenicity,<br />
acetylcholine esterase activity or non-specific baseline<br />
toxicity) and performed with enriched water samples are
Table 1 – Wastewater quality summary (mg/L).<br />
SPM COD P NH4–N NO3–N pH<br />
Final sedimentation<br />
Median 4.8 17 0.19 0.1 8.3<br />
10% percentile 3.6 15 0.17 0.04 5.4 8.12<br />
90% percentile 8.96 23.8 0.22 0.27 10.1 8.4<br />
n 37 23 37 37 2 23<br />
Ozonation and sand filtration<br />
Median 2 15 0.17 0.04 9.8 8.3<br />
10% percentile 1.6 12.2 0.14 0.04 7.28 8.02<br />
90% percentile 2.4 19.8 0.19 0.15 12.48 8.4<br />
n 37 23 37 37 37 23<br />
suitable tools for toxicity characterization of WW as shown by<br />
Escher et al. (2008). These methods were shown to be highly<br />
sensitive even for identification and characterization of low<br />
toxic water samples. Nevertheless, as typical screening tools<br />
they are not designed to replace chronic in vivo tests of whole<br />
effluent samples. To allow for a more comprehensive and<br />
integrative assessment of the potentially hazardous impact of<br />
WW ozonation we applied the fish early life stage toxicity test<br />
(FELST) with rainbow trout (Oncorhynchus mykiss).<br />
2. Materials and methods<br />
Ozonated and conventionally treated WW were tested in<br />
parallel on site at the wastewater treatment plant (WWTP)<br />
Wüeri (Regensdorf, Switzerland) with the fish early life stage<br />
toxicity test (FELST) in a flow-through system.<br />
2.1. Characterization of the wastewater treatment plant<br />
The municipal WWTP Wüeri operates experimentally with<br />
a full scale ozonation reactor after final sedimentation and<br />
with a sand filtration step after the ozone reactor. Table 1<br />
shows the WW quality parameters. The water parameters<br />
after the final sedimentation and after the ozone reactor are<br />
on the same level and therefore exemplarily shown for final<br />
sedimentation water. Low ammonium and phosphate<br />
concentrations indicate that the treatment plant is already<br />
working well. The dissolved organic carbon (DOC) ranged from<br />
5.4 to 5.9 mg/L and the pH was nearly constant in all test<br />
waters. The WWTP serves for a population equivalent of<br />
25,000. The median discharge in the experimental period was<br />
6190 m 3 /day (10th percentile: 4430 m 3 /day, 90th percentile:<br />
10,500, n ¼ 109) and the applied ozone concentration was in<br />
a range between 0.4 and 1 mg O 3/mg DOC.<br />
water research 44 (2010) 439–448 441<br />
2.2. Experimental setup<br />
WW from three different sampling points of serial treatment<br />
steps was tested (Fig. 1): after the final sedimentation (FS),<br />
after the ozone reactor (O) and after additional sand filtration<br />
(OS).<br />
Test waters were passively transported through high-grade<br />
steel pipes to aerated high-grade steel reservoir tanks. The<br />
retention time in the pipes was adjusted to be at least 45 min<br />
to avoid ozone residuals reaching the exposure vessels.<br />
Indeed no ozone was detected by indigo blue method (Bader<br />
and Hoigne, 1981) in the reservoir tank during maximum<br />
required flow-through and during maximum applied ozone<br />
concentration (1 mg O3/mg DOC). From reservoir tanks test<br />
waters were transported through polytetrafluoroethylene<br />
tubes via a peristaltic pump (IPC24, Ismatec, Wertheim–<br />
Mondfeld, Germany) to the exposure vessels each equipped<br />
with a passive discharge device and tempered using<br />
a temperature-controlled water bath. Constant temperature<br />
conditions in the water bath were achieved with a flowthrough<br />
cooling system (Van der Heijden, Dörentrup,<br />
Germany). The flow-through rates in the exposure vessels<br />
ranged from 11 mL per minute up to 44 mL per minute (2–8 fold<br />
water exchange per day in the exposure vessels) depending on<br />
the fish size to match the loading rate criteria (OECD, 1992b).<br />
Reconstituted water according to OECD guideline 203 (1992a)<br />
was used as control water (C).<br />
The FELST was performed with the rainbow trout (Oncorhynchus<br />
mykiss) according to OECD guideline 210 (1992b) with<br />
a constant water temperature of 10 2 C as well as darkness<br />
for embryo development and 12 2 C with 12/12 h light/dark<br />
photoperiod post hatch. Sixty newly fertilized eggs per replicate<br />
were exposed to the test waters for 65 days in high-grade<br />
steel 10 L tanks. One test series was performed with unfiltered<br />
WW and a second with membrane filtered WW (pore size:<br />
0.4 mm, Kubota Corp., Osaka, Japan) to minimize microbial<br />
impacts. Macromolecules and organic compounds were not<br />
retained. However, it has to be considered that membrane<br />
filtration additionally removes suspended particulate matter<br />
and consequently all particle bound pollutants. The filter<br />
membranes were placed in the reservoir vessels.<br />
A third test was performed with yolk-sac fry (5 days post<br />
hatch, 30 larvae per exposure vessel) and non-filtered test<br />
waters because we postulated a reduced sensitivity to pathogen<br />
contamination of the larvae compared to the egg stage.<br />
The test duration was 64 days. All tests were performed with<br />
undiluted WW to increase the probability to detect differences<br />
between the treatment groups. With the beginning of swim up<br />
(the swim up process marks a developmental transition from<br />
larval stage to juvenile fish stage and is characterized with the<br />
Fig. 1 – Sampling points at the wastewater treatment plant. Abbreviations: FS, after final sedimentation; O, after the ozone<br />
reactor; OS, after sand filtration.
442<br />
beginning of exogenous ingestion) the fish were fed four times<br />
per day (trout starter, 4% body weight per day).<br />
The tests were run with four (C) and three (FS, O, OS)<br />
replicates in the first FELST with unfiltered WW and two (C,<br />
OS) and three replicates (FS, O) per test water in the remaining<br />
two tests. The latter were performed in parallel and therefore<br />
it was not possible to use consistently three replicates per<br />
WW-group due to limited space capacities. In each test replicates<br />
were placed randomized in the water bath. Observations<br />
on egg coagulation, hatching, mortality, swim up, malformation<br />
and abnormal behaviour were recorded daily. Fish were<br />
humanely killed by MS222 (tricaine methanesulfonate, Sigma–<br />
Aldrich, St. Louis, USA) overdose. Individual fish were blotted<br />
dry, weight and body length were determined. Afterwards fish<br />
were frozen in liquid nitrogen and stored at 80 C until<br />
vitellogenin detection in whole body homogenates.<br />
2.3. Vitellogenin detection<br />
Whole body homogenates of 11 fish (from the third test with<br />
yolk-sac fry) per replicate were prepared as described by<br />
Holbech et al. (2006) with slight modifications. Aliquots of 0.3 g<br />
frozen fish, excised between head and pectoral fin, were<br />
mixed with 10 fold of the body weight of homogenisation<br />
buffer (50 mM Tris–HCl pH 7.4; 1% protease inhibitor cocktail<br />
(P 8340, Sigma–Aldrich, St. Louis, USA)) and homogenated<br />
with a dispersing apparatus (T18 basic Ultra–Turrax Ò , IKA,<br />
Staufen, Germany). The homogenate was centrifuged for<br />
30 min at 20,000 rcf and the supernatant was used for vitellogenin<br />
analysis. Vitellogenin (VTG) was detected with<br />
a rainbow trout vitellogenin ELISA test kit (Biosense, Bergen,<br />
Norway) using a 1:20 dilution.<br />
2.4. Statistical analysis<br />
Complete statistical analyses (Kruskal–Wallis with Dunn’s<br />
post test, Fisher’s exact test, non-linear regression with variable<br />
slope) were performed using GraphPad Prism version 5.0<br />
for windows (GraphPad software, San Diego, CA, USA). All<br />
given error values indicate the standard error (SE) and in all<br />
figures error bars display the SE. Kruskal–Wallis with Dunn’s<br />
post test was chosen to test on significant differences because<br />
data were not normally distributed in all cases. For quantal<br />
data Fisher’s exact test was applied (mortality, swim up,<br />
coagulation). When toxicity endpoints could be analysed<br />
individually for each test animal (biomass, body length,<br />
vitellogenin), the collected data are presented and statistically<br />
evaluated on a per specimen basis.<br />
3. Results<br />
3.1. FELST with unfiltered wastewater<br />
Unfiltered WW caused an increased coagulation rate of the<br />
exposed eggs in the FELST (Fig. 2). The eggs exposed to WW<br />
after final sedimentation (FS) were completely coagulated<br />
after 18 days with a 50% coagulation time of 12.1 days. The<br />
coagulation of the eggs exposed to WW after the ozone reactor<br />
(O) was considerably delayed compared to FS. However the<br />
water research 44 (2010) 439–448<br />
Fig. 2 – Cumulative coagulation of Oncorhynchus mykiss<br />
eggs (mean values ± SE) exposed to differently treated<br />
wastewaters. Abbreviations: C, control water; FS, final<br />
sedimentation; O, ozonation; OS, ozonation and sand<br />
filtration; SE, standard error. Time scale is logarithmized.<br />
After 40 days all treatment groups differ significantly from<br />
each other (Fisher’s exact test: p < 0.001; n [ 240 (C), 180<br />
(FS, O, OS)).<br />
coagulation after 40 days achieves still an average rate of<br />
87.2 4.8% with a 50% coagulation time of 17.6 days. The<br />
lowest coagulation rate in the WW treatments occurred after<br />
sand filtration (OS; 64.4 4.0% after 40 days; 50% coagulation<br />
time: 25.8 days). After 10–15 days exposure, fungus mycelia<br />
(first appearing in the FS vessels) were observed in all WW<br />
exposure vessels. Mycelia were found on and between eggs as<br />
well as vorticellas on the eggs whereas the reference water (C)<br />
remained observably free from mycelia and vorticellas.<br />
Hatching success in the control group (53.4 5.2%) did not<br />
meet validity criteria (>66%; OECD guideline 210).<br />
3.2. FELST with membrane filtered wastewater<br />
To exclude microbial impairment the second FELST was performed<br />
with membrane filtered WW and eggs were obtained<br />
from another fish hatchery.<br />
All test vessels remained free from microbial contamination<br />
throughout the test duration. The egg coagulation rates in<br />
the WW treatment groups of this experimental series were<br />
reduced compared to the first FELST with a maximum of 25%<br />
(O, OS) and only 20.1 1.1% after FS (Fig. 3A). Egg coagulation<br />
was significantly increased and hatching success significantly<br />
decreased in the WW treatment groups compared to the<br />
control ( p < 0.05, Fisher’s exact). The coagulation rates were<br />
slightly but not significantly increased in O and OS compared<br />
to FS (Fig. 3A). The hatching progress was slightly delayed in O<br />
compared to FS and OS but hatching success achieved at least<br />
75% in all treatments (Fig. 3B). The control group met the<br />
validity criteria according to OECD guideline 210 (egg coagulation:<br />
10.0 1.7%, hatching success: 90.0 1.7%).
Fig. 3 – Cumulative coagulation of eggs (A) and cumulative hatching of larvae (B) from Oncorhynchus mykiss (mean<br />
values ± SE) exposed to differently treated and membrane filtered wastewaters. Abbreviations: C, control water; FS, final<br />
sedimentation; O, ozonation; OS, ozonation and sand filtration; SE, standard error. Time scale is logarithmized. After 40 and<br />
36 days, respectively all wastewater treatment groups are significantly different from the control (Fisher’s exact test:<br />
p < 0.01–0.05; n [ 120 (C, OS), 180 (FS, O)).<br />
Fig. 4 shows the cumulative swim up of hatched fish<br />
beginning after 45 days of exposure. The swim up is considerably<br />
delayed in all WW treatment groups compared to the<br />
control. This effect is most notable in O, even if compared to<br />
Fig. 4 – Cumulative swim up of Oncorhynchus mykiss larvae<br />
(mean values ± SE) exposed to differently treated and<br />
membrane filtered wastewaters. Abbreviations: C, control<br />
water; FS, final sedimentation; O, ozonation; OS, ozonation<br />
and sand filtration; SE, standard error. Time scale is<br />
logarithmized. After 64 days the O group differs<br />
significantly from the other treatment groups (Fisher’s<br />
exact test: p < 0.001–0.01; n [ 95 (C), 116 (FS), 103 (O), 75<br />
(OS)).<br />
water research 44 (2010) 439–448 443<br />
FS and OS. At the end of the experiment only 83.3 1.6% of the<br />
fish swam up in O while 100% swam up in C and FS and<br />
97.6 2.4% in OS. Hereby the swim up success after 64 days is<br />
significantly decreased in O compared to the other treatment<br />
groups ( p < 0.01, Fisher’s exact). The 50% swim up time in the<br />
control is 47.4 1.0 days, which is only slightly increased in FS<br />
and OS (50.2 1.0; 49.3 1.0) but obviously increased in O<br />
(57.5 1.04).<br />
The biomass as well as the body length of fish is significantly<br />
decreased ( p < 0.001, Kruskal–Wallis with Dunn’s post<br />
test) in all WW treatments compared to the control (Fig. 5).<br />
Both endpoints in fish exposed to O are furthermore significantly<br />
decreased compared to the FS group ( p < 0.001) and<br />
significantly decreased compared to OS ( p < 0.05).<br />
Generally the mortality is comparatively low in all WW<br />
treatment groups (Fig. 6), as they are fulfilling the validity<br />
criteria for controls according to OECD guideline 210 (survival<br />
after hatch 70%). Nevertheless the mortality in the WW<br />
treatment groups is slightly increased compared to C. The<br />
highest and significantly increased mortality rate was detected<br />
in the ozonated water (24.0 4.0%, p < 0.05, Fisher’s exact).<br />
3.3. Fish test starting with yolk-sac fry & vitellogenesis<br />
The fish test starting with the yolk-sac stage revealed no<br />
statistically significant differences in development between<br />
WW treatments and the control. The biomass exhibited no<br />
major deviations between exposure groups (
444<br />
Fig. 5 – Biomass (A) and body length (B) in percentage relative to the control (0.28 ± 0.01 g and 34.1 ± 0.2 mm, respectively) of<br />
Oncorhynchus mykiss (mean values ± SE) exposed to differently treated and membrane filtered wastewaters. Abbreviations:<br />
C, control water; FS, final sedimentation; O, ozonation; OS, ozonation and sand filtration; SE, standard error. Significant<br />
differences to the control are indicated with white asterisks, between treatments with black asterisks (Kruskal–Wallis with<br />
Dunn’s post test: +, p < 0.05; +++, p < 0.001; n [ 73–114).<br />
Kruskal–Wallis with Dunn’s post test) compared to fish maintained<br />
in reference water (11.0 6.6 ng/mL) whereas it was<br />
significantly decreased in the O and OS groups (4.6 1.9 and<br />
4.8 1.9 ng/mL, respectively; p 0.01) compared to FS (Fig. 7).<br />
4. Discussion<br />
4.1. FELST with unfiltered wastewater<br />
Unfiltered WW led to an increased coagulation rate of the<br />
exposed eggs in the FELST (Fig. 2). High coagulation rates<br />
Fig. 6 – Mortality of Oncorhynchus mykiss post hatch (mean<br />
values ± SE) exposed to differently treated and membrane<br />
filtered wastewaters. Abbreviations: C, control water; FS,<br />
final sedimentation; O, ozonation; OS, ozonation and sand<br />
filtration; SE, standard error. Significant differences to<br />
control are indicated with white asterisks (Fisher’s exact<br />
test: +, p < 0.05; n [ 108 (C), 143 (FS), 135 (O), 90 (OS)).<br />
water research 44 (2010) 439–448<br />
presumably are a result of microbial contamination as the<br />
reference water (C) remained free from mycelia and vorticellas<br />
whereas in the WW treatment groups fungi mycelia and<br />
vorticellas were found after 10–15 days exposure. However,<br />
the hatching success in the control (53.4 5.2%) did not<br />
comply with the validity criteria (>66%) according to OECD<br />
guideline (1992b) indicating that egg quality was not sufficient<br />
to run a valid test. The disinfectant effect of ozonation (Tyrrell<br />
et al., 1995) is likely to be responsible for the delayed egg<br />
Fig. 7 – Whole body vitellogenin concentration of<br />
Oncorhynchus mykiss (mean values ± SE) after 60 days<br />
exposure to differently treated wastewaters starting with<br />
the yolk-sac stage. Abbreviations: FS, final sedimentation;<br />
O, ozonation; OS, ozonation and sand filtration; C, control<br />
water; SE, standard error. Significant differences to control<br />
are indicated with white asterisks, between treatments<br />
with black asterisks (Kruskal–Wallis with Dunn’s post test:<br />
+, p < 0.05; ++, p < 0.01; n [ 22 (C, OS) – 33 (FS, O)).
coagulation in the O group compared to FS. Ozonation is also<br />
known to produce high amounts of assimilable organic<br />
carbons (AOC) which may have allowed a fast regrowth of<br />
microorganisms (Huang et al., 2005), resulting in still high<br />
coagulation rates in the O group. Sand filtration reduces the<br />
amount of suspended particulate matter (SPM) and of AOC<br />
(Wang and Summers, 1996). This may have reduced microbial<br />
development and resulting coagulation rates.<br />
4.2. FELST with membrane filtered wastewater<br />
Exposure to membrane filtered WW revealed a distinct delay<br />
of the swim up process in the O group compared to FS and<br />
OS (Fig. 4) accompanied by a significant decrease in body<br />
weight and body length (Fig. 5). The reduced biomass and<br />
body length after ozonation is most likely a result of the<br />
delayed development because fish start exogenous ingestion<br />
at the onset of the swim up process. Consequently trout that<br />
swim up later may suffer from developmental disadvantages.<br />
Oxidative byproducts in ozonated WW may have impeded<br />
embryonic and/or larval development of fish. Possibly<br />
because sand filtration is able to remove ozonation metabolites<br />
(e.g. aldehydes, glyoxal, AOC; Wang and Summers, 1996)<br />
the effect in OS was reduced. However, no single compound<br />
could be clearly identified for the retardation effect. Possibly<br />
the sum of aldehydes (e.g. formaldehyde, glyoxal, acetaldehyde),<br />
carboxylic acids (e.g. formate), ketones and brominated<br />
organic compounds formed due to ozonation (Huang<br />
et al., 2005; Wert et al., 2007) caused these effects. Unfortunately,<br />
no chronic toxicity data for juvenile rainbow trout are<br />
available in literature for these compounds. Nevertheless,<br />
Wang and Summers (1996) were able to demonstrate that<br />
sand filtration efficiently removes aldehydes produced<br />
during ozone application, supporting this assumption.<br />
Besides, the formation of more complex organic metabolites<br />
evoked by ozonation may result in an increased toxicity<br />
compared to chemical precursors as it has already been<br />
documented for polycyclic aromatic hydrocarbons (PAHs). In<br />
the experiments of Luster-Teasley et al. (2002, 2005) chrysene<br />
and pyrene and byproducts as a result of ozonation were<br />
examined for their ability to disrupt gap junctional intercellular<br />
communication (GJIC), an indicator for tumor<br />
promoting properties. Among the transformation products,<br />
aldehydic compounds exhibited an increased toxicity<br />
compared to the precursor substance while their carboxylic<br />
structure analogue showed no GJIC disrupting activity. For<br />
the antiepileptic drug carbamazepine McDowell et al. (2005)<br />
identified three new oxidation products of unknown toxicity<br />
after ozone treatment. Furthermore, a possible increase in<br />
mutagenicity after ozonation of WW, as observed by Monarca<br />
et al. (2000) with the Ames test, verifies the potential of<br />
ozonation to produce toxic oxidation byproducts. In this<br />
context conversion of the widely used fungicide tolylfluanide<br />
into the carcinogen N-nitrosodimethylamine during ozonation<br />
was discovered by Schmidt and Brauch (2008). Based on<br />
these results it is likely that ozonation of WW leads to an<br />
increased number of unknown and potentially toxic metabolites<br />
depending on the composition of WW at the outset and<br />
the post treatment (e.g. sand filtration). Petala et al. (2006)<br />
demonstrated with the Vibrio fischeri bioluminescence test,<br />
water research 44 (2010) 439–448 445<br />
that toxicity originating from ozonated WW decreases with<br />
increasing storage time, indicating a rapid decomposition of<br />
toxic metabolites.<br />
In spite of the developmental retardation and the reduced<br />
biomass no significant differences in mortality rates were<br />
observed between the WW treatment groups (Fig. 6). Mortality<br />
was highest directly after ozonation (O, 24.0 4.0%). Considering<br />
the developmental retardation and the decreased<br />
biomass of fish exposed to ozonated water the increased<br />
mortality of O group specimens compared to FS and OS is<br />
noticeable and, although not significant, potentially a result of<br />
a reduced fitness of the fish in the ozonated water.<br />
It might be assumed that the retarded development is<br />
a consequence of an unspecific and general impairment of the<br />
fish’s health condition. This may increase their sensitivity<br />
towards environmental and anthropogenic stressors, thus<br />
leading to an increased mortality. Furthermore, under field<br />
conditions the delayed development possibly increases the<br />
risk for the fish to fall prey to predators since before swim up<br />
the larvae rest on the bottom, not capable to abscond effectively.<br />
Based on the assumption that the compounds causing<br />
adverse effects after ozonation are readily degradable it is<br />
possible that the detoxication after the ozone reactor will<br />
occur in the river as well. However, the discharge of ozonated<br />
WW without sand filtration would bear the risk to endanger<br />
fish populations in a considerable range of the receiving river,<br />
depending on WW load and flow velocity. Therefore, ozonation<br />
should always be followed by a sand filtration and further<br />
studies should focus on whether sand filtration is capable to<br />
remove oxidation byproducts sufficiently.<br />
4.3. Fish test starting with yolk-sac fry & vitellogenesis<br />
The fish test with non-filtered WW starting with the yolk-sac<br />
stage resulted in negligible and insignificant deviations of<br />
development and biomass between treatment groups. Ozonation<br />
had no effect on these toxicity endpoints. These results<br />
indicate that yolk-sac fry 5 days post hatch are less sensitive<br />
to ozonation metabolites, compared to fish embryos and<br />
newly hatched fish, respectively. Nevertheless, the amount of<br />
suspended particulate matter in the exposure vessels and in<br />
the Teflon tubes was accompanied by an increased biofilm<br />
formation. This may have contributed to an increased detoxication<br />
of the ozonated WW due to (bio-) degradation of<br />
oxidation byproducts.<br />
The significant increase of vitellogenin content in fish<br />
exposed to FS compared to the reference water indicates an<br />
environmentally relevant contamination of the WW with<br />
estrogenic active compounds (Fig. 7). After ozonation the VTG<br />
content decreased even below the control level. The formation<br />
of antiestrogenic compounds is not likely because the VTG<br />
decrease compared to control level is not significant and<br />
phenols, as an important functional group interacting with<br />
the estrogen receptor (Nishihara et al., 2000), are particularly<br />
susceptible to ozone attack (von Gunten, 2003). These results<br />
confirm the high efficiency of ozonation to eliminate estrogenic<br />
contamination in WW as already shown by Huber et al.<br />
(2004) with an in vitro test system. Based on these results it<br />
can be assumed that ozonation is well suited to reduce the<br />
estrogenic burden of WW below environmental relevance.
446<br />
Regarding ecosystem health it should be considered that<br />
estrogen and xenoestrogen concentrations detected in the<br />
environment have been shown to impair the sustainability of<br />
wild fish populations (Kidd et al., 2007), arguing for the<br />
establishment of an end of pipe technique for estrogen<br />
removal to protect fish populations in surface waters receiving<br />
high amounts of WW.<br />
4.4. Ozonation vs. alternatives<br />
Currently there is a controversy regarding the appropriate<br />
advanced WW treatment techniques to reduce contamination<br />
of the aquatic ecosystem with micropollutants. Alternative<br />
processes other than ozonation also deliver promising results<br />
with regard to the removal of contaminants. Wastewater<br />
treatment with powdered activated carbon (PAC) achieves<br />
reduction rates of 75–90% for pharmaceuticals (including<br />
X-ray contrast media) with PAC dosages of 10–20 mg/L<br />
(Püttmann et al., 2008). The main advantage of PAC treatment<br />
is that a reduced chemical concentration is equivalent to<br />
removal of chemicals whereas ozonation leads basically to<br />
a transformation of the compounds. According to cost estimations<br />
PAC treatment is expected to be about 30% more<br />
expensive than ozonation (Joss et al., 2008). However, the<br />
main disadvantage of PAC treatment might be the need for the<br />
disposal of the used and contaminated carbon. Membrane<br />
filtration is suitable for micropollutants retention but as<br />
a result of considerably higher requirement for energy and<br />
technical equipment economically not competitive with<br />
ozonation or activated carbon treatment (Joss et al., 2008).<br />
Overall, end of pipe techniques are presumably a suitable<br />
solution to reduce toxicity of hazardous WW in a mediumterm<br />
perspective. Nevertheless in the long term source control<br />
strategies such as wastewater separation (e.g. urine separation),<br />
ecologically correct disposal of drugs by end users, reuse<br />
or recycling by the pharmaceutical industry or alternative<br />
medical treatments to drug therapies (Daughton, 2003; Joss<br />
et al., 2006) could offer environmentally friendly and affordable<br />
options.<br />
5. Conclusions<br />
The sanitizing effect of ozonation was confirmed. Coagulation<br />
rates of eggs exposed to ozonated wastewater were on<br />
a lower level compared to those exposed to conventionally<br />
treated wastewater. Disinfection is presumably only efficient<br />
when linked to sand filtration because of rapid<br />
recovery of microorganisms due to increased assimilable<br />
organic carbon formation as a result of ozonation.<br />
Membrane filtered wastewater (for removal of microorganisms)<br />
reveals developmental retardation directly after<br />
ozonation. After sand filtration this adverse effect disappears.<br />
The reduced biomass and body length in fish exposed<br />
to ozonated wastewater is most probably a result of the<br />
formation of toxic oxidation byproducts.<br />
The mortality in the ozonated wastewater was significantly<br />
increased as a result of retarded development. Impairment of<br />
the fish’s health condition may increase the sensitivity<br />
towards environmental and anthropogenic stressors.<br />
water research 44 (2010) 439–448<br />
Developmental retardation might increase the risk for the fish<br />
to fall prey to predators because swim up stage is delayed.<br />
Ozonation should not be applied without appropriate<br />
barrier for oxidation byproducts. Effectiveness of sand<br />
filtration for removal of oxidation byproducts should be<br />
further evaluated.<br />
Reduction of vitellogenin content in fish exposed to ozonated<br />
wastewater on control level confirms the suitability of<br />
this technique to reduce estrogenic activity, possibly below<br />
environmental relevance.<br />
Acknowledgements<br />
The authors should like to express their gratitude to the staff<br />
from WWTP Wüeri for their technical cooperation and<br />
Adriano Joss from the EAWAG for helpful suggestions and<br />
technical support. Ulrike Schulte-Oehlmann is acknowledged<br />
for critical comments and helpful suggestions on the manuscript.<br />
This study was part of the EU project Neptune (contract<br />
no 036845, SUSTDEV-2005-3.II.3.2) within the Energy, Global<br />
Change and Ecosystems Programme of the Sixth Framework<br />
(FP6-2005-Global-4) and co-funded by Bundesamt für Umwelt<br />
(BAFU), Bern (CH) within the Strategy MicroPoll Programme<br />
(contract no 05.0013.PJ/F471-0916).<br />
references<br />
Andreozzi, R., Campanella, L., Fraysse, B., Garric, J., Gonnella, A.,<br />
Lo Giudice, R., Marotta, R., Pinto, G., Pollio, A., 2004. Effects of<br />
advanced oxidation processes (AOPs) on the toxicity of<br />
a mixture of pharmaceuticals. Water Science and Technology<br />
50 (5), 23–28.<br />
Bader, H., Hoigne, J., 1981. Determination of ozone in water by the<br />
indigo method. Water Research 15 (4), 449–456.<br />
Brian,J.V.,Harris,C.A.,Scholze,M.,Backhaus,T.,Booy,P.,<br />
Lamoree,M.,Pojana,G.,Jonkers,N.,Runnalls,T.,Bonfa,A.,<br />
Marcomini, A., Sumpter, J.P., 2005. Accurate prediction of<br />
the response of freshwater fish to a mixture of estrogenic<br />
chemicals. Environmental Health Perspectives 113 (6),<br />
721–728.<br />
Daughton, C.G., 2003. Cradle-to-cradle stewardship of drugs for<br />
minimizing their environmental disposition while<br />
promoting human health. II. Drug disposal, waste reduction,<br />
and future directions. Environmental Health Perspectives<br />
111 (5), 775–785.<br />
Daughton, C.G., Ternes, T.A., 1999. Pharmaceuticals and personal<br />
care products in the environment: agents of subtle change?<br />
Environmental Health Perspectives 107, 907–938.<br />
Escher, B.I., Bramaz, N., Quayle, P., Rutishauser, S.,<br />
Vermeirssen, E.L.M., 2008. Monitoring of the ecotoxicological<br />
hazard potential by polar organic micropollutants in sewage<br />
treatment plants and surface waters using a mode-of-action<br />
based test battery. Journal of Environmental Monitoring 10 (5),<br />
622–631.<br />
European Commission, 2000. Directive 2000/60/EC of the<br />
European Parliament and of the Council of 23 October 2000<br />
establishing a framework for Community action in the field<br />
of water policy. Official Journal of the European<br />
Communities L327. Retrieved from: http://eur-lex.europa.eu/
LexUriServ/LexUriServ.do?uri¼OJ: L:2000:327:0001:0072:EN:<br />
PDF [12.02.2009].<br />
Gagne, F., Andre, C., Cejka, P., Gagnon, C., Blaise, C., 2007.<br />
Toxicological effects of primary-treated urban wastewaters,<br />
before and after ozone treatment, on freshwater mussels<br />
(Elliptio complanata). Comparative Biochemistry and Physiology<br />
C – Toxicology & Pharmacology 145 (4), 542–552.<br />
Holbech, H., Kinnberg, K., Petersen, G.I., Jackson, P., Hylland, K.,<br />
Norrgren, L., Bjerregaard, P., 2006. Detection of endocrine<br />
disrupters: evaluation of a fish sexual development<br />
test (FSDT). Comparative Biochemistry and Physiology<br />
C – Toxicology & Pharmacology 144 (1), 57–66.<br />
Huang, W.J., Fang, G.C., Wang, C.C., 2005. The determination<br />
and fate of disinfection by-products from ozonation of<br />
polluted raw water. Science of the Total Environment 345<br />
(1-3), 261–272.<br />
Huber, M.M., Ternes, T.A., von Gunten, U., 2004. Removal of<br />
estrogenic activity and formation of oxidation products during<br />
ozonation of 17a-ethinylestradiol. Environmental Science &<br />
Technology 38 (19), 5177–5186.<br />
Huber, M.M., Gobel, A., Joss, A., Hermann, N., Loffler, D.,<br />
McArdell, C.S., Ried, A., Siegrist, H., Ternes, T.A., von<br />
Gunten, U., 2005. Oxidation of pharmaceuticals during<br />
ozonation of municipal wastewater effluents: a pilot study.<br />
Environmental Science & Technology 39 (11), 4290–4299.<br />
Jahnel, J., Neamtu, M., Schudoma, D., Frimmel, F.H., 2006.<br />
Scientific risk assessment of considered water relevant<br />
substances (Bestimmung von Umweltqualitaetsnormen fuer<br />
potenziell gewaesserrelevante Stoffe). Acta Hydrochimica et<br />
Hydrobiologica 34 (4), 389–397.<br />
Jensen, G.K., 2007. European Parliament increases number of<br />
chemicals earmarked for priority action by European<br />
Commission. The Lancet Oncology 8 (5), 382–383.<br />
Jobling, S., Sheahan, D., Osborne, J.A., Matthiessen, P.,<br />
Sumpter, J.P., 1996. Inhibition of testicular growth in rainbow<br />
trout (Oncorhynchus mykiss) exposed to estrogenic<br />
alkylphenolic chemicals. Environmental Toxicology and<br />
Chemistry 15 (2), 194–202.<br />
Jobling, S., Coey, S., Whitmore, J.G., Kime, D.E., Van Look, K.J.W.,<br />
McAllister, B.G., Beresford, N., Henshaw, A.C., Brighty, G.,<br />
Tyler, C.R., Sumpter, J.P., 2002. Wild intersex roach (Rutilus<br />
rutilus) have reduced fertility. Biology of Reproduction 67 (2),<br />
515–524.<br />
Joss, A., Klaschka, U., Knacker, T., Liebig, M., Lienert, J., Ternes, T.A.,<br />
Wennmalm, A., 2006. Source control, source separation. In:<br />
Ternes, T.A., Joss, A. (Eds.), Human Pharmaceuticals, Hormones<br />
and Fragrances. The Challenge of Micropollutants in Urban<br />
Water Management. ISBN 1843390930, pp. 353–384.<br />
Joss, A., Siegrist, H., Ternes, T.A., 2008. Are we about to upgrade<br />
wastewater treatment for removing organic micropollutants?<br />
Water Science and Technology 57 (2), 251–255.<br />
Kidd, K.A., Blanchfield, P.J., Mills, K.H., Palace, V.P., Evans, R.E.,<br />
Lazorchak, J.M., Flick, R.W. (2007). Collapse of a fish population<br />
after exposure to a synthetic estrogen. Proceedings of the<br />
National Academy of Sciences of the United States of America<br />
104(21), 8897–8901.<br />
Larsson, D.G.J., Adolfsson-Erici, M., Parkkonen, J.,<br />
Pettersson, M., Berg, A.H., Olsson, P.E., Forlin, L., 1999.<br />
Ethinylestradiol – an undesired fish contraceptive? Aquatic<br />
Toxicology 45 (2-3), 91–97.<br />
Luster-Teasley, S.L., Yao, J.J., Herner, H.H., Trosko, J.E., Masten, S.J.,<br />
2002. Ozonation of chrysene: evaluation of byproduct mixtures<br />
and identification of toxic constituent. Environmental Science<br />
& Technology 36 (5), 869–876.<br />
Luster-Teasley, S.L., Ganey, P.E., DiOrio, M., Ward, J.S.,<br />
Maleczka, R.E., Trosko, J.E., Masten, S.J., 2005. Effect of<br />
byproducts from the ozonation of pyrene: biphenyl-2,2 0 ,6,6 0 -<br />
tetracarbaldehyde and biphenyl-2,2 0 ,6,6 0 -tetracarboxylic acid<br />
water research 44 (2010) 439–448 447<br />
on gap junction intercellular communication and neutrophil<br />
function. Environmental Toxicology and Chemistry 24 (3),<br />
733–740.<br />
McDowell, D.C., Huber, M.M., Wagner, M., Von Gunten, U.,<br />
Ternes, T.A., 2005. Ozonation of carbamazepine in drinking<br />
water: identification and kinetic study of major oxidation<br />
products. Environmental Science & Technology 39 (20),<br />
8014–8022.<br />
Moltmann, J.F., Liebig, M., Knacker, T., Keller, M., Scheurer, M.,<br />
Ternes, T., 2007. Relevance of Endocrine Disrupting<br />
Substances and Pharmaceuticals in Surface Waters<br />
(Gewässerelevanz endokriner Stoffe und Arzneimittel). Final<br />
report. Federal Environment Agency. Funding code 20524205.<br />
Monarca, S., Feretti, D., Collivignarelli, C., Guzzella, L., Zerbini, I.,<br />
Bertanza, G., Pedrazzani, R., 2000. The influence of different<br />
disinfectants on mutagenicity and toxicity of urban<br />
wastewater. Water Research 34 (17), 4261–4269.<br />
Nakada, N., Shinohara, H., Murata, A., Kiri, K., Managaki, S.,<br />
Sato, N., Takada, H., 2007. Removal of selected<br />
pharmaceuticals and personal care products (PPCPs) and<br />
endocrine-disrupting chemicals (EDCs) during sand filtration<br />
and ozonation at a municipal sewage treatment plant. Water<br />
Research 41 (19), 4373–4382.<br />
Nishihara, T., Nishikawa, J., Kanayama, T., Dakeyama, F., Saito, K.,<br />
Imagawa, M., Takatori, S., Kitagawa, Y., Hori, S., Utsumi, H.,<br />
2000. Estrogenic activities of 517 chemicals by yeast two-hybrid<br />
assay. Journal of Health Science 46 (4), 282–298.<br />
OECD, 1992a. Fish Acute Toxicity Test. OECD Guidelines for the<br />
Testing of Chemicals. No. 203. Organisation for Economic<br />
Cooperation and Development, Paris.<br />
OECD, 1992b. Fish Early Life Stage Toxicity Test. OECD Guidelines<br />
for the Testing of Chemicals. No. 210. Organisation for<br />
Economic Cooperation and Development, Paris.<br />
Parrott, J.L., Blunt, B.R., 2005. Life-cycle exposure of fathead<br />
minnows (Pimephales promelas) to an ethinylestradiol<br />
concentration below 1 ng/L reduces egg fertilization success<br />
and demasculinizes males. Environmental Toxicology 20 (2),<br />
131–141.<br />
Petala, M., Samaras, P., Zouboulis, A., Kungolos, A.,<br />
Salkellaropoulos, G., 2006. Ecotoxicological properties of<br />
wastewater treated using tertiary methods. Environmental<br />
Toxicology 21 (4), 417–424.<br />
Püttmann, W., Keil, F., Oehlmann, J., Schulte-Oehlmann, U., 2008.<br />
Strategy to reduce pharmaceuticals in drinking water –<br />
technical approach (Wassertechnische Strategien zur<br />
Reduzierung der Trinkwasserbelastung durch<br />
Arzneimittelwirkstoffe). Umweltwissenschaften und<br />
Schadstoff-Forschung 20 (3), 209–226.<br />
Purdom, C.E., Hardiman, P.A., Bye, V.V.J., Eno, N.C., Tyler, C.R.,<br />
Sumpter, J.P., 1994. Estrogenic effects of effluents from sewage<br />
treatment works. Chemical Ecology 8 (4), 275–285.<br />
Rodgers-Gray, T.P., Jobling, S., Kelly, C., Morris, S., Brighty, G.,<br />
Waldock, M.J., Sumpter, J.P., Tyler, C.R., 2001. Exposure of<br />
juvenile roach (Rutilus rutilus) to treated sewage effluent<br />
induces dose-dependent and persistent disruption in gonadal<br />
duct development. Environmental Science & Technology 35<br />
(3), 462–470.<br />
Schmidt, C.K., Brauch, H.J., 2008. N, N-dimethosulfamide as<br />
precursor for N-nitrosodimethylamine (NDMA) formation<br />
upon ozonation and its fate during drinking water treatment.<br />
Environmental Science & Technology 42 (17), 6340–6346.<br />
Schwaiger, J., Ferling, H., Mallow, U., Wintermayr, H.,<br />
Negele, R.D., 2004. Toxic effects of the non-steroidal antiinflammatory<br />
drug diclofenac Part 1: histopathological<br />
alterations and bioaccumulation in rainbow trout. Aquatic<br />
Toxicology 68 (2), 141–150.<br />
Schwarzenbach, R.P., Escher, B.I., Fenner, K., Hofstetter, T.B.,<br />
Johnson, C.A., von Gunten, U., Wehrli, B., 2006. The challenge
448<br />
of micropollutants in aquatic systems. Science 313 (5790),<br />
1072–1077.<br />
Ternes, T.A., Stumpf, M., Müller, J., Haberer, K., Wilken, R.D.,<br />
Servos, M., 1999. Behavior and occurrence of estrogens in<br />
municipal sewage treatment plants – I. Investigations in<br />
Germany, Canada and Brazil. Science of the Total<br />
Environment 225 (1-2), 81–90.<br />
Tixier, C., Singer, H.P., Oellers, S., Müller, S.R., 2003. Occurrence<br />
and fate of carbamazepine, clofibric acid, diclofenac,<br />
ibuprofen, ketoprofen, and naproxen in surface waters.<br />
Environmental Science & Technology 37 (6), 1061–1068.<br />
Triebskorn, R., Casper, H., Heyd, A., Eikemper, R., Köhler, H.R.,<br />
Schwaiger, J., 2004. Toxic effects of the non-steroidal antiinflammatory<br />
drug diclofenac Part II. Cytological effects in<br />
liver, kidney, gills and intestine of rainbow trout (Oncorhynchus<br />
mykiss). Aquatic Toxicology 68 (2), 151–166.<br />
water research 44 (2010) 439–448<br />
Tyrrell, S.A., Rippey, S.R., Watkins, W.D., 1995. Inactivation of<br />
bacterial and viral indicators in secondary sewage<br />
effluents, using chlorine and ozone. Water Research 29<br />
(11), 2483–2490.<br />
von Gunten, U., 2003. Ozonation of drinking water: Part I.<br />
Oxidation kinetics and product formation. Water Research 37<br />
(7), 1443–1467.<br />
Wang, J.Z., Summers, R.S., 1996. Biodegradation behavior of<br />
ozonated natural organic matter in sand filters. Revue des<br />
Sciences de l’Eau 9 (1), 3–16.<br />
Wert, E.C., Rosario-Ortiz, F.L., Drury, D.D., Snyder, S.A., 2007.<br />
Formation of oxidation byproducts from ozonation of<br />
wastewater. Water Research 41 (7), 1481–1490.<br />
Ying, G.G., Williams, B., Kookana, R., 2002. Environmental fate of<br />
alkylphenols and alkylphenol ethoxylates – a review.<br />
Environment International 28 (3), 215–226.
The role of organic matter in the removal of emerging<br />
trace organic chemicals during managed aquifer recharge<br />
T. Rauch-Williams, C. Hoppe-Jones, J.E. Drewes*<br />
Advanced Water Technology Center (AQWATEC), Colorado School of Mines, Environmental Science and Engineering Division,<br />
Golden, CO 80401-1887, United States<br />
article info<br />
Article history:<br />
Received 20 April 2009<br />
Received in revised form<br />
11 August 2009<br />
Accepted 20 August 2009<br />
Available online 27 August 2009<br />
Keywords:<br />
Trace organic chemicals (TOrC)<br />
Groundwater recharge<br />
Effluent organic matter<br />
Managed aquifer recharge<br />
Riverbank filtration<br />
Biotransformation<br />
Co-metabolism<br />
Primary substrate<br />
1. Introduction<br />
abstract<br />
Managed aquifer recharge (MAR) systems, such as riverbank<br />
filtration (RBF) and soil aquifer treatment (SAT), are widely<br />
used natural processes for drinking water augmentation<br />
projects using source water that might be impaired by<br />
wastewater discharge. Previous studies have demonstrated<br />
that MAR systems are effective in dampening and reducing<br />
water research 44 (2010) 449–460<br />
Available at www.sciencedirect.com<br />
journal homepage: www.elsevier.com/locate/watres<br />
* Corresponding author.<br />
E-mail address: jdrewes@mines.edu (J.E. Drewes).<br />
0043-1354/$ – see front matter ª 2009 Elsevier Ltd. All rights reserved.<br />
doi:10.1016/j.watres.2009.08.027<br />
This study explored the effect of different bulk organic carbon matrices on the fate of trace<br />
organic chemicals (TOrC) during managed aquifer recharge (MAR). Infiltration through<br />
porous media was simulated in biologically active column experiments under aerobic and<br />
anoxic recharge conditions. Wastewater effluent derived organic carbon types, differing in<br />
hydrophobicity and biodegradability (i. e., hydrophobic acids, hydrophilic carbon, organic<br />
colloids), were used as feed substrates in the column experiments. These carbon substrates<br />
while fed at the same concentration differed in their ability to support soil biomass growth<br />
during porous media infiltration. Removal of degradable TOrC (with the exception of<br />
diclofenac and propyphenazone) was equal or better under aerobic versus anoxic porous<br />
media infiltration conditions. During the initial phase of infiltration, the presence of<br />
biodegradable organic carbon (BDOC) enhanced the decay of degradable TOrC by<br />
promoting soil biomass growth, suggesting that BDOC served as a co-substrate in a cometabolic<br />
transformation of these contaminants. However, unexpected high removal<br />
efficiencies were observed for all degradable TOrC in the presence of low BDOC concentrations<br />
under well adopted oligotrophic conditions. It is hypothesized that removal under<br />
these conditions is caused by a specialized microbial community growing on refractory<br />
carbon substrates such as hydrophobic acids. Findings of this study reveal that the<br />
concentration and character of bulk organic carbon present in effluents affect the degradation<br />
efficiency for TOrC during recharge operation. Specifically aerobic, oligotrophic<br />
microbiological soil environments present favorable conditions for the transformation of<br />
TOrC, including rather recalcitrant compounds such as chlorinated flame retardants.<br />
ª 2009 Elsevier Ltd. All rights reserved.<br />
the concentrations of dissolved organic carbon (DOC) as well<br />
as various trace organic contaminants (TOrC) that might be<br />
present in impaired source waters (Drewes and Fox, 1999;<br />
Brauch et al., 2000; Grünheid et al., 2005). The presence of<br />
TOrC has become a key concern for drinking water augmentation<br />
projects during the past decade (Kolpin et al., 2002;<br />
Heberer, 2002; Focazio et al., 2008). Although adverse human<br />
health effects caused by these compounds at concentrations
450<br />
commonly observed in impaired water resources are very<br />
unlikely (Schwab et al., 2005), minimizing exposure of<br />
wastewater derived contaminants in these projects is desired.<br />
Previous research on the fate of TOrC during MAR has<br />
primarily focused on collecting anecdotal and site specific<br />
information on their occurrence and removal (Drewes et al.,<br />
2002; Montgomery-Brown et al., 2003; Grünheid et al., 2005;<br />
Dillon et al., 2008). Studies delineating the mechanisms and<br />
boundary conditions for the transformation of wastewater<br />
derived TOrC during MAR are lacking.<br />
Previous studies demonstrated that the type and bioavailability<br />
of effluent organic matter (EfOM) controls the extent of<br />
soil biomass growth in MAR systems (Rauch and Drewes, 2004;<br />
Rauch and Drewes, 2005). EfOM may consequently effect the<br />
degradation of TOrC by serving as a co-substrate in microbiologically<br />
facilitated transformations (Stratton et al., 1983).<br />
The diversity and expression of the soil microbial community<br />
also depends on composition and concentration of the organic<br />
carbon substrate controlling trophic cycles in the subsurface<br />
(Preuß and Nehrkorn, 1996; Szewzyk et al., 1998). The<br />
composition of EfOM (i.e. in terms of its bioavailability) is<br />
primarily determined by the degree of wastewater treatment<br />
employed (Drewes and Fox, 1999), which can vary widely from<br />
primary to conventional to advanced wastewater treatment.<br />
As a result, effluent qualities fed to MAR systems can vary in<br />
biodegradable dissolved organic carbon (BDOC) concentrations<br />
from less than 1 up to 15 mg/L or more. As a consequence,<br />
soil microbial communities growing on different<br />
levels of BDOC can differ widely in total biomass and diversity.<br />
The objectives of this research were to investigate the role<br />
of 1) abiotic vs. biotic conditions, 2) BDOC and 3) the type of<br />
organic carbon matrices on the removal of select TOrC, such<br />
as pharmaceutical residues, personal care products, and<br />
household chemicals, during MAR.<br />
2. Methodology<br />
2.1. Target organic contaminants<br />
Compounds selected for this study represent small molecular<br />
weight organic chemicals (180 to 360 Dalton) that are hydrophilic<br />
at neutral pH regimes as indicated by an octanol/water<br />
partition coefficient at pH 7 (log DpH¼7 of less than 2.6). TOrC<br />
with these properties have a high potential to migrate into<br />
groundwater and are not expected to be adsorbed onto porous<br />
media. The molecular structures and physicochemical properties<br />
of the target compounds chosen for this study are presented<br />
in Table 1. These compounds cover a wide range of<br />
biodegradability as previously reported for soil/water systems.<br />
The anticonvulsants carbamazepine and primidone have<br />
been classified as recalcitrant during wastewater treatment<br />
and MAR in earlier studies (Heberer, 2002; Drewes et al., 2003;<br />
Clara et al., 2004). The chlorinated phosphate esters tris<br />
(1-chloroisopropyl)-phosphate (TCPP) and tris(2-chloroethyl)phosphate<br />
(TCEP) are two widely used flame retardants and<br />
are persistent in the aquatic environment (Heberer et al., 2001;<br />
Fries and Püttmann, 2003). During bank filtration in Germany,<br />
however, TCPP and TCEP exhibited a significant reduction,<br />
which was attributed to biotransformation in the aquifer<br />
water research 44 (2010) 449–460<br />
(Heberer et al., 2003). Amy and Drewes (2006) also reported<br />
removal of TCEP to concentrations below the detection limit<br />
after two years of subsurface travel in an MAR facility supporting<br />
that chlorinated flame retardants can be biotransformed<br />
under anoxic conditions. Propyphenazone is<br />
a poorly biodegradable analgesic that persists during RBF<br />
(Heberer et al., 2003) and SAT (Drewes et al., 2003). For diclofenac,<br />
a popular analgesic drug, low removal due to biodegradation<br />
or adsorption was reported (Buser et al., 1998; Möhle<br />
et al., 1999) unless soils contain a high organic carbon content<br />
(Drillia et al., 2003). Several studies report a faster degradation<br />
of diclofenac under anoxic conditions as compared to aerobic<br />
conditions (Zwiener and Frimmel, 2003; Hua et al., 2003).<br />
Opposing results were reported by Schmidt et al. (2004) in that<br />
diclofenac was almost completely removed during aerobic<br />
bank filtration but recalcitrant during anaerobic recharge.<br />
Ibuprofen, ketoprofen, and naproxen are common analgesics<br />
that are well degradable during wastewater treatment (Buser<br />
et al., 1999; Zwiener and Frimmel, 2003; Carballa et al., 2004)<br />
and during soil infiltration (Sedlak and Pinkston, 2001; Drewes<br />
et al., 2003). Gemfibrozil is a commonly prescribed blood lipid<br />
regulator found in unconfined shallow aquifers impacted by<br />
wastewater infiltration in the low to moderate ng/L-concentration<br />
range (Heberer and Stan, 1997; Drewes and Shore,<br />
2001; Heberer, 2002). Gemfibrozil was removed below the limit<br />
of detection within a few weeks during groundwater recharge<br />
using SAT (Drewes et al., 2003).<br />
2.2. Analytical methods<br />
2.2.1. GC/MS analysis<br />
The following TOrC were analyzed by gas chromatography<br />
coupled with mass spectrometry (GC/MS) using a HP 6890<br />
gas chromatograph and a HP 5973 quadrupole mass spectrometer<br />
from Agilent Technologies (Waldbronn, Germany)<br />
adopting a method published by Reddersen and Heberer<br />
(2003): gemfibrozil, primidone, diclofenac sodium salt,<br />
carbamazepine, ketoprofen, naproxen, phenacetine, tris<br />
(2-chloroethyl)phosphate (TCEP) (Sigma Aldrich Chemicals),<br />
tris(chloropropyl)phosphate (TCPP), and propyphenazone<br />
(Pfaltz & Bauer, Inc.). Stock solutions were prepared by dissolving<br />
the compounds in milli-Q water adjusted to a pH of<br />
10, during sonification and in the dark. A volume of 500 mL<br />
of sample was collected and filtered (0.45 mm, Whatman)<br />
prior to solid phase extraction. 10,11-dihydrocarbamazepine<br />
(Sigma Aldrich Chemicals) and 2-(m-chlorophenoxy) propionic<br />
acid (Sigma Aldrich Chemicals) were used as surrogate<br />
standards. The limits of detection (LofD) and limit of<br />
quantification (LoQ) for the target compounds ranged from 5<br />
to 10 ng/L and from 10 to 50 ng/L, respectively. Only parent<br />
target compounds were investigated in this study, metabolites<br />
and conjugates were not analyzed.<br />
2.2.2. HPLC analysis<br />
A high performance liquid chromatography (HPLC) HP 1100<br />
(Agilent Technologies) combined with a UV diode array detector<br />
(DAD) was used to quantify concentrations of primidone, phenacetine,<br />
carbamazepine, naproxen, diclofenac, and ibuprofen<br />
in the lower mg/L range during adsorption breakthrough tests.<br />
The samples were directly injected without filtration or other
Table 1 – Physical–chemical properties of target organic compounds.<br />
Compound Structure pKa log P<br />
(log DpH7)<br />
Carbamazepine<br />
(anticonvulsant)<br />
Diclofenac sodium<br />
(analgesic/antiinflammatory)<br />
Gemfibrozil (blood<br />
lipid regulator)<br />
Ibuprofen<br />
(analgesic/antiinflammatory)<br />
Ketoprofen<br />
(analgesic/antiinflammatory)<br />
Naproxen<br />
(analgesic/antiinflammatory)<br />
Phenacetine<br />
(antipyretic)<br />
Primidone<br />
(anticonvulsant)<br />
Propyphenazone<br />
(analgesic)<br />
water research 44 (2010) 449–460 451<br />
Expected<br />
degradability<br />
13.9 2.67 Persistent<br />
4.0<br />
4.8<br />
4.4<br />
4.2<br />
4.0<br />
n/a<br />
12.3<br />
n/a<br />
4.06<br />
(1.06)<br />
4.39<br />
(2.19)<br />
3.72<br />
(1.12)<br />
2.81<br />
(0.01)<br />
3.00<br />
(0.00)<br />
1.63<br />
(1.63)<br />
0.4<br />
(0.4)<br />
1.74<br />
(1.74)<br />
Degradable<br />
(anoxic)<br />
Relatively<br />
persistent<br />
Degradable<br />
(better<br />
aerobic)<br />
Degradable<br />
Degradable<br />
Relatively<br />
persistent<br />
Persistent<br />
Poor/<br />
persistent<br />
References<br />
Mersmann et al.,<br />
2002; Schmidt<br />
et al., 2004<br />
Buser et al., 1998;<br />
Möhle et al., 1999;<br />
Drillia et al., 2003;<br />
Zwiener and Frimmel,<br />
2003, Hua et al.,<br />
2003; Schmidt et al., 2004<br />
Heberer and Stan, 1997;<br />
Drewes and Shore, 2001;<br />
Heberer, 2002; Drewes et al., 2003<br />
Buser et al., 1999;<br />
Zwiener and Frimmel,<br />
2003; Carballa et al., 2004;<br />
Sedlak and Pinkston, 2001;<br />
Drewes et al., 2003<br />
Buser et al., 1999;<br />
Zwiener and Frimmel,<br />
2003; Carballa et al., 2004;<br />
Sedlak and Pinkston, 2001;<br />
Drewes et al., 2003<br />
Buser et al., 1999;<br />
Zwiener and Frimmel,<br />
2003; Carballa et al., 2004;<br />
Sedlak and Pinkston, 2001;<br />
Drewes et al., 2003<br />
Buser et al., 1999;<br />
Zwiener and Frimmel,<br />
2003; Carballa et al., 2004;<br />
Sedlak and Pinkston, 2001;<br />
Drewes et al., 2003<br />
Buser et al., 1999;<br />
Zwiener and Frimmel,<br />
2003; Carballa et al., 2004;<br />
Sedlak and Pinkston, 2001;<br />
Drewes et al., 2003<br />
Heberer et al., 2003;<br />
Drewes et al., 2003<br />
(continued on next page)
452<br />
Table 1 (continued)<br />
Compound Structure pKa log P<br />
(log DpH7)<br />
Tris(2chloroethyl)phosphate<br />
(TCEP) (flame<br />
retardant)<br />
sample preparation. The mobile phase consisted of two<br />
solvents: 1) type 1 water (adjusted to pH 2.5 with o-phosphoric<br />
acid (HPLC grade, Fisher Scientific) and buffered with 25 mmol<br />
potassium phosphate monobasic (Fisher Scientific)), and 2)<br />
acetonitrile (HPLC grade, Mallinckrodt ChromAR HPLC) with the<br />
same concentration of o-phosphoric acid as solvent 1). Sample<br />
injection (100 mL) occurred in triplicates for each sample at a flow<br />
rate of 2 mL/min. A mobile phase gradient was applied during<br />
the run for solvent 2 (acetonitrile) of 30% at t ¼ 0 min., 80% at<br />
t ¼ 12 min., and 30% at t ¼ 15 min. with a post-run time of 7 min.<br />
for column cleaning (30% of solvent 2). The detection wavelengths<br />
(band width 10 nm) were set to 205 nm at t ¼ 0min.for<br />
quantification of primidone, phenacetine, and carbamazepine,<br />
230 nm at t ¼ 6.8 min. (naproxen), and 205 nm at 7.8 min. for<br />
detection of diclofenac and ibuprofen with a reference wavelength<br />
of 330 nm (band width 60 nm) for all compounds. Standards<br />
were run in the range of 5–500 mg/L. The detection limits<br />
for each compound were 5 mg/L.<br />
2.2.3. DOC/UV absorbance/nitrate/ammonium<br />
A Sievers 800 Total Organic Carbon Analyzer (GE, Boulder)<br />
was used for DOC quantification after microfiltration<br />
(0.45 mm, Whatman) (Standard Method 5310C). UV absorbance<br />
(UVA) measurements were conducted at 254 nm using<br />
a Beckman Coulter DU 800 Spectrophotometer after 0.45 mm<br />
filtration (Standard method 5910B). The specific UV absorbance<br />
(SUVA) was calculated as the ratio of UVA and DOC.<br />
Ammonium and nitrate concentrations were measured<br />
using the Nessler and the chromotropic acid method (HACH)<br />
with a detection range of 0.02–2.0 mg/l and 0.2–30 mg/L,<br />
respectively.<br />
2.2.4. Biomass<br />
Soil biomass was determined as total viable biomass (i.e.<br />
viable, not necessarily active bacteria) using phospholipid<br />
extraction (PLE) as described in Rauch and Drewes (2005).<br />
Analyses were conducted in triplicates from the top-soil<br />
(0–2 cm, infiltration zone) of the columns.<br />
2.2.5. Wastewater effluent<br />
Secondary treated wastewater effluent served as the feed to<br />
column systems and as source for subsequent organic carbon<br />
fractionation. The secondary treated effluent was collected<br />
from a local wastewater treatment plant employing nitrification<br />
and partial denitrification. The average DOC concentration<br />
of the effluent was 8.74 1.44 mg/L.<br />
water research 44 (2010) 449–460<br />
n/a<br />
0.48<br />
(0.48)<br />
n/a not applicable.<br />
pKa and log P values calculated by software ACD.<br />
log D pH¼7 values calculated using equations proposed by Scherrer and Howard, (1977).<br />
Expected<br />
degradability<br />
Relatively<br />
persistent<br />
2.2.6. Organic carbon fractionation<br />
Organic matter (less than 1 mm in size as defined for this study)<br />
of the secondary effluent sample was isolated into three<br />
organic fractions: colloidal organic carbon, hydrophobic acids<br />
(HPO-A), and hydrophilic carbon (HPI) following the procedure<br />
described in Rauch and Drewes (2005) with some modifications.<br />
The sample was concentrated by vacuum rotary evaporation<br />
at 45 C (concentration factor: 40). The concentrate<br />
was separated by dialysis (Spectra/Por, Spectrum, 6000–<br />
8000 Dalton) at pH 4–5 into organic colloids and DOC. The<br />
permeate of the dialysis (containing DOC) was collected and<br />
further separated into HPI and HPO-A by XAD-8 fractionation<br />
according to Leenheer et al. (2000) using a capacity factor of<br />
k 0 ¼ 4. A carbon mass balance was performed for each carbon<br />
fractionation based on UVA and DOC measurements for<br />
quality control. In average, the secondary treated effluent<br />
contained 16 percent colloidal organic carbon, 38 percent<br />
HPO-A, 37 percent HPI, and 8 percent hydrophobic neutrals<br />
(HPO-N) (Rauch and Drewes, 2004). Organic carbon isolates<br />
were diluted to 3 mg/L DOC using milli-Q water, adjusted for<br />
ion strength, macro nutrient (nitrogen, phosphate) and micro<br />
nutrient concentrations (Rauch and Drewes, 2004) and<br />
promptly utilized as column feed waters.<br />
2.3. Experimental set-up<br />
References<br />
Heberer et al., 2001;<br />
Heberer et al., 2003,<br />
Amy and Drewes, 2006;<br />
Fries and Püttmann, 2003<br />
2.3.1. Anoxic column system<br />
The columns (PC system) were operated to study TOrC removal<br />
during simulated MAR under anoxic conditions and up to 3–4<br />
weeks of retention time in the subsurface. The anoxic columns<br />
consisted of four 1-m plexiglass columns (15 cm i.d.) in series<br />
filled with aquifer material (d50 ¼ 0.8 mm, foc ¼ 0.003%) (Fig. SI-1,<br />
Supplemental Information). The column system was operated<br />
in flow-through mode at a loading rate of 0.065 m/d under<br />
saturated, anoxic (denitrifying) flow conditions (Table 2). The<br />
hydraulic retention time of the system using four columns in<br />
series was previously determined as 25 days. The columns had<br />
been continuously fed with secondary or tertiary treated effluents<br />
for over 6 years and for 5 months with the secondary<br />
treated effluent quality employed in this study prior to spiking of<br />
TOrC. The column influent was regularly purged with nitrogen<br />
gas to keep dissolved oxygen concentrations below 0.5 mg/L.<br />
Samples were collected once to twice a week from column<br />
influent and the four column effluents, respectively, and<br />
analyzed for DOC, UV absorbance, pH, conductivity and TOrC<br />
concentrations.
Table 2 – Operational parameters and water quality changes of columns used in this study. a<br />
Label Feed water<br />
matrix<br />
PC Anoxic columns<br />
(EfOM)<br />
C1 Hydrophobic acids<br />
(HPO-A)<br />
C2 Hydrophilic carbon<br />
(HPI)<br />
C3 Effluent organic<br />
matter (EfOM)<br />
C4 Organic colloids<br />
(Org. Colloids)<br />
Abiotic Abiotic column,<br />
Type I water<br />
Length<br />
(m)<br />
4 1m¼<br />
4m<br />
2.3.2. Organic carbon fraction columns<br />
Four columns (C1–C4) were operated to examine the effect of<br />
different organic carbon matrices on soil biomass growth and<br />
TOrC removal during aerobic conditions and short retention<br />
times (less than 1 day) in the subsurface. This system consisted<br />
of four parallel plexiglass columns (L ¼ 30 cm, i.d. ¼ 5 cm) filled<br />
with silica sand (d50 ¼ 0.65 mm, foc ¼ 0.004%). The columns<br />
were operated under saturated, aerobic flow conditions at<br />
a loading rate of 0.9 m/d representing a retention time of about<br />
19 h in the porous media. The system had been acclimated for<br />
over two years to the infiltration of the four organic carbon<br />
fractions derived from secondary effluent (HPO-A, HPI, organic<br />
colloids, EfOM), respectively. Approximately six weeks prior to<br />
spiking TOrC to the column systems, the feed water concentrations<br />
of all four fractions was adjusted to 3 mg/L as DOC.<br />
Samples for DOC, UVA and pH were taken once to twice a week<br />
from the four column influents and effluents. Three consecutive<br />
TOrC spiking events were conducted. For each event,<br />
hydraulically corresponding samples of column influents and<br />
effluents were collected, respectively, after a complete breakthrough<br />
of the spike solution was observed using conductivity<br />
as a conservative tracer (typically after 24 h). After the 1st<br />
spiking event (day 0), the 2nd and 3rd spikes occurred on day<br />
11 and 22, respectively. Set-up and operation conditions of the<br />
different columns employed in this study are summarized in<br />
Table 2. All small column systems received feed waters that<br />
were adjusted to a pH of 8.0 0.2 to minimize adsorption<br />
effects of acidic TOrC onto porous media.<br />
2.3.3. Adsorption test<br />
An additional column was operated under abiotic conditions<br />
to assess the role of physical adsorption of selected TOrC onto<br />
porous media. The column set-up and operation was the same<br />
as described for the biologically active columns C1–C4. The<br />
adsorption column was filled with aquifer material from an<br />
RBF site (d50 ¼ 0.55 mm, bulk density ¼ 1.81 g/cm 3 , effective<br />
porosity ¼ 0.36, fOC ¼ 0.066%). Six TOrC were spiked into<br />
secondary effluent in a concentration range of 175–226 mg/L<br />
adjusted to a pH of 7.0 (feed water). To minimize biotransformation<br />
during this experiment, the column was kept<br />
Organic Carbon Predominant<br />
Feed water Effluent BOC<br />
Redox Condition<br />
(mg/L) (mg/L) (mg/L)<br />
6.9 See Fig. 1 for<br />
DOC column profile<br />
abiotic by adding sodium azide at a concentration of 2 mM.<br />
Sodium bromide was added as a conservative tracer and<br />
bromide was measured by ion chromatography (Dionex,<br />
Sunnyvale, CA) in the column effluent. The column was<br />
operated in single flow-through mode under saturated flow<br />
conditions at a hydraulic loading rate of 0.33 m/d. Effluent<br />
samples were collected using a fraction collector and analyzed<br />
by HPLC analysis. Retardation factors (Rd) were fitted to the<br />
step input breakthrough curves for each compound using the<br />
software program CXTFIT (Toride et al., 1995). Adsorption<br />
coefficients (K d) were calculated according to the following<br />
relationship<br />
Rd ¼ rKd<br />
þ 1 (1)<br />
3<br />
where Rd is the retardation factor, r is the sand bulk density,<br />
and 3 is the sand porosity.<br />
3. Results and discussion<br />
3.1. Adsorption behavior of target TOrC<br />
In order to assess whether physical retardation would be<br />
a contributing factor in the removal of select TOrC in the<br />
column experiments, an abiotic column experiment was<br />
conducted to study sorption of the compounds onto the virgin<br />
porous media used in the columns. None of the six TOrC<br />
included in this experiment exhibited significant retardation<br />
during flow through porous media as compared to the<br />
conservative tracer bromide (indicated by R d values very close<br />
to 1, Table 3). These findings reveal that sorption did not play<br />
an important role in the attenuation of these compounds onto<br />
the porous media used in the columns.<br />
3.2. Organic carbon removal<br />
Total Viable<br />
Soil Biomass b<br />
(nmol PO4 3 /g d.w soil)<br />
Anoxic 17.3 1.3<br />
0.3 3.1 0.9 2.7 0.02 0.4 0.9 Aerobic 3.3 0.4<br />
0.3 2.8 0.4 2.6 0.1 0.3 0.4 Aerobic 10.5 5.3<br />
0.3 3.0 0.4 2.7 0.4 0.4 0.4 Aerobic 20.6 2.5<br />
0.3 3.1 0.5 2.2 0.3 0.8 0.5 Aerobic 27.2 1.8<br />
0.3
454<br />
Table 3 – Retardation and sorption coefficients of select<br />
TOrC derived during abiotic column study.<br />
TOrC Rd Kd (mL/g) Mass recovery (%)<br />
Naproxen 1.30 0.08 94.2<br />
Ibuprofen 1.10 0.03 92.7<br />
Phenacetine 1.02 n.a. n.a.<br />
Diclofenac 1.41 0.01 93.6<br />
Primidone 1.00
Table 4 – Summary of TOrC concentrations observed in column studies.<br />
TOrC (ng/L) Feed<br />
water<br />
1-m<br />
Column<br />
effluent<br />
Anoxic Column HPO-A (C1) HPI (C2) EfOM (C3) Org. colloids (C4)<br />
2-m<br />
Column<br />
effluent<br />
3-m<br />
Column<br />
effluent<br />
4-m<br />
column<br />
effluent<br />
Column<br />
influent<br />
Column<br />
effluent<br />
Column<br />
influent<br />
Column<br />
effluent<br />
Column<br />
influent<br />
Column<br />
effluent<br />
Column<br />
influent<br />
Carbamazepine 279–296 (n ¼ 3) n.a. n.a. n.a. 318 n.a. n.a. n.a. n.a.<br />
Diclofenac sodium 307–416 (n ¼ 2) 75 134 110 80–120 (n ¼ 2) 604 538 n.a. n.a. 710 682<br />
538 190 70
456<br />
Fig. 2 – Removal of TOrC in anoxic column system.<br />
first-order degradation kinetics are indicative of organic<br />
carbon degradation at concentrations below the half-velocity<br />
coefficient Ks, or under mass transport limitations that keep<br />
biodegradation rates below the biological capacity. Zero-order<br />
reaction kinetics are indicative of degradation below the biological<br />
degradation potential, where either substrate concentrations<br />
of trace contaminants are significantly higher than<br />
the affinity for the specific substrate, or the limiting substrate<br />
is delivered at a constant rate (Simoni et al., 2001; Alexander,<br />
1999). It should be noted that, while the column results<br />
ng/L removed in column<br />
900<br />
800<br />
700<br />
600<br />
500<br />
400<br />
300<br />
200<br />
100<br />
0<br />
Gemfibrozil Ketoprofen Naproxen<br />
3.3 10.4 27.2<br />
HPO-A HPI Org. Colloids<br />
3- -1<br />
nmol org-PO4 g<br />
Fig. 3 – Relationship between TOrC removal, soil biomass,<br />
and organic carbon substrate in columns C1, C2, and C4<br />
(1st spike event).<br />
water research 44 (2010) 449–460<br />
indicate a linear removal profile for ketoprofen, naproxen,<br />
gemfibrozil, and ibuprofen, the removal profile might also<br />
have been affected by a change in soil microbial community<br />
composition with column depth caused by changing organic<br />
carbon substrate composition and concentrations along the<br />
flow path. Thus, definite conclusions about the kinetic order<br />
are not possible without additional experiments.<br />
Due to the low ambient concentrations of trace organic<br />
contaminants (occurring in the ng/L range) it is unlikely that<br />
energy from TOrC metabolism is sufficient for biomass<br />
maintenance and growth. It was hypothesized that at trace<br />
levels, TOrC were transformed by co-metabolism (co-utilization).<br />
In this context the term co-metabolism describes<br />
a transformation of a trace compound, which is by itself<br />
unable to support cell replication, and requires the presence of<br />
another transformable organic compound (primary carbon<br />
source) that allows microorganisms to obtain energy for their<br />
metabolism and growth (Brandt et al., 2003; Stratton et al.,<br />
1983; Clara et al., 2005). Because the primary substrate<br />
(biodegradable bulk organic carbon) was present in the<br />
column experiment in concentrations two orders of magnitude<br />
above common concentrations of TOrC in the environment,<br />
co-metabolism offers a possible explanation for the<br />
observed removal of the target TOrC, that were fed at trace<br />
concentrations. This hypothesis was further tested by examining<br />
the effect of organic carbon subgroups present in EfOM<br />
and differing in biodegradability on soil biomass growth and<br />
TOrC removal.<br />
3.6. Effect of aqueous organic carbon matrices<br />
on TOrC removal<br />
In order to further study the effect of soil biomass and aqueous<br />
organic carbon matrices on the removal of TOrC, aerobic<br />
column experiments were conducted using different organic<br />
carbon bulk fractions (EfOM, HPO-A, HPI, and organic colloids)<br />
isolated from secondary treated effluent. The different organic<br />
bulk substrates were fed at similar DOC concentrations to<br />
different column systems and were able to support different<br />
amounts of viable soil biomass (Table 2). During three consecutive<br />
spiking tests, hydraulically corresponding influent and<br />
effluent samples were collected from each column (Fig. 3)with<br />
the exception of the column fed with organic colloids for which<br />
only two spiking tests were conducted.<br />
The results of the first spiking test for the HPI and organic<br />
colloid columns are presented in Fig. 3. Findings suggest that<br />
the more viable soil biomass was present on the column<br />
media, the more complete was the removal for gemfibrozil,<br />
ketoprofen, and naproxen. In other terms the concentration of<br />
biodegradable effluent derived organic carbon substrate<br />
limited the transformation of TOrC. (Phenacetine was<br />
removed below the quantification limit in all columns and<br />
removal could, thus, not be related to biomass activity in the<br />
columns.) Organic colloids (BDOC content 0.86 0.48 mg/L)<br />
appeared to have promoted the degradation of TOrC by<br />
serving as the primary substrate and establishing a relatively<br />
higher microbial population that was able to co-metabolize<br />
TOrC better as compared to HPI organic carbon. TOrC removal<br />
in the HPI column was possibly lower because HPI substrate<br />
(BDOC content 0.25 0.41 mg/L) supported less biomass
emoval %<br />
100<br />
90<br />
80<br />
70<br />
60<br />
50<br />
40<br />
30<br />
20<br />
10<br />
0<br />
growth to metabolize TOrCs. During longer exposure times to<br />
the contaminants the performance of the HPI column for<br />
TOrC removal improved. In the 2nd and 3rd spiking event,<br />
ketoprofen, phenacetine, gemfibrozil, and naproxen exhibited<br />
increasingly better removal exceeding 95 percent 22 days after<br />
the first spike (Fig. 4) suggesting that the microbial community<br />
had adapted to metabolizing TOrC.<br />
In accordance with the performance of the anoxic column<br />
system (Fig. 2), ibuprofen was equally efficient removed (>85%)<br />
in the presence of all organic carbon fractions and in all three<br />
spiking events (Table 4 and Fig. 4) under both aerobic and anoxic<br />
conditions. This confirms previously reported findings that<br />
ibuprofen is very well degradable under various field conditions<br />
(Sedlak and Pinkston, 2001; Drewes et al., 2003). In contrast to the<br />
anoxic column system, propyphenazone was not eliminated in<br />
any of the four short columns(Table 4). Consistent with previous<br />
studies, propyphenazone was found to require longer retention<br />
times (>1 week) to exhibit partial removal under MAR conditions<br />
(Drewes et al., 2003; Heberer et al., 2003). The anticonvulsant<br />
drugs were not eliminated in any of the columns<br />
investigated (Table 4). Based on these study results, effluentderived<br />
BDOC does not seem to be a suitable substrate to stimulate<br />
co-metabolic decay of the selected anticonvulsant drugs<br />
(i.e. primidone, carbamazepine) under environmentally relevant<br />
bulk and TOrC concentrations. Whereas TCPP was not<br />
removed in the anoxic columns, this flame retardant seems to<br />
be removed in the aerobic HPO-A, HPI, and EfOM column after<br />
a lag time of two to three weeks after starting to feed this<br />
compound to the column, possibly indicating biological acclimatization<br />
leading to some degree of attenuation of this<br />
compound (Table 4). However, results are inconsistent when<br />
comparing influent and effluent concentrations in the different<br />
columns, therefore further studies are recommended to solidify<br />
this observation.<br />
Given that the first column of the anoxic system was<br />
operated at a longer residence time (about 6 days) compared to<br />
the aerobic columns (less than 1 day), it is noticeable that most<br />
degradable TOrC (i.e. gemfibrozil, ibuprofen, ketoprofen,<br />
naproxen) were significantly better transformed under<br />
aerobic conditions as compared to anoxic recharge<br />
87<br />
1st spike 2nd spike (day 11 after 1st spike)<br />
3rd spike (day 22 after 1st spike)<br />
561<br />
925<br />
water research 44 (2010) 449–460 457<br />
206<br />
544<br />
700<br />
550<br />
506<br />
Ketoprofen Naproxen Phenacetine Gemfibrozil Ibuprofen<br />
508<br />
113<br />
200<br />
243<br />
490 495 499<br />
Fig. 4 – Microbial adaptation to TOrC in HPI column (values on top of bars represent removal in ng/L).<br />
conditions. Diclofenac exhibited consistent removal during<br />
anoxic conditions but not during aerobic recharge, which is<br />
consistent with findings reported by Zwiener and Frimmel<br />
(2003) and Hua et al. (2003).<br />
3.7. Removal of TOrC in oligotrophic environments<br />
Although the HPO-fraction provided little soil biomass growth<br />
(
458<br />
Because oligotrophic organisms are more likely to be prevalent<br />
in native environments characterized by low-carbon<br />
fluxes than in wastewater treatment plants, the degradation<br />
of TOrC may be more efficient during soil percolation than in<br />
wastewater treatment plants (Daughton and Ternes, 1999).<br />
It is anticipated that oligotrophic organisms reside in MAR<br />
systems in soil depths below the immediate infiltration zone<br />
(which in contrast is likely dominated by microbial organisms<br />
growing primarily on easily degradable EfOM). Further investigations<br />
directed at the interrelation between effluent derived<br />
organic carbon and the succession of biocommunities in soil<br />
based on trophic carbon sequences seem key to the understanding<br />
of mechanisms and limitations to TOrC removal in<br />
MAR systems.<br />
4. Conclusions<br />
Hydrophilic, small-molecular size TOrC can survive conventional<br />
and advanced wastewater treatment processes and are<br />
one of the key concerns in drinking water augmentation<br />
projects using impaired source waters. More research is<br />
needed to better understand the trophodynamic role of organic<br />
carbon on the removal of TOrC in MAR applications. In this<br />
study we conducted column experiments to illuminate key<br />
factors for the metabolic removal of twelve TOrC during soil<br />
infiltration. Experiments were designed to resemble natural<br />
environments (flow through systems, long biological acclimatization<br />
times, complex aqueous carbon compositions, environmentally<br />
relevant bulk and TOrC concentrations) in order<br />
to reveal interactions between bulk organic carbon and TOrC<br />
removal that are relevant for full-scale MAR conditions.<br />
Removal efficiencies varied widely among the target<br />
compounds but were generally higher during aerobic soil<br />
infiltration with the exception of diclofenac that was faster<br />
degraded during anoxic recharge. Propyphenazone required<br />
a longer residence time for removal than was provided in the<br />
aerobic columns employed in this study. The biological reactions<br />
for all other degradable TOrC were fast at ambient<br />
concentrations (ng/L level) and completed within the first 2 m<br />
of porous media infiltration. Threshold concentrations below<br />
which further degradation was negligible were observed for<br />
acidic TOrC under anoxic conditions in the lower ng/L-range.<br />
Intensive research has been conducted on stimulating the<br />
degradation of certain pollutants by artificially adding carbon<br />
substrates (Horvath, 1973, Semprini, 1997). Findings of this<br />
study demonstrated that naturally available organic matter in<br />
aqueous environments can promote the degradation of<br />
certain TOrC by serving as a secondary carbon substrate for<br />
their removal. The composition of effluent-derived bulk<br />
organic carbon had a strong effect on the biological removal of<br />
the target TOrC and is based on several mechanisms. BDOC,<br />
prevalent in form of colloidal and hydrophilic carbon, stimulates<br />
soil biomass growth and induces secondary substrate<br />
utilization of TOrC. Soil biomass in aerobic systems was able<br />
to adapt to most acidic drugs showing an improved removal<br />
over time. Soil biomass growth occurred even in response to<br />
the infiltration of recalcitrant hydrophobic acids. This organic<br />
carbon substrate induced an oligotrophic microbial community<br />
that was more effective in removing TOrC than during<br />
water research 44 (2010) 449–460<br />
copiotrophic metabolism in the presence of higher concentrations<br />
of BDOC. Consistent with previous observations, the<br />
anticonvulsant drugs carbamazepine and primidone exhibited<br />
no removal under any conditions examined in this study.<br />
Acknowledgments<br />
Partial funding for this study was provided by the Gwangju<br />
Institute of Technology in Korea. We are thankful for analytical<br />
support during this study provided by Dr. Thomas Heberer<br />
at the Technical University in Berlin, Germany, Matt<br />
Oedekoven, and Stephan Wagner.<br />
Appendix. Supplementary data<br />
Supplementary data associated with this article can be found<br />
in the online version at doi:10.1016/j.watres.2009.08.027.<br />
references<br />
Amy, G., Drewes, J.E., 2006. Soil-aquifer treatment (SAT) as<br />
a natural and sustainable wastewater reclamation/reuse<br />
technology: fate of wastewater effluent organic matter (EfOM)<br />
and trace organic compounds. Environmental Monitoring and<br />
Assessment 129 (1–3), 19–26.<br />
Alexander, M., 1999. Biodegradation and Bioremediation.<br />
Academic Press, San Diego, California.<br />
Brandt, B.W., Van Leeuwen, I.M.M., Kooijman, S.A.L.M., 2003.<br />
A general model for multiple substrate biodegradation.<br />
Application to co-metabolism of structurally non-analogous<br />
compounds. Water Research 37, 4843–4854.<br />
Brauch, H.-J., Sacher, F., Denecke, E., Tacke, T., 2000. Effectiveness<br />
of bank filtration for the removal of polar organic trace<br />
contaminants. GWF Wasser Abwasser 141 (4) 2260–234 (in<br />
German).<br />
Buser, H.R., Poiger, T., Müller, M.D., 1998. Occurrence and fate of<br />
the pharmaceutical drug diclofenac in surface waters: rapid<br />
photodegradation in a lake. Environmental Science and<br />
Technology 32 (22), 3449–3456.<br />
Buser, H.R., Poiger, T., Müller, M.D., 1999. Occurrence and<br />
environmental behavior of the chiral pharmaceutical drug<br />
ibuprofen in surface waters and in wastewater.<br />
Environmental Science and Technology 33 (15), 2529–2535.<br />
Carballa, M., Omil, F., Lema, J.M., Lllompart, M., Jares, C.G.,<br />
Rodriguez, I., Gomez, M., Ternes, T., 2004. Behaviour of<br />
pharmaceuticals, cosmetics and hormones in a sewage<br />
treatment plant. Water Research 38, 2918–2926.<br />
Clara, M., Strenn, B., Kreuzinger, N., 2004. Carbamazepine as<br />
a possible anthropogenic marker in the aquatic environment:<br />
investigations on the behaviour of carbamazepine in<br />
wastewater treatment and during groundwater infiltration.<br />
Water Research 38, 947–954.<br />
Clara, M., Kreuzinger, N., Strenn, B., Gans, O., Kroiss, H., 2005. The<br />
solids retention time – a suitable design parameter to evaluate<br />
the capacity of wastewater treatment plants to remove<br />
micropollutants. Water Research 39, 97–106.<br />
Daughton, C.G., Ternes, T.A., 1999. Pharmaceuticals and personal<br />
care products in the environment: agents of subtle change?<br />
Environmental Health Perspectives 107 (Suppl. 6) December<br />
1999.
Dillon, P., Page, D., Vanderzalm, J., Pavelic, P., Toze, S., Bekele, E.,<br />
Sidhu, J., Prommer, H., Higginson, S., Regel, R., Rinck-<br />
Pfeiffer, S., Purdie, M., Pitman, C., Wintgens, T., 2008. A critical<br />
evaluation of combined engineered and aquifer treatment<br />
systems in water recycling. Water Science and Technology 57<br />
(5), 753–762.<br />
Drewes, J.E., Fox, P., 1999. Fate of natural organic matter (NOM)<br />
during groundwater recharge using reclaimed water. Water<br />
Science and Technology 40 (9), 241–248.<br />
Drewes, J.E., Shore, L.S., 2001. Concerns about pharmaceuticals in<br />
water reuse, groundwater recharge, and animal waste. In:<br />
American Chemical Society (Ed.), Pharmaceuticals and<br />
personal care products in the environment. Symp. Ser, vol.<br />
791, pp. 206–228.<br />
Drewes, J.E., Heberer, T., Reddersen, K., 2002. Fate of<br />
pharmaceuticals during indirect potable reuse. Water Science<br />
& Technology 46 (3), 73–80.<br />
Drewes, E.J., Heberer, T., Rauch, T., Reddersen, K., 2003. Fate of<br />
pharmaceuticals during groundwater recharge. Ground Water<br />
Monitoring and Remediation 23 (3), 64–72.<br />
Drillia, P., Stamatelatou, K., Lyberatos, G., 2003. Mobility of<br />
Pharmaceuticals in Soil, 8th Conference on Environmental<br />
Science and Technology, Lemnos, Greece.<br />
Focazio, M.J., Kolpin, D.W., Barnes, K.K., Furlong, E.T., Meyer, M.T.,<br />
Zaugg, S.D., Barber, L.B., Thurman, M.E., 2008. A national<br />
reconnaissance for pharmaceuticals and other organic<br />
wastewater contaminants in the United States–II untreated<br />
drinking water sources. Science of the Total Environment 402<br />
(2–3), 201–216.<br />
Fries, E., Püttmann, W., 2003. Monitoring of the three<br />
organophosphate esters TBP, TCEP and TBEP in river water<br />
and ground water (Oder, Germany). Journal of Environmental<br />
Monitoring 5 (2), 346–352.<br />
Grünheid, S., Amy, G., Jekel, M., 2005. Removal of bulk dissolved<br />
organic carbon (DOC) and trace organic compounds by bank<br />
filtration and artificial recharge. Water Research 39, 3219–3228.<br />
Heberer, T., Stan, H.-J., 1997. Determination of clofibric acid and<br />
N-(phenylsulphonyl)-sarcosine in sewage, river and drinking<br />
water. International Journal of Environmental Analytical<br />
Chemistry 67, 113–124.<br />
Heberer, T., Verstraeten, I.M., Meyer, M.T., Mechlinski, A.,<br />
Reddersen, K., 2001. Occurrence and fate of pharmaceuticals<br />
during bank filtration–preliminary results from investigations in<br />
Germany and the United States. Water Resources Update 20, 4–17.<br />
Heberer, T., 2002. Occurrence, fate and removal of<br />
pharmaceutical residues in the aquatic environment: a review<br />
of recent research data. Toxicology Letters 131, 5–17.<br />
Heberer, T., Mechlinski, A., Fanck, B., 2003. NASRI – Occurrence<br />
and Fate of Pharmaceuticals During Bank Filtration,<br />
Conference Wasser Berlin, KompetenzZentrum Wasser Berlin,<br />
pp. 41–49.<br />
Horvath, R.S., 1973. Enhancement of co-metabolism of<br />
chlorobenzoates by the co-substrate enrichment technique.<br />
Applied Microbiology 25 (6), 961–963.<br />
Hua, J., An, P., Winter, J., Gallert, C., 2003. Elimination of COD,<br />
microorganisms and pharmaceuticals from sewage by<br />
trickling through sandy soil below leaking sewers. Water<br />
Research 37, 4395–4404.<br />
Kolpin, D.W., Furlong, E.T., Meyer, M.T., Thurman, E.M., Zaugg, S.<br />
D., Barber, L.B., Buxton, H.T., 2002. Pharmaceuticals,<br />
hormones, and other organic wastewater contaminants in U.<br />
S. streams, 1999–2000: a national reconnaissance.<br />
Environmental Science & Technology 36, 1202–1211.<br />
Leenheer, J.A., Croue, J.P., Benjamin, M., Korshin, G.V., Hwang, C..,<br />
Bruchet, A., Aiken, G.R., 2000. Comprehensive isolation of<br />
natural organic matter from water for spectral characterizations<br />
and reactivity testing. In: Barett, S.E., Krasner, S.W., Amy, G.L.<br />
water research 44 (2010) 449–460 459<br />
(Eds.), Natural Organic Matter and Disinfection By-Products, vol.<br />
761. ACS Symposium Series, pp. 68–83.<br />
Margesin, R., Schinner, F., 2001. Bioremediation (natural<br />
attenuation and biostimulation) of diesel-oil-contaminated<br />
soil in an alpine glacier skiing area. Applied and<br />
Environmental Microbiology 67 (7), 2127–3133.<br />
McCarty, P.L., Reinhard, M., Rittmann, B.E., 1981. Trace organics in<br />
groundwater. Environmental Science and Technology 15 (1), 40–51.<br />
Mersmann, P., Scheytt, T., Heberer, T., 2002. Column experiments<br />
on the transport of pharmaceutically active compounds in the<br />
saturated zone. Acta Hydrochimica et Hydrobiologica 30 (5–6),<br />
275–284 (in German).<br />
Möhle, E., Kempter, C., Kern, A., Metzger, J.W., 1999.<br />
Examination of the degradation of drugs in municipal sewage<br />
plants using liquid chromatography-electrospray mass<br />
spectrometry. Acta Hydrochimica et Hydrobiologica 27 (6),<br />
430–436 (in German).<br />
Montgomery-Brown, J., Drewes, J.E., Fox, P., Reinhard, M., 2003.<br />
Behavior of alkylphenol polyethoxlate metabolites during soil<br />
aquifer treatment. Water Research 37, 3672–3681.<br />
Preuß, G., Nehrkorn, A., 1996. Succession of microbial<br />
communities during bank filtration and artificial groundwater<br />
recharge Artificial Recharge of Groundwater. Proceedings of<br />
an International Symposium, A.-L. Kivimäki and T. Suokko,<br />
eds., Helsinki, Finland, June 3–5, 1996, Nordic Hydrological<br />
Programme, pp. 215–221.<br />
Rauch, T., Drewes, J.E., 2004. Assessing the removal potential of<br />
soil-aquifer treatment systems for bulk organic matter. Water<br />
Science & Technology 50 (2), 245–253.<br />
Rauch, T., Drewes, J.E., 2005. A novel approach for quantifying<br />
bulk organic carbon removal in groundwater recharge<br />
systems. Journal of Environmental Engineering 131 (6),<br />
909–923.<br />
Reddersen, K., Heberer, T., 2003. Multi-compound methods for<br />
the detection of pharmaceutical residues in various waters<br />
applying solid phase extraction (SPE) and gas chromatography<br />
with mass spectrometric (GC-MS) detection. Journal of<br />
Separation Sciences 26, 1443–1450.<br />
Rittmann, B.E., McCarty, P.L., 2001. Environmental Biotechnology.<br />
Principles and Applications. McGraw-Hill, Boston.<br />
Scherrer, R.A., Howard, S.M., 1977. Use of distribution coefficients<br />
in quantitative structure-activity relationships. Journal of<br />
Medical Chemistry 20 (1), 53–58.<br />
Schmidt, C.K., Lange, F.T., Brauch, H.-J., 2004. Assessing the<br />
Impact of Different Redox Conditions and Residence Times on<br />
the Fate of Organic Micropollutants during Riverbank<br />
Filtration, Fourth International Conference on<br />
Pharmaceuticals and Endocrine Disrupting Chemicals in<br />
Water, October 13–15, Minneapolis, Minnesota.<br />
Schwab, B.W., Hayes, E.P., Fiori, J.M., Mastrocco, F.J., Roden, N.M.,<br />
Cragin, D., Meyerhoff, R.D., D’aco, V.J., Anderson, P.D., 2005.<br />
Human pharmaceuticals in US surface waters: a human<br />
health risk assessment. Regulatory Toxicology and<br />
Pharmacology 42 (3), 296–312.<br />
Sedlak, D.L., Pinkston, K.E., 2001. Factors affecting the<br />
concentrations of pharmaceuticals released to the aquatic<br />
environment. Journal of Contemporary Water Research and<br />
Education 120, 56–64. September.<br />
Semprini, L., 1997. Strategies for the aerobic co-metabolism of<br />
chlorinated solvents. Current Opinion in Biotechnology 8 (3),<br />
296–308.<br />
Simoni, S.F., Schäfer, A., Harms, H., Zehnder, A.J.B., 2001. Factors<br />
affecting mass transfer limited biodegradation in saturated<br />
porous media. Journal of Contaminant Hydrology 50, 99–120.<br />
Stratton, R.G., Namkung, E., Rittmann, B.E., 1983. Secondary<br />
utilization of trace organic by biofilms on porous media.<br />
Journal American Water Works Association 75, 463–469.
460<br />
Szewzyk, U., Kalmbach, S., Manz, W., La˚ ngmark, J., and<br />
Stenström, T.-A., 1998. Artificial groundwater as the future<br />
water supply of greater Stockholm IV: Phylogenetic<br />
diversity of the unsaturated zones of experimental sand<br />
filter columns. Artificial recharge of groundwater.<br />
Proceedings of the third international symposium on<br />
artificial recharge of groundwater–Tisar 98, J.H. Peters, ed.,<br />
Amsterdam, Netherlands, 21–25 September 1998,<br />
Rotterdam, pp. 391–393.<br />
Toride, N., Leij, F.J., van Genuchten, M.T., 1995. The CXTFIT Code<br />
for Estimating Transport Parameters from Laboratory or Field<br />
water research 44 (2010) 449–460<br />
Tracer Experiments. Version 2.0. U.S. Salinity Laboratory Res.<br />
Rep. 137, U.S. Salinity, Riverside, CA.<br />
Van der Meer, J.R., Roelofsen, W., Schraa, G., Zehnder, A.J.B., 1987.<br />
Degradation of low concentrations of dichlorobezenes and 1,2,4thrichlorobenzene<br />
by Pseudomonas sp. strain P51 in nonsterile<br />
soil columns. FEMS Microbiology Letters 45 (6), 333–341.<br />
Zwiener, C., Frimmel, F.H., 2003. Short-term tests with a pilot<br />
sewage plant and biofilm reactors for the biological<br />
degradation of the pharmaceutical compounds clofibric acid,<br />
ibuprofen, and diclofenac. Science of the Total Environment<br />
309, 201–211.
Toxicological relevance of emerging contaminants for<br />
drinking water quality<br />
Merijn Schriks a, *, Minne B. Heringa a , Margaretha M.E. van der Kooi a , Pim de Voogt a,b ,<br />
Annemarie P. van Wezel a<br />
a KWR Watercycle Research Institute, P.O. Box 1072, 3430 BB Nieuwegein, The Netherlands<br />
b Institute for Biodiversity and Ecosystem Dynamics, University of Amsterdam, Nieuwe Achtergracht 166, 1018 WV,<br />
Amsterdam, The Netherlands<br />
article info<br />
Article history:<br />
Received 17 May 2009<br />
Received in revised form<br />
14 August 2009<br />
Accepted 21 August 2009<br />
Available online 26 August 2009<br />
Keywords:<br />
Risk assessment<br />
Drinking water<br />
(Provisional) drinking water<br />
guideline value<br />
Threshold of Toxicological<br />
Concern (TTC)<br />
1. Introduction<br />
abstract<br />
Due to anthropogenic activities, freshwater systems worldwide<br />
are confronted with thousands of compounds. In the<br />
European Union, for example, there are more than 100 000<br />
registered chemicals (EINECS), of which 30 000–70 000 are in<br />
daily use. About 300 million tons of synthetic compounds<br />
annually used in industrial and consumer products, partially<br />
find their way to natural waters (Schwarzenbach et al., 2006).<br />
A major contribution to chemical contamination originates<br />
from wastewater discharges that impact surface water quality<br />
water research 44 (2010) 461–476<br />
Available at www.sciencedirect.com<br />
journal homepage: www.elsevier.com/locate/watres<br />
* Corresponding author. Tel.: þ31 30 6069564.<br />
E-mail address: merijn.schriks@kwrwater.nl (M. Schriks).<br />
0043-1354/$ – see front matter ª 2009 Elsevier Ltd. All rights reserved.<br />
doi:10.1016/j.watres.2009.08.023<br />
The detection of many new compounds in surface water, groundwater and drinking water<br />
raises considerable public concern, especially when human health based guideline values<br />
are not available it is questioned if detected concentrations affect human health. In an<br />
attempt to address this question, we derived provisional drinking water guideline values<br />
for a selection of 50 emerging contaminants relevant for drinking water and the water<br />
cycle. For only 10 contaminants, statutory guideline values were available. Provisional<br />
drinking water guideline values were based upon toxicological literature data. The<br />
maximum concentration levels reported in surface waters, groundwater and/or drinking<br />
water were compared to the (provisional) guideline values of the contaminants thus<br />
obtained, and expressed as Benchmark Quotient (BQ) values. We focused on occurrence<br />
data in the downstream parts of the Rhine and Meuse river basins. The results show that<br />
for the majority of compounds a substantial margin of safety exists between the maximum<br />
concentration in surface water, groundwater and/or drinking water and the (provisional)<br />
guideline value. The present assessment therefore supports the conclusion that the<br />
majority of the compounds evaluated pose individually no appreciable concern to human<br />
health.<br />
ª 2009 Elsevier Ltd. All rights reserved.<br />
with incompletely removed organic contaminants (Kolpin<br />
et al., 2004; Snyder et al., 2001). Additional contamination<br />
comes from diffuse agricultural activities, in which over 140<br />
million tons of fertilizers and several million tons of pesticides<br />
are applied each year, and from atmospheric deposition. Such<br />
contamination can become an increasing problem for<br />
drinking water supplies, especially since the European REACH<br />
legislation may drive producers to develop newly designed<br />
less lipophilic/bioaccumulative chemicals that will be inherently<br />
more difficult to remove by traditional drinking water<br />
treatment techniques.
462<br />
Recently, Loos et al. (2009) presented an EU-wide monitoring<br />
study on 35 organic compounds in European river<br />
waters in concentrations up to 40 mg/L. In addition, we have<br />
shown the occurrence of emerging polar contaminants,<br />
such as benzotriazoles and metabolites of illicit drugs (e.g.<br />
benzoylecgonine, desalkylflurazepam and 9-carbonic acid-d-<br />
9-THC) in groundwaters and surface waters in the<br />
Netherlands (Hogenboom et al., 2009; Van Leerdam<br />
et al., 2009).<br />
Many of these emerging contaminants raise considerable<br />
toxicological and public concern, especially when human<br />
health based guideline values are unavailable. At present,<br />
the World Health Organization (WHO) and the US Environmental<br />
Protection Agency (US EPA) have derived approximately<br />
125 statutory guideline values for contaminants in<br />
drinking water (Cotruvo, 1988; US EPA, 2006; WHO, 2006).<br />
However, the potential health effects of many emerging<br />
contaminants present in the water cycle and the potential<br />
human health concern associated with direct water ingestion<br />
have not been evaluated and statutory standards are<br />
not available. Therefore, we proposed earlier to assess the<br />
potential human health concern of unknown non-genotoxic<br />
compounds (lacking structural alerts that raise concern for<br />
potential genotoxicity) by comparing the environmental or<br />
drinking water concentration to a TTC (Threshold of Toxicological<br />
Concern) derived target value (Mons et al., 2008).<br />
The TTC concept was developed in the context of food<br />
safety to obtain a first clue on risks of unregulated chemicals<br />
present at low levels. The TTC thus derived is based on the<br />
molecular structure of the chemical involved and its related<br />
mode of toxic action (Munro et al., 1996). Assuming a daily<br />
intake of 2 l/day of drinking water, and a maximum contribution<br />
of 10% from drinking water to the total exposure –<br />
both of which are standard assumptions for deriving<br />
drinking water-quality guidelines (WHO, 2006) –TTCbased<br />
target values proposed for drinking water are 0.1 mg/L for<br />
non-genotoxic compounds and 0.01 mg/L for genotoxicants<br />
(Mons et al., 2008). This value is based on a TTC level of<br />
1.5 mg/person/d (0.15 mg/person/d for substances containing<br />
structural alerts that raise concern for potential genotoxicity<br />
with an acceptable lifetime cancer risk of 10 6 ) for<br />
compounds in food (Kroes et al., 2004). Since the TTC based<br />
value is rather conservative, an over-estimation of the<br />
actual risk may be the result. For emerging contaminants,<br />
a more profound human health based assessment may<br />
therefore be very valuable. The objectives of the present<br />
study were twofold. The first objective was to collect existing<br />
drinking water guideline values for a selection of 50<br />
emerging contaminants relevant for the water cycle. If<br />
existing guideline values were not available, provisional<br />
guideline values were derived with the aid of relevant<br />
toxicological literature data. The second aim was to<br />
compare the maximum concentration levels reported in<br />
surface water, groundwater and/or drinking water to the<br />
(provisional) guideline values of the contaminants thus<br />
obtained, and express this as a Benchmark Quotient (BQ)<br />
value (further abbreviated as ‘‘BQ value’’). The present study<br />
does not attempt to quantify mixture interactions, since for<br />
compounds with an unknown mode of action there is no<br />
accepted methodology for such an assessment.<br />
water research 44 (2010) 461–476<br />
2. Materials and methods<br />
The toxicological assessment of the compounds presented in<br />
this paper comprises of a tiered approach in five consecutive<br />
steps (Fig. 1). First, the compounds to be assessed were<br />
selected. Second, n-octanol–water partition coefficients<br />
(log K ow) were obtained and compounds with a log K ow > 3<br />
were excluded from further assessment. This log K ow cut off<br />
value is applied as a default threshold; for compounds with<br />
a log K ow above 3 it is less likely that they pass drinking water<br />
treatment plants (Westerhoff et al., 2005). Third, if available,<br />
statutory drinking water guideline values were obtained from<br />
websites of competent authorities; else provisional guideline<br />
values were derived with the aid of toxicological data relevant<br />
for humans as reported in literature. Fourth, measured<br />
maximum surface water, groundwater and/or drinking water<br />
concentrations were obtained from various sources and<br />
compared to (provisional) guideline values. Finally, a BQ value<br />
was calculated from the maximum concentrations reported<br />
and the (provisional) drinking water guideline values<br />
obtained. These steps are described in more detail below.<br />
2.1. Selection of compounds for assessment<br />
A priority list representing a broad range of chemical classes<br />
was formulated with more than 100 compounds of interest.<br />
The arguments for inclusion were (i) questions related to<br />
toxicity posed by Dutch drinking water companies, (ii) potential<br />
low removal efficiency during drinking water production,<br />
(iii) appearance in recent literature and (iv) occurrence in<br />
surface waters, groundwaters and drinking water as determined<br />
by ourselves and others in various screening studies.<br />
2.2. Collection of compound-specific data<br />
2.2.1. n-Octanol–water partition coefficients (log K ow)<br />
All log Kow values were obtained with the aid of the estimation<br />
program KOWWIN (US EPA, v1.67). An exception was<br />
made for perfluoroctane sulfonate (PFOS) and perfluoroctanoic<br />
acid (PFOA), for which accurate log K ow values<br />
cannot be calculated with estimation software. For these<br />
compounds the log Kow values were obtained from a database<br />
(Krop and de Voogt, 2008).<br />
2.2.2. Toxicological data<br />
As illustrated in Fig. 1 (step 3), the first step was to obtain existing<br />
statutory drinking water guideline values from e.g. the US EPA<br />
(URL1) and the WHO (URL2). If not available, the second step was<br />
to obtain an established (by an (inter)national organization)<br />
Tolerable Daily Intake (TDI), Acceptable Daily Intake (ADI) or<br />
Reference Dose (RfD) and subsequently a provisional drinking<br />
water guideline value was derived as further described in<br />
Section 2.3. If not available, in a third step toxicity data collection<br />
focused primarily on established (by an (inter)national organization)<br />
lowest/no observed (adverse) effect levels (LO/NO(A)ELs)<br />
and subsequently a TDI was calculated as further described in<br />
Section 2.3. Finally, in a fourth step, miscellaneous toxicological<br />
information was collected and a TDI was calculated accordingly.<br />
In the case of insufficient human relevant toxicological data the
compound of interest was not further evaluated and removed<br />
from the list. To facilitate the interpretation for which<br />
compounds the toxicity database is strong and less strong, all<br />
compounds were categorized (Table 2); (A) representing<br />
compounds with a statutory drinking water guideline value, (B)<br />
representing compounds with an established TDI, ADI or RfD,<br />
(C) representing compounds for which the TDI was calculated<br />
with an established LO(A)EL or NO(A)EL and (D) representing<br />
compounds for which the TDI was calculated with miscellaneous<br />
toxicological information.<br />
TDIs, ADIs, RfDs and/or other chronic toxicity data were<br />
sourced from peer-reviewed scientific papers and from other<br />
sources such as ‘‘grey literature’’. In addition, a literature<br />
search was performed on the internet and/or toxicological<br />
relevant data were obtained from the US EPA IRIS database,<br />
the European Chemicals Bureau (ECB), the Organization for<br />
Economic Cooperation and Development (OECD), the Dutch<br />
National Institute for Health and the Environment (RIVM), the<br />
US Food and Drug Administration (US FDA), the Joint Meeting<br />
FAO/WHO Meetings on Pesticide Residues (JMPR), the Dutch<br />
Expert Committee for Occupational Standards (DECOS), the<br />
Dutch board for authorization of plant protection products<br />
and biocides (CTGB), the Scientific Committee on Occupational<br />
Exposure Limits (SCOEL), the US National Toxicology<br />
Program (NTP), the Joint FAO/WHO Committee on Food<br />
Additives (JECFA), the Food and Agriculture Organization<br />
(FAO), the Danish veterinary and food administration, the<br />
European Union (EU), the US National Research Council (NRC),<br />
the EFSA scientific panel on contaminants in the food chain<br />
(CONTAM) and the Hazardous Substance Data Bank (HSDB).<br />
water research 44 (2010) 461–476 463<br />
Fig. 1 – Flow diagram of the assessment conducted in the present study. Abbreviations: log K ow, n-octanol–water partition<br />
coefficient; GLV, drinking water guideline value; ADI, acceptable daily intake; RfD, reference dose; TDI, tolerable daily intake;<br />
LO/NO(A)EL, lowest/no observed (adverse) effect level; numbers correspond to consecutive steps as described Section 2.<br />
Compound categories: see Section 2.2.2.<br />
2.2.3. Occurrence data<br />
Collection of occurrence data focused primarily on maximum<br />
concentrations of compounds measured in the downstream<br />
parts of the Rhine and Meuse river basins during the past<br />
decade. If not available, maximum concentrations in other<br />
surface waters and/or groundwaters were sought. The<br />
primary source of occurrence data of compounds in surface<br />
waters were the annual reports of the Dutch Association of<br />
River Water Companies (RIWA) and the German Association<br />
of River Water Companies (ARW). Occurrence data of<br />
compounds in drinking water were obtained from the REWAB<br />
data set (restricted water-quality data from the Dutch water<br />
companies). If not available, alternative resources were<br />
searched such as ‘‘grey literature’’, unpublished data, or<br />
publicly available sources such as RIVM, the Dutch ministry of<br />
transport, public works and water management (Rijkswaterstaat)<br />
and the WHO. Additional data on the occurrence of<br />
compounds was obtained from peer-reviewed scientific<br />
papers and an internet based literature search.<br />
2.3. Derivation of provisional drinking water guideline<br />
values<br />
A drinking water guideline value represents the concentration<br />
of a constituent that does not exceed tolerable risk to the<br />
health of the consumer over a lifetime (WHO, 2006). In some<br />
cases, an odour-threshold value may be much lower than the<br />
health based guideline value. To calculate a provisional health<br />
based guideline value, the general methodology was applied
464<br />
Table 1 – List of compounds assessed in the present study.<br />
No. Compound CAS log K ow No. Compound CAS Log K ow<br />
1 1,4-Dioxane 2<br />
2 2,6-Dichlorobenzamide (BAM) 3<br />
3 4-Methylbenzenesulfonamide<br />
(p-toluenesulfonamide,<br />
4-tolylsulfonamide) 2<br />
4 Acetylsalicylate<br />
(aspirin, acetyl salicylic acid) 6<br />
5 Alpha-amino-3-hydroxy-<br />
5-methyl-4-isoxazole propionic<br />
acid (AMPA) 3<br />
6 Amidotrizoic acid (diatrizoic acid) 1<br />
7 Bentazone 3<br />
8 Benzene 2<br />
9 Benzothiazole 2<br />
10 Benzotriazole (1H-benzotriazole) 2<br />
11 Bis(2-chloroisopropyl)ether (BCIPE) 2<br />
12 Carbamazepine 6<br />
13 Carbendazim 3<br />
14 Chloridazon (pyrazon) 3<br />
15 Clofibric acid 6<br />
16 Dichlorophenoxyacetic acid (2,4-D) 3<br />
17 Diethyl phthalate 2<br />
18 Diethyl toluamide (DEET) 3<br />
19 Diethylamine (DEA) 2<br />
20 Diethylene glycol dimethyl ether<br />
(diglyme, bis(2-methoxy ethyl)ester)) 2<br />
21 Diethylene triamine penta acetic<br />
acid (DTPA) 2<br />
22 Dimethenamid 3<br />
23 Dimethylamine (DMA) 2<br />
24 Diuron 3<br />
25 Ethyl tert-butyl ether (ETBE) 4<br />
123-91-1 0.3 a<br />
2008-58-4 0.8 a<br />
70-55-3 0.9 b<br />
50-78-2 1.2 a<br />
77521-29-0 2.2 b<br />
117-96-4 1.4 b<br />
25057-89-0 2.3 a<br />
71-43-2 2.1 a<br />
95-16-9 2.0 a<br />
95-14-7 1.4 a<br />
108-60-1 2.5 a<br />
298-46-4 2.5 a<br />
10605-21-7 1.5 a<br />
1698-60-8 1.1 a<br />
882-09-7 2.6 a<br />
94-75-7 2.8 a<br />
84-66-2 2.4 a<br />
134-62-3 2.2 a<br />
109-89-7 0.6 a<br />
111-96-6 0.4 a<br />
as described by Van Leeuwen (2000) and the WHO (2006). For<br />
compounds without a statutory drinking water guideline<br />
value, first the Tolerable Daily Intake (TDI) was determined.<br />
The point of departure (POD) for calculating the TDI was<br />
mostly a chronic LO(A)EL, NO(A)EL, benchmark dose level,<br />
maximum tolerated dose (MTD) or lowest effective safe dose.<br />
In case only inhalatory toxicity data could be found, a routeto-route<br />
extrapolation was carried out according to toxicological<br />
methods as described by Stokinger and Woodward<br />
(1958). An appropriate safety factor to extrapolate between<br />
species (inter-species differences), inter-individual differences<br />
(intraspecies differences), exposure route/duration and<br />
quality of the data was utilized as part of the TDI calculation<br />
(Van Leeuwen and Vermeire, 2007). Secondly, a drinking water<br />
equivalent level (DWEL) was calculated by multiplying the TDI<br />
water research 44 (2010) 461–476<br />
67-43-6 4.2 b<br />
87674-68-8 2.2 a<br />
124-40-3 0.4 a<br />
330-54-1 2.7 a<br />
637-92-3 1.9 b<br />
26 Ethylenediamine tetra<br />
acetic acid (EDTA) 2<br />
27 Glyphosate 3<br />
28 Imidacloprid 3<br />
29 Iohexol 1<br />
30 Iomeprol (iomeron) 1<br />
31 Iopamidol 1<br />
32 Iopromide 1<br />
33 Isoproturon 3<br />
34 Methyl tert-butyl ether (MTBE) 4<br />
35 Metoprolol 6<br />
36 n-Butylbenzenesulphonamide 2<br />
37 Nicosulfuron 3<br />
38 n-Nitrosodimethylamine<br />
(NDMA) 2<br />
39 p,p 0 -Sulfonyldiphenol 2<br />
40 Perfluoroctanesulfonate<br />
(PFOS) (potassium-salt) 5<br />
41 Perfluorooctanoic acid<br />
(PFOA) 5<br />
42 Phenazone 6<br />
43 Simazine 3<br />
44 Sulfamethoxazole 6<br />
45 Tolyltriazole 2<br />
46 Trichloroethene 2<br />
47 Triethylphosphate<br />
(ethylphosphate) (TEP) 2<br />
48 Triphenylphosphine<br />
oxide (TPPO) 2<br />
49 Tris(2-chloroethyl)<br />
phosphate (TCEP) 2<br />
50 Urotropine (methenamine,<br />
hexamine) 2<br />
60-00-4 3.9 b<br />
1071-83-6 4.0 a<br />
138261-41-3 0.6 a<br />
66108-95-0 3.1 a<br />
78649-41-9 1.4 b<br />
62883-00-5 2.4 a<br />
73334-07-3 2.1 a<br />
34123-59-6 2.9 a<br />
1634-04-4 0.9 a<br />
37350-58-6 1.9 a<br />
3622-84-2 2.3 b<br />
111991-09-4 1.2 b<br />
62-75-9 0.6 a<br />
80-09-1 1.7 b<br />
2795-39-3 1.08 c<br />
335-67-1 2.8 c<br />
60-80-0 0.4 a<br />
122-34-9 2.2 a<br />
723-46-6 0.9 a<br />
29385-43-1 NA<br />
79-01-6 2.4 a<br />
78-40-0 0.8 a<br />
791-28-6 2.8 a<br />
115-96-8 1.4 a<br />
100-97-0 4.2 b<br />
NA not available. Chemical categories: 1 Iodinated contrast media; 2 Miscellaneous organic compounds; 3 Miscellaneous pesticides; 4 Oxygenated<br />
gasoline additives; 5 Perfluorinated organic compounds; 6 Pharmaceuticals.<br />
a log K ow values were derived from US EPA’s KOWWIN experimental database.<br />
b log Kow values were calculated with the aid of US EPA’s KOWWIN.<br />
c log Kow values were obtained from Krop and de Voogt (2008).<br />
by a typical average body weight of 70 kg and division by<br />
a daily water consumption of 2 l. Finally, to account for the<br />
fraction of the TDI allocated to drinking water, the DWEL was<br />
multiplied by an allocation factor to give the provisional<br />
guideline value. In most cases, when there was insufficient<br />
exposure information to derive chemical-specific allocation<br />
factors, a default allocation factor of 10% was used.<br />
2.4. Evaluation of water-quality data in the context of<br />
human health<br />
Of each compound the maximum concentration level reported<br />
in surface waters, groundwater and/or drinking water was<br />
compared to its (provisional) guideline value and was<br />
expressed as a BQ value (concentration in water divided by
Table 2 – Parameters used for derivation of (provisional) drinking water guideline values.<br />
Compound Point of departure (POD) Category a SUF b TDI, ADI or RfD<br />
(mg/kg bw/d)<br />
1,4-Dioxane POD is an oral slope factor as derived by US EPA (1988a) of 0.011 per mg/kg bw/d from a 110-week study<br />
in rats (NCI, 1978).<br />
2,6-Dichlorobenzamide (BAM) POD is a NOAEL of 4.5 mg/kg bw/d for decreased body weight in both sexes and increased liver weight<br />
in males as derived by the Danish EPA (2004) from a study in which dogs were fed BAM in the diet for 2<br />
years (Wilson and Thorpe, 1971), with an uncertainty factor of 300 (100 for inter- and intraspecies<br />
variation and 3 for uncertainties in the dataset).<br />
4-Methylbenzenesulfonamide<br />
(p-toluenesulfonamide,<br />
4-tolylsulfonamide)<br />
Acetylsalicylate (aspirin, acetyl<br />
salicylic acid)<br />
Alpha-amino-3-hydroxy-5-methyl-<br />
4-isoxazole propionic acid (AMPA)<br />
POD is a NOAEL of 300 mg/kg bw/d for reproductive effects (decrease lactation index and litter weight at<br />
birth) as derived from a GLP compliant study in which rats were administered 4methylbenzenesulfonamide<br />
via oral gavage for 42 days (OECD/SIDS, 1994), with an uncertainty factor<br />
of 400 (100 for inter- and intraspecies variation and 4 for extrapolation to chronic exposure).<br />
POD is a LOEL of 10 mg/person from a human study as derived by the Dutch National Institute for<br />
Health and the Environment (RIVM) (Versteegh et al., 2007).<br />
POD is a NOAEL of 32 mg/kg bw/d as derived by the WHO (2005a) from a 26-month toxicity study in rats<br />
(study reference unknown).<br />
Amidotrizoic acid (diatrizoic acid) POD is the highest therapeutic dose of 50 mg/person/d (0.71 mg/kg bw/d) as derived by the Dutch<br />
National Institute for Health and the Environment (RIVM) (Versteegh et al., 2007).<br />
Bentazone POD is a NOAEL of 9 mg/kg bw/d as derived by the WHO (1998) from a 2-year dietary toxicity study in<br />
rats (study reference unknown).<br />
Benzene POD is a risk estimate as derived by the WHO (2003a) from a 2-year gavage study in rats and mice (NTP,<br />
1986).<br />
Benzothiazole POD is a NOEL of 5.1 mg/kg bw/d as derived by the WHO/JECFA (2003) from a study in which rats were<br />
administered benzothiazole in the diet for 90 days (Morgareidge, 1971). Daily observations revealed no<br />
treatment related effects in histopathological/haematological parameters, body weight, food<br />
consumption and liver/kidney weights. An uncertainty factor of 200 was applied (100 for inter- and<br />
intraspecies variation and 2 for extrapolation to chronic exposure).<br />
Benzotriazole (1H-benzotriazole) POD is a LOAEL of 295 mg/kg bw/d for histological changes in the liver, decreased body weight gain and<br />
inflammation of the prostate/uterus as derived by the Dutch Expert Committee for Occupational<br />
Standards (DECOS, 2000) from a study in which rats were administered benzotriazole in the diet for 78<br />
weeks (BIBRA Toxicology International, 1995), with an uncertainty factor of 1000 (100 for inter- and<br />
intraspecies variation and 10 for extrapolation of a LOAEL to a NOAEL).<br />
Bis(chloroisopropyl)ether (BCIPE) POD is a NOAEL of 35.8 mg/kg bw/d as derived by the US EPA (1989) from a 24-month chronic toxicity<br />
study in mice (Mitsumori et al., 1979).<br />
Carbamazepine POD is a maximum tolerated dose (MTD) of 250 mg/kg bw/d as derived by Snyder et al. (2008) from a 2year<br />
study in rats showing evidence of carcinogenicity (Singh et al., 2005).<br />
Carbendazim POD is a NOAEL of 2.5 mg/kg bw/d as derived by the WHO/JECFA (1995) from a 2-year study in dogs<br />
(Sherman, 1972).<br />
Chloridazon (pyrazon) POD is a NOAEL of 5.4 mg/kg bw/d for adverse histopathological/haematological changes, decreased<br />
food intake, lower body weight gain and higher organ weights (liver, kidney, thyroid gland) as derived<br />
from a study in which rats were orally administered chloridazon (method of administration<br />
unspecified) for 7 weeks (ECB, 2000b), with an uncertainty factor of 100 (inter- and intraspecies<br />
variation).<br />
Clofibric acid POD is a LOEL of 1 mg/kg bw/d as derived by the Dutch National Institute for Health and the<br />
Environment (RIVM) (Versteegh et al., 2007) from an 8-week oral study in humans (Larsen et al., 1994).<br />
(Provisional)<br />
guideline<br />
value (mg/L)<br />
B NA NA 30 d<br />
C 300 0.015 52.5<br />
D 400 0.75 2600<br />
B 20 0.007 25<br />
A 100 0.3 900 c<br />
B 10 NA 250 000<br />
A 100 0.1 300 c<br />
A NA NA 10 c,d<br />
C 200 0.026 90<br />
C 1000 0.295 1000<br />
B 1000 0.04 140<br />
B NA 0.00034 1<br />
B 100 0.03 105<br />
D 100 0.054 189<br />
B 100 0.01 30<br />
(continued on next page)<br />
water research 44 (2010) 461–476 465
Table 2 (continued)<br />
Compound Point of departure (POD) Category a SUF b TDI, ADI or RfD<br />
(mg/kg bw/d)<br />
Dichlorophenoxyacetic acid (2,4-D) POD is a NOAEL of 1 mg/kg bw/d as derived by the WHO (2003b) from a 1-year study of toxicity in dogs<br />
and a 2-year study of toxicity and carcinogenicity in rats (study reference unknown).<br />
Diethyl phthalate POD is a NOAEL of 750 mg/kg bw/d as derived by US EPA (1988b) from a 16-week toxicity study in rats<br />
(Brown et al., 1978).<br />
Diethyl toluamide (DEET) POD is a NOEL of 100 mg/kg/bw/d based on clinical signs, reduced haemoglobin/haematocrit levels and<br />
histological changes in liver, lymph nodes and uterus as derived by the California Environmental<br />
Protection Agency (California EPA, 2000) from a study in which beagle dogs were orally administered<br />
DEET (gelatin capsules) for 1 year (Goldenthal, 1994), with an uncertainty factor of 56 (14 for inter- and<br />
intraspecies variation and 4 for extrapolation to chronic exposure).<br />
Diethylamine (DEA) POD is a LOAEL of 75 mg/m 3 for reduced mean body weights and adverse histopathological effects<br />
(lesions of the nasal mucosa) as derived by the scientific committee on occupational exposure limits<br />
(SCOEL, 2002) from a study in which rats were exposed to DEA via the inhalatory route for 24 weeks<br />
(6.5 h/d, 5d/wk) (Lynch et al., 1986), with an uncertainty factor of 50 (5 for the absence of human data<br />
and a NOAEL and 10 for route-to-route extrapolation uncertainties).<br />
Diethylene glycol dimethyl ether<br />
(diglyme, bis(2-methoxy ethyl)ester))<br />
Diethylene triamine penta acetic acid<br />
(DTPA)<br />
POD is a NOAEL 25 mg/kg bw/d for developmental effects (adversely affected implants per liter and<br />
decreased weight gain) as derived by the WHO (2002) from a study in which rabbits were administered<br />
diglyme via oral gavage for 13 days (NTP, 1987), with an uncertainty factor of 500 (100 for inter- and<br />
intraspecies variation and 5 for uncertainties in the dataset).<br />
POD is a NOAEL of 100 mg/kg bw/d for developmental effects (increased fetal deformations) as derived<br />
from a study according to OECD guideline 414 in which rats were administered DTPA (in its sodium<br />
form) via oral gavage during day 6–15 of pregnancy (ECB, 2000c), with an uncertainty factor of 1000 (100<br />
for inter- and intraspecies variation and 10 for extrapolation to chronic exposure).<br />
Dimethenamid POD is a NOAEL of 7 mg/kg bw/d as derived by the WHO/JMPR (2005) from a 24-month study in rats<br />
given diets containing racemic dimethenamid (study reference unknown).<br />
Dimethylamine (DMA) POD is a LOAEL of 19 mg/m 3 for concentration-related lesions in the respiratory/olfactory mucosa as<br />
derived by the scientific committee on occupational exposure limits (SCOEL, 1991) from a study in<br />
which rats and mice were exposed to DMA via the inhalatory route for 2 years (6 h/d, 5d/wk) (CIIT,<br />
1990), with an uncertainty factor of 50 (5 for the absence of human data and a NOAEL and 10 for routeto-route<br />
extrapolation uncertainties).<br />
Diuron POD is a NOEL of 0.625 mg/kg bw/d as derived by US EPA (1988c) from a 2-year feeding study in dogs<br />
(DuPont, 1964).<br />
Ethyl tert-butyl ether (ETBE) POD is a NOAEL of 500 ppm (29.1 mg/kg bw/d h ) for testes degeneration as derived from a study in which<br />
rats were exposed to ETBE via the inhalatory route for 13 weeks (Medinsky et al., 1999), with an<br />
uncertainty factor of 200 (100 for inter- and intraspecies variation and 2 for extrapolation to chronic<br />
exposure).<br />
Ethylenediamine tetra acetic Acid<br />
(EDTA)<br />
POD is a NOAEL of 250 mg/kg bw/d (190 mg/kg bw/d as the free acid) as derived by the WHO/JECFA<br />
(1973) from a 2-year toxicity study in rats (study reference unknown).<br />
Glyphosate POD is a NOAEL of 32 mg/kg bw/d as derived by the WHO (2005a) from a 26-month study of toxicity in<br />
rats fed technical-grade glyphosate (study reference unknown).<br />
Imidacloprid POD is a NOAEL of 5.7 mg/kg bw/d as derived by the WHO/JMPR (2001) from a 2-year study of toxicity<br />
and carcinogenicity in rats (study reference unknown).<br />
Iohexol POD is a safe dose of 75 g/person/d (1.07 g/kg bw/d) as derived by the Dutch National Institute for<br />
Health and the Environment (RIVM) (Versteegh et al., 2007).<br />
(Provisional)<br />
guideline<br />
value (mg/L)<br />
A 100 0.01 30 c<br />
B 1000 0.8 2800<br />
C 56 1.8 6250<br />
C 50 2.14 f<br />
750<br />
C 500 0.05 175<br />
D 1000 0.1 350<br />
B 100 0.07 245<br />
C 50 0.54 g<br />
190<br />
B 300 0.002 7<br />
D 200 0.15 525 e<br />
A 100 1.9 600 b,c<br />
A 100 0.3 900 c<br />
B 100 0.06 210<br />
B 10 NA 375 000<br />
466<br />
water research 44 (2010) 461–476
Iomeprol (iomeron) POD is a NOEL of 2 g Iodine/kg bw/d (equal to 4 g iomeprol/kg bw/d) for adverse effects on liver and<br />
kidney (non-lipid cytoplasmic vacuolization of hepatocytes and renal tubular epithelium cells) as<br />
derived from a study in which dogs were intravenously exposed to iomeprol for 28 days (Morisetti et al.,<br />
1994), with an uncertainty factor of 2100 (35 for inter- and intraspecies, 10 for route-to-route<br />
extrapolation and 6 for extrapolation to chronic exposure).<br />
D 2100 1.9 6700<br />
Iopamidol POD is a safe dose of 83 g/person/d (1.19 g/kg bw/d) as derived by the Dutch National Institute for<br />
Health and the Environment (RIVM) (Versteegh et al., 2007).<br />
B 10 NA 415 000<br />
Iopromide POD is a safe dose of 50 g/person/d (0.71 g/kg bw/d) as derived by the Dutch National Institute for<br />
Health and the Environment (RIVM) (Versteegh et al., 2007).<br />
B 10 NA 250 000<br />
Isoproturon POD is a NOAEL of 3 mg/kg bw/d as derived by the WHO (2003c) from a 90-day study in dogs and a 2-year<br />
feeding study in rats (study reference unknown).<br />
A 1000 0.003 9 c<br />
Methyl tert-butyl ether (MTBE) POD is a NOAEL of 300 mg/kg bw/d as derived by the Dutch National Institute for Health and the<br />
Environment (RIVM) (Swartjes et al., 2004) from a 90-day oral toxicity study in rats (Robinson et al.,<br />
1990).<br />
B 1000 0.3 9400 e<br />
Metoprolol POD is a LOEL of 100 mg/person/d (1.42 mg/kg bw/d) as derived by the Dutch National Institute for<br />
Health and the Environment (RIVM) (Versteegh et al., 2007).<br />
B 100 0.014 50<br />
n-Butylbenzenesulphonamide POD is a NOAEL of 50 mg/kg bw/d for adverse treatment related macro- or microscopic effects (liver<br />
enlargement, hepatocyte hypertrophy, thymic atrophy and lymphocytolysis) as derived from a study<br />
according to OECD guideline 407 (GLP compliant) in which rats were administered nbutylbenzenesulphonamide<br />
via oral gavage for 28 days (Proviron Fine Chemicals, 2003), with an<br />
uncertainty factor of 600 (100 for inter- and intraspecies differences, 6 for extrapolation to chronic<br />
exposure).<br />
D 600 0.083 292<br />
Nicosulfuron POD is a NOAEL of 199 mg/kg bw/d as derived from a 2-year study in rats (URL3). B 1000 0.2 700<br />
n-Nitrosodimethylamine (NDMA) POD is a tumor dose (TD05, dose level that causes 5% in increase in tumour incidence over the<br />
background) of 18 mg/kg bw/d as derived by the WHO (2008) from a detailed 2-year cancer dose–<br />
response study in rats (Peto et al., 1991a,b).<br />
A NA NA 0.1 c,d<br />
p,p 0 -Sulfonyldiphenol POD is a NOEL of 10 mg/kg bw/d for histopathological effects (hyperplasia of the mucosal epithelium,<br />
hypertrophy of hepatocytes), decreased food consumption/body weight gain and increased liver<br />
weights as derived from a study according to OECD guideline 421 in which male and female rats were<br />
administered p,p 0 -sulfonyldiphenol via oral gavage for respectively 45 days and from 14 days before<br />
mating to day 3 of lactation (Mitsubishi Chemical Safety Institute Ltd, date of study unknown), with an<br />
uncertainty factor of 600 (100 for inter- and intraspecies differences, 6 for extrapolation to chronic<br />
exposure).<br />
Perfluoroctane sulfonate (PFOS) POD is a NOAEL of 0.03 mg/kg bw/d as derived by the Scientific Panel on Contaminants in the Food<br />
Chain (CONTAM) (EFSA, 2008) from a 182-day study in Cynomolgus monkeys (Seacat et al., 2002).<br />
Perfluorooctanoic acid (PFOA) POD is a benchmark dose level (BMDL10) of 0.3 mg/kg bw/d as derived by the Scientific Panel on<br />
Contaminants in the Food Chain (CONTAM) (EFSA, 2008) from a number of studies in mice and male<br />
rats (study references unknown).<br />
Phenazone POD is a LOEL of 250 mg/person/day (3.57 mg/kg bw/d) as derived by the Dutch National Institute for<br />
Health and the Environment (RIVM) (Versteegh et al., 2007).<br />
Simazine POD is a NOAEL of 0.52 mg/kg bw/d as derived by the WHO (1996) from a 2-year combined chronic<br />
toxicity/oncogenicity study in rats (Ciba-Geigy, 1988; unpublished study submitted to WHO).<br />
Sulfamethoxazole POD is a NOAEL of 25 mg/kg/d as derived by Schwab et al. (2005) from a 60-week study in rats (Swarm<br />
et al., 1973).<br />
Tolyltriazole POD is a NOAEL of 150 mg/kg bw/d for observed mild apathy as derived from a study in which rats were<br />
administered tolyltriazole via oral gavage for 29 days (Benzotriazoles Coalition, 2001; ECB, 2000a), with<br />
an uncertainty factor of 600 (100 for inter- and intraspecies variation and 6 for extrapolation to chronic<br />
exposure).<br />
D 600 0.017 60<br />
B 200 0.00015 0.5<br />
B 200 0.0015 5.3<br />
B 100 0.036 125<br />
A 1000 0.52 2 c<br />
B 200 0.13 440<br />
D 600 0.25 875<br />
(continued on next page)<br />
water research 44 (2010) 461–476 467
Table 2 (continued)<br />
Compound Point of departure (POD) Category a SUF b TDI, ADI or RfD<br />
(mg/kg bw/d)<br />
Trichloroethene POD is a benchmark dose level (BMDL10) of 0.146 mg/kg bw/d as derived by the WHO (2005b)/Health<br />
Canada (2003) from a developmental toxicity study in rats (Dawson et al., 1993).<br />
Triethylphosphate (ethylphosphate)<br />
(TEP)<br />
POD is a NOEL of 335 mg/kg bw/d for fertility effects (effects on litter size) as derived from a study in<br />
which rats were administered TEP via the food for a unknown period (OECD/SIDS, 1998), with an<br />
uncertainty factor of 600 (100 for inter- and intraspecies variation and 6 for extrapolation to chronic<br />
exposure).<br />
Triphenylphosphine oxide (TPPO) POD is a NOAEL of 8 mg/kg bw/d for salivation, vomiting, diarrhea, histopathological (liver damage and<br />
skeletal muscle atrophy)/haematological (elevated GPT, GOT and alkaline phosphatase activities,<br />
reduced haemoglobin/haematocrit levels) parameters as derived from a study in which dogs were<br />
administered TPPO via the food for 3 months (ECB, 2000d), with an uncertainty factor of 1000 (100 for<br />
inter- and intraspecies variation and 10 for extrapolation to chronic exposure).<br />
Tris(2-chloroethyl)phosphate (TCEP) POD is a NOAEL of 22 mg/kg bw/d for increased relative liver and kidney weights as derived from<br />
a study in which rats were administered TCEP via oral gavage for 16 weeks (NTP, 1991), with an<br />
uncertainty factor of 1000 (10 for inter- and intraspecies variation and 10 for extrapolation to chronic<br />
exposure and uncertainty in genotoxic potential).<br />
Urotropine (methenamine, hexamine) POD is a NOAEL of 15 mg/kg bw/d as derived by the WHO/JECFA (1974) from a teratogenicity study<br />
(exposure from the fourth to fifty-sixth day after mating) in dogs (Hurni and Ohder, 1973).<br />
(Provisional)<br />
guideline<br />
value (mg/L)<br />
A 100 0.0015 20 c<br />
D 600 0.56 1950<br />
D 1000 0.008 28<br />
D 1000 0.022 77<br />
B 100 0.15 500<br />
NA not available.<br />
a Categories: A) statutory drinking water guideline value available; B) established TDI, ADI or RfD available; C) TDI calculated with a established LO(A)EL or NO(A)EL; D) TDI calculated with<br />
miscellaneous toxicological information.<br />
b Uncertainty factors.<br />
c WHO drinking water guidelines (WHO, 2006).<br />
d Based on a specific cancer risk level of 10 5 .<br />
e The odour-threshold value for drinking water preparation is 15 mg/L for MTBE and w1 mg/L for ETBE (Swartjes et al., 2004; Van Wezel et al., 2009).<br />
f Using the Stokinger–Woodward approach (Stokinger and Woodward, 1958), a TDI of 150 mg/person/d (2.14 mg/kg bw/d) can be calculated from the 8-h threshold limit value (15 mg/m 3 ) assuming<br />
100% oral/inhalatory absorption and a 8-h total workshift ventilation of 10 m 3 .<br />
g Using the Stokinger–Woodward approach (Stokinger and Woodward, 1958), a TDI of 40 mg/kg/person/d (0.54 mg/kg bw/d) can be calculated from the 8-hour threshold limit value (3.8 mg/m 3 )<br />
assuming 100% oral/inhalatory absorption and a 8-h total workshift ventilation of 10 m 3 .<br />
h A TDI of 0.15 mg/kg bw/day can be calculated from the NOAEL (500 ppm ¼ 29.1 mg/kg/bw/d) assuming 100% oral/inhalatory absorption, a specific lung retention of 25% and a minute ventilation<br />
volume for rat of 45 mL.<br />
468<br />
water research 44 (2010) 461–476
guideline value) in order (i) to provide perspective on what the<br />
occurrence of emerging contaminants might signify to human<br />
health and (ii) to help prioritize further investigations. A BQ<br />
value of 1 represents a (drinking) water concentration equal to<br />
the (provisional) guideline value. Compounds with a BQ value<br />
of 1 in drinking water may be of potential human health<br />
concern if the water were to be consumed over a lifetime<br />
period. Compounds with a BQ value 0.1 in drinking water<br />
were identified as those that may warrant further investigation;<br />
this is consistent with various US State and Federal<br />
practices (Toccalino, 2007). For compounds found in surface<br />
waters and groundwater the BQ value threshold to carry out<br />
an additional assessment was set at an arbitrary value of 0.2,<br />
since these source waters are purified in drinking water<br />
treatment plants which provides extra safety. Compounds in<br />
surface waters/groundwater or drinking water with a BQ value<br />
of 0.2 or 0.1 respectively, are presumed to present no<br />
appreciable concern to human health.<br />
3. Results<br />
3.1. Selection of compounds<br />
For only 50 compounds out of the original list, statutory<br />
guideline values or useful toxicity and occurrence data could<br />
be found. These compounds constitute the final list, which<br />
includes compounds from various groups such as iodinated<br />
contrast media, pharmaceuticals, oxygenated gasoline additives,<br />
perfluorinated organic compounds, miscellaneous<br />
organic compounds and pesticides (Table 1). Natural and<br />
synthetic steroid hormones such as 17b-estradiol, 17a-ethynylestradiol<br />
and estrone were not included in this assessment,<br />
as they are removed relatively easily in drinking water<br />
purification processes (Nghiem et al., 2004).<br />
3.2. (Provisional) drinking water guideline values<br />
For 10 compounds WHO statutory drinking water guideline<br />
values were available and these compounds were classified<br />
as category A. For the remaining 40 compounds a provisional<br />
guideline value was established with the aid of toxicological<br />
data. An established TDI, ADI or RfD was available for 22<br />
compounds (category B). In 7 cases when there was no TDI,<br />
ADI or RfD available, an established NO(A)EL or LO(A)EL was<br />
used to calculate a TDI and subsequently a provisional<br />
drinking water guideline value (category C). For the remaining<br />
11 compounds, miscellaneous toxicological data was<br />
used to calculate a TDI and subsequently a provisional<br />
drinking water guideline value (category D). As tabulated in<br />
Table 2, (provisional) guideline values ranged from<br />
0.0001 mg/L for NDMA to 415 mg/L for the iodinated contrast<br />
medium iopamidol. All iodinated contrast media had relatively<br />
high provisional guideline values, ranging from 6.7 mg/<br />
L (iomeprol) to 415 mg/L (iopamidol). In the cases of MTBE<br />
and ETBE, the human health based guideline values were at<br />
least one order of magnitude higher than the corresponding<br />
odour-threshold based guideline values of respectively 15 mg/<br />
L and w1 mg/L (Swartjes et al., 2004; Van Wezel et al., 2009).<br />
For two compounds (DMA and DEA) the provisional guideline<br />
water research 44 (2010) 461–476 469<br />
values were established by route-to-route extrapolation of<br />
inhalatory LOAELs to oral LOAELs. Since benzene was evaluated<br />
to be genotoxic/carcinogenic (IARC group I) and NDMA<br />
(IARC group 2A) and 1,4-dioxane (IARC group 2B) are respectively<br />
suspected non-genotoxic and genotoxic carcinogens,<br />
their corresponding (provisional) guideline values are<br />
provided as an upper bound lifetime cancer risk to an individual<br />
of 10 5 , or the odds that one case of cancer would<br />
result for every 100 000 persons subjected to continuous<br />
exposure over a 70-year lifetime.<br />
3.3. Concentration of compounds in surface waters,<br />
groundwaters and drinking water<br />
The maximum concentrations of compounds reported in<br />
surface waters and/or groundwaters are summarized in<br />
Table 3. Measured maximum surface water concentrations<br />
were available for 37 of the 50 compounds in the annual<br />
reports of RIWA and ARW. For two compounds (MTBE and<br />
clofibric acid) maximum concentrations in Dutch groundwater<br />
are reported. For the remaining compounds, the<br />
maximum concentration reported in surface waters was<br />
taken from other sources (see Section 2.2.3).<br />
The six compounds with the highest reported maximum<br />
concentrations in surface waters were EDTA (29 mg/L), DTPA<br />
(12.2 mg/L), p,p 0 -sulfonyldiphenol (10 mg/L), urotropine (10 mg/<br />
L), 1,4-dioxane (10 mg/L) and AMPA (5 mg/L), whereas in<br />
groundwater a relatively high concentration was found for<br />
MTBE (27.3 mg/L) showing the environmental relevance of this<br />
compound. The highest maximum concentration of iodinated<br />
contrast media in surface waters was reported for<br />
iomeprol (0.97 mg/L).<br />
Table 3 also summarizes the maximum concentrations of<br />
compounds reported in drinking water. Data on the occurrence<br />
of compounds in drinking water were relatively scarce,<br />
and limited to 35 compounds. For 18 compounds, drinking<br />
water concentrations were obtained from the Dutch REWAB<br />
database and for 17 compounds drinking water concentrations<br />
were taken from reports by others. Drinking water<br />
concentrations for the remaining compounds could not be<br />
found. The highest maximum concentration reported was for<br />
EDTA (13.6 mg/L), followed by DTPA (9 mg/L), metoprolol<br />
(2.1 mg/L) and BCIPE (1.9 mg/L).<br />
3.4. Comparison of compound concentrations to<br />
(provisional) guideline values (BQ value)<br />
For all compounds found in surface waters, groundwaters<br />
and drinking water the calculated BQ value was
Table 3 – Reported concentrations in surface waters, groundwaters and drinking water and comparison to (provisional) drinking water guideline values expressed as<br />
Benchmark Quotient (BQ) values.<br />
Compound Surface waters and groundwaters Drinking water<br />
Max conc (mg/L) (number<br />
of measurements, year)<br />
Source Ref BQ<br />
value a<br />
Max conc (mg/L) (number<br />
of measurements, year)<br />
Source Ref BQ<br />
value a<br />
1,4-Dioxane 10 (NA, 1997) SW, NL (4) 0.3 0.5 (NA) UDW, NL (4) 0.02<br />
2,6-Dichlorobenzamide (BAM) 0.05 (40, 2002-2006) SW, NL (11) 0.001 0.23 (14, 2002) FDW, USA (10) 0.004<br />
4-Methylbenzenesulfonamide<br />
(p-toluenesulfonamide,<br />
4-tolylsulfonamide)<br />
0.06 (20, 2005) SW, NL (13) 0.00002 NA<br />
Acetylsalicylate (aspirin, acetyl<br />
salicylic acid)<br />
0.065 (NA, 2007) SW, NL (15) 0.003 0.12 (12, 2007) FDW, NL (15) 0.005<br />
Alpha-amino-3-hydroxy-5-methyl-<br />
4-isoxazole propionic acid (AMPA)<br />
5 (499, 2005) SW, NL (11) 0.006 1.1 (6, 2001) FDW, NL (10) 0.001<br />
Amidotrizoic acid (diatrizoic acid) 0.63 (189, 2006) SW, GER (2) 0.000003 0.25 (6, 2006) FDW, NL (10) 0.000001<br />
Bentazone 0.1 (126, 2007) SW, NL (11) 0.0003 0.28 (11, 2006) FDW, NL (10) 0.0009<br />
Benzene 0.74 (116, 2001) SW, NL (11) 0.07 0.96 (12, 2005) TW, NL (10) 0.1<br />
Benzothiazole 0.03 (3, 2008) SW, NL (6) 0.0003 0.01 (10, 2007) FDW, NL (14) 0.0001<br />
Benzotriazole (1H-benzotriazole) 0.54 (11, 2007) SW, NL (11) 0.0005 0.2 (10, 2007) FDW, NL (14) 0.0002<br />
Bis(chloroisopropyl)ether (BCIPE) 2.9 (15, 1984-1985) GW, NL (6) 0.02 1.9 (9, 1982-1984) FDW, NL (6) 0.01<br />
Carbamazepine 0.227 (263, 2003) SW, NL (11) 0.2 0.03 (2, 2007) FDW, NL (15) 0.03<br />
Carbendazim 1.5 (111, 2006) SW, BE (11) 0.01 NA<br />
Chloridazon (pyrazon) 0.3 (68, 2002) SW, BE (11) 0.002 NA<br />
Clofibric acid 0.091 (NA, 2007) BFGW, NL (15) 0.003 0.14 (2, 2007) FDW, NL (15) 0.005<br />
Dichlorophenoxyacetic acid (2,4-D) 0.2 (34, 2006) SW, NL (11) 0.007 0.11 (5, 2002) TW, NL (10) 0.004<br />
Diethyl phthalate 0.9 (8, 2005) SW, NL (11) 0.0003 NA<br />
Diethyl toluamide (DEET) 0.06 (36, 2005) SW, NL (11) 0.00001 0.03 (1, 2005) TW, NL (10) 0.000005<br />
Diethylamine (DEA) 0.29 (38, 2007) SW, NL (11) 0.0004 NA<br />
Diethylene glycol dimethyl<br />
ether (diglyme, bis(2-methoxy ethyl)ester))<br />
3.64 (11, 2007) SW, NL (11) 0.02 0.15 (13, 2007) UDW, NL (1) 0.0009<br />
Diethylene triamine penta acetic acid (DTPA) 12.2 (53, 2005) SW, NL (11) 0.03 9 (2, 2001) FDW, NL (10) 0.03<br />
Dimethenamid 0.12 (4, 2005) SW, NL (9) 0.0005 NA<br />
Dimethylamine (DMA) 0.34 (42, 2005) SW, NL (11) 0.002 NA<br />
Diuron 0.68 (386, 2002) SW, BE (11) 0.1 0.08 (2, 2005) TW, NL (10) 0.01<br />
Ethyl tert-butyl ether (ETBE) 1.2 (97, 2006) SW, GER (2) 0.002 (1.2 b ) NA<br />
Ethylenediamine tetra acetic Acid (EDTA) 29 (192, 2005) SW, NL (11) 0.05 13.6 (7, 2001) FDW, NL (10) 0.02<br />
Glyphosate 1.2 (291, 2006) SW, NL (11) 0.001 0.46 (3, 2006) TW, NL (10) 0.0005<br />
Imidacloprid 0.06 (1, 2007) SW, NL (11) 0.0003 NA<br />
Iohexol 0.5 (180, 2005) SW, NL (11) 0.000001 0.06 (1, 2007) FDW, NL (15) 0.0000002<br />
Iomeprol (iomeron) 0.97 (172, 2007) SW, NL (11) 0.0001 0.01 (1, 2006) FDW, NL (10) 0.000001<br />
Iopamidol 0.714 (188, 2006) SW, NL (11) 0.000002 0.1 (6, 2006) FDW, NL (10) 0.0000002<br />
Iopromide 0.56 (186, 2004) SW, NL (11) 0.000002 0.04 (2, 2007) FDW, NL (15) 0.0000002<br />
Isoproturon 0.31 (256, 2002) SW, NL (11) 0.03 0.02 (1, 2004) TW, NL (10) 0.002<br />
Methyl tert-butyl ether (MTBE) 27.3 (14, 2003-2005) GW, NL (7) 0.003 (1.8 b ) 1.25 (27, 2006) FDW, NL (10) 0.0001 (0.08 b )<br />
Metoprolol 0.2 (114, 2006) SW, NL (11) 0.004 2.1 (2, 2005) FDW, NL (10) 0.04<br />
n-Butylbenzenesulphonamide 0.78 (300-400, 2004-2006) SW, NL (6) 0.003 0.05 (2, 2004) TW, NL (13) 0.0002<br />
Nicosulfuron 0.17 (5, 2007) SW, NL (11) 0.0002 NA<br />
n-Nitrosodimethylamine (NDMA) 0.0071 (38, 2006) SW, NL (11) 0.07 0.002 (21, 2007) UDW, NL (5) 0.02<br />
470<br />
water research 44 (2010) 461–476
p,p0-Sulfonyldiphenol 10 (300, 2005-2006) SW, BE (6) 0.1 NA<br />
Perfluoroctane sulfonate (PFOS) 0.11 (NA, 2008) SW, NL (8) 0.2 0.02 (NA, 2005/2006) TW, GER (12) 0.04<br />
Perfluorooctanoic acid (PFOA) 0.647 (NA, 2005/2006) SW, GER (13) 0.1 0.52 (NA, 2005/2006) TW, GER (12) 0.1<br />
Phenazone 0.11 (26, 2005) SW, NL (12) 0.0009 0.03 (8, 2005) FDW, NL (15) 0.0002<br />
Simazine 0.13 (124, 2006) SW, BE (11) 0.07 0.06 (4, 2004) FDW, NL (10) 0.03<br />
Sulfamethoxazole 0.11 (170, 2005) SW, NL (11) 0.0003 0.03 (4, 2007) FDW, NL (15) 0.00007<br />
Tolyltriazole 0.29 (11, 2007) SW, NL (11) 0.0003 NA<br />
Trichloroethene 1.35 (206, 2006) SW, NL (11) 0.07 1.75 (8, 2005) FDW, NL (10) 0.09<br />
Triethylphosphate (ethylphosphate) (TEP) 0.189 (25, 2007) SW, NL (11) 0.0001 NA<br />
Triphenylphosphine oxide (TPPO) 0.344 (88, 2003) SW, NL (11) 0.01 0.13 (6, 2007) FDW, NL (6) 0.005<br />
Tris(2-chloroethyl)phosphate (TCEP) 0.29 (8, 2006) SW, NL (11) 0.004 NA<br />
Urotropine (methenamine, hexamine) 10 (NA, 1997-1998) SW, GER (3) 0.02 NA<br />
Sources: bank-filtrated groundwater (BFGW); groundwater (GW); surface water (SW); finished drinking water (FDW); tap water (TW); unspecified drinking water (UDW). Locations: Belgium (BE);<br />
Germany (GER); the Netherlands (NL); United States of America (USA). References: (1) Anonymous data Dutch drinking water companies; (2) ARW database; URL4; (3) Brauch et al., 2000; (4) ECB, 2002; (5)<br />
Kleinneijenhuis and Puijker, 2008; (6) KWR internal data (KWR, 2009); (7) De Voogt et al., 2008; (8) Loos et al., 2009; (9) Mout et al., 2007; (10) REWAB database, 2009; (11) RIWA database; URL5 (12)<br />
Skutlarek et al., 2006; (13) van Beelen, E.S.E. (HWL, the Netherlands), personal communication; (14) Van Leerdam et al., 2009; (15) Versteegh et al., 2007. NA: Not available; an absence of measured<br />
concentrations above the detection limit.<br />
a BQ value: ratio of reported maximum concentration to the (provisional) guideline value (cf. Table 2).<br />
b Based on an odour-threshold value (cf. Table 2).<br />
water research 44 (2010) 461–476 471<br />
there is a concern for drinking water production, because<br />
consumers will not accept odourous drinking water. The BQ<br />
values of the remaining compounds calculated for surface<br />
waters and groundwater ranged between 0.1 (p,p 0 -sulphonylphenol,<br />
diuron) and 0.000001 (iohexol). For all iodinated<br />
contrast media the concentration in surface waters was at<br />
least three orders of magnitude less than the (provisional)<br />
guideline value. For drinking water, the BQ values of these<br />
compounds were even lower (Fig. 2B). For two compounds<br />
(benzene and PFOA) found in drinking water, the BQ value<br />
was equal to 0.1 indicating that additional assessments such<br />
as establishing trends may be warranted. However, for 15<br />
compounds occurrence data in drinking water were not<br />
available and therefore the human health concern associated<br />
with drinking water consumption due to presence of any of<br />
these compounds remains unknown.<br />
4. Discussion<br />
Rapid new developments in analytical chemistry lead to the<br />
detection and quantification of many emerging contaminants<br />
in drinking water and its environmental sources (surface<br />
water and groundwater). Since toxicological information is<br />
often absent, such compounds are a growing concern for<br />
drinking water companies and their customers.<br />
The present study attempts to address potential human<br />
health concern associated with water containing emerging<br />
contaminants. The 50 compounds included in this study<br />
represent a broad range of chemical classes for which<br />
maximum concentrations in surface waters, groundwater<br />
and/or drinking water were obtained in the downstream parts<br />
of the Rhine and Meuse basins. The results as presented in<br />
Fig. 2 indicate that a substantial margin exists between the<br />
(provisional) guideline value and the maximum concentrations<br />
of most compounds reported in surface waters,<br />
groundwaters and/or drinking water.<br />
The compounds evaluated with a relatively high BQ value<br />
(i.e. a high potential human health concern) and a known<br />
carcinogenic action are 1,4-dioxane, benzene and NDMA. The<br />
(provisional) guideline values for 1.4-dioxane (30 mg/L), benzene<br />
(10 mg/L) and NDMA (0.1 mg/L) used in the present study are<br />
based on a specific cancer risk level of 10 5 . However, when<br />
applying a specific risk level of 10 6 ,asiscommonpracticeinthe<br />
Netherlands, the provisional guidelines value would be 3 mg/L,<br />
1 mg/L and 0.01 mg/L, respectively. This would result in BQ values<br />
much higher than the arbitrary thresholds for surface waters<br />
and drinking water employed in the present study. This indicates<br />
that very low concentrations of these compounds in<br />
drinking water could lead to a potential carcinogenic effect, and<br />
we conclude that for these compounds it is important to<br />
monitor trends in their (environmental) occurrence.<br />
Furthermore, Fig. 2 illustrates that the (provisional) guideline<br />
values of the majority of non-genotoxic compounds are at least<br />
two orders of magnitude above the Threshold of Toxicological<br />
Concern (TTC)-based drinking water target value for non-genotoxic<br />
compounds (0.1 mg/L). In addition, the (provisional)<br />
guideline values (expressed as a specific risk level of 10 6 )ofthe<br />
three compounds with known carcinogenic action (1,4-dioxane,<br />
benzene and NDMA) are equal (NDMA) or much higher (1,4
472<br />
Fig. 2 – Comparison of compound concentrations in (A) surface/groundwaters and (B) drinking water to (provisional)<br />
guideline values. Benchmark Quotient (BQ) thresholds are indicated with dashed lines. Threshold of Toxicological Concern<br />
(TTC) based target value for non-genotoxic compounds (0.1 mg/L) is indicated with a dotted line. Numbers correspond to<br />
compounds as tabulated in Table 1.<br />
dioxane and benzene) than the TTC derived target value of<br />
0.01 mg/L for genotoxic compounds. This illustrates that the TTC<br />
based drinking water target value may be a conservative value<br />
ideally suited for exposure based waiving of compounds for<br />
which there is no sufficient toxicological information, which<br />
can be followed up by a more data-intensive evaluation.<br />
Two perfluorinated organic compounds were evaluated in<br />
the present study. For PFOA in drinking water a BQ value equal<br />
to the arbitrary threshold of 0.1 was calculated, whereas for<br />
PFOS in surface waters a BQ value of 0.2 was calculated. These<br />
persistent compounds are becoming a global problem, and<br />
PFOA and PFOS have already been detected in the ng/L range<br />
in, e.g. European and Japanese tapwaters (Ericson et al., 2007;<br />
Loos et al., 2007; Norimitsu et al., 2004). Recently, Skutlarek<br />
et al. (2006) observed at sampling site Neheim (river Ruhr<br />
catchment, a tributary of the river Rhine, Germany)<br />
water research 44 (2010) 461–476<br />
concentrations of PFOA of 0.65 mg/L in Lake Moehne, and<br />
0.53 mg/L in corresponding drinking water, respectively. The<br />
authors concluded that water treatment steps may not<br />
effectively eliminate perfluorinated compounds to a sufficient<br />
extent, although approximately 50% of the waterworks at the<br />
Ruhr river are equipped with activated carbon filters. Hence,<br />
more research should be devoted to the behavior of perfluorinated<br />
organic compounds in drinking water treatment<br />
processes.<br />
Several structurally related iodinated contrast media<br />
(iopamidol, iohexol, iomeprol and iopromide) were evaluated<br />
in the present research. However, their calculated BQ values<br />
are much lower than the BQ threshold above which further<br />
investigations would be warranted. Iopromide, for example, is<br />
a relatively non-toxic compound with a reported safe dose<br />
(intravenous) of 50 g/person/d (Versteegh et al., 2007). Despite
the absence of human health effects, these compounds may<br />
deserve further attention from an environmental impact<br />
point of view. Since environmental sublethal effects of<br />
iodinated contrast media to organisms are largely unknown,<br />
taken together with high persistence and environmental<br />
presence at relatively high concentrations, additional environmental<br />
assessments may be necessary.<br />
For compounds with a low (provisional) guideline value as<br />
identified in the present study (e.g. carbamazapine), additional<br />
environmental monitoring may be warranted to characterize<br />
concentrations and to establish trends in their occurrence. As<br />
shown by Walraven and Laane (2009), river flow rates may<br />
influence contaminant concentrations seasonally, thus<br />
resulting in substantially varying BQ values. For example, it can<br />
be observed that the riverine concentration of the fuel<br />
oxygenate MTBE is highly dependent on the flow of the river<br />
Meuse. Similar patterns may occur for other compounds,<br />
resulting in (temporarily) exceedance of the BQ threshold.<br />
The evaluation as presented here supports the conclusion<br />
that the majority of the selected compounds as found in<br />
surface waters, groundwater and drinking water do not pose<br />
an appreciable concern to human health. This finding of no<br />
adverse effect to human health from exposure to trace<br />
quantities of compounds (e.g. pharmaceuticals) in surface<br />
waters and/or drinking water is supported by other results<br />
reported in the literature. Kingsbury et al. (2008) recently<br />
evaluated the potential health effects of 148 organic<br />
compounds in source water and finished water. The authors<br />
showed that the annual mean concentration of all compounds<br />
detected in finished water was less than the established<br />
human health benchmarks. Furthermore, Snyder et al. (2008)<br />
arrived at the same findings after evaluating human health<br />
effects associated with potential drinking water exposure of<br />
a suite of 62 indicator pharmaceuticals and potential endocrine<br />
disrupting compounds.<br />
Despite the absence of any concern to human health,<br />
drinking water remains a major point of consumer concern<br />
and some residual uncertainties need further exploration. For<br />
example, drinking water guideline values are developed using<br />
toxicity information for single compounds. Hence, the longterm<br />
cumulative dose-additive or synergistic effects of low<br />
concentrations of contaminants co-occurring as mixtures on<br />
human health and potentially sensitive sub-populations<br />
remain currently unknown. Understanding and implementing<br />
of such information is important for the development of<br />
future (enforceable) guideline values. Finally, the relatively<br />
large data gap on occurrence of compounds in drinking water<br />
should compel further research and assessment, especially<br />
for those compounds with a low (provisional) guideline value.<br />
5. Major conclusions<br />
For most compounds evaluated in the present assessment,<br />
a substantial margin exists between the (provisional)<br />
guideline value and the maximum concentrations in<br />
surface waters, groundwaters and/or drinking water.<br />
The TTC based drinking water target values (0.1 mg/L and<br />
0.01 mg/L for non-carcinogenic compounds, respectively) as<br />
proposed earlier are the conservative values they are meant<br />
water research 44 (2010) 461–476 473<br />
to be. They are optimally suited to provide exposure based<br />
waiving.<br />
The concentrations in drinking water of compounds such as<br />
MTBE, ETBE, 1,4-dioxane, NDMA and benzene should be<br />
monitored closely, since their guideline values are easily<br />
exceeded.<br />
Alkylated perfluorinated compounds such as PFOA and<br />
PFOS are environmentally persistent compounds and their<br />
increasing occurrence in (the sources of) drinking water<br />
should be monitored closely.<br />
For compounds with a very low (provisional) guideline value<br />
(e.g. mutagenic and carcinogenic compounds) it is important<br />
to better establish trends in their environmental occurrence.<br />
From a toxicological point of view iodinated contrast media<br />
as present in drinking water, such as amidotrizoic acid<br />
iopamidol, iohexol and iopromide, are not a direct concern<br />
for human health. However, further environmental assessment<br />
may be necessary, especially since the sublethal<br />
(ecological) effects of these compounds are largely unknown.<br />
Better understanding of the potential mixture effects of<br />
emerging compounds present in drinking water is important<br />
for the development of future guideline values.<br />
Acknowledgements<br />
The authors wish to acknowledge the significant contribution<br />
to this work by Leo Puijker (KWR) and Ruud Jansen. Thanks<br />
also to Dr. Corine Houtman (HWL, the Netherlands) for her<br />
valuable comments on the manuscript. The research<br />
described was funded by the Joint Research Programme of the<br />
Dutch Water utilities (BTO).<br />
references<br />
Benzotriazoles Coalition, 2001. Benzotriazoles category<br />
justification and testing rationale. Synthetic Organic Chemical<br />
Manufacturers Organization. http://www.epa.gov/oppt/<br />
chemrtk/pubs/summaries/benzo/c13456.pdf (accessed June<br />
2009).<br />
BIBRA Toxicology International, 1995. Toxicity Profile of<br />
Benzotriazole. BIBRA Toxicology International, Carshalton<br />
(Surrey), UK.<br />
Brauch, H.J., Sacher, F., Denecke, E., Tacke, T., 2000. Efficiency of<br />
river-bank filtration for the removal of polar organic<br />
compounds (Werksamkeit der Uferfiltration fuer die<br />
Enfernung von polaren organischen Spurenstoffen). Wasser-<br />
Abwasser 141 (4), 226–234.<br />
Brown, D., Butterworth, K.R., Gaunt, I.F., Grasso, P., Gangolli, S.D.,<br />
1978. Short-term oral toxicity study of diethyl phthalate in the<br />
rat. Food Cosmet. Toxicol. 16, 415–422.<br />
California EPA (California Environmental Protection Agency),<br />
2000. N,N-diethyl-meta-toluamide (DEET) risk<br />
characterization document. Department of pesticide<br />
regulation. Document number RCD 00-01, Sacramento,<br />
California, USA.<br />
CIIT (Chemical Industry Institute of Toxicology), 1990. Twenty<br />
four month final report inhalation toxicity of dimethylamine<br />
in F344 rats and B6C3F1 mice. CIIT Archives, Docket #11957.<br />
Ciba-Geigy, 1988. Combined chronic toxicity/oncogenicity study<br />
in rats. Basle, Ciba-Geigy, 1988 (unpublished study submitted<br />
to WHO).
474<br />
Cotruvo, J.A., 1988. Drinking water standards and risk<br />
assessment. Regul. Toxicol. Pharmacol. 8 (3), 288–299.<br />
Danish EPA (Danish Environmental Protection Agency), 2004.<br />
Evaluation of Health Hazards by Exposure to BAM (2,6-<br />
Dichlorobenzamide) and Risk Characterisation of Drinking<br />
Water Exposure. Environmental Project Nr. 943 2004.<br />
Dawson, B.V., Johnson, P.D., Goldberg, S.J., Ulreich, J.B., 1993.<br />
Cardiac teratogenesis of halogenated<br />
hydrocarbon-contaminated drinking water. J. Am. Coll.<br />
Cardiol. 21 (6), 1466–1472.<br />
De Voogt, P., Vink, C., Puijker, L.M., 2008. MTBE and ETBE in Dutch<br />
groundwater (MTBE en ETBE in Nederlands grondwater), KWR<br />
08.043, KWR, Nieuwegein, The Netherlands.<br />
DuPont de Nemours and Company, Inc., 1964. Report MRID No.<br />
00017763, 00080899.<br />
DECOS (Dutch Expert Committee for Occupational Standards),<br />
2000. 1,2,3-benzotriazole. Health-based recommended<br />
occupational exposure limit. No. 2000/14OSH, The Hague, The<br />
Netherlands.<br />
ECB (European Chemicals Bureau), 2000a. IUCLID dataset on<br />
benzotriazoles.<br />
ECB (European Chemicals Bureau), 2000b. IUCLID dataset on<br />
Chloridazon.<br />
ECB (European Chemicals Bureau), 2000c. IUCLID dataset on<br />
Diethylene triamine penta acetic acid.<br />
ECB (European Chemicals Bureau), 2000d. IUCLID dataset on<br />
triphenylphosphine oxide.<br />
ECB (European Chemicals Bureau), 2002. European Risk<br />
Assessment Report 1,4-dioxane.<br />
EFSA (European Food Safety Authority), 2008. Opinion of the<br />
scientific panel on contaminants in the food chain on<br />
perfluorooctane sulfonate (PFOS), perfluorooctanoic acid<br />
(PFOA) and their salts. EFSA J. 53, 1–131.<br />
Ericson, I., Nadal, M., van Bavel, B., Lindström, G., Domingo, J.L.,<br />
2007. Levels of perfluorochemicals in water samples from<br />
Catalonia, Spain: is drinking water a significant contribution<br />
to human exposure? Environ. Sci. Pollut. Res. 15, 614–619.<br />
Goldenthal, E.I. (International Research and Development Corp.),<br />
1994. Evaluation of DEET in a one-year chronic oral toxicity<br />
study in dogs. DEET Joint Venture/Chemical Specialties<br />
Manufacturers Assoc. DPR vol. 50191-164, Rec. No. 131265.<br />
Health Canada, 2003. Benchmark Dose for TCE in Drinking Water.<br />
Health Canada, Healthy Environments and Consumer Safety<br />
Branch, Biostatistics Unit, April, Ottawa, Ontario.<br />
Hogenboom, A.C., Van Leerdam, J.A., De Voogt, P., 2009. Accurate<br />
mass screening and identification of emerging contaminants<br />
in environmental samples by liquid chromatography–hybrid<br />
linear ion trap Orbitrap mass spectrometry. J. Chromatogr. A<br />
1216 (3), 510–519.<br />
Hurni, H., Ohder, H., 1973. Reproduction study with formaldehyde<br />
and hexamethylenetetramine in beagle dogs. Food Cosmet.<br />
Toxicol. 11 (3), 459–462.<br />
Kingsbury, J.A., Delzer, C.D., Hopple, J.A., 2008. Anthropogenic<br />
organic compounds in source water of nine community water<br />
systems that withdraw from streams, 2002–2005, USGS,<br />
Reston, Virginia, USA, Report 2008–5208.<br />
Kleinneijenhuis, A.J., Puijker, L.M., 2008. Monitoring program for<br />
DMSA, DMST, N,N-dimethylsulfamide and its ozone reaction<br />
product NDMA in untreated and ozone-treated water. TNO<br />
report 7686, Delft, The Netherlands.<br />
Kolpin, D.W., Skopec, M., Meyer, M.T., Furlong, E.T., Zaugg, S.D.,<br />
2004. Urban contribution of pharmaceuticals and other<br />
organic wastewater contaminants to streams during differing<br />
flow conditions. Sci. Total Environ. 328 (1–3), 119–130.<br />
Kroes, R., Renwick, A.G., Cheeseman, M., Kleiner, J.,<br />
Mangelsdorf, I., Piersma, A., Schilter, B., Schlatter, J.,<br />
Van Schothorst, F., Vos, J.G., Wurtzen, G., 2004. Structurebased<br />
thresholds of toxicological concern (TTC): guidance for<br />
water research 44 (2010) 461–476<br />
application to substances present at low levels in the diet.<br />
Food Chem. Toxicol. 42, 65–83.<br />
Krop, H. de Voogt, P., 2008. Physicochemical parameters and<br />
source markers of PFAS. Report IVAM ISO doc O/0818,<br />
Amsterdam, The Netherlands, pp. 1–147.<br />
KWR Watercycle Research Institute, 2009. KWR internal data,<br />
Nieuwegein, the Netherlands.<br />
Larsen, M.L., Illingworth, D.R., O’Malley, J.P., 1994. Comparative<br />
effects of gemfibrozil and clofibrate in type III<br />
hyperlipoproteinemia. Atherosclerosis 106, 235–240.<br />
Loos, R., Wollgast, J., Huber, T., Hanke, G., 2007. Polar herbicides,<br />
pharmaceutical products, PFOS, PFOA, and nonylpheniol and<br />
its carboxylates and ethoxylates in surface and tap waters<br />
around Lake Maggiore in Northern Italy. Anal. Bioanal. Chem.<br />
387, 1469–1478.<br />
Loos, R., Gawlik, B.M., Locoro, G., Rimaviciute, E., Contini, S.,<br />
Bidoglio, G., 2009. EU-wide survey of polar organic persistent<br />
pollutants in European river waters. Environ. Pollut. 157 (2),<br />
561–568.<br />
Lynch, D.W., Moorman, W.J., Stober, P., Lewis, T.R., Iverson, W.O.,<br />
1986. Subchronic inhalation of diethylamine vapour in Fisher-<br />
344 rats: organ toxicity. Fund. Appl. Toxicol. 6 (3), 559–565.<br />
Medinsky, M.A., Wolf, D.C., Cattley, R.C., Wong, B., Janszen, D.B.,<br />
Farris, G.M., Wright, G.A., Bond, J.A., 1999. Effects of<br />
a thirteen-week inhalation exposure to ethyl tertiary<br />
butyl ether on Fischer-344 rats and CD-1 mice. Toxicol. Sci.<br />
51, 108–118.<br />
Mitsubishi Chemical Safety Institute Ltd. (date of study unknown).<br />
4,4’-sulfonyldiphenol. Summary of repeated dose and<br />
developmental studies. Kashima Laboratory, Ibaraki, Japan.<br />
Mitsumori, K., Usui, T., Takahashi, K., Shirasu, Y., 1979. Twentyfour<br />
month chronic toxicity studies of dichlorodiisopropyl<br />
ether in mice. J. Pestic. Sci. 4 (3), 323–335.<br />
Mons, M., Van der Hoek, J.P., Stoks, P., Van der Kooij, D., 2008.<br />
Vision on and guideline values for xenobiotic substances in<br />
drinking water (Visie op en streefwaarden voor<br />
milieuvreemde stoffen in drinkwater). H 2O 4, 1–4 (in Dutch).<br />
Morgareidge, K., 1971. 90-Day feeding study with benzothiazole in<br />
rats. Unpublished report from Food and Drug Research<br />
Laboratories, Inc., Maspeth, New York, USA. Submitted to<br />
WHO by Flavor and Extract Manufacturers’ Association of the<br />
United States.<br />
Morisetti, A., Tirone, P., Luzzani, F., de Haen, C., 1994.<br />
Toxicological safety assessment of iomeprol, a new X-ray<br />
contrast agent. Eur. J. Radiol. 1 (18 Suppl), S21–31.<br />
Mout, L., Jansen, L., van der Pijll, S., Nierop, S., 2007. Implementation<br />
program dimethenamid-p. Clean sources, now and in the future<br />
(uitvoeringsprogramma dimethenamid-p. Schone bronnen, nu<br />
en in de toekomst: tweede reeks knelpunten (http://www.<br />
schonebronnen.nl/). The Hague, The Netherlands.<br />
Munro, I.C., Ford, R.A., Kennepohl, E., Sprenger, J.G., 1996.<br />
Correlation of structural class with no-observed effect levels:<br />
a proposal for establishing a threshold of concern. Food Chem.<br />
Toxicol 34, 829–867.<br />
NCI (National Cancer Institute), 1978. Bioassay of 1,4-dioxane for<br />
possible carcinogenicity, CAS No. 123-91-1. NCI<br />
Carcinogenesis Technical Report Series No. 80. DHEW<br />
Publication No. (NIH) PB-285-711, Bethesda, Maryland, USA.<br />
Nghiem, L.D., Schafer, A.I., Elimelech, M., 2004. Removal of<br />
natural hormones by nanofiltration membranes:<br />
measurement, modeling, and mechanisms. Environ. Sci.<br />
Technol. 38 (6), 1888–1896.<br />
Norimitsu, S., Kouji, H., Kayoko, I., Kazuaki, S., Takeo, Y., Akio, K.,<br />
2004. Perfluorooctanoate and perfluorooctane sulfonate<br />
concentrations in surface water in Japan. J. Occup. Health 46,<br />
49–59.<br />
NTP (National Toxicology Program), 1986. Toxicology and<br />
carcinogenesis studies of benzene in F344/N rats and B6C3F1
mice (gavage studies). Research Triangle Park, NC, USA, US<br />
Department of Health and Human Services. (Technical report<br />
series No. 289).<br />
NTP (National Toxicology Program), 1987. Teratologic evaluation<br />
of diethylene glycol dimethyl ether (CAS No. 111-96-6)<br />
administered to New Zealand White rabbits on gestation days<br />
6 through 19. Research Triangle Park, NC, USA, National<br />
Institute of Environmental Health Sciences, National<br />
Toxicology Program (NTP-87-108; PB 87–209532).<br />
NTP (National Toxicology Progam), 1991. Toxicology and<br />
carcinogenesis studies of tris(2-chloroethyl)phosphate in<br />
F344/N rats and B6C3F1 mice. Technical report series No. 391.<br />
NTP TR 391, Research Triangle Park, NC, USA.<br />
OECD/SIDS (Organization for Economic Development and<br />
Cooperation/Screening Information DataSet), 1994. p-<br />
Toluenesulfonamide. UNEP publications, Nairobi, Kenya.<br />
OECD/SIDS (Organization for Economic Development and<br />
Cooperation/Screening Information DataSet), 1998.<br />
Triethylphosphate. UNEP publications, Nairobi, Kenya.<br />
Peto, R., Gray, R., Brantom, P., Grasso, P., 1991a. Effects on 4080<br />
rats of chronic ingestion of N-nitrosodiethylamine or Nnitrosodimethylamine:<br />
a detailed dose–response study.<br />
Cancer Res. 51, 6415–6451.<br />
Peto, R., Gray, R., Brantom, P., Grasso, P., 1991b. Dose and time<br />
relationships for tumor induction in the liver and esophagus<br />
of 4080 inbred rats by chronic ingestion of Nnitrosodiethylamine<br />
or N-nitrosodimethylamine. Cancer Res.<br />
51, 6452–6469.<br />
Proviron Fine Chemicals, 2003. IUCLID dataset on Nbutylbenzenesulphonamide.<br />
REWAB Database, 2009. KWR Watercycle Research Institute,<br />
Nieuwegein, The Netherlands.<br />
Robinson, M., Bruner, R.H., Olson, G.R., 1990. Fourteen- and<br />
ninety-day oral toxicity studies of methyl-t-butyl ether in<br />
Sprague-Dawley rats. J. Am. Coll. Toxicol. 9, 525–540.<br />
Schwarzenbach, R.P., Escher, B.I., Fenner, K., Hofstetter, T.B.,<br />
Johnson, C.A., Von Gunten, U., Wehrli, B., 2006. The challenge<br />
of micropollutants in aquatic systems. Science 313 (5790),<br />
1072–1077.<br />
SCOEL (Scientific Committee on Exposure Limits), 1991. No 11b.<br />
Recommendations of the SCOEL for DMA.<br />
SCOEL (Scientific Committee on Exposure Limits), 2002. No 91.<br />
Recommendations of the SCOEL for DEA.<br />
Schwab, B.W., Hayes, E.P., Fiori, J.M., Mastrocco, F.J., Roden, N.M.,<br />
Cragin, D., Meyerhoff, R.D., D’Aco, V.J., Anderson, P.D., 2005.<br />
Human pharmaceuticals in US surface waters: a human<br />
health risk assessment. Regul. Toxicol. Pharmacol. 42 (3),<br />
296–312.<br />
Seacat, A.M., Thomford, P.J., Hansen, K.J., Olsen, G.W., Case, M.T.,<br />
Butenhoff, J.L., 2002. Subchronic toxicity studies on<br />
perfluorooctanesulfonate potassium salt in cynomolgus<br />
monkeys. Toxicol. Sci. 68 (1), 249–264.<br />
Sherman, H., 1972. Long-term feeding studies in rats and dogs<br />
with 2-benzimadazole carbamic acid, methyl ester (INE-965)<br />
(50% and 70% MBC wettable powder formulations).<br />
Unpublished report from E.I. DuPont de Nemours and Co., Inc.,<br />
Haskell Laboratory, Newark, Delaware, USA.<br />
Singh, G., Driever, P.H., Sander, J.W., 2005. Cancer risk in people<br />
with epilepsy: the role of antiepileptic drugs. Brain 128 (Pt 1),<br />
7–17.<br />
Skutlarek, D., Exner, M., Farber, H., 2006. Perfluorinated<br />
surfactants in surface and drinking waters. Environ. Sci.<br />
Pollut. Res. 13 (5), 299–307.<br />
Snyder, S.A., Villeneuve, D.L., Snyder, E.M., Giesy, J.P., 2001.<br />
Identification and quantification of estrogen receptor agonists<br />
in wastewater effluents. Environ. Sci. Technol. 35 (18),<br />
3620–3625.<br />
Snyder, S.A., Trenholm, R.A., Snyder, E.M., Bruce, G.M.P.,<br />
Hemming, J.D.C., 2008. Toxicological Relevance of EDCs and<br />
water research 44 (2010) 461–476 475<br />
Pharmaceuticals in Drinking Water. Awwa Research<br />
Foundation, Denver, USA, p. 484.<br />
Stokinger, H.E., Woodward, R.L., 1958. Toxicologic methods for<br />
establishing drinking water standards. J. Am. Water Works<br />
Ass. 50, 515–529.<br />
Swarm, R.L., Roberts, G.K., Levy, A.C., Hines, L.R., 1973.<br />
Observations on the thyroid gland in rats following the<br />
administration of sulfamethoxazole and trimethoprim.<br />
Toxicol. Appl. Pharm. 24 (3), 351–363.<br />
Swartjes, F.A., Baars, A.J., Fleuren, R.H.L.J., Otte, P.F., 2004. Risk<br />
limits for MTBE (methyl tertiair-butyl ether) in soil, sediment,<br />
groundwater, surface water, drinking water and for drinking<br />
water production (risicogrenzen voor MTBE (Methyl tertiair-<br />
Butyl Ether) in bodem, sediment, grondwater,<br />
oppervlaktewater, drinkwater en voor drinkwaterbereiding)<br />
RIVM report 711701039/2004, Bilthoven, The Netherlands.<br />
Toccalino, P.L., 2007. Development and application of healthbased<br />
screening levels for use in water-quality assessments.<br />
US Geological Survey Scientific Investigations Report 2007-<br />
5106, 12 p, Reston, Virginia, USA.<br />
URL1: http://www.epa.gov/waterscience/criteria/drinking/<br />
dwstandards.pdf, accessed June 2009.<br />
URL2: http://www.who.int/water_sanitation_health/dwq/<br />
guidelines/en/, accessed June 2009.<br />
URL3: http://www.ctb.agro.nl/ctb_files/11995_01.html, accessed<br />
May 2009 (In Dutch).<br />
URL4: http://www.iawr.org/index.php, publications, accessed<br />
January 2009.<br />
URL5: http://www.riwa-rijn.org/riwa_en.html, publications,<br />
accessed January 2009.<br />
US EPA (United States Environmental Protection Agency) IRIS<br />
database (1988a). 1,4-dioxane.<br />
US EPA (United States Environmental Protection Agency) IRIS<br />
database, 1988b. Diethyl phthalate.<br />
US EPA (United States Environmental Protection Agency) IRIS<br />
database, 1988c. Diuron.<br />
US EPA (United States Environmental Protection Agency) IRIS<br />
database, 1989. Bis(chloroisopropyl)ether.<br />
US EPA (United States Environmental Protection Agency), 2006.<br />
National Primary Drinking Water Standards. http://www.epa.<br />
gov/safewater/consumer/pdf/mcl.pdf (accessed August 2009).<br />
Van Leerdam, J.A., Hogenboom, A.C., Kooi van der, M.E., de<br />
Voogt, P., 2009. Determination of polar 1-H-benzotriazoles and<br />
benzothialzoles in water by solid-phase extraction and liquid<br />
chromatography LTQ Orbitrap mass spectrometry. Int. J. Mass<br />
Spectrom. 282 (3), 99–107.<br />
Van Leeuwen, F.X., 2000. Safe drinking water: the toxicologist’s<br />
approach. Food Chem. Toxicol. 38 (1 Suppl), S51–S58.<br />
Van Leeuwen, C.J., Vermeire, T., 2007. Risk Assessment of<br />
Chemicals, second ed., Springer, ISBN 978-4020-6101-1.<br />
Van Wezel, A.P., Puijker, L., Vink, C., Versteegh, A., de Voogt, P.,<br />
2009. Odour and flavour thresholds of gasoline additives<br />
(MTBE, ETBE and TAME) and their occurrence in Dutch<br />
drinking water collection areas. Chemosphere 76, 672–676.<br />
Versteegh, J., Van Der Aa, N., Dijkman, E., 2007. Pharmaceuticals<br />
in drinking water and sources of drinking water - results of<br />
monitoring campaign 2005/2006 (geneesmiddelen in<br />
drinkwater en drinkwaterbronnen - resultaten van het<br />
meetprogramma 2005/2006), 53, RIVM, Bilthoven, The<br />
Netherlands. Number: R(T)-1-67.<br />
Walraven, N., Laane, R.W., 2009. Assessing the discharge of<br />
pharmaceuticals along the Dutch coast of the North sea. Rev.<br />
Environ. Contam. Toxicol. 199, 1–18.<br />
Westerhoff, P., Yoon, Y., Snyder, S., Wert, E., 2005. Fate of<br />
endocrine-disruptor, pharmaceutical, and personal care<br />
product chemicals during simulated drinking water treatment<br />
processes. Environ. Sci. Technol. 37 (17), 6649–6663.<br />
WHO (World Health Organization), 1996. Simazine in drinkingwater.<br />
Background document for development of WHO
476<br />
Guidelines for drinking-water quality. WHO/SDE/WSH/03.04/<br />
42, Geneva, Switzerland.<br />
WHO (World Health Organization), 1998. Guidelines for<br />
drinking water quality. Second edition, volume 1<br />
recommendations: addendum, ISBN 92 4 154514 3, Geneva,<br />
Switzerland.<br />
WHO (World Health Organization), 2002. Diethylene glycol<br />
dimethyl ether. Concise International Chemical Assessment<br />
Document 41, Geneva, Switzerland.<br />
WHO (World Health Organization), 2003a. Benzene in drinking<br />
water. WHO report WHO/SDE/WSH/03.04/24, Geneva,<br />
Switzerland.<br />
WHO (World Health Organization), 2003b. 2,4-D in drinking water.<br />
WHO report WHO/SDE/WSH/03.04/70, Geneva, Switzerland.<br />
WHO (World Health Organization), 2003c. Isoproturon in<br />
Drinking-water. Background document for development of<br />
WHO Guidelines for Drinking-water Quality WHO/SDE/WSH/<br />
03.04/37, Geneva, Switzerland.<br />
WHO (World Health Organization), 2005a. Glyphosate and AMPA<br />
in drinking-water. Background document for development of<br />
WHO Guidelines for Drinking-water Quality WHO/SDE/WSH/<br />
03.04/97, Geneva, Switzerland.<br />
WHO (World Health Organization), 2005b. Trichloroethene in<br />
Drinking-water, Background document for development of<br />
WHO Guidelines for Drinking-water Quality. WHO/SDE/WSH/<br />
05.08/22, Geneva, Switzerland.<br />
WHO (World Health Organization), 2006. Guidelines for drinking<br />
water quality, First Adendum to third ed., Recommendations,<br />
vol. 1, 2006, Geneva, Switzerland.<br />
WHO (World Health Organization), 2008. N-<br />
Nitrosodimethylamine in drinking-water. Background<br />
water research 44 (2010) 461–476<br />
document for preparation of WHO Guidelines for drinkingwater<br />
quality. Geneva, World Health Organization (WHO/HSE/<br />
AMR/08.03/8), Geneva, Switzerland.<br />
WHO/JECFA (World Health Organization/joint FAO/WHO expert<br />
committee on food additives), 1973. Food Additives Series 32;<br />
Toxicological Evaluation of Certain Food additives and<br />
Contaminants. Prepared by the forty-first meeting of the joint<br />
FAO/WHO expert committee.<br />
WHO/JECFA (World Health Organization/joint FAO/WHO expert<br />
committee on food additives), 1974. Food Additives Series No.<br />
5; Seventeenth Report of the Joint FAO/WHO Expert<br />
Committee on Food Additives, World Health Organization<br />
technical report series, 1974, No. 539; FAO Nutrition Meetings<br />
Report Series, 1974, No. 53.<br />
WHO/JECFA (World Health Organization/joint FAO/WHO expert<br />
committee on food additives), 1995. 892. Carbendazim<br />
(Pesticide residues in food: 1995 evaluations Part II<br />
Toxicological & environmental).<br />
WHO/JECFA (World Health Organization/joint FAO/WHO expert<br />
committee on food additives), 2003. WHO Food Additives<br />
Series No. 50; sulfur-containing heterocyclic compounds.<br />
WHO/JMPR (World Health Organization/Joint Meeting on<br />
Pesticide Residues), 2001. Toxicological evaluations<br />
imidacloprid. Pesticide residues in food.<br />
WHO/JMPR (World Health Organization/Joint Meeting on<br />
Pesticide Residues), 2005. Dimethenamid-p. pp. 189–239.<br />
Wilson, A.B., Thorpe, E., 1971. Toxicity studies on the ‘‘Prefix’’<br />
residue 2,6-dichlorobenzamide: two year oral experiment with<br />
dogs. Tunstall Laboratory. Report number: Lab Project<br />
Number: T507531; Group research report: TLGR.0028.71, pp.<br />
1–28.
Monitoring the biological activity of micropollutants during<br />
advanced wastewater treatment with ozonation and activated<br />
carbon filtration<br />
M. Macova a , B.I. Escher a, *, J. Reungoat b , S. Carswell a,c , K. Lee Chue a ,<br />
J. Keller b , J.F. Mueller a<br />
a<br />
University of Queensland, National Research Centre for Environmental Toxicology (EnTox), 39 Kessels Road, Brisbane, QLD 4108, Australia<br />
b<br />
University of Queensland, Advanced Water Management Centre (AWMC), Brisbane, QLD 4072, Australia<br />
c<br />
Queensland Health Forensic and Scientific Services, Organics Laboratory, 39 Kessels Road, Brisbane, QLD 4108, Australia<br />
article info<br />
Article history:<br />
Received 13 May 2009<br />
Received in revised form<br />
28 August 2009<br />
Accepted 7 September 2009<br />
Available online 16 September 2009<br />
Keywords:<br />
Bioassays<br />
Baseline toxicity<br />
Phytotoxicity<br />
Estrogenicity<br />
Genotoxicity<br />
Toxic equivalency concept<br />
water research 44 (2010) 477–492<br />
Available at www.sciencedirect.com<br />
journal homepage: www.elsevier.com/locate/watres<br />
abstract<br />
* Corresponding author. Tel.: þ61 (0) 7 3274 9180; fax: þ61 (0) 7 3274 9003.<br />
E-mail address: b.escher@uq.edu.au (B.I. Escher).<br />
0043-1354/$ – see front matter ª 2009 Elsevier Ltd. All rights reserved.<br />
doi:10.1016/j.watres.2009.09.025<br />
A bioanalytical test battery was used to monitor the removal efficiency of organic micropollutants<br />
during advanced wastewater treatment in the South Caboolture Water Reclamation<br />
Plant, Queensland, Australia. This plant treats effluent from a conventional sewage treatment<br />
plant for industrial water reuse. The aqueous samples were enriched using solid-phase<br />
extraction to separate some organic micropollutants of interest from metals, nutrients and<br />
matrix components. The bioassays were chosen to provide information on groups of chemicals<br />
with a common mode of toxic action. Therefore they can be considered as sum indicators<br />
to detect certain relevant groups of chemicals, not as the most ecologically or human health<br />
relevant endpoints. The baseline toxicity was quantified with the bioluminescence inhibition<br />
test using the marine bacterium Vibrio fischeri. The specific modes of toxic action that were<br />
targeted with five additional bioassays included aspects of estrogenicity, dioxin-like activity,<br />
genotoxicity, neurotoxicity, and phytotoxicity. While the accompanying publication discusses<br />
the treatment steps in more detail by drawing from the results of chemical analysis as well as<br />
the bioanalytical results, here we focus on the applicability and limitations of using bioassays<br />
for the purpose of determining the treatment efficacy of advanced water treatment and for<br />
water quality assessment in general. Results are reported in toxic equivalent concentrations<br />
(TEQ), that is, the concentration of a reference compound required to elicit the same response<br />
as the unknown and unidentified mixture of micropollutants actually present. TEQ proved to<br />
be useful and easily communicable despite some limitations and uncertainties in their derivation<br />
based on the mixture toxicity theory. The results obtained were reproducible, robust<br />
and sensitive. The TEQ in the influent ranged in the same order of magnitude as typically seen<br />
in effluents of conventional sewage treatment plants. In the initial steps of the treatment<br />
chain, no significant degradation of micropollutants was observed, and the high levels of<br />
dissolved organic carbon probably affected the outcome of the bioassays. The steps of coagulation/flocculation/dissolved<br />
air flotation/sand filtration and ozonation decreased the effectbased<br />
micropollutant burden significantly.<br />
ª 2009 Elsevier Ltd. All rights reserved.
478<br />
1. Introduction<br />
Water quantity and water quality issues have led many<br />
countries to explore water reuse and water reclamation as<br />
alternatives to provide water for direct use in industrial<br />
applications as well as agricultural and gardening use, and<br />
for indirect use to supplement water reservoirs. Australia is<br />
amongst the countries most affected by uneven distribution<br />
of water resources and droughts. Therefore, various water<br />
treatment and desalination plants have been put into<br />
operation, among them the Water Reclamation Plant in<br />
South Caboolture, Queensland, Australia (van Leeuwen<br />
et al., 2003).<br />
Besides hygiene issues, organic micropollutants are of<br />
major concern because classical wastewater treatment with<br />
biological activated sludge only partially removes hazardous<br />
chemicals. Furthermore there is the potential that persistent<br />
and toxic micropollutants and transformation products could<br />
build up during the recycling process.<br />
Urban and agricultural wastewaters contain a high<br />
number of organic micropollutants, including pharmaceuticals<br />
and personal care products, hormones and industrial<br />
chemicals, biocides and pesticides (Schwarzenbach et al.,<br />
2006). A large number of studies have investigated the fate<br />
and removal of various individual and specific groups of<br />
micropollutants by treatment processes like biological activated<br />
sludge treatment (Ternes et al., 2004; Joss et al., 2005),<br />
ozonation and advanced oxidation processes (von Gunten,<br />
2003; Huber et al., 2004; Huber et al., 2005; Lee et al., 2008) or<br />
activated carbon treatment (Sanchez-Polo et al., 2006; Snyder<br />
et al., 2007) but little is known how these treatment processes<br />
change the composition and biological activity of the<br />
micropollutants.<br />
Toxicity testing may provide complementary information<br />
to chemical analysis on the sum of micropollutants present in<br />
treated water. However, whole-effluent toxicity testing, which<br />
is a valuable tool for hazard assessment of industrial and<br />
municipal effluents (Chapman, 2000), is inappropriate to<br />
evaluate the fate of micropollutants during advanced water<br />
treatment because acute toxicity tests are not sensitive<br />
enough for purified water and long-term chronic tests are too<br />
expensive and time-consuming for routine monitoring. In<br />
addition, the physical treatment of wastewater alters the<br />
matrix and electrolyte composition and it becomes virtually<br />
impossible to differentiate the effect caused by micropollutants<br />
from that induced by the matrix, such as organic<br />
matter content, pH, ionic strength or nutrient content.<br />
Therefore, over the last years, we have developed<br />
a framework for bioanalytical quantification of organic<br />
micropollutants, which has three main cornerstones: Firstly,<br />
the aqueous samples are enriched using solid-phase extraction<br />
(SPE) to separate the organic micropollutants of interest<br />
from metals and matrix components. This SPE method was<br />
previously validated for recovery and extraction yield for<br />
selected chemicals including pesticides and pharmaceuticals<br />
(Bengtson Nash et al., 2005a; Escher et al., 2005; Escher et al.,<br />
2008a). Secondly, a battery of bioanalytical methods were<br />
selected that cover a non-specific bioassay, the bioluminescence<br />
inhibition of the marine bacterium Vibrio fischeri and<br />
water research 44 (2010) 477–492<br />
five bioassays for specific modes of toxic action that are<br />
indicative for groups of chemicals of particular relevance for<br />
human and environmental health, including aspects of<br />
estrogenicity, dioxin-like activity, genotoxicity, neurotoxicity,<br />
and phytotoxicity (Muller et al., 2007; Muller et al., 2008;<br />
Escher et al., 2008a; Escher et al., in press). Table 1 summarizes<br />
the key features of the bioassays and more details are<br />
given in the Supporting Information.<br />
Finally,we developed a coherent data evaluation method that<br />
is based on the toxic equivalency concept (Villeneuve et al., 2000;<br />
Escher et al., 2008b). Thus, the effects at various steps of the<br />
treatment chain can be easily compared and treatment efficiency<br />
can be compared between the different effect endpoints.<br />
Bioassays in combination with SPE have been successfully<br />
applied by several groups to evaluate various water treatment<br />
options (Escher et al., 2008a; Cao et al., 2009). In particular, the<br />
bioluminescence inhibition test with Vibrio fischeri and the<br />
Salmonella based test for genotoxicity have been widely used<br />
despite their recognized limitations (Johnson, 2005; Petala<br />
et al., 2006; Petala et al., 2008). The present work is more<br />
extensive covering several additional toxicity endpoints,<br />
allowing high-throughput monitoring applications while<br />
remaining cost-efficient.<br />
Here we present this framework for the first time to evaluate<br />
all steps of enhanced water treatment in the South<br />
Caboolture Water Reclamation Plant (van Leeuwen et al.,<br />
2003). This paper focuses on the advantages and limitations in<br />
the application of the bioanalytical test battery, and the<br />
interpretation and significance of the results. In the accompanying<br />
paper, the results obtained from the bioanalytical<br />
tools are compared with chemical analytical data for a variety<br />
of pharmaceuticals and pesticides and conclusions for<br />
advanced wastewater treatment are drawn (Reungoat et al.,<br />
2010).<br />
2. Materials and methods<br />
2.1. Samples and sites<br />
Four consecutive sets of 24-hour composite samples were<br />
collected on 11-07-08, 22-07-08, 27-07-08 and 06-08-08 using<br />
refrigerated autosamplers at 10 sites of the South Caboolture<br />
Water Reclamation Plant (Table 2).<br />
2.2. Solid phase extraction<br />
Immediately after sampling, concentrated HCl (36%) was<br />
added to a final concentration of 5 mM for preservation. It was<br />
demonstrated in earlier work that a pharmaceutical cocktail<br />
in a wastewater matrix had highest recoveries for HLB at pH 3<br />
(Escher et al., 2005).<br />
Samples were extracted using 1 g OASIS Ò HLB solid phase<br />
material in 20 mL cartridges (Waters, Australia) following<br />
centrifugation (only S1 sample) and filtration with a glass fibre<br />
filter. After conditioning the cartridges with 10 mL methanol<br />
and 20 mL of 5 mM HCl in MilliQ water, a known volume of<br />
sample was percolated under vacuum. The cartridges were<br />
dried overnight under vacuum and were eluted with 10 mL
Table 1 – Bioanalytical test battery (adapted from Muller (2008)).<br />
Assay Measured by Target mode of<br />
toxic action<br />
Bioluminescence<br />
inhibition test<br />
Acetylcholinesterase<br />
(AChE)<br />
Inhibition Assay<br />
Imaging-PAM<br />
Assay<br />
methanol and 10 mL hexane:acetone (1:1). All eluates were<br />
evaporated to dryness and reconstituted with 0.5 mL of ethanol.<br />
2.3. Bioanalytical tools<br />
The bioassays were performed as previously reported (Escher<br />
et al., 2005; Muller et al., 2007; Muller et al., 2008; Escher et al.,<br />
2008a; Escher et al., in press) using the methods cited in Table<br />
1 and described in detail in the Supporting Information.<br />
2.4. Relative enrichment factor of the samples REF<br />
For each sample, the enrichment factor of the solid phase<br />
extraction was calculated using (Eq. 1) which represents the<br />
enrichment of the extract compared to the source water.<br />
enrichment factorSPE ¼ Vwater<br />
Reduction in luminescence of<br />
the naturally bioluminescent<br />
marine bacteria Vibrio fischeri<br />
Product of the enzymatic<br />
reaction (colorimetric)<br />
Increased Chl a fluorescence,<br />
inversely proportional to PS II<br />
photosynthetic yield<br />
E-SCREEN Cellular proliferation in<br />
estrogenic dependent cell line<br />
AhR-CAFLUX Induction of green<br />
fluorescent protein under the<br />
control of Ah receptor;<br />
(fluorescence)<br />
umuC assay Induction of ß-galactosidase<br />
enzyme as an indicator of<br />
DNA damage;<br />
Product of the enzymatic<br />
reaction (colorimetric)<br />
Vextract<br />
Sample volume varied from 2.0 to 2.5 L. The final volume of<br />
each extract was 0.5 mL therefore the enrichment factor SPE<br />
was between 3500 and 5500 depending on the sample.<br />
An aliquot of the enriched sample extracts were then<br />
added to the microtiter plate of the respective bioassay and<br />
serially diluted by a test medium to obtain a concentrationeffect<br />
curve. A dilution factor of each bioassay was calculated<br />
using (Eq. 2). The dilution factors ranged from 0.05 to 0.007 for<br />
the highest dose depending on the bioassay.<br />
water research 44 (2010) 477–492 479<br />
Effect on energy<br />
status through<br />
cytotoxicity and/or<br />
baseline toxicity<br />
Inhibition of the<br />
enzyme, which<br />
hydrolyses the<br />
neurotransmitter<br />
acetylcholine<br />
PS II derived<br />
photosynthesis<br />
inhibition<br />
(1)<br />
Chemicals that<br />
give response<br />
Literature reference<br />
All chemicals International Standard<br />
Organisation, 1998;<br />
Johnson, 2005;<br />
Farré et al., 2006; Escher<br />
et al., 2008a<br />
Organophosphates<br />
and carbamate<br />
insecticides<br />
Triazine and phenylurea<br />
herbicides<br />
Estrogenicity Estrogens,<br />
estrogenic industrial<br />
chemicals<br />
Dioxin-like activity;<br />
Aryl hydrocarbon<br />
(Ah) receptor<br />
activation<br />
Genotoxicity;<br />
DNA damage<br />
(SOS-response)<br />
Polychlorinated<br />
dibenzodioxins/<br />
furans (PCDD/F)<br />
and biphenyls (PCB)<br />
or polycyclic<br />
aromatic<br />
hydrocarbons (PAH)<br />
Chlorinated<br />
byproducts,<br />
aromatic amines,<br />
polycyclic aromatic<br />
hydrocarbons (PAH)<br />
Ellman et al., 1961;<br />
Deutsche Norm, 1995;<br />
Hamers et al., 2000<br />
Schreiber et al., 2002;<br />
Schreiber et al., 2007<br />
Soto et al., 1995; Körner<br />
et al., 1999<br />
Nagy et al., 2002; Zhao<br />
and Denison, 2004<br />
Oda et al., 1985;<br />
Reifferscheid et al., 1991;<br />
International Standard<br />
Organisation, 2000<br />
volume of extract added to bioassay<br />
dilution factorbioassay ¼<br />
total volume of bioassay<br />
(2)<br />
The final relative enrichment factor REF is the combination<br />
of the enrichment of the extraction and the dilution in the<br />
bioassay (Eq. 3) (Escher et al., 2006; Muller et al., 2007) and<br />
represents the enrichment or dilution of the original sample<br />
in each bioassay (Table 3). The REF is equivalent to concentration<br />
and is expressed in the units [Lwater sample/Lbioassay].<br />
REF ¼ dilution factorbioassay enrichment factorSPE (3)<br />
2.5. Evaluation of the concentration-effect curves<br />
Due to the nature of the endpoints, the six bioanalytical tests<br />
of the battery could not be evaluated with a single data evaluation<br />
model (Table 3). In general, all concentration effect<br />
curves of the reference compounds and sample extracts followed<br />
a log-logistic function (Eq. 4), which was fitted using<br />
Prism 5.0 software (GraphPad, San Diego, CA, USA). Adjustable<br />
parameters are the minimum and maximum of the effect,<br />
min and max, the slope s, and the effect concentration<br />
inducing 50% of the maximum effect, EC50. If the concentration-effect<br />
curve dropped sharply at the range of higher doses<br />
indicating that the sample has a cytotoxic effect, the cytotoxic<br />
concentrations were excluded from the concentration-effect<br />
assessment.
480<br />
Table 2 – Sample details. Samples were each taken after the described process. More details are given in Reungoat et al.<br />
(2010). TOC/DOC data represent the average of two samplings (22-07-08 and 06-08-08).<br />
Sample ID Site description TOC (mg/L) DOC (mg/L)<br />
effect¼min<br />
þ<br />
max min<br />
1þ10 s ðlog EC 50 log ðconcentrationof reference compoundor REF of sampleÞÞ<br />
Three bioassays, the inhibition of bioluminescence of Vibrio<br />
fischeri, inhibition of AChE and inhibition of PSII photosynthetic<br />
yield, all produced effects between 0% and 100%, therefore the<br />
maximum effect in (Eq. 4) could be set to 100% (Table 3).<br />
Two bioassays, the AhR-CAFLUX and the E-SCREEN, yield<br />
information on the induction of a receptor or proliferation of<br />
cells and there is no absolute maximum effect. Dependent on<br />
the mode of action, the maximum effect can be higher or<br />
lower than the maximum effect of the reference compounds.<br />
The extent of difference of maximally attainable effect yields<br />
some additional information on the mode of action and the<br />
type of compounds that induce the effect. Details of each<br />
assay are discussed below. The curve fit is then performed in<br />
such a way that the maximum effect in equation 4 is an<br />
additional fitting parameter.<br />
Finally the umuC assay exhibited a linear concentrationeffect<br />
curve. At low effect levels the log-logistic concentration<br />
effect curve (Eq. 4) is congruent with a linear concentrationeffect<br />
curve. Unlike the other assays, where effects were<br />
compared to a reference compound, genotoxicity is expressed<br />
as the concentration that induces the threshold of induction<br />
defined by the EN ISO guideline (International Standard<br />
Organisation, 2000). This way of data evaluation needs further<br />
refinement in the future because it is not directly comparable<br />
to the results in the other bioassays and the effect concentration<br />
cannot be translated into TEQ.<br />
The detection limit of the bioassays was defined as three<br />
times the standard deviation of the response using the lowest<br />
concentration of the standard that induced an effect significantly<br />
different from the control.<br />
2.6. Estimating equivalent concentrations<br />
in the samples<br />
For an easy-to-follow reporting of the effect, in all but one<br />
bioassay, the effects were reported in terms of toxic equivalent<br />
(4)<br />
concentrations (Villeneuve et al., 2000; Escher et al., 2008b). The<br />
equivalent concentration represents the concentration of<br />
the reference compound required to produce the same effect as<br />
the mixture of different compounds in the sample.<br />
Equivalent concentrations were calculated from concentration-effect<br />
curves of the reference compound and the<br />
samples. Both reference compounds and samples generally<br />
followed a sigmoidal log–concentration-effect curve (Eq. 4),<br />
which is equivalent to a linear function with respect to nonlogarithmic<br />
concentration at small effect levels. Sample<br />
extracts are normally composed of a mixture of unknown<br />
substances at unknown concentrations, so the concentration<br />
unit of the sample in the concentration effect curve is based<br />
on the Relative Enrichment Factor (REF) in units of<br />
[Lwater sample/Lbioassay].<br />
Toxic equivalent concentrations (TEQ) were calculated as<br />
the ratio of EC50 values of the reference compound to the EC50<br />
of the sample (Eq. 5).<br />
TEQ ¼ EC50ðreference compoundÞ<br />
EC50ðsampleÞ<br />
2 g<br />
3<br />
Lbioassay 4 ¼<br />
g<br />
5 ð5Þ<br />
L water sample<br />
L bioassay<br />
Lwater sample<br />
Since the EC50(reference compound) is expressed in concentration<br />
units such as [g/Lbioassay] and the EC50(sample) in<br />
concentration units [Lwater sample/Lbioassay], representing the<br />
enrichment or dilution (REF) of the sample required to elicit the<br />
50% effect, the toxic equivalent concentration is expressed in<br />
[g/Lwater sample].<br />
In principle, any effect level can be used to derive TEQ,<br />
provided that the concentration-effect curves are parallel i.e.<br />
have the same slope, which is in fact one of the prerequisites for<br />
the validity of the toxic equivalency concept (Villeneuve et al.,<br />
2000). Therefore, if the effect of the sample in the bioassay does<br />
not reach 50%, toxic equivalency can be estimated using EC20 of<br />
the sample and EC20 of the corresponding reference compound.<br />
2.7. QA/QC<br />
avg sd avg sd<br />
Blank MilliQ water n.a. n.a. n.a. n.a.<br />
S1 Influent (Effluent from WWTP) 20 4 17 4<br />
S2 Denitrification 24 n.a. 21 n.a.<br />
S3 Pre-Ozonation 21 7 21 4<br />
S4 Coagulation/flocculation/DAFF 11 1 11 0.5<br />
S5 Main ozonation 9.9 0.9 10 0.9<br />
S6 Activated carbon filtration 6.6 0.6 6.6 0.9<br />
S7 Effluent 7.1 0.5 6.8 0.7<br />
C3 Biological sand filter fed with S4 water 8.5 1 8.8 0.8<br />
C4 Biological activated carbon filter fed with S4 water 4.0 0.4 4.2 0.04<br />
C5 Biological activated carbon filter fed with S5 water 3.6 0.07 4.1 0.08<br />
avg - average; sd - standard deviation; n.a. - not applicable.<br />
water research 44 (2010) 477–492<br />
To determine background levels of contamination associated<br />
with the extraction process as well as assessing any effect of
Table 3 – Evaluation of the concentration effect curves and expression of the bioanalytical results.<br />
Assay Concentration-effect<br />
assessment<br />
Baseline Toxicity –<br />
Bioluminescence<br />
inhibition in<br />
Vibrio<br />
fischeri<br />
Neurotoxicity –<br />
AChE<br />
Phytotoxicity –<br />
Max-I-PAM<br />
Estrogenicity –<br />
E-SCREEN<br />
Ah-Receptor –<br />
AhR-CAFLUX<br />
Genotoxicity –<br />
Umu<br />
Top: 100<br />
Bottom: 0<br />
Slope: 1<br />
Top: w<br />
Bottom: w<br />
Slope: 1<br />
Top: at least<br />
100 and then w<br />
Bottom: w<br />
Slope: 1<br />
the solvent, 2.0 L of MilliQ water were extracted identically to<br />
the samples and analysed in all bioassays as a procedural<br />
blank.<br />
For quality control and assurance purposes two replicates<br />
of randomly selected samples (2 sites per each sampling; for<br />
details see Supporting information Table S1–S6) were<br />
collected and analysed in all bioassays on a different day than<br />
the first replicate to evaluate repeatability of the SPE and the<br />
bioassays. Each sample extract was tested in the bioassays in<br />
duplicates or triplicates per run, depending on the assay.<br />
All individual data sets are summarised in the Supporting<br />
information Table S1–S6. Since the variability among the four<br />
sampling events was not higher than the variability between<br />
the sample replicates collected at the same time, we report in<br />
the main manuscript the average sd of four sample extracts<br />
at four different consecutive sampling events (n ¼ 4, whith<br />
exception of S2, where n ¼ 3). Analysis of variance (ANOVA)<br />
was used to test the differences among the average TEQ of<br />
samples S1-C5, including the blank, in comparison to S1.<br />
3. Results and discussion<br />
REF<br />
range<br />
3.1. Baseline toxicity – bioluminescence inhibition<br />
in vibrio fischeri<br />
Reference compound;<br />
Positive control<br />
0.4–38 Virtual baseline<br />
toxicant;<br />
Phenol a<br />
The EC50 of phenol was 130 37 mg/L over 8 plates in 3 different<br />
repetitions of the assay, confirming the reproducibility and<br />
repeatability of the assay. While in all other assays, the reference<br />
compound served two purposes, as quality control and to<br />
define the reference for the TEQ, in the bioluminescence inhibition<br />
test, phenol was only used for quality control.<br />
EC 50 of the<br />
reference<br />
compound<br />
12 mg/L;<br />
130 mg/L<br />
Result<br />
expression<br />
Baseline toxicity<br />
equivalent<br />
concentrations<br />
(baseline-TEQ)<br />
0.04–115 Parathion 120 mg/l Parathion equivalent<br />
concentrations (PTEQ)<br />
1.5–258 Diuron 16 mg/L Diuron equivalent<br />
concentrations<br />
(DEQ)<br />
0.0006–55 17ß-Estradiol (E2) 2 ng/L Estradiol equivalent<br />
concentrations (EEQ);<br />
Relative proliferative<br />
effect (RPE)<br />
0.004–55 2,3,7,8-<br />
Tetrachlorodibenzodioxin<br />
(TCDD)<br />
Linear regression 2.2–110 (-S9) 2-amino<br />
anthracene a<br />
(þS9) 4-nitroquinoline-<br />
N-oxide a<br />
a Only used as positive control, not as reference compound.<br />
b Quantification limit; w variable; n.a. - not applicable.<br />
water research 44 (2010) 477–492 481<br />
8 ng/L TCDD equivalent<br />
concentrations<br />
(TCDDEQ)<br />
n.a. REF that elicits genotoxic<br />
effect with threshold<br />
Induction Ratio of<br />
1.5<br />
Detection limit<br />
20% inhibition<br />
of bioluminescence<br />
0.3 mg/L<br />
0.01 mg/L<br />
0.02 ng/L<br />
0.04 ng/L<br />
Maximum REF<br />
achieved in the<br />
assay b<br />
Toxicity of the sample was expressed in baseline toxicity<br />
equivalents (baseline-TEQ) derived from a baseline toxicity<br />
QSAR (quantitative structure-activity relationship) (Fig. 1A)<br />
using a virtual compound with log K ow of 3 and molecular<br />
weight of 300 g/mol as a reference, which equates to an EC50 of<br />
12 mg/L (Escher et al., 2008b). The rationale behind this choice<br />
is the fact that there is not a singular positive control, as every<br />
chemical exhibits baseline toxicity. For specifically acting<br />
compounds the baseline effect is marginal and cannot be<br />
resolved if the single chemical is tested. However, in a mixture<br />
with a large number of chemicals with a variety of specific<br />
modes of action, as is the case in a wastewater sample, the<br />
baseline toxicity might actually dominate the overall mixture<br />
effect (Warne and Hawker, 1995). Thus, baseline toxicity will<br />
provide an integrative measure of the combination of chemicals<br />
that act together in the mixture and each chemical’s<br />
contribution is essentially weighted by its hydrophobicity only<br />
(Escher and Schwarzenbach, 2002). Fig. 1A depicts the relationship<br />
between hydrophobicity and effect concentration on<br />
a log-log plot and how the EC50 of a virtual baseline toxicant is<br />
derived (note that the hydrophobicity scale is based on liposome-water<br />
partitioning not octanol-water partitioning),<br />
therefore the virtual baseline toxicant with log K ow of 3 is<br />
situated slightly higher than 3 on the log-hydrophobicity<br />
scale.<br />
Samples were tested for baseline toxicity at 8 different<br />
concentrations after serial dilution 1:2, with the REF ranging<br />
from 0.4 to 38. All previously tested baseline toxicants<br />
exhibited a common slope of 1 (Escher et al., 2008b), and<br />
therefore the slope of the samples was fixed to 1, too. Effect<br />
concentrations (EC50) of the samples used to calculate the<br />
toxic equivalent concentrations were expressed in REFs,
482<br />
Fig. 1 – Concentration-effect curves of the reference compounds and selected samples tested in bioluminescence inhibition<br />
assay (A, B), AChE inhibition assay (C, D) and I-PAM (E, F) assay. Bottom and top of each curve was fixed to 0 and 100,<br />
respectively. Quality of the fit expressed as R 2 was >0.90, except for S7-effluent sample tested in AChE assay, where the R 2<br />
was 0.74. (A) Baseline toxicity QSAR for the bioluminescence inhibition test with Vibrio fischeri and derivation of the EC50 of<br />
the virtual baseline toxicant that is used as reference compound in this assay (data and QSAR from Escher et al. (2008b)).<br />
Error bars indicate SD.<br />
representing the enrichment or dilution of the sample<br />
required to elicit the 50% effect, and are summarised in Table 4.<br />
The EC 50 of the blank was 59, which means that if MilliQ<br />
water were enriched 59 times, the blank extract would elicit<br />
50% bioluminescence inhibition. This EC 50 was extrapolated<br />
from experimental data, where the highest measured REF was<br />
38, which caused a 25% effect. It is not reasonable to enrich the<br />
samples to a higher REF because the enrichment of impurities<br />
of solvents and materials or contamination during the<br />
enrichment procedure would mask the measurable effect.<br />
The influent, denitrification and ozonation samples all<br />
exhibited EC50 in the range of 4–5.6, i.e. the samples had to be<br />
water research 44 (2010) 477–492<br />
enriched by approximately five times to see an effect. The<br />
concentration window for a visible effect is relatively narrow.<br />
A sample that is not enriched (REF ¼ 1) or even diluted would<br />
not give any visible response. The necessary enrichment is not<br />
the only reason for using SPE for sample preparation:<br />
a majority of matrix compounds are removed by SPE, such as<br />
salts and particulates. All organisms are very sensitive to<br />
variable electrolyte concentration. In this assay too little salt<br />
would decrease the bioluminescence of the marine bacterium<br />
(Escher et al., 2008a), and salting up is difficult because the salt<br />
content of water samples is unknown. In contrast to salts and<br />
particulate matter, we assume that dissolved organic carbon
Table 4 – Sample enrichment required to produce a 50% effect in each assay expressed as the average ± sd of four 24 h<br />
sampling events.<br />
ID Site description Bioluminescence inhibition AChE I-PAM<br />
(DOC), in particular the low molecular weight fraction, is<br />
partially retained by SPE.<br />
Baseline-TEQ increased slightly from influent (2.3 mg/L) to<br />
denitrification (2.8 mg/L) and after pre–ozonation (3.2 mg/L).<br />
However the stepwise increase was not statistically significant<br />
(ANOVA, p ¼ 0.05). Significant decrease by more than<br />
a factor of two in TEQ was observed after coagulation/flocculation/DAFF<br />
(1.4 mg/L). Baseline-TEQ after the main ozonation,<br />
activated carbon filtration and in the effluent ranged<br />
water research 44 (2010) 477–492 483<br />
EC50 (REF) EC50 (REF) EC50 (REF)<br />
avg sd avg sd avg sd<br />
Blank 59 0.81 BDL (
484<br />
Fig. 2 – Relationship between DOC concentration and<br />
baseline-TEQ. The black diamonds are the data from this<br />
study (from Tables 1 and 5). The empty squares refer to<br />
a previous study performed in Switzerland (all data refer to<br />
secondary effluent) (Escher et al., in press). Error bars<br />
indicate SD.<br />
matter that is co-extracted with SPE. Assimilable organic carbon<br />
may be taken up by microorganisms and contribute to bioluminescence<br />
inhibition, while higher molecular-weight fractions<br />
of organic matter may bind micropollutants, potentially<br />
causing a decrease in their bioavailability and thus decrease the<br />
response. Note also that organic micropollutants contribute to<br />
DOC in DOC measurements. Therefore it is virtually impossible<br />
to differentiate between assimilable organic carbon and organic<br />
micropollutants.<br />
The TEQ of the influent of the reclamation plant, which is the<br />
effluent of a conventional wastewater treatment plant (WWTP),<br />
was higher than of the secondary effluent of a WWTP in<br />
Switzerland (Escher et al., 2008a; Escher et al., in press). It must<br />
also be noted that the DOC levels were significantly lower in the<br />
Swiss WWTP effluent than in the present study. In fact,<br />
the baseline-TEQ of the Swiss study (Escher et al., in press)fallon<br />
the lower left corner of Fig. 2, but they fit the correlation. This<br />
observation supports the conclusion that DOC contributed to<br />
the effects measured in this bioassay.<br />
3.2. Neurotoxicity – acetylcholinesterase<br />
inhibition assay<br />
The concentration-effect curve of the reference compound,<br />
parathion, followed a log-logistic function with the slope 1 and<br />
the EC50 of 120 32 mg/L (Fig. 1C), using a commercially available<br />
acetylcholinesterase enzyme from Electrophorus electricus.<br />
The EC50 of parathion in our work was higher than data<br />
reported in literature: 27 mg/L (Escher et al., in press) and 26 mg/L<br />
(Escher et al., 2008a) with the use of bovine acetylcholinesterase.<br />
The difference in the sensitivity of the assay might be<br />
due to a different origin of acetylcholinesterase enzyme.<br />
The samples were tested at 12 different concentrations<br />
after serial dilution 1:2 with the REF range from 0.04 to 115.<br />
The best-fit slope of the samples’ concentration-effect curve<br />
was 1 (Fig. 1D) as it was for the reference compound. Effective<br />
concentrations required to elicit 50% inhibition are summarised<br />
in Table 4. Influent samples had to be enriched 52 times to<br />
elicit 50% inhibition, in contrast to effluent, which would<br />
water research 44 (2010) 477–492<br />
require an enrichment of 1100 times to elicit the same effect.<br />
As discussed above, we did not apply such high enrichment<br />
factors but had to extrapolate the EC50. The effect of the blank<br />
and samples treated with activated carbon was less than 20%,<br />
defined as the detection limit of the assay. Therefore the<br />
effective concentrations were not extrapolated from the<br />
concentration-effect curves.<br />
Results of the acetylcholinesterase inhibition assay were<br />
expressed as parathion equivalent concentrations (PTEQ) and<br />
are summarised in Table 5. PTEQ of the samples showed firstly<br />
a slight increase after pre-ozonation to 4.2 mg/L in comparison<br />
with the influent PTEQ of 3.1 mg/L, and then a significant<br />
decrease to 1.9 mg/L after the main ozonation. Activated<br />
carbon filtration further reduced PTEQ below the detection<br />
limit (
measurement is somewhat compromised, when several nonspecifically<br />
acting chemicals in the mixture mask the phytotoxic<br />
effect (Escher et al., 2008a). In fact, when single baseline<br />
toxicants (Escher et al., 2008b) were tested for the inhibition of<br />
photosynthesis yield they showed a response when<br />
measuring Y after 2 hours incubation, albeit at higher<br />
concentrations than for the corresponding growth inhibition<br />
endpoint, and the curve was steeper with a slope of 1.9. Since<br />
it is impossible to quantitatively differentiate between the<br />
contributions from phytotoxicity and cytotoxicity (unless<br />
cytotoxicity is so overwhelming that it makes the photosynthesis<br />
yield measurements impossible), we used the slope of<br />
the reference compound diuron to fit the samples even<br />
though the quality of fit was sub-optimal.<br />
The phytotoxic response of the samples after 2 hours of<br />
incubation expressed as diuron equivalent concentration DEQ<br />
is summarised in Table 5. Denitrification significantly<br />
increased the DEQ (0.29 mg/L) in comparison with the influent<br />
DEQ of 0.11 mg/L. The pre-ozonation step itself had no significant<br />
effect (0.34 mg/L). After coagulation/flocculation/DAFF,<br />
the DEQ decreased by 85% to 0.05 mg/L. DEQ decreased in small<br />
steps after the main ozonation and activated carbon filtration<br />
stages and in the effluent samples. No significant additional<br />
decrease in DEQ was observed after sand filter or biological<br />
activated carbon filters. All DEQ after coagulation/flocculation/DAFF<br />
are close to the detection limit and are based on<br />
water research 44 (2010) 477–492 485<br />
Fig. 4 – Concentration-effect curves of selected samples and corresponding standards tested in E–SCREEN (A, B) and AhR-<br />
CAFLUX (C, D) assay. The slope of the curves was fixed to 1. Bottom of the curves was an adjustable parameter in both<br />
assays. The top of the curve was adjustable in the E–SCREEN assay, while in AhR-CAFLUX, the top was set to the level of at<br />
least 100% of the TCDD maximal effect. Quality of the fit expressed as R 2 ranged for selected samples and reference<br />
compounds from 0.96 to 0.99, except for Blank and S7-effluent samples tested in E-SCREEN assay, where the R 2 was 0.3 and<br />
0.4, respectively due to a very low response of both samples. Error bars indicate SD.<br />
a single data point, therefore should be interpreted with some<br />
caution.<br />
The initial slight but not statistically significant increase in<br />
TEQ from sample S1 to S3 is unexpected at first sight. It is<br />
interesting to note, however, that this small increase was<br />
common for baseline toxicity and slightly less clear also for<br />
neurotoxicity (Figs. 2 and 3). The increase in baseline toxicity<br />
was explained by an increase of DOC, which is presumably<br />
partially co-extracted at low pH. The small increase in DEQ<br />
could also be caused by baseline toxicants interfering with the<br />
measurement of the photosynthesis yield. We have not<br />
observed such an effect in our previous work (Escher et al.,<br />
2008a; Escher et al., in press). However, we have never<br />
encountered DOC concentrations exceeding 10 mg/L in any of<br />
our previous work. It is unclear how much of the DOC is coextracted<br />
with the micropollutants by SPE. The effect of DOC<br />
needs to be investigated in future work but is beyond the<br />
scope of the present study. Possibly, extraction at a higher pH<br />
might reduce the fraction of co-extracted DOC due to the<br />
higher negative charge of DOC under these conditions.<br />
Previous work on comparing herbicides concentrations<br />
with DEQ often showed a relatively good correlation and<br />
a substantial fraction (> 50% of the response) could be<br />
explained by the measured herbicides (Bengtson Nash et al.,<br />
2005b; Bengtson Nash et al., 2006; Escher et al., 2006; Muller<br />
et al., 2008). There were some cases, such as samples collected
486<br />
Table 6 – Sample enrichment required to produce a defined effect in each assay: 50% effect in E–SCREEN assay and 20% of<br />
maximal TCDD effect in AhR-CAFLUX assay. Results represent the average ± sd of 4 samplings.<br />
ID Site description E-SCREEN AhR-CAFLUX<br />
from the Thames River, where the biological DEQ were much<br />
higher than the chemically analysed DEQ (Bengtson Nash<br />
et al., 2006). The authors suggested this difference was due to<br />
the presence of additional herbicides that were not analysed<br />
with the analytical chemical methodology, but given the new<br />
evidence it is also possible that additional cytotoxicity by nonspecifically<br />
acting micropollutants and DOC could have<br />
contributed to the discrepancy.<br />
3.4. Estrogenic activity: E-SCREEN assay<br />
In the E-SCREEN, bottom and top of the concentration-effect<br />
curve (Eq. 4), both adjustable parameters, refer to the<br />
minimum or maximum stimulatory response, i.e., proliferation<br />
of the estrogen-dependent cells in E–SCREEN. The slopes<br />
of the concentration-effect curves of the reference compound,<br />
17ß-estradiol, and the samples were fixed to 1 (Figs. 4A and B).<br />
The maximum of the curve yields information on the mode of<br />
action, thus being able to distinguish among full agonist,<br />
water research 44 (2010) 477–492<br />
Before Clean-up After Clean-up<br />
RPE Mode of Action EC50 (REF) EC20 TCDD (REF) EC20 TCDD (REF)<br />
avg sd avg sd avg sd avg sd<br />
Blank 0.05 – Not estrogenic BDL (
Table 7 – Sample enrichment required to produce<br />
a genotoxic effect [ Induction Ratio of 1.5 (IR1.5) in umuC<br />
assay without and with metabolic activation. Results<br />
represent the average ± sd of 4 samplings.<br />
ID Site description umuC -S9 umuC þ S9<br />
EC IR1.5 (REF) EC IR1.5 (REF)<br />
avg sd avg sd<br />
Blank >78 >78<br />
S1 Influent (Effluent from WWTP) 5.2 2 17 3<br />
S2 Denitrification 6.9 2 22 3<br />
S3 Pre-ozonation 6.3 2 18 3<br />
S4 Coagulation/flocculation/DAFF 15 7 28 8<br />
S5 Main ozonation 78 20 >87<br />
S6 Activated carbon filtration >84 >84<br />
S7 Effluent >69 >73<br />
C3 Biological sand<br />
filter fed with S4 water<br />
24 13 34 16<br />
C4 Biological activated<br />
carbon filter<br />
fed with S4 water<br />
>80 >80<br />
C5 Biological activated<br />
carbon filter<br />
fed with S5 water<br />
>79 >79<br />
(1667 times dilution) to 55 (55 times enrichment of the original<br />
water sample). Sample REF required to produce a 50% effect<br />
(EC50) in the E–SCREEN assay ranged from 0.27 (equivalent to 4<br />
times dilution of the original water sample) to 6.7 (equivalent to<br />
6.7 times enrichment of the original water sample) (Table 3). The<br />
wide range of REF windows allows one to monitor a wide range<br />
of samples collected throughout the treatment chain.<br />
The estrogenicity of the samples was quantified with two<br />
parameters: the relative proliferative effect (RPE) and the<br />
estradiol equivalent concentration (EEQ). RPE of the samples,<br />
representing the ratio of the maximum proliferation induced<br />
by test samples compared to 17b-estradiol (also referred to as<br />
relative efficacy), are summarised in Table 6. The samples from<br />
the influent up to ozonation were fully or partially agonistic,<br />
while the remainder of the samples were not classified estrogenic<br />
and no EEQ could be derived for these samples.<br />
When a sample was classified as either a partial or full<br />
agonist, an estradiol equivalent concentration (EEQ) was<br />
calculated from the EC 50. Influent to the reclamation plant<br />
(representing the effluent of the WWTP) showed full agonistic<br />
activity (RPE > 0.8) with the estradiol equivalent concentration<br />
EEQ of 6 ng/L (Table 5). This was comparable with the EEQs<br />
reported for WWTP effluent in the literature: 2 ng/L) and the resulting effluent of the ozonation contained<br />
0.6 ng/L EEQ (Escher et al., in press).<br />
Estrogenicity was further markedly altered after activated<br />
carbon filtration. Estrogenic effect of the sample treated with<br />
activated carbon (S6) was less than 20% of the estradiol effect<br />
(RPE < 0.2), which corresponds to the detection limit of the<br />
assay, therefore the EEQ was less than 0.02 ng/L.<br />
Estrogenicity (RPE) of the effluent sample was slightly<br />
increased, but did not reach 50% effect of estradiol, therefore<br />
the sample was classified as not estrogenic and the EEQ was<br />
not quantified. The estrogenic effect was below quantification<br />
limit. Taking into account the REF of the sample in this assay,<br />
the EEQ of the effluent was less than 0.06 ng/L.<br />
Besides the main treatment chain, additional treatment<br />
processes were assessed in this study: sand filtration of the<br />
coagulation/flocculation/DAFF effluent (C3) and biological<br />
activated carbon filtration of the coagulation/flocculation/<br />
DAFF effluent (C4) and main ozonation effluent (C5). Sand<br />
filtration significantly decreased the EEQ of the coagulation/<br />
flocculation/DAFF effluent from 9.8 ng/L to 0.63 ng/L. Biological<br />
activated carbon filtration decreased the estrogenic effect<br />
of both effluents to below detection limit (
488<br />
reproducibility and repeatability of the assay and is in good<br />
agreement with the previous levels from the literature of<br />
5.9 ng/L (Nagy et al., 2002) and 3.9 ng/L (Zhao and Denison,<br />
2004).<br />
TCDD was chosen as the reference compound as it is one of<br />
the most potent agonists for this assay. However, high affinity<br />
ligands for the AhR also include dibenzofurans and biphenyls as<br />
well as a variety of polycyclic aromatic hydrocarbons, benzoflavones<br />
and other chemicals (Denison and Nagy, 2003). To<br />
assess the contribution of chemicals other than dioxins, furans<br />
and dioxin-like PCBs to the potency estimates, sulfuric acid silica<br />
gel clean-up was performed in this study, which removed most<br />
organic chemicals (e.g. PAHs) except persistent chemicals such<br />
as dioxins, furans and PCBs. Original sample extracts, as well as<br />
the extracts cleaned with sulfuric acid silica gel were tested with<br />
the AhR-CAFLUX assay at 5 different concentrations after serial<br />
dilution 1:10, with the REF ranging from 55 to 0.004. The response<br />
of many samples, especially after clean-up, did not reach 50%<br />
TCDD response. Therefore effective concentrations of the<br />
samples required to elicit 20% effect of TCDD (EC 20 TCDD) were<br />
interpolated from concentration-effect curves (Table 6). The EC 20<br />
TCDD of the samples ranged from 1.4 to 8.2. The effective<br />
concentration of the same samples was significantly altered<br />
after clean up and ranged from 14 to 22, i.e. samples after acid<br />
silica gel clean up had to be enriched approximately 3–10 times<br />
more to elicit the same effect.<br />
The response in the AhR-CAFLUX assay was expressed as the<br />
2,3,7,8-TCDD-equivalent concentration (TCDDEQ). TCDDEQ was<br />
significantly decreased by the main ozonation to 0.44 ng/L in<br />
comparison with the TCDDEQ of 0.83 ng/L in the influent. Sand<br />
filtration did not change the TCDDEQ of the coagulation/flocculation/DAFF<br />
effluent, while treatment with the biological activated<br />
carbon filters reduced the TCDD equivalent concentration<br />
to 0.33 ng/L a level not significantly different from the blank.<br />
The TCDDEQ of the samples after the sulfuric acid silica gel<br />
clean-up were reduced to levels not significantly different<br />
from the blank (Table 5). These results indicate that the<br />
observed effect in the AhR-CAFLUX assay prior to the sulfuric<br />
acid silica gel clean-up step may be attributed to chemicals<br />
other than dioxins, furans and PCBs.<br />
3.6. Genotoxicity-UmuC assay<br />
The umuC genotoxicity assay can detect both cytotoxic and<br />
genotoxic effects. The extracts were tested both with (þS9) and<br />
without ( S9) exogenous metabolic activation. Response on<br />
this assay is determined as an induction ratio (IR) (Fig. 5). An<br />
IR 1.5 is considered genotoxic, providing the sample is not<br />
cytotoxic (growth < 0.5). For samples that demonstrate significant<br />
genotoxic response, the effect concentrations EC IR 1.5<br />
were derived from the linear concentration-effect curve as<br />
depicted in Fig. 5. Effect concentrations are in dimensionless<br />
units of REF and represent how many times the samples must<br />
be concentrated or diluted to elicit a threshold IR of 1.5 in the<br />
assay (Table 7). Results are expressed as 1/ECIR 1.5 therefore<br />
a higher number represents a higher genotoxic effect (Table 5).<br />
Unfortunately no TEQ could be derived for this endpoint but in<br />
theory this should be possible and in the future we will work on<br />
implementing the TEQ concept for this assay. This is somehow<br />
water research 44 (2010) 477–492<br />
problematic because of the frequently observed cytotoxicity of<br />
the samples, which can only be partially accounted for.<br />
Overall, the umuC test without metabolic activation gave<br />
higher responses than the test with metabolic activation<br />
indicating that some of the micropollutants that are responsible<br />
for genotoxicity are detoxified by the liver enzyme fraction<br />
S9. As in all other bioassays, the denitrification and preozonation<br />
did not affect the activity in the umuC test but<br />
coagulation/flocculation/DAFF decreased the genotoxicity<br />
without S9 by 58% and that with S9 by 35%. Subsequent main<br />
ozonation completely removed the effect for the test with prior<br />
S9 treatment and drastically reduced it in the absence of S9.<br />
4. Conclusion<br />
A similar test battery has been previously used for various<br />
applications in wastewater treatment efficiency and water<br />
quality assessment. To apply the test battery to a treatment<br />
train with very subtle steps and for relatively pure water was<br />
a challenge. This challenge was met by the following measures:<br />
A comprehensive framework was developed that centres<br />
around the bioassays but links to sample preparation with<br />
SPE prior to testing and to the data evaluation scheme.<br />
Quality control was implemented and all effects were<br />
related to reference compounds, thereby correcting for<br />
slight day-to-day variability of the test results.<br />
The use of SPE allows a relatively non-specific extraction of<br />
water samples to concentrate a wide range of relevant<br />
pollutants to levels where they can exhibit a measurable<br />
response, while matrix components such as salts and<br />
metals are removed from the sample during SPE.<br />
The detection window of the bioassays is adjustable. For the<br />
application in the present study the concentration range of<br />
the sample in the bioassay ranged between 1700 times dilution<br />
and 250 times enrichment of the sample. The detection<br />
window depends on the constitution of the sample and the<br />
type of bioassay and is only limited by the effect of the blank<br />
at very high enrichment and nonspecific cytotoxicity of the<br />
sample for each of the assays. By dynamically changing<br />
the relative enrichment window of the sample throughout<br />
the study one can adapt the various bioassays to a wide<br />
variety of samples and results are robust and reproducible.<br />
Results are reported as toxic equivalent concentrations,<br />
a format that allows for comparison to results from chemical<br />
analysis. This is also a key feature of the bioanalytical toolsthey<br />
are no direct indicators for environmental health but<br />
they are indicators of the presence of mixtures of chemicals,<br />
accounting for the mixture toxicity of the unidentified<br />
chemicals in an environmental sample. To report bioanalytical<br />
results as toxic equivalent concentrations is better<br />
suited for hazard assessment and extrapolation than the<br />
commonly reported % effect of the water sample in a given<br />
bioassay because it is too difficult to extrapolate effects while<br />
it is possible to compare effect concentrations of chemicals<br />
across different level of biological organisation. So for<br />
example, if a sample elicits 0.5 mg/L DEQ one can compare<br />
this value with the environmental quality standard for
diuron to estimate what potential long-term effects could be<br />
expected from such an exposure.<br />
The study also highlighted some difficulties that remain and<br />
some lessons are to be learned. One of the weak points of the<br />
approach is the sample preparation with SPE. However, this is<br />
a common problem for chemical analysis and can only be circumvented<br />
if there are recovery standards for every single<br />
compound to be analysed. Such an approach is not possible for<br />
bioassays because spiked chemicals would add to the mixture<br />
toxicity. A shortcoming of the SPE method that was noticed here<br />
for the first time is the co-extraction of an unknown fraction of<br />
dissolved organic matter, an issue that requires further attention<br />
in the future.<br />
In addition, volatile compounds will evaporate during SPE<br />
but they would also cross contaminate in a 96 well plate, so it<br />
is a procedural advantage to have removed the volatile<br />
chemicals prior to toxicity testing. It must also be noted that<br />
oxidation by-products formed during ozonation are often<br />
more hydrophilic than their parent compound and might<br />
therefore not be extracted efficiently by SPE.<br />
In addition, one must be aware that mixture effects are<br />
assessed when applying bioanalytical tools to complex samples<br />
and there is always the problem that the subtle effect of newly<br />
formed by-products might be shielded by other micropollutants<br />
present in the mixture. Some of the oxidation byproducts<br />
stemming from ozonation are very reactive, for example aldehydes<br />
and other electrophiles formed from ß-blockers during<br />
ozonation (Benner and Ternes, 2009), others are very hydrophilic<br />
organics such as NDMA (N-nitrosodimethylamine) or<br />
inorganics such as bromate (Hollender et al., in press) noneof<br />
these would be caught with the bioanalytical tools used here but<br />
require targeted chemical analysis for quantification. However,<br />
ozonation often only mildly oxidizes organic micropollutants<br />
and many resulting transformation products are likely to retain<br />
some of their toxic potential (Escher et al., 2008c) or lose their<br />
specific receptor-binding effect but retain non-specific baseline<br />
toxicity (Lee et al., 2008) and will thus contribute to the mixture<br />
toxicity in the bioanalytical tools applied in the present study.<br />
While the results are discussed in more detail and<br />
compared to chemical analysis in the accompanying paper<br />
(Reungoat et al., 2010), some general conclusions can be<br />
drawn:<br />
The influent of the water reclamation plant had similar<br />
levels of effects for the different endpoints as we have<br />
encountered in previous studies for wastewater after<br />
secondary treatment (Escher et al., 2008a; Escher et al., in<br />
press).<br />
The effluent of the water reclamation plant had strongly<br />
reduced effect levels compared to the influent, reducing the<br />
baseline-TEQ by 79%, the DEQ by 76%, the estrogenicity to<br />
below the detection limit of 0.02 ng/L EEQ, the PTEQ by 88%<br />
and the genotoxicity to below the detection limit. Only the<br />
AhR-CAFLUX assay did not yield any information on the<br />
removal efficiency of the treatment because the level of<br />
persistent inducers of the arylhydrocarbon receptor<br />
remained close to the detection limit. This was not too<br />
unexpected given that these are typically very hydrophobic<br />
compounds, which are usually adsorbed and not present in<br />
water research 44 (2010) 477–492 489<br />
the water column. In contrast, the non-persistent inducers<br />
of the AhR were clearly reduced by a factor of two during<br />
ozonation.<br />
The most efficient steps with respect to removal of the toxic<br />
responses in all selected assays proved to be the coagulation/flocculation/DAFF,<br />
the main ozonation and (biological)<br />
activated carbon filtration steps.<br />
This paper demonstrates the first comprehensive application<br />
of the bioanalytical tool framework on an advanced water<br />
treatment plant. One future goal would be to automate the<br />
assays to a degree that they can be used routinely by<br />
commercial laboratories or water companies themselves. At<br />
this stage, there remain some scientific questions to be<br />
resolved, including (1) the sample treatment and dosing, (2)<br />
extension of the battery and (3) data interpretation. (1) SPE and<br />
dosing by solvent spiking will always change the composition<br />
of an undefined mixture. A combination of passive sampling to<br />
a biomimetic polymer and passive dosing directly from the<br />
polymer into the bioassay might be a way to overcome this<br />
limitation. (2) A battery of bioassays will never be comprehensive<br />
because of the myriads of receptors and regulatory<br />
pathways in an organism. Nevertheless the battery would<br />
benefit from inclusion of additional endpoints, such as chemically<br />
induced immunosuppression or even other examples of<br />
mode-of-action categories already included, e.g. other endocrine<br />
endpoints. (3) At this stage we can say how much of<br />
a toxic equivalent concentrations is removed by a certain<br />
treatment step and we can give comparative results but we will<br />
never be able to deduce ecological or health consequences<br />
from that. This discussion must be lead independently. One<br />
solution would be to compare the TEQ of the mixture with the<br />
defined Environmental Water Quality Criteria (EQS) for single<br />
substances of the Water Framework Directive (European<br />
Commission, 2006) or guideline values for drinking water (e.g.<br />
NHMRC–NRMMC, 2004) but not for all reference compounds<br />
are such EQS and guideline values available.<br />
Acknowledgements<br />
This work was co-funded by the Urban Water Security Research<br />
Alliance under the Enhanced Treatment Project, the CRC Water<br />
Quality and Treatment Project No. 2.0.2.4.1.1. Dissolved Organic<br />
Carbon Removal by Biological Treatment and by EnTox. The<br />
National Research Centre for Environmental Toxicology (EnTox)<br />
is a joint venture of the University of Queensland and Queensland<br />
Health Forensic and Scientific Services (QHFSS).<br />
The authors acknowledge the Moreton Bay Water for access to<br />
the South Caboolture Water Reclamation Plant; Ray McSweeny<br />
and Paul McDonnell (Moreton Bay Water) for their help during<br />
sampling and Chris Pipe-Martin (Ecowise) for providing information<br />
on the South Caboolture Water Reclamation Plant.<br />
The authors are grateful to Renee Muller (Gold Coast<br />
Water), Fred Leusch (Griffith University/EnTox), Karen<br />
Kennedy (EnTox) and Pam Quayle (EnTox) for constructive<br />
discussions and for help with the data analysis method, Marita<br />
Goodwin (EnTox) for running various assays and for helpful<br />
comments Christina Carswell (QHFSS) for extracting the
490<br />
samples and Chris Paxman (EnTox) for technical assistance.<br />
We would like to thank also Michael Denison (University of<br />
California Davis, USA) for providing the H4G1.1c2 cells, Georg<br />
Reifferscheid (German Federal Institute of Hydrology,<br />
Germany) for providing the bacteria Salmonella typhimurium<br />
TA1535/pSK1002, and Ana Soto (Tufts University, USA) for<br />
providing the MCF7-BOS cells.<br />
Appendix. Supporting information<br />
Supporting information related to this article can be found at<br />
doi:10.1016/j.watres.2009.09.025.<br />
references<br />
Bengtson Nash, S.M., Schreiber, U., Ralph, P.J., Mueller, J.F., 2005a.<br />
The combined SPE: ToxY-PAM phytotoxicity assay;<br />
application and appraisal of a novel biomonitoring tool for the<br />
aquatic environment. Biosensors and Bioelectronics 20 (7),<br />
1443–1451.<br />
Bengtson Nash, S.M., McMahon, K., Eaglesham, G., Mueller, J.F.,<br />
2005b. Application of a novel phytotoxicity assay for the<br />
detection of herbicides in Hervey Bay and the Great Sandy<br />
Straits. Marine Pollution Bulletin 51, 351–360.<br />
Bengtson Nash, S.M., Goddard, J., Muller, J.F., 2006. Phytotoxicity<br />
of surface waters of the Thames and Brisbane river estuaries:<br />
a combined chemical analysis and bioassay approach for the<br />
comparison of two systems. Biosensors and Bioelectronics 21,<br />
2086–2093.<br />
Benner, J., Ternes, T.A., 2009. Ozonation of propranolol: formation<br />
of oxidation products. Environmental Science & Technology<br />
43, 5086–5093.<br />
Cao, N., Yang, M., Zhang, Y., Hu, J., Ike, M., Hirotsuji, J., Matsui, H.,<br />
Inoue, D., Sei, K., 2009. Evaluation of wastewater reclamation<br />
technologies based on in vitro and in vivo bioassays.<br />
Ecotoxicology and Environmental Safety 407 (5), 1588–1597.<br />
Chapman, P.M., 2000. Whole effluent toxicity testing – usefulness,<br />
level of protection, and risk assessment. Environmental<br />
Toxicology and Chemistry 19 (1), 3–13.<br />
Denison, M.S., Nagy, S.R., 2003. Activation of the aryl hydrocarbon<br />
receptor by structurally diverse exogenous and endogenous<br />
chemicals. Annual Reviews in Pharmacology and Toxicology<br />
43, 309–334.<br />
Deutsche Norm, 1995. In: DIN, N.W.N.i. (Ed.), Suborganismische<br />
Verfahren: Teil 1: Bestimmung von Cholinesterase-<br />
Hemmenden Organophosphat- und Carbamat-Pestiziden<br />
(Cholinesterase-Hemmtest) DIN 38415-1: 1995-02, Berlin,<br />
Germany.<br />
Ellman, G.L., Courtney, K.D., Andres, V., Featherstone, R.M., 1961. A<br />
new and rapid colorimetric determination of acetylcholinesterase<br />
activity. Biochemical Pharmacology 7, 88–95.<br />
Escher, B., Schwarzenbach, R.P., 2002. Mechanistic studies on<br />
baseline toxicity and uncoupling as a basis for modeling<br />
internal lethal concentrations in aquatic organisms. Aquatic<br />
Sciences 64, 20–35.<br />
Escher, B.I., Bramaz, N., Maurer, M., Richter, M., Sutter, D., von<br />
Känel, C., Zschokke, M., 2005. Screening test battery for<br />
pharmaceuticals in urine and wastewater. Environmental<br />
Toxicology and Chemistry 24 (3), 750–758.<br />
Escher, B.I., Quayle, P., Muller, R., Schreiber, U., Mueller, J., 2006.<br />
Passive sampling of herbicides combined with effect analysis<br />
in algae using a novel high-throughput phytotoxicity assay<br />
water research 44 (2010) 477–492<br />
(Maxi-Imaging- PAM). Journal Environmental Monitoring 8,<br />
456–464.<br />
Escher, B.I., Bramaz, N., Quayle, P., Rutishauser, S., Vermeirssen, E.,<br />
2008a. Monitoring of the ecotoxicological hazard potential by<br />
polar organic micropollutants in sewage treatment plants and<br />
surface waters using a mode-of-action based test battery.<br />
Journal Environmental Monitoring 10, 622–631.<br />
Escher, B.I., Bramaz, N., Mueller, J.F., Quayle, P., Rutishauser, S.,<br />
Vermeirssen, E., 2008b. Toxic equivalent concentrations<br />
(TEQs) for baseline toxicity and specific modes of action as<br />
a tool to improve evaluation of ecotoxicity tests on<br />
environmental samples. Journal Environmental Monitoring<br />
10, 612–621.<br />
Escher, B.I., Baumgartner, R., Lienert, J., Fenner, K., 2008c.<br />
Predicting the ecotoxicological effects of transformation<br />
products. SpringerLink online. In: Boxall, A. (Ed.), Handbook of<br />
Environmental Chemistry – Degradation of Synthetic<br />
Chemicals in the Environment. Springer, Heidelberg,<br />
Germany. doi:10.1007/6998_2_015. http://www.springerlink.<br />
com/content/48r2t364q183q654/?p ¼ ec460152bfd<br />
14906adc89378463e4997&pi ¼ 5 published Wednesday,<br />
April 30, 2008.<br />
Escher, B.I., Bramaz, N. and Ort, C., 2009. Monitoring the<br />
treatment efficiency of a full scale ozonation on a<br />
sewage treatment plant with a mode-of-action based test<br />
battery. Journal Environmental Monitoring, doi: 10.1039/<br />
b907093a.<br />
European Commission, 2006. Proposal for a Directive of the<br />
European Parliament and the Council on Environmental<br />
Quality Standards in the Field of Water Policy and Amending<br />
Directive 2000/60/EC COM(2006) 398 Final. COM(2006) 398<br />
Final. European Parliament, Committee on the Environment,<br />
Public Health and Food Safety, Brussels.<br />
Farré, M., Martinez, E., Hernando, M., Fernandez-Alba, A., Fritz, J.I.,<br />
Unruh, E., Mihail, O., Sakkas, V., Morbey, A., Albanis, T.A.,<br />
Brito, F., Hansen, P.D., Barcelo, D., 2006. European ring<br />
exercise on water toxicity using different bioluminescence<br />
inhibition tests based on Vibrio fischeri, in support to the<br />
implementation of the water framework directive. Talanta,<br />
323–333.<br />
GWRC, 2008. In: Leusch, F.D.L. (Ed.), Tools to Detect Estrogenic<br />
Activity in Environmental Waters. Global Water Research<br />
Coalition, London, UK, p. 84.<br />
Hamers, T., Molin, K.R.J., Koeman, J.H., Murk, A.J., 2000. A smallvolume<br />
bioassay for quantification of the esterase inhibiting<br />
potency of mixtures of organophosphates and carbamate<br />
insceticides in rainwater: development and optimization.<br />
Toxicological Sciences 58, 60–67.<br />
Hollender, J., Zimmermann, S.G., Koepke, S., Krauss, M.,<br />
McArdell, C.S., Ort, C., Singer, H., von Gunten, U., Siegrist, H.,<br />
2009. Elimination of organic micropollutants in a municipal<br />
wastewater treatment plant upgraded with a full-scale postozonation<br />
followed by sand filtration. Environmental Science<br />
& Technology, Articles ASAP (Article), in press. doi: 10.1021/<br />
es9014629.<br />
Huber, M.M., Ternes, T.A., von Gunten, U., 2004. Removal of<br />
estrogenic activity and formation of oxidation products during<br />
ozonation of 17 alpha-ethinylestradiol. Environmental<br />
Science & Technology 38, 5177–5186.<br />
Huber, M.M., Göbel, A., Joss, A., Hermann, N., Löffler, D.,<br />
McArdell, C.S., Ried, A., Siegrist, H., Ternes, T.A., von<br />
Gunten, U., 2005. Oxidation of pharmaceuticals during<br />
ozonation of municipal wastewater effluents: a pilot study.<br />
Environmental Science & Technology 39, 4290–4299.<br />
International Standard Organisation 1998 Water Quality–<br />
determination of the inhibitory effect of water samples on the<br />
light emission of Vibrio fischeri (luminescent bacteria test), EN<br />
ISO 11348-3, Geneva, Switzerland.
International Standard Organisation 2000 Water quality –<br />
Determination of the genotoxicity of water and waste water<br />
using the umu-test, EN ISO 13829(2000) and 38415-3 (1996),<br />
Geneva, Switzerland.<br />
Johnson, B.T., 2005. Microtox acute toxicity test. In: Blaise, C.,<br />
Ferard, J.-F. (Eds.), Small-Scale Freshwater Toxicity<br />
Investigations. Toxicity Test Methods, vol. 1. Springer.<br />
Joss, A., Keller, E., Alder, A.C., Gobel, A., McArdell, C.S., Ternes, T.,<br />
Siegrist, H., 2005. Removal of pharmaceuticals and fragrances<br />
in biological wastewater treatment. Water Research 39 (14),<br />
3139–3152.<br />
Körner, W., Hanf, V., Schuller, W., Kempter, C., Metzger, J.,<br />
Hagenmaier, H., 1999. Development of a sensitive E-screen<br />
assay for quantitative analysis of estrogenic activity in<br />
municipal sewage plant effluents. Ecotoxicology and<br />
Environmental Safety 225 (1–2), 33–48.<br />
Körner, W., Bolz, U., Sussmuth, W., Hiller, G., Schuller, W.,<br />
Hanf, V., Hagenmaier, H., 2000. Input/output balance of<br />
estrogenic active compounds in a major municipal sewage<br />
plant in Germany. Chemosphere 40 (9–11), 1131–1142.<br />
Körner, W., Spengler, P., Bolz, U., Schuller, W., Hanf, V.,<br />
Metzger, J.W., 2001. Substances with estrogenic activity in<br />
effluents of sewage treatment plants in southwestern<br />
Germany. 2. Biological analysis. Environmental Toxicology<br />
and Chemistry 20 (10), 2142–2151.<br />
Lee, Y., Escher, B.I., von Gunten, U., 2008. Efficient removal of<br />
estrogenic activity during oxidative treatment of waters<br />
containing steroid estrogens. Environmental Science &<br />
Technology 42 (17), 6333–6339.<br />
Legler, J., Zeinstra, L.M., Schuitemaker, F., Lanser, P.H., Bogerd, J.,<br />
Brouwer, A., Vethaak, A.D., De Voogt, P., Murk, A.J., Van der<br />
Burg, B., 2002. Comparison of in vivo and in vitro reporter gene<br />
assays for short-term screening of estrogenic activity.<br />
Environmental Science & Technology 36 (20), 4410–4415.<br />
Leusch, F.D., Chapman, H.F., van den Heuvel, M.R., Tan, B.L.,<br />
Gooneratne, S.R., Tremblay, L.A., 2006. Bioassay-derived<br />
androgenic and estrogenic activity in municipal sewage in<br />
Australia and New Zealand. Ecotoxicology and Environmental<br />
Safety 65 (3), 403–411.<br />
Muller, R., Tang, J., Thier, R., Mueller, J., 2007. Combining passive<br />
sampling and toxicity testing for evaluation of mixtures of<br />
polar organic chemicals in sewage treatment plant effluents.<br />
Journal Environmental Monitoring 9, 104–109.<br />
Muller, R. 2008 Passive sampling and bioanalysis of organic<br />
chemicals in surface water. Ph.D. thesis. University of<br />
Queensland, St. Lucia, Qld, Australia, 201pp. Accessible at<br />
http://library.uq.edu.au/search~S7?/XRenee þ<br />
Muller&searchscope ¼ 7&SORT ¼ AZ/XRenee þ<br />
Muller&searchscope ¼ 7&SORT ¼ AZ&extended ¼ 0&SUBKEY<br />
¼ Renee%20Muller/21%22C23%22C23%22CB/frameset&FF ¼<br />
XRenee þ Muller&searchscope ¼ 27&SORT ¼<br />
AZ&21%22C21%22C<br />
Muller, R., Schreiber, U., Escher, B.I., Quayle, P., Bengtson Nash, S.,<br />
Mueller, J., 2008. Rapid exposure assessment of PSII herbicides<br />
in surface water using a novel chlorophyll a fluorescence<br />
imaging assay. Science Total Environment 401, 51–59.<br />
Murk, A.J., Legler, J., van Lipzig, M.M.H., Meerman, J.H.N.,<br />
Belfroid, A.C., Spenkelink, A., van der Burg, B., Rijs, G.B.J.,<br />
Vethaak, D., 2002. Detection of estrogenic potency in<br />
wastewater and surface water with three in vitro bioassays.<br />
Environmental Toxicology and Chemistry 21 (1), 16–23.<br />
Nagy, S., Sanborn, J., Hammock, B., Denison, M., 2002.<br />
Development of a green fluorescent protein-based cell<br />
bioassay for the rapid and inexpensive detection and<br />
characterization of Ah receptor agonists. Toxicological<br />
Sciences 65, 200–210.<br />
NHMRC–NRMMC, 2004. Australian Drinking Water Guidelines<br />
(ADWG). National Health and Medical Research Council<br />
water research 44 (2010) 477–492 491<br />
(NHMRC) in Collaboration With the Natural Resource<br />
Management Ministerial Council (NRMMC), Commonwealth<br />
of Australia. www.nhmrc.gov.au/publications/synopses/_<br />
files/adwg_11_06.pdf.<br />
Oda, J., Nakamura, S.I., Oki, I., Kato, T., Shinagawa, H., 1985.<br />
Evaluation of the new system (umu-test) for the detection of<br />
environmental mutagens and carcinogens. Mutatation<br />
Research 147, 219–229.<br />
Petala, M., Samaras, P., Zouboulis, A., Kungolos, A.,<br />
Salkellaropoulos, G., 2006. Ecotoxicological properties of<br />
wastewater treated using tertiary methods. Environmental<br />
Toxicology 21 (4), 417–424.<br />
Petala, M., Samaras, P., Zouboulis, A., Kungolos, A.,<br />
Sakellaropoulos, G.P., 2008. Influence of ozonation on the in<br />
vitro mutagenic and toxic potential of secondary effluents.<br />
Water Research 42 (20), 4929–4940.<br />
Reifferscheid, G., Heil, J., Oda, Y., Zahn, R.K., 1991. A microplate<br />
version of the SOS/umu-test for rapid detection of genotoxins<br />
and genotoxic potentials of environmental samples. Mutation<br />
Research 253 (3), 215–222.<br />
Reungoat, J., Macova, M., Escher, B.I., Carswell, S., Mueller, J.F.,<br />
Keller, J., 2010. Removal of micropollutants and reduction of<br />
biological adverse effects in a full scale reclamation plant<br />
using ozonation and activated carbon filtration. Water<br />
Research 44 (2), 625–637.<br />
Rutishauser, B.V., Pesonen, M., Escher, B.I., Ackermann, G.E.,<br />
Aerni, H.-R., Suter, M.J.-F., Eggen, R.I.L., 2004. Comparative<br />
analysis of estrogenic activity in sewage treatment plant<br />
effluents involving three in vitro assays and chemical analysis of<br />
steroids. Environmental Toxicology and Chemistry 23, 857–864.<br />
Sanchez-Polo, M., Salhi, E., Rivera-Utrilla, J., von Gunten, U., 2006.<br />
Combination of ozone with activated carbon as an alternative<br />
to conventional advanced oxidation processes. Ozone-Science<br />
& Engineering 28 (4), 237–245.<br />
Schreiber, U., Mueller, J.F., Haugg, A., Gademann, R., 2002. New<br />
type of dual-channel PAM chlorophyll fluorometer for highly<br />
sensitive water toxicity biotests. Photosynthesis Research 74,<br />
317–330.<br />
Schreiber, U., Quayle, P., Schmidt, S., Escher, B.I., Mueller, J., 2007.<br />
Methodology and evaluation of a highly sensitive algae<br />
toxicity test based on multiwell chlorophyll fluorescence<br />
imaging. Biosensors and Bioelectronics 22, 2554–2563.<br />
Schwarzenbach, R.P., Escher, B.I., Fenner, K., Hofstetter, T.B.,<br />
Johnson, C.A., von Gunten, U., Wehrli, B., 2006. The challenge<br />
of micropollutants in aquatic systems. Science 313 (5790),<br />
1072–1077.<br />
Snyder, S.A., Adham, S., Redding, A.M., Cannon, F.S., DeCarolis, J.,<br />
Oppenheimer, J., Wert, E.C., Yoon, Y., 2007. Role of<br />
membranes and activated carbon in the removal of endocrine<br />
disruptors and pharmaceuticals. Desalination 202, 156–181.<br />
Soto, A.M., Sonnenschein, C., Chung, K.L., Fernandez, M.F., Olea, N.<br />
, Serrano, F.O., 1995. The E-SCREEN assay as a tool to identify<br />
estrogens: an update on estrogenic environmental pollutants.<br />
Environmental Health Perspectives 103 (Suppl. 7), 113–122.<br />
Tan, B.L.L., Hawker, D.W., Muller, J.F., Leusch, F.D.L., Tremblay, L.A.,<br />
Chapman, H.F., 2007. Comprehensive study of endocrine<br />
disrupting compounds using grab and passive sampling at<br />
selected wastewater treatment plants in South East Queensland,<br />
Australia. Environment International 33 (5), 654–669.<br />
Ternes, T.A., Joss, A., Siegrist, H., 2004. Scrutinizing pharmaceuticals<br />
and personal care products in wastewater treatment.<br />
Environmental Science & Technology 38 (17), 393A–399A.<br />
van Leeuwen, J., Pipe-Martin, C., Lehmann, R.M., 2003. Water<br />
reclamation at South Caboolture, Queensland, Australia.<br />
Ozone: Science & Engineering 25, 107–120.<br />
Villeneuve, D.L., Blankenship, A.L., Giesy, J.P., 2000. Derivation and<br />
application of relative potency estimates based on in-vitro<br />
bioassays. Environmental Toxicology Chemistry 19, 2835–2843.
492<br />
von Gunten, U., 2003. Ozonation of drinking water: part I.<br />
Oxidation kinetics and product formation. Water Research 37<br />
(7), 1443–1467.<br />
Warne, M.S.J., Hawker, D.W., 1995. The number of components<br />
in a mixture determines whether synergistic and<br />
antagonistic or additive toxicity predominate – the funnel<br />
water research 44 (2010) 477–492<br />
hypothesis. Ecotoxicology and Environmental Safety 31 (1),<br />
23–28.<br />
Zhao, B., Denison, M., 2004. Dvelopment and characterization of<br />
a green fluorescence protein-based rat cell bioassay system<br />
for detection of Ah receptor ligands. Organohalogen<br />
Compounds 66, 3332–3337.
An evaluation of a pilot-scale nonthermal plasma advanced<br />
oxidation process for trace organic compound degradation<br />
Daniel Gerrity*, Benjamin D. Stanford, Rebecca A. Trenholm, Shane A. Snyder<br />
Applied Research and Development Center, Southern Nevada Water Authority, River Mountain Water Treatment Facility, P.O. Box 99954,<br />
Las Vegas, NV 89193-9954, USA<br />
article info<br />
Article history:<br />
Received 22 May 2009<br />
Received in revised form<br />
3 September 2009<br />
Accepted 10 September 2009<br />
Available online 17 September 2009<br />
Keywords:<br />
Advanced oxidation process (AOP)<br />
Nonthermal plasma (NTP)<br />
Trace organic compound<br />
Pharmaceutical<br />
Endocrine disrupting compound<br />
(EDC)<br />
1. Introduction<br />
abstract<br />
Wastewater-derived contaminants, including pharmaceuticals<br />
and personal care products (PPCPs), endocrine disrupting<br />
compounds (EDCs), and other trace organic compounds, have<br />
been a significant aspect of environmental research efforts for<br />
decades (Snyder et al., 2003). However, adverse impacts on<br />
aquatic ecosystems and increased public awareness of trace<br />
organic compounds in water supplies have stimulated recent<br />
interest in this issue (Snyder et al., 2003). Due to rapid population<br />
growth and increasingly stressed water supplies,<br />
particularly in semi-arid regions, many cities are turning to<br />
alternative sources of drinking water such as indirect potable<br />
reuse. In some systems, wastewater effluent is having<br />
a greater impact on reservoir water quality due to drought and<br />
overdraft (Benotti et al., 2009a). Over time, these trends may<br />
water research 44 (2010) 493–504<br />
Available at www.sciencedirect.com<br />
journal homepage: www.elsevier.com/locate/watres<br />
* Corresponding author. Tel.: þ1 (702) 856 3666; fax: þ1 (702) 856 3647.<br />
E-mail address: Daniel.Gerrity@snwa.com (D. Gerrity).<br />
0043-1354/$ – see front matter ª 2009 Elsevier Ltd. All rights reserved.<br />
doi:10.1016/j.watres.2009.09.029<br />
This study evaluated a pilot-scale nonthermal plasma (NTP) advanced oxidation process<br />
(AOP) for the degradation of trace organic compounds such as pharmaceuticals and<br />
potential endocrine disrupting compounds (EDCs). The degradation of seven indicator<br />
compounds was monitored in tertiary-treated wastewater and spiked surface water to<br />
evaluate the effects of differing water qualities on process efficiency. The tests were also<br />
conducted in batch and single-pass modes to examine contaminant degradation rates and<br />
the remediation capabilities of the technology, respectively. Values for electrical energy per<br />
order (EEO) of magnitude degradation ranged from
494<br />
ozone, granular activated carbon (GAC), nanofiltration/reverse<br />
osmosis, and advanced oxidation) can achieve significant<br />
removal or transformation (Ternes et al., 2002; Westerhoff<br />
et al., 2005; Snyder et al., 2007). Advanced oxidation processes<br />
(AOPs) are often used to achieve further reductions in trace<br />
organics, target recalcitrant compounds, and comply with<br />
certain water reuse regulations (e.g., California Department of<br />
Public Health Title 22 requirements for NDMA and 1,4dioxane).<br />
The most common AOPs consist of UV/peroxide and<br />
ozone/peroxide, but emerging technologies such as<br />
nonthermal plasma (NTP) provide viable alternatives.<br />
This study focuses on a novel pilot-scale NTP reactor<br />
developed by Aquapure Technologies, Ltd. (Upper Galilee,<br />
Israel). The NTP uses high-voltage electrical pulses across<br />
fiber-like electrodes to form a corona discharge, which in turn<br />
generates UV light, ozone, and hydroxyl radicals (Locke et al.,<br />
2006). The light generated by the corona discharge spans<br />
wavelengths of approximately 250 nm to 1,000 nm with<br />
multiple peaks at wavelengths associated with radical<br />
formation (Locke et al., 2006). Although the light is lowintensity<br />
(Locke et al., 2006), it has been credited with<br />
a capacity for NDMA destruction (Even-Ezra et al., 2009),<br />
which is generally inefficient in ozone-based oxidation<br />
processes (Mitch et al., 2003).<br />
NTP is considered to be highly efficient because little energy<br />
is lost in heating the surrounding fluid, which allows the<br />
energy to be focused on the excitation of electrons (Pekarek,<br />
2003). In contrast to the electron beam technology, which<br />
introduces electrons from an external source, the NTP used in<br />
this study employs a corona discharge that excites electrons in<br />
the ambient air directly above the target water matrix<br />
(Pekarek, 2003). Other NTP configurations are also available,<br />
including those that create a corona discharge in aerosols or<br />
within the target water matrix (Pekarek, 2003; Locke et al.,<br />
2006). The level of treatment can be adjusted by changing the<br />
frequency (e.g., from 500 to 1,000 Hz) or voltage (up to 40 kV) of<br />
the electrical pulses. The ionizing capability of the electrical<br />
pulses generates singlet oxygen atoms, which in turn form<br />
ozone and hydroxyl radicals. One of the most significant<br />
benefits of NTP is the fact that oxidants can be generated<br />
without the addition of costly chemicals or UV lamps, which<br />
require cleaning and are hindered by high turbidity and matrix<br />
absorbance. However, the NTP technology has limited<br />
commercial availability and has not been tested in full-scale<br />
drinking water and wastewater treatment applications so its<br />
long-term reliability, efficiency, and effectiveness are still<br />
unknown. In addition, the NTP technology is currently limited<br />
to low-flow applications due to its thin-film configuration.<br />
Although plasma-based technologies have been studied<br />
extensively for the degradation of volatile organic compounds<br />
(VOCs) (Masuda et al., 1995; Hwang and Jo, 2005; Ighigeanu<br />
et al., 2005; Thevenet et al., 2007; Kim et al., 2008), many of<br />
these studies have focused on air quality and/or bench-scale<br />
configurations. Due to limitations associated with conventional<br />
AOPsdprimarily the need for chemical additiond<br />
studies of novel AOPs for water and wastewater treatment<br />
applications are gaining popularity. A recent pilot-scale study<br />
utilizing the Aquapure technology evaluated the efficacy of<br />
NTP in degrading trichloroethylene (TCE), N-nitrosodimethylamine<br />
(NDMA), 1,4-dioxane, and methyl tert-butyl<br />
water research 44 (2010) 493–504<br />
ether (MTBE) in batch experiments and single-pass remediation<br />
(Even-Ezra et al., 2009). The current study uses the same<br />
pilot-scale system to evaluate the degradation of PPCPs and<br />
EDCs in water and wastewater. Despite the lack of PPCP/EDC<br />
data for plasma-based technologies, other AOPs, particularly<br />
UV/peroxide (Huber et al., 2003; Pereira et al., 2007; Snyder<br />
et al., 2007; Benotti et al., 2009b), ozone/peroxide (Acero and<br />
von Gunten, 2001; Snyder et al., 2007), and photocatalysis<br />
(Benotti et al., 2009b; Westerhoff et al., 2009), have been tested<br />
extensively with respect to trace organic degradation. In their<br />
review of NTP applications, Locke et al. (2006) discussed the<br />
need for comparisons of NTP with more common AOPs to<br />
determine whether there is a net benefit associated with<br />
employing NTP technologies. Among other research needs,<br />
Locke et al. (2006) specifically identified energy consumption<br />
and applicability to different waste matrices as useful points<br />
of comparison. To this end, the objective of this study is to<br />
expand the NTP knowledge base by evaluating its efficacy and<br />
energy efficiency in degrading seven PPCPs and EDCs in<br />
tertiary-treated wastewater and spiked surface water.<br />
2. Experimental section<br />
2.1. Selection of wastewater-derived contaminants<br />
In order to narrow the scope of this research, a subset of the<br />
numerous compounds detected in previous occurrence<br />
studies was selected for evaluation. The indicator compounds<br />
were selected based on their magnitude and frequency of<br />
occurrence in water and wastewater (Snyder et al., 2007),<br />
varying physical/chemical characteristics and resulting<br />
susceptibility to treatment (Ternes et al., 2002; Westerhoff<br />
et al., 2005; Snyder et al., 2007), and ease of analytical methods<br />
(Trenholm et al., 2009). The seven compounds targeted in this<br />
study include meprobamate (anti-anxiety), dilantin (also<br />
known as phenytoin; anticonvulsant), primidone (anticonvulsant),<br />
carbamazepine (anticonvulsant), atenolol (betablocker),<br />
trimethoprim (antibiotic), and atrazine (herbicide<br />
and suspected EDC). Their structures and characteristics are<br />
described in Snyder et al. (2007). Although these compounds<br />
have generated considerable interest in the research, treatment,<br />
and regulatory arenas, only atrazine is currently regulated<br />
by the United States (U.S.) Environmental Protection<br />
Agency (EPA) at a maximum contaminant level (MCL) of 3 mg/L.<br />
2.2. Nonthermal plasma pilot unit<br />
The pilot-scale unit depicted in Fig. 1 is an ‘‘electrode-to-plate’’<br />
NTP prototype (Locke et al., 2006) developed by Aquapure<br />
Technologies, Ltd.; the pilot skid contains two reactors connected<br />
in series. Water flows in a thin film (z5 mm) along the<br />
stainless steel ground electrode (anode) where it is exposed to<br />
high-voltage electrical pulses. The electrical pulses from the<br />
generator have frequencies ranging from 500 to 1,000 Hz,<br />
maximum voltage of 8.0 kV, maximum current of 100 A,<br />
maximum energy of 1 J, and rise time of approximately 18 ns<br />
(Locke et al., 2006; Even-Ezra et al., 2009). Depending on the<br />
applied voltage and frequency, each reactor draws between 0.4<br />
and 1.0 kW, and the external pumping and controls draw
approximately 2.8 kW. The electrical pulses generate a corona<br />
discharge that stretches approximately 1 cm from the carbon<br />
fiber-like cathode to the water surface. The distance between<br />
the carbon electrode and water surface can be adjusted at<br />
multiple points along the reactor to level the system and<br />
optimize the process for a particular water matrix. The 1-cm<br />
distance was selected to achieve a consistent corona discharge<br />
and minimize sparking within the reactor, which would be<br />
destructive to the carbon fibers. Despite the harsh conditions<br />
experienced by the gas-phase electrode, no visible wear was<br />
detected on the carbon fibers over the duration of the experiments.<br />
With continuous, long-term operation of the NTP<br />
reactor, it is likely that the carbon fibers would eventually<br />
require replacement, but this life span is currently unknown.<br />
After traveling through the first reactor, the water is<br />
collected in a storage tank before being pumped into the<br />
ozone injector system or the second reactor. The ozone<br />
injector system combines a portion of the treated water<br />
from the first reactor with ozone-rich air (z2 g/m 3 ) from<br />
the reactor headspace using a Venturi inductor. For the<br />
single-pass configuration, this mixture is collected in<br />
another storage tank for additional ozone contact time<br />
before being mixed and discharged with effluent from the<br />
first reactor. Excess air from the ozone contactor is then<br />
drawn out of the system by vacuum and quenched in an<br />
ozone destruct unit. The hydraulic residence times in the<br />
reactor, storage tank, and ozone contactor are approximately<br />
0.5 min, 4–7.5 min, and 1.5 min, respectively. The<br />
hydraulic residence times in the reactor and storage tank<br />
depend on the process flow rate, whereas the residence<br />
time in the ozone contactor remains relatively constant<br />
because the injector flow rate (z40 L/min) is controlled by<br />
water research 44 (2010) 493–504 495<br />
Fig. 1 – Schematic of NTP pilot unit. The bottom figure illustrates a profile view of the electrodes within the reactor.<br />
a separate pump. The process is then repeated in the<br />
second reactor before being pumped out of the pilot skid.<br />
For the batch configuration, the effluent pump is replaced<br />
by a recirculation pump that allows the water to cycle<br />
continuously through a single reactor and ozone contactor.<br />
Additional details related to the NTP reactor used in this<br />
study are provided in Even-Ezra et al. (2009).<br />
2.3. Experimental design<br />
The study was divided into three experiments: (1) a batch<br />
experiment with 150 L of tertiary-treated wastewater (i.e.,<br />
grit removal, primary settling, activated sludge, secondary<br />
settling, and dual media filtration) at a recirculation rate of<br />
8.0 L/min, (2) a single-pass experiment with tertiary-treated<br />
wastewater at a flow rate of 15.5 L/min, and (3) a single-pass<br />
experiment with spiked surface water at a flow rate of<br />
11.4 L/min. For experiment (3), spiked surface water (Colorado<br />
River water from Lake Mead, NV) was pumped into the<br />
reactor from a 3,000-gallon polyethylene water tank<br />
(American Tank Company, Windsor, CA). The spiking solution<br />
was prepared by dissolving neat standards in water in<br />
order to avoid the introduction of radical scavenging<br />
solvents such as methanol. The spiking solution was added<br />
to the 3,000-gallon tank followed by recirculation for 24 h to<br />
facilitate homogenization. The two water types (wastewater<br />
and surface water) were selected to provide an assessment<br />
of NTP treatment efficiency for matrices with different<br />
organic loadings. The initial pH, total alkalinity, total organic<br />
carbon (TOC), UV254 absorbance, and trace organic<br />
compound concentrations for the three experiments are<br />
provided in Table 1. In all tables, ‘‘Batch’’ refers to the batch
496<br />
Table 1 – Initial water quality conditions.<br />
Parameter Batch (1) WW (2) a,c<br />
SSW (3) b,c<br />
pH 7.0 7.0 8.0<br />
Total Alkalinity 126 mg/L as 126 mg/L as 137 mg/L as<br />
CaCO3 CaCO3 CaCO3 TOC 5.9 mg-C/L 5.6 mg-C/L 2.6 mg-C/L<br />
UV254 Abs. 0.119 cm 1<br />
N/A 0.036 cm 1<br />
Meprobamate 275 ng/L 256 25 e ng/L 933 15 ng/L<br />
Dilantin 119 ng/L 165 10 ng/L 1,005 74 ng/L<br />
Primidone 124 ng/L 146 6 ng/L 1,375 96 ng/L<br />
Carbamazepine 176 ng/L 219 9 ng/L 1,600 115 ng/L<br />
Atenolol 378 ng/L 413 8 ng/L 1,018 57 ng/L<br />
Trimethoprim 36 ng/L 39 2 ng/L 1,200 82 ng/L<br />
Atrazine d<br />
N/A N/A 1,118 159 ng/L<br />
a WW (2) ¼ Single-pass wastewater experiment.<br />
b SSW (3) ¼ Single-pass spiked surface water experiment.<br />
c Four influent samples were taken throughout the experiment to<br />
calculate a representative ambient concentration and evaluate<br />
temporal variability in target compound concentrations.<br />
d Atrazine was spiked in the SSW experiment but was not detected<br />
in the wastewater.<br />
e Errors represent 1 standard deviation.<br />
wastewater experiment, ‘‘WW’’ refers to the single-pass<br />
wastewater experiment, and ‘‘SSW’’ refers to the single-pass<br />
spiked surface water experiment.<br />
The generator settings for the NTP pilot differed for each of<br />
the experiments, as described in Table 2. For the batch<br />
experiment (1), a single reactor/generator was operated at<br />
a frequency of 500 Hz and a voltage of 8.0 kV. To assess the<br />
degradation rate, ten samples were collected at generator<br />
(AOP-specific) energy consumption values ranging from 0 to<br />
7.3 kWh/m 3 . For the single-pass wastewater experiment (2),<br />
four different generator scenarios provided treatment levels<br />
ranging from 0.7 to 1.8 kWh/m 3 for the generators and 3.8 to<br />
4.8 kWh/m 3 for the overall pilot skid. Samples were collected<br />
from the effluent line exiting the pilot skid. Residual oxidants<br />
in each sample were quenched with calcium thiosulfate.<br />
During the wastewater experiments, the dissolved ozone<br />
concentration in the NTP reactor effluent ranged from<br />
approximately 0.05 mg/L to 0.10 mg/L depending on the<br />
generator settings. The single-pass surface water experiment<br />
(3) incorporated nine different generator scenarios ranging<br />
from 0.6 to 2.6 kWh/m 3 for the generators and 4.8 to 6.8 kWh/<br />
m 3 for the pilot skid. For the surface water, samples were<br />
taken from Reactor 1, Ozone Tank 1, Reactor 2, and the pilot<br />
skid effluent to evaluate the extent of contaminant degradation<br />
in each system component. During the surface water<br />
experiment, the dissolved ozone concentrations in the reactor<br />
effluent ranged from 0.25 to 0.50 mg/L depending on the<br />
generator settings.<br />
2.4. Analytical methods<br />
Ozone gas concentrations in the reactor headspace were<br />
determined by online measurements with an OMC20 ozone<br />
monitor and IN-2000-L2-LC ozone analyzer (IN USA, Norwood,<br />
MA). Dissolved ozone residuals were measured with the<br />
indigo trisulfonate method (Yates and Stenstrom, 2000). TOC<br />
and UV 254 absorbance were analyzed according to standard<br />
methods 5310B and 5910, respectively. Excitation-emission<br />
matrices (EEM) were also developed for the single-pass<br />
wastewater samples using a QuantaMaster UV-Vis QM4<br />
Steady State Spectrofluorometer (Photon Technology International,<br />
Inc., Birmingham, NJ).<br />
Trace organic compounds were extracted and analyzed<br />
using on-line solid phase extraction and liquid chromatography<br />
with tandem mass spectrometry (SPE-LC-MS/MS)<br />
according to Trenholm et al. (2009). All standards and reagents<br />
were of the highest purity commercially available. All<br />
Table 2 – Experimental test conditions.<br />
Experiment Generator 1 Generator 2 Total generator<br />
energy<br />
(1) Batch a<br />
Frequency<br />
(Hz)<br />
Voltage<br />
(kV)<br />
water research 44 (2010) 493–504<br />
Energy<br />
(kWh/m 3 )<br />
Frequency<br />
(Hz)<br />
Voltage<br />
(kV)<br />
Energy<br />
(kWh/m 3 )<br />
Total skid<br />
energy<br />
(kWh/m 3 ) (kWh/m 3 )<br />
Off Off 0.00 500 8.0 N/A N/A N/A<br />
(2) WW a – 1 Off Off 0.00 500 8.0 0.71 0.71 3.75<br />
WW – 2 800 8.0 1.04 Off Off 0.00 1.04 4.08<br />
WW – 3 600 8.0 0.87 500 8.0 0.71 1.58 4.62<br />
WW – 4 800 8.0 1.04 500 8.0 0.71 1.75 4.79<br />
(3) SSW b – 1 Off Off 0.00 600 5.9 0.63 0.63 4.78<br />
SSW – 2 Off Off 0.00 800 5.9 0.85 0.85 5.01<br />
SSW – 3 500 8.0 0.97 Off Off 0.00 0.97 5.12<br />
SSW – 4 Off Off 0.00 1000 5.9 1.07 1.07 5.23<br />
SSW – 5 1000 8.0 1.53 Off Off 0.00 1.53 5.68<br />
SSW – 6 500 8.0 0.97 600 5.9 0.63 1.60 5.75<br />
SSW – 7 500 8.0 0.97 1000 5.9 1.07 2.04 6.19<br />
SSW – 8 1000 8.0 1.53 600 5.9 0.63 2.16 6.31<br />
SSW – 9 1000 8.0 1.53 1000 5.9 1.07 2.60 6.75<br />
a The batch and WWTP experiments were conducted with ambient concentrations of trace organics in tertiary-treated wastewater.<br />
b The SSW experiments were conducted with raw surface water (Lake Mead) spiked with a suite of trace organics.
pharmaceuticals were obtained from Sigma-Aldrich (St. Louis,<br />
MO, USA). Meprobamate-d3 and trimethoprim-d9 were<br />
obtained from Toronto Research Chemicals (Ontario, Canada).<br />
Phenytoin-d10 and atenolol-d7 were obtained from C/D/N<br />
Isotopes (Pointe-Claire, Canada). Carbamazepine-d10 and primidone-d5<br />
were obtained from Cambridge Isotope Laboratories<br />
(Andover, MA, USA). All solvents were trace analysis grade<br />
from Burdick and Jackson (Muskegon, MI). Reagent water was<br />
obtained using a Milli-Q Ultrapure Water Purification System<br />
(Millipore, Bedford, MA, USA). All concentrated stocks were<br />
prepared in methanol and stored at 20 C, while mixed<br />
spiking solutions were prepared in reagent water and stored<br />
at 4 C.<br />
On-line SPE was performed using a SymbiosisÔ Pharma<br />
(Spark Holland, Emmen, the Netherlands) automated SPE in<br />
XLC mode. Briefly, a 10-mL volume of sample was measured<br />
in a volumetric flask at which time isotopically-labeled<br />
standards were spiked at 100 ng/L. This provided sufficient<br />
sample volume for duplicates, matrix spikes, and dilutions, if<br />
necessary. A 1.5-mL aliquot of each sample was then<br />
transferred to a 2-mL autosampler vial, although only 1.0 mL<br />
was used for extractions. Extractions were performed using<br />
Waters Oasis HLB Prospekt cartridges (30 mm, 2.5 mg,<br />
10 1 mm, 96 tray) (Milford, MA). Prior to sample loading,<br />
each cartridge was sequentially conditioned with 1 mL of<br />
dichloromethane (DCM), MTBE, methanol, and reagent water<br />
(Milli-Q). Samples were loaded onto the SPE cartridge at<br />
1 mL/min after which the cartridge was washed with 1 mL of<br />
reagent water. After sample loading, the analytes were<br />
eluted from the SPE cartridge to the LC column with 200 mL<br />
methanol, using the LC peak focusing mode. A 5-mM<br />
ammonium acetate solution and methanol gradient was<br />
used for LC mobile phases with a flow rate of 800 mL/min.<br />
Analytes were separated using a 150 4.6 mm Luna C18(2)<br />
with a 5-mm particle size (Phenomenex, Torrance, CA).<br />
Method reporting limits (MRLs) were chosen at 3 to 5 times<br />
the method detection limit (MDL). MRLs were 10 ng/L for all<br />
compounds, except for atenolol, which was 25 ng/L. All<br />
analytes were quantified using isotope dilution. Stringent<br />
QA/QC protocols (i.e., matrix spikes, duplicate samples, field<br />
blanks, and laboratory blanks) were followed throughout the<br />
duration of the experiment. Using the on-line SPE method,<br />
the concentrations of duplicate samples rarely varied by<br />
greater than 5%.<br />
3. Results and discussion<br />
3.1. Experiment 1 – batch experiment with tertiarytreated<br />
wastewater<br />
Contaminant degradation profiles during the batch wastewater<br />
experiment are provided in Fig. 2. The initial ambient<br />
concentrations for the target contaminants ranged from<br />
36 ng/L for trimethoprim to as high as 378 ng/L for atenolol<br />
(Table 1). Although included in the analytical method, atrazine<br />
was below the method reporting limit (MRL) for both<br />
wastewater experiments. As indicated in Table 2, the generator<br />
was operated at a frequency of 500 Hz, a voltage of 8.0 kV,<br />
and a recirculation rate of 8.0 L/min.<br />
water research 44 (2010) 493–504 497<br />
C/C 0<br />
1.0<br />
0.8<br />
0.6<br />
0.4<br />
0.2<br />
0.0<br />
0 1 2 3 4 5 6 7 8<br />
Generator Energy Consumption (kWh/m 3 )<br />
Meprobamate Dilantin Primidone Carbamazepine Atenolol Trimethoprim<br />
Fig. 2 – Contaminant degradation profiles from the batch<br />
experiment (1) with ambient concentrations of trace<br />
organic compounds in tertiary-treated wastewater. This<br />
graph illustrates the relationship between contaminant<br />
degradation and generator (AOP-specific) energy<br />
consumption. The dashed lines represent samples that<br />
were below the MRL for those compounds.<br />
The degradation of each detected contaminant was<br />
assumed to be pseudo first order based on a plot of ln(C/C 0)<br />
versus generator energy consumption. Degradation rates, 95%<br />
confidence intervals, and R 2 values are provided in Table 3.<br />
Due to the low initial trimethoprim concentration, the extent<br />
of degradation was limited, as indicated by the dashed line<br />
representing the MRL in Fig. 2. Despite the limited degradation,<br />
it is apparent that trimethoprim and carbamazepine are<br />
highly susceptible to NTP treatment. This result is consistent<br />
with previous studies assessing oxidation strategies for trace<br />
organic compounds. Westerhoff et al. (2005) reported nearly<br />
100% degradation of trimethoprim and carbamazepine with<br />
standard chlorination and ozonation practices. Dilantin and<br />
atenolol degraded more slowly during the NTP treatment,<br />
which is also consistent with oxidation by chlorine or ozone<br />
(Westerhoff et al., 2005). Finally, meprobamate has been<br />
shown to be highly resistant to chlorination and ozonation in<br />
comparison to other trace organic compounds (Westerhoff<br />
et al., 2005). This conclusion is supported by the NTP results in<br />
that meprobamate required nearly double the energy to<br />
Table 3 – Pseudo first order rate constants for batch<br />
experiment (1) in tertiary-treated wastewater.<br />
Compound k (m 3 /kWh) 95% CI b<br />
R 2<br />
Meprobamate 0.36 0.03 0.98 9<br />
Dilantin 0.52 0.15 0.92 5<br />
Primidone 0.40 0.15 0.88 5<br />
Carbamazepine 1.03 0.35 0.93 4<br />
Atenolol 0.61 0.22 0.88 5<br />
Trimethoprim 0.71 N/A 0.92 2<br />
Atrazine a<br />
N/A N/A N/A N/A<br />
a Atrazine concentrations were below MRL in all samples.<br />
b 95% CI ¼ 95% confidence interval.<br />
N
498<br />
achieve a specified level of degradation. Primidone was not<br />
quite as resistant as meprobamate, but its degradation was<br />
slower and more variable than the other compounds.<br />
With respect to bulk organic parameters, there was no<br />
significant change in TOC over the duration of the experiment.<br />
The consistent TOC level was expected considering that<br />
mineralization is highly energy intensive in hydroxyl radicaldominated<br />
processes (Gerrity et al., 2009), and high CT values<br />
are required in ozone-dominated processes (Rosal et al., 2009).<br />
The limited energy consumption, low dissolved ozone<br />
concentrations, and limited ozone contact time were likely<br />
insufficient to induce measurable organic mineralization in<br />
the NTP reactor.<br />
Although there was no reduction in TOC, UV254 absorbance<br />
was reduced by more than 65%; this change was correlated to<br />
the degradation of target compounds in Fig. 3 as a potential<br />
surrogate measure of process efficacy. Quantifying trace<br />
organic compounds requires specialized equipment and<br />
training, and the analytical methods are very expensive and<br />
time-consuming. For this reason, it is generally impractical for<br />
water and wastewater utilities to employ trace contaminant<br />
monitoring programs for their systems. Implementation of<br />
this type of surrogate framework (i.e., reduction in UV 254<br />
absorbance in lieu of trace contaminant degradation) would<br />
allow utilities to assess the performance of certain treatment<br />
processes for emerging contaminants. For oxidation<br />
processes, UV254 absorbance is an easily measured parameter<br />
that appears to show consistent correlations with the degradation<br />
of trace organic compounds (Wert et al., 2009). Fig. 3<br />
provides support for this surrogate framework because it<br />
demonstrates strong correlations between the degradation of<br />
the target compounds and reduction in UV 254 absorbance. The<br />
slopes of the regression lines indicate that relative contaminant<br />
degradation (i.e., percent degradation) is actually more<br />
rapiddmore than twice as fast for nearly all compoundsdthan<br />
the reduction in UV254 absorbance. The slopes<br />
also show a general agreement with the degradation profiles<br />
% Degradation of Trace Organic Compounds<br />
100%<br />
80%<br />
60%<br />
40%<br />
20%<br />
water research 44 (2010) 493–504<br />
in Fig. 2. Although trimethoprim was expected to degrade<br />
rapidly, its slope of 6.25 is likely an outlier and artifact of its<br />
small sample size. With the exception of trimethoprim, the<br />
magnitude of the slopes and the negative intercepts are<br />
similar to those reported in the Wert et al. (2009) ozonation<br />
study. The negative intercepts indicate that there are initial<br />
reductions in UV254 absorbance that precede degradation of<br />
the target compounds. This may be attributable to initial<br />
oxidant/radical scavenging by the bulk organic matter.<br />
Values for electrical energy per order (EEO) of magnitude<br />
degradation (Bolton et al., 1996), which is often measured in<br />
kWh/m 3 -log, are also provided in Table 4. Since the batch<br />
experiment involved a continuous treatment, the EEO values<br />
were calculated based on first order degradation rates and<br />
linear regression. In other words, a linear regression was<br />
generated for ln(C/C0) against generator energy consumption,<br />
and the EEO was calculated as ln(0.1) divided by the slope of<br />
the linear regression. The EEO values ranged from 2.2 kWh/m 3<br />
for carbamazepine to 6.4 kWh/m 3 for meprobamate. It is<br />
important to note that the batch EEO values may be overestimates<br />
because the batch system requires approximately<br />
10–15 min to achieve its maximum ozone concentration after<br />
starting the generator.<br />
3.2. Experiment 2 – single-pass experiment with<br />
tertiary-treated wastewater<br />
Contaminant degradation profiles for the single-pass wastewater<br />
experiment are illustrated in Fig. 4. The initial ambient<br />
concentrations for the target contaminants were similar to<br />
the batch experiment and ranged from 39 ng/L for trimethoprim<br />
to as high as 413 ng/L for atenolol (Table 1). As indicated<br />
in Table 2, this experiment involved a single-reactor<br />
configuration and both reactors operating in series at a flow<br />
rate of 15.5 L/min. All scenarios for this experiment involved<br />
voltages of 8.0 kV. The reactor consumed 0.7 kWh/m 3 with<br />
Generator 2 operating at a frequency of 500 Hz during the<br />
Trimethoprim 6.25 -90% 1.00 2<br />
0%<br />
0% 20% 40% 60% 80% 100%<br />
% Reduction in UV 254 Absorbance<br />
Linear Regression Models:<br />
Contaminant Slope Intercept R 2 N<br />
Meprobamate 1.56 -14% 0.97 8<br />
Dilantin 2.09 -22% 0.99 4<br />
Primidone 2.08 -28% 0.98 4<br />
Carbamazepine 2.75 -30% 1.00 3<br />
Atenolol 2.68 -44% 0.99 4<br />
Fig. 3 – Correlation between the degradation of target compounds and UV 254 absorbance during the batch experiment (1).<br />
This analysis excludes samples that were below the MRL and the final five carbamazepine samples (>3 kWh/m 3 ; see Fig. 2)<br />
due to high variability near the MRL. The slopes were all significant at the 0.03 level or better, and intercepts were all<br />
significant at the 0.07 level or better.
Table 4 – AOP-specific EEO values (kWh/m 3 -log).<br />
Contaminant (1) (2) (3) UV c<br />
Batch<br />
SSW<br />
a<br />
Min.<br />
WW b<br />
Med.<br />
WW b<br />
Max.<br />
WW b<br />
Min.<br />
SSW b<br />
Med.<br />
SSW b<br />
Max.<br />
SSW b<br />
WW-1 scenario, and the reactor consumed 1.0 kWh/m 3 with<br />
Generator 1 operating at a frequency of 800 Hz during the<br />
WW-2 scenario. WW-3 and WW-4 involved both reactors<br />
operating in series at 600/500 Hz (1.6 kWh/m 3 ) and 800/500 Hz<br />
(1.8 kWh/m 3 ), respectively. The energy consumption values<br />
for the entire skid, which include energy related to the<br />
generators, pumps, and controls, are also provided for each<br />
scenario.<br />
The relative degradation rates were identical to those of<br />
the batch experiment. For example, carbamazepine and<br />
trimethoprim experienced the most rapid degradation, and<br />
primidone and meprobamate were the most recalcitrant<br />
compounds. At the highest level of energy consumption<br />
(1.8 kWh/m 3 ), every compound except meprobamate experienced<br />
at least 70% degradation. The evaluation of<br />
C/C 0<br />
UV/H 2O 2 c,d<br />
SSW<br />
UV/TiO 2 c,e<br />
SSW<br />
Meprobamate 6.4 4.6 10 14 2.1 3.5 5.3 6.6 1.0 6.8<br />
Dilantin 4.4 2.7 3.1 3.5 1.1 2.0 3.1 2.1 1.0 2.2<br />
Primidone 5.8 2.8 4.3 4.8 1.1 2.2 3.3 3.7 0.6 3.9<br />
Carbamazepine 2.2 1.3 1.8 2.1
500<br />
Fig. 5 – Excitation-emission matrices as a function of generator energy consumption for the single-pass experiment (2) with<br />
tertiary-treated wastewater. The regions are described as follows: (I) Aromatic protein, (II) aromatic protein II, (III) fulvic<br />
acid-like, (IV) soluble microbial byproduct-like, and (V) humic acid-like (Chen et al., 2003).<br />
scope of this study, but future studies should address this<br />
issue in greater detail.<br />
3.3. Experiment 3 – single-pass experiment with spiked<br />
surface water<br />
Contaminant degradation profiles, including that of atrazine,<br />
for the single-pass spiked surface water experiment are<br />
illustrated in Fig. 6. The initial spiked concentrations were<br />
much higher than the ambient wastewater concentrations as<br />
all of the compounds were spiked at greater than 900 ng/L<br />
(Table 1). Similar to the single-pass wastewater experiment,<br />
this experiment involved a single-reactor configuration and<br />
both reactors operating in series at a flow rate of 11.4 L/min<br />
(Table 2). For this experiment, Generator 1 was operated at<br />
a voltage of 8.0 kV, and Generator 2 was operated at a voltage<br />
of 5.9 kV. The generator and overall skid energy consumption<br />
values ranged from 0.6 to 2.6 kWh/m 3 and 4.8 to 6.8 kWh/m 3 ,<br />
respectively.<br />
The relative degradation rates were similar to those of the<br />
wastewater experiments. For example, meprobamate was the<br />
most difficult to degrade, while trimethoprim and carbamazepine<br />
rapidly dropped below the MRL in all treated samples.<br />
The lower organic loading of the surface water (i.e., lower<br />
initial TOC concentration and UV254 absorbance), which<br />
water research 44 (2010) 493–504<br />
resulted in a higher residual ozone concentration (0.25 to<br />
0.50 mg/L), likely contributed to the improved degradation. In<br />
contrast to the wastewater experiments, atenolol degradation<br />
was more similar to trimethoprim and carbamazepine, while<br />
primidone degradation was more similar to that of dilantin.<br />
As expected, the degradation of atrazine, which is considered<br />
a recalcitrant compound (Westerhoff et al., 2005), was almost<br />
identical to that of meprobamate. For all of the compounds<br />
with concentrations above the MRL, the degradation peaked at<br />
about 70% for the most recalcitrant compounds and 85% for<br />
the moderately recalcitrant compounds after reaching an<br />
energy consumption of 1.6 kWh/m 3 . At the highest energy<br />
level, the effluent concentrations of the residual compounds<br />
ranged from 140 ng/L for dilantin to 390 ng/L for atrazine.<br />
During Experiment 3, there was also greater fluctuation for<br />
several time points. Since the samples represent discrete,<br />
rather than continuous (i.e., batch), treatment levels, one<br />
would anticipate greater variability, but there was still a relatively<br />
consistent decrease in contaminant concentration over<br />
the range of energy values tested.<br />
The EEO values for all of the compounds were dramatically<br />
lower in the surface water experiment than the wastewater<br />
experiments, presumably due to the lower scavenging<br />
capacity of the surface water. Meprobamate and atrazine had<br />
the highest EEOs with maximum values of 5.3 and 6.3 kWh/
C/C 0<br />
1<br />
0.8<br />
0.6<br />
0.4<br />
0.2<br />
0<br />
0.0 0.5 1.0 1.5 2.0 2.5 3.0<br />
m 3 -log, respectively. The maximum values generally occurred<br />
during the experiments when one or both of the generators<br />
were operating at 1,000 Hz. Although operating both generators<br />
at 1,000 Hz achieved the highest level of degradation,<br />
increasing the pulse frequency generally yielded diminishing<br />
rates of return with respect to efficiency. Operating Generator<br />
2 at a frequency of 600 Hz provided the most efficient treatment<br />
conditions for all of the target compounds. However,<br />
Generator 1 had been operated immediately prior to that<br />
scenario, so the residual ozone from Reactor 1 likely contributed<br />
to this increased efficiency. Although this overestimates<br />
the anticipated degree of degradation, it indicates that it may<br />
be beneficial to incorporate ozone contact tanks before and<br />
after each reactor to maximize degradation. Based on this<br />
Generator Energy Consumption (kWh/m 3 )<br />
Meprobamate Dilantin Primidone Carbamazepine Atenolol Trimethoprim Atrazine<br />
Fig. 6 – Contaminant degradation profiles from the single-pass experiment (3) with spiked surface water. This graph<br />
illustrates the relationship between contaminant degradation and generator (AOP-specific) energy consumption.<br />
Concentration (ng/L)<br />
1600<br />
1400<br />
1200<br />
1000<br />
800<br />
600<br />
400<br />
200<br />
0<br />
water research 44 (2010) 493–504 501<br />
observation, the median values are likely the most<br />
appropriate EEO estimates since they always corresponded<br />
to both reactors operating simultaneously. For the<br />
compounds with reportable concentrations, the median<br />
EEO values ranged from 1.0 to 3.7 kWh/m 3 for atenolol and<br />
atrazine, respectively.<br />
A subset of samples was also analyzed to assess the<br />
contribution of the system components (i.e., reactor vs. ozone<br />
contactor) to contaminant degradation. The results from one<br />
set of samples (SSW-8) are provided in Fig. 7; the other sample<br />
sets yielded similar trends and therefore are not shown. Based<br />
on these results, it appears that the ozone contactors provided<br />
the majority of the degradation for the more recalcitrant<br />
compounds, including meprobamate, dilantin, primidone,<br />
Influent Reactor 1 Ozone 1 Reactor 2 Effluent<br />
Sample Location<br />
Meprobamate Dilantin Primidone Carbamazepine Atenolol Trimethoprim Atrazine<br />
Fig. 7 – Comparison of the pilot system components in degrading the target compounds during the single-pass experiment<br />
(3) with spiked surface water. This data refers to samples from SSW-8 with generator and skid energy consumption values<br />
of 2.2 and 6.3 kWh/m 3 , respectively. The small increase in concentrations from Ozone 1 to Reactor 2 is explained by mixing<br />
of Reactor 1 and Ozone 1 effluent prior to passage into Reactor 2 (see Fig. 1).
502<br />
% Degradation of Trace Organic Compounds<br />
100%<br />
80%<br />
60%<br />
40%<br />
20%<br />
Linear Regression Models:<br />
Contaminant Slope Intercept R 2<br />
N<br />
Meprobamate 2.40 0 0.72 9<br />
Dilantin 1.35 45% 0.84 9<br />
Primidone 1.52 40% 0.81 9<br />
Atrazine 2.27 0 0.74 9<br />
0%<br />
0% 5% 10% 15% 20% 25% 30% 35%<br />
atenolol, and atrazine. However, carbamazepine and<br />
trimethoprim experienced rapid degradation in the reactor<br />
itself with the additional oxidation pathways (i.e., UV light,<br />
hydroxyl radicals, and ozone).<br />
The spiked surface water experiment was also evaluated<br />
based on general water quality parameters. For example, the<br />
initial bromide concentration in the surface water was<br />
approximately 62 mg/L, and additional analyses indicated that<br />
bromate increased only slightly (maximum of 2.4 mg/L) above<br />
the ambient surface water concentration of 1.7 mg/L. Similar<br />
to the wastewater experiments, there was no significant<br />
change in TOC over the duration of the experiment, but the<br />
reduction in UV254 absorbance was correlated to the degradation<br />
of the target contaminants in Fig. 8. In a manner<br />
similar to Fig. 3, Fig. 8 illustrates strong positive correlations<br />
between contaminant degradation and reduction in UV 254<br />
absorbance. Furthermore, the degradation of target contaminants<br />
was still more rapid than the reduction in UV 254<br />
absorbance. In contrast to the batch wastewater experiment<br />
and data provided in Wert et al. (2009), the regression lines<br />
for dilantin and primidone have positive intercepts, while the<br />
intercepts for meprobamate and atrazine were found to be<br />
insignificant ( p > 0.09). Wert et al. (2009) proposed that the<br />
correlations between contaminant degradation and UV254<br />
absorbance were independent of wastewater quality, but it<br />
appears that the correlations differ when the matrices<br />
themselves are drastically different (i.e., wastewater vs.<br />
surface water). Specifically, the positive intercepts for<br />
dilantin and primidone in the surface water experiment<br />
indicate that these target compounds started to degrade<br />
(z40–45%) prior to any reduction in UV 254 absorbance. In<br />
addition to contributing to their smaller slopes, this may<br />
indicate that there was less oxidant/radical scavenging and<br />
that the bulk organic matter was more recalcitrant in the<br />
surface water compared to the wastewater. It is also interesting<br />
to note that meprobamate has the smallest absolute<br />
intercept for the wastewater experiment and an insignificant<br />
water research 44 (2010) 493–504<br />
% Reduction in UV254 Absorbance<br />
Meprobamate Dilantin Primidone Atrazine<br />
Fig. 8 – Correlation between the degradation of detected target compounds and UV254 absorbance during the single-pass<br />
experiment (3) with spiked surface water. The slopes were all significant at the 0.001 level, the intercepts for primidone and<br />
dilantin were significant at the 0.001 level, and the intercepts for meprobamate and atrazine were considered insignificant<br />
( p > 0.09; regressions for meprobamate and atrazine were forced through the origin).<br />
intercept for the surface water experiment, and atrazine also<br />
has an insignificant intercept for the surface water experiment.<br />
Therefore, the initial degradation of these highly<br />
resistant compounds is more similar to the initial degradation<br />
of the bulk organic matter. For the surface water<br />
experiments, this supports the theory that the aromatic<br />
organic matter in the surface water was more recalcitrant<br />
than that of the wastewater. Based on these observations, it<br />
appears that the UV 254 surrogate framework is still valid, but<br />
there may be differences in the correlations for dissimilar<br />
water matrices.<br />
Table 4 provides EEO values for the three experiments in<br />
this study, and it also provides EEO values from Benotti et al.<br />
(2009b), which evaluated the efficacy of photolysis, UV/H2O2,<br />
and photocatalysis in degrading PPCPs and EDCs. Benotti et al.<br />
(2009b) provides an interesting point of comparison since the<br />
water matrix used for that study (spiked surface water from<br />
Lake Mead), target compounds, and analytical methods were<br />
identical. As indicated by the data, NTP was generally more<br />
efficient than photolysis and photocatalysis for each of the<br />
compounds, but was either comparable or slightly less efficient<br />
than UV/H2O2 at a peroxide dose of 10 mg/L. It is<br />
important to note that the NTP technology does not require<br />
UV lamps or costly peroxide feeds to generate the oxidative<br />
species. Therefore, NTP may be a viable alternative to the<br />
more common UV/H2O2 process despite the slightly higher<br />
energy requirements for some compounds. Although the NTP<br />
achieved comparable efficiencies to those of UV/H 2O 2 in these<br />
experiments, identification of the best alternative for<br />
a particular application also depends on capital costs and<br />
reliability. Since the NTP technology has not been fully<br />
commercialized, it is difficult to compare its capital costs and<br />
reliability with more established technologies. Future evaluations<br />
of large-scale NTP reactors, including long-term<br />
studies, will provide valuable information in determining how<br />
NTP compares to more established technologies on a life-cycle<br />
basis.
4. Conclusions<br />
With respect to treatment efficacy, the pilot-scale NTP system<br />
demonstrated rapid degradation of the target compounds.<br />
The relative degradation rates were consistent between the<br />
three phases of the study with carbamazepine and trimethoprim<br />
as the most susceptible compounds and meprobamate,<br />
primidone, and atrazine as the most recalcitrant compounds.<br />
Atenolol and dilantin were generally more resistant to<br />
oxidation than carbamazepine and trimethoprim. These<br />
similar relative degradation rates justify the use of indicator<br />
compounds in evaluations of PPCP and EDC treatment. The<br />
EEO values for the optimal scenarios were generally less than<br />
5 kWh/m 3 -log for all of the target compounds, particularly in<br />
the spiked surface water with a lower organic loading. With<br />
process optimization (i.e., optimization of electrode height for<br />
each water matrix) and limited modifications to the system,<br />
including more effective use of ambient ozone, the EEO values<br />
could likely be reduced further.<br />
More common AOPs, including UV/H 2O 2 and ozone/H 2O 2,<br />
require significant chemical addition and residual H 2O 2<br />
quenching, which represent a significant portion of their<br />
operational costs. The primary benefit of the NTP AOP is the<br />
ability to generate UV light, ozone, and hydroxyl radicals<br />
without chemical addition or the use of UV lamps. However,<br />
the true capital costs and reliability of large-scale NTP reactors<br />
for water and wastewater treatment are still relatively unclear.<br />
Also, since oxidation by molecular ozone may be the dominant<br />
mechanism for many compounds, future studies should<br />
provide direct comparisons of NTP with conventional ozone<br />
generators to compare system performance. These expanded<br />
studies are necessary before one can fully determine the<br />
applicability of the novel NTP technology for water and<br />
wastewater treatment. However, these preliminary results<br />
indicate that NTP may be a viable alternative to more common<br />
AOPs due to its chemical-free operation and comparable<br />
energy requirements for trace contaminant degradation.<br />
Acknowledgments<br />
We would like to thank the SNWA Applied Research and<br />
Development Center staff, particularly Janie Zeigler, Shannon<br />
Ferguson, Christy Meza, and Elaine Go, for assisting with the<br />
various analyses. We would also like to thank Dvir Solnik, Itay<br />
Even-Ezra, Gal Bitan, and Shahar Nuriel of Aquapure Technologies<br />
for use of the nonthermal plasma pilot unit and for<br />
providing crucial training and guidance during the study.<br />
references<br />
Acero, J.L., von Gunten, U., 2001. Characterization of oxidation<br />
processes: ozonation and the AOP O 3/H 2O 2. J. AWWA 93 (10),<br />
90–100.<br />
Benotti, M.J., Stanford, B.D., Snyder, S.A., 2009a. Increased loading<br />
of pharmaceuticals and endocrine disrupting compounds to<br />
source drinking water due to climate change. J. Environ. Qual.<br />
Accepted.<br />
water research 44 (2010) 493–504 503<br />
Benotti, M.J., Stanford, B.D., Wert, E.C., Snyder, S.A., 2009b.<br />
Evaluation of a photocatalytic reactor membrane pilot system<br />
for the removal of pharmaceuticals and endocrine disrupting<br />
compounds from water. Water Res. 43 (6), 1513–1522.<br />
Benotti, M.J., Trenholm, R.A., Vanderford, B.J., Holaday, J.C.,<br />
Stanford, B.D., Snyder, S.A., 2009c. Pharmaceuticals and<br />
endocrine disrupting compounds in U.S. drinking water.<br />
Environ. Sci. Technol. 43 (3), 597–603.<br />
Bolton, J.R., Bircher, K.G., Tumas, W., Tolman, C.A., 1996. Figuresof-merit<br />
for the technical development and application of<br />
advanced oxidation processes. J. Adv. Oxid. Technol. 1 (1),<br />
13–17.<br />
Chen, W., Westerhoff, P., Leenheer, J.A., Booksh, K., 2003.<br />
Fluorescence excitation-emission matrix regional integration<br />
to quantify spectra for dissolved organic matter. Environ. Sci.<br />
Technol. 37 (24), 5701–5710.<br />
Even-Ezra, I., Mizrahi, A., Gerrity, D., Snyder, S., Salveson, A.,<br />
Lahav, O., 2009. Application of a novel plasma-based advanced<br />
oxidation process for efficient and cost effective destruction of<br />
refractory organics in tertiary effluents and contaminated<br />
groundwater. Desalin. Water Treat. Accepted.<br />
Gerrity, D., Mayer, B., Ryu, H., Crittenden, J., Abbaszadegan, M.,<br />
2009. A comparison of pilot-scale photocatalysis and<br />
enhanced coagulation for disinfection byproduct mitigation.<br />
Water Res. 43, 1597–1610.<br />
Huber, M.M., Canonica, S., Park, G., von Gunten, U., 2003.<br />
Oxidation of pharmaceuticals during ozonation and advanced<br />
oxidation processes. Environ. Sci. Technol. 37, 1016–1024.<br />
Hwang, Y.H., Jo, Y.M., 2005. Decomposition of odorous gases in<br />
a pilot-scale nonthermal plasma reactor. J. KOSAE 21 (2),<br />
57–65.<br />
Ighigeanu, D., Martin, D., Zissulescu, E., Macarie, R., Oproiu, C.,<br />
Cirstea, E., Iovu, H., Calinescu, I., Iacob, N., 2005. SO 2 and NO x<br />
removal by electron beam and electrical discharge induced<br />
non-thermal plasmas. Vacuum 77, 493–500.<br />
Kim, S.D., Cho, J., Kim, I.S., Vanderford, B.J., Snyder, S.A., 2007.<br />
Occurrence and removal of pharmaceuticals and endocrine<br />
disruptors in South Korean surface, drinking, and waste<br />
waters. Water Res. 41, 1013–1021.<br />
Kim, H., Han, J., Sakaguchi, Y., Minami, W., 2008. Simultaneous<br />
oxidation of NO x and SO 2 by a new non-thermal plasma<br />
reactor enhanced by catalyst and additive. Plasma Sci.<br />
Technol. 10 (1), 53–56.<br />
Kolpin, D.W., Furlong, E.T., Meyer, M.T., Thurman, E.M., Zaugg, S.<br />
D., Barber, L.B., Buxton, H.T., 2002. Pharmaceuticals,<br />
hormones, and other organic wastewater contaminants in U.<br />
S. streams, 1999–2000: A national reconnaissance. Environ.<br />
Sci. Technol. 36, 1202–1211.<br />
Locke, B.R., Sato, M., Sunka, P., Hoffmann, M.R., Chang, J.S., 2006.<br />
Electrohydraulic discharge and nonthermal plasma for water<br />
treatment. Ind. Eng. Chem. Res. 45, 882–905.<br />
Masuda, S., Hosokawa, S., Tu, X., Wang, Z., 1995. Novel plasma<br />
chemical technologiesdPPCP and SPCP for control of gaseous<br />
pollutants and air toxics. J. Electrostatics 34, 415–438.<br />
Mitch, W.A., Sharp, J.O., Trussell, R.R., Valentine, R.L., Alvarez-<br />
Cohen, L., Sedlak, D.L., 2003. N-nitrosodimethylamine (NDMA)<br />
as a drinking water contaminant: a review. Environ. Eng. Sci.<br />
20 (5), 389–404.<br />
Pekarek, S., 2003. Non-thermal plasma ozone generation. Acta<br />
Polytech. 43 (6), 47–51.<br />
Pereira, V.J., Linden, K.G., Weinberg, H.S., 2007. Evaluation of UV<br />
irradiation for photolytic and oxidative degradation of<br />
pharmaceutical compounds in water. Water Res. 41 (19),<br />
4413–4423.<br />
Rosal, R., Rodriguez, A., Perdigon-Melon, J.A., Petre, A., Garcia-<br />
Calvo, E., 2009. Oxidation of dissolved organic matter in the<br />
effluent of a sewage treatment plant using ozone combined<br />
with hydrogen peroxide (O 3/H 2O 2). Chem. Eng. J. 149, 311–318.
504<br />
Snyder, S.A., Wert, E.C., Lei, H., Westerhoff, P., Yoon, Y., 2007.<br />
Removal of EDCs and Pharmaceuticals in Drinking and<br />
Reuse Treatment Processes. Water Research Foundation,<br />
Denver, CO.<br />
Snyder, S.A., Westerhoff, P., Yoon, Y., Sedlak, D.L., 2003.<br />
Pharmaceuticals, personal care products, and endocrine<br />
disruptors in water: implications for the water industry.<br />
Environ. Eng. Sci. 20 (5), 449–469.<br />
Ternes, T.A., 1998. Occurrence of drugs in German sewage<br />
treatment plants and rivers. Water Res. 32 (11), 3245–3260.<br />
Ternes, T.A., Meisenheimer, M., McDowell, D., Sacher, F.,<br />
Brauch, H., Haist-Gulde, B., Preuss, G., Wilme, U., Zulei-<br />
Seibert, N., 2002. Removal of pharmaceuticals during drinking<br />
water treatment. Environ. Sci. Technol. 36, 3855–3863.<br />
Thevenet, F., Guaitella, O., Puzenat, E., Herrmann, J.M.,<br />
Rousseau, A., Guillard, C., 2007. Oxidation of acetylene by<br />
photocatalysis coupled with dielectric barrier discharge. Catal.<br />
Today 122, 186–194.<br />
water research 44 (2010) 493–504<br />
Trenholm, R.A., Vanderford, B.J., Snyder, S.A., 2009. On-line SPE<br />
LC-MS/MS analysis of pharmaceutical indicators in water:<br />
a green alternative to conventional methods. Talanta 79,<br />
1425–1432.<br />
Wert, E.C., Rosario-Ortiz, F.L., Snyder, S.A., 2009. Using ultraviolet<br />
absorbance and color to assess pharmaceutical oxidation<br />
during ozonation of wastewater. Environ. Sci. Technol. 43,<br />
4858–4863.<br />
Westerhoff, P., Yoon, Y., Snyder, S., Wert, E., 2005. Fate of<br />
endocrine-disruptor, pharmaceutical, and personal care<br />
product chemicals during simulated drinking water treatment<br />
processes. Environ. Sci. Technol. 39, 6649–6663.<br />
Westerhoff, P., Moon, H., Minakata, D., Crittenden, J., 2009.<br />
Oxidation of organics in retentates from reverse osmosis<br />
wastewater reuse facilities. Water Res. 43 (16), 3992–3998.<br />
Yates, R.S., Stenstrom, M.K., 2000. Gravimetric sampling<br />
procedure for aqueous ozone concentrations. Water Res. 34<br />
(4), 1413–1416.
Single-walled carbon nanotubes dispersed in aqueous media<br />
via non-covalent functionalization: Effect of dispersant<br />
on the stability, cytotoxicity, and epigenetic toxicity<br />
of nanotube suspensions<br />
Alla L. Alpatova a , Wenqian Shan a , Pavel Babica b , Brad L. Upham b,c ,<br />
Adam R. Rogensues a , Susan J. Masten a , Edward Drown d,1 , Amar K. Mohanty d,2 ,<br />
Evangelyn C. Alocilja e , Volodymyr V. Tarabara a, *<br />
a<br />
Department of Civil and Environmental Engineering, Michigan State University, East Lansing, MI 48824, USA<br />
b<br />
Center for Integrative Toxicology, Michigan State University, East Lansing, MI 48824, USA<br />
c<br />
Department of Pediatrics and Human Development, Michigan State University, East Lansing, MI 48824, USA<br />
d<br />
School of Packaging, Michigan State University, East Lansing, MI 48824, USA<br />
e<br />
Department of Biosystems and Agricultural Engineering, Michigan State University, East Lansing, MI 48824, USA<br />
article info<br />
Article history:<br />
Received 2 June 2009<br />
Received in revised form<br />
14 September 2009<br />
Accepted 17 September 2009<br />
Available online 30 September 2009<br />
Keywords:<br />
Single-walled carbon nanotubes<br />
Dispersion<br />
Non-covalent functionalization<br />
Cytotoxicity<br />
Epigenetic toxicity<br />
water research 44 (2010) 505–520<br />
Available at www.sciencedirect.com<br />
journal homepage: www.elsevier.com/locate/watres<br />
abstract<br />
As the range of applications for carbon nanotubes (CNTs) rapidly expands, understanding<br />
the effect of CNTs on prokaryotic and eukaryotic cell systems has become an important<br />
research priority, especially in light of recent reports of the facile dispersion of CNTs in<br />
a variety of aqueous systems including natural water. In this study, single-walled carbon<br />
nanotubes (SWCNTs) were dispersed in water using a range of natural (gum arabic,<br />
amylose, Suwannee River natural organic matter) and synthetic (polyvinyl pyrrolidone,<br />
Triton X-100) dispersing agents (dispersants) that attach to the CNT surface non-covalently<br />
via different physiosorption mechanisms. The charge and the average effective hydrodynamic<br />
diameter of suspended SWCNTs as well as the concentration of exfoliated SWCNTs<br />
in the dispersion were found to remain relatively stable over a period of 4 weeks. The<br />
cytotoxicity of suspended SWCNTs was assessed as a function of dispersant type and<br />
exposure time (up to 48 h) using general viability bioassay with Escherichia coli and using<br />
neutral red dye uptake (NDU) bioassay with WB-F344 rat liver epithelia cells. In the E. coli<br />
viability bioassays, three types of growth media with different organic loadings and salt<br />
contents were evaluated. When the dispersant itself was non-toxic, no losses of E. coli and<br />
WB-F344 viability were observed. The cell viability was affected only by SWCNTs dispersed<br />
using Triton X-100, which was cytotoxic in SWCNT-free (control) solution. The epigenetic<br />
toxicity of dispersed CNTs was evaluated using gap junction intercellular communication<br />
(GJIC) bioassay applied to WB-F344 rat liver epithelial cells. With all SWCNT suspensions<br />
except those where SWCNTs were dispersed using Triton X-100 (wherein GJIC could not be<br />
measured because the sample was cytotoxic), no inhibition of GJIC in the presence of<br />
SWCNTs was observed. These results suggest a strong dependence of the toxicity of<br />
* Corresponding author. Tel.: þ517 432 1755; fax: þ517 355 0250.<br />
E-mail address: tarabara@msu.edu (V.V. Tarabara).<br />
1<br />
Present address: Department of Chemical Engineering and Materials Science, Michigan State University, East Lansing, MI 48824, USA.<br />
2<br />
Present address: Department of Plant Agriculture, University of Guelph, 50 Stone Rd. E., Guelph, Ontario, N1 G 2W1 Canada.<br />
0043-1354/$ – see front matter ª 2009 Elsevier Ltd. All rights reserved.<br />
doi:10.1016/j.watres.2009.09.042
506<br />
1. Introduction<br />
The discovery (Iijima, 1991) and subsequent extensive characterization<br />
of carbon nanotubes (CNTs) have revealed a class<br />
of materials with extraordinary electrical, mechanical, and<br />
thermal properties (Tasis et al., 2006). The wider application of<br />
CNTs in electronic, optical, sensing, and biomedical fields has<br />
been impeded by the low solubility of as-produced CNTs in<br />
polar liquids and by the strong tendency of CNTs to aggregate<br />
due to hydrophobic–hydrophobic interactions (Lin et al., 2004).<br />
Dispersing CNTs and ensuring long-term stability of CNTs<br />
suspended in a liquid medium have proved especially challenging<br />
for aqueous systems.<br />
1.1. Dispersing CNTs in aqueous media<br />
Recent efforts on the development of efficient and facile<br />
methods of dispersing CNTs in aqueous media have been<br />
focused on the hydrophilization of CNT with molecules that<br />
bind to the CNT surface non-covalently (O’Connell et al., 2001;<br />
Bandyopadhyaya et al., 2002; Star et al., 2002; Islam et al., 2003;<br />
Moore et al., 2003; Didenko et al., 2005; Wang et al., 2005;<br />
Bonnet et al., 2007; Grossiord et al., 2007; Hyung et al., 2007; Liu<br />
et al., 2007). Such non-covalent functionalization has great<br />
promise as the modification-induced changes in the electronic<br />
and mechanical properties of CNTs are minimized<br />
(Yang et al., 2007). Various surfactants (Islam et al., 2003;<br />
Moore et al., 2003; Grossiord et al., 2007), synthetic polymers<br />
(e.g., polyvinyl pyrrolidone (O’Connell et al., 2001; Didenko<br />
et al., 2005), poly(ethylene glycol) (Vaisman et al., 2006), polyphosphazene<br />
(Park et al., 2006)), natural organic matter<br />
(NOM) (Hyung et al., 2007; Liu et al., 2007; Saleh et al., 2009),<br />
biomolecules (e.g., proteins (Karajanagi et al., 2006; Zong et al.,<br />
2007), aminoacids (Georgakilas et al., 2002), DNA (Enyashin<br />
et al., 2007)), and carbohydrates (e.g., cyclodextrines (Dodziuk<br />
et al., 2003), amylose (Bonnet et al., 2007), starch (Star et al.,<br />
2002), and GA (Bandyopadhyaya et al., 2002)) have been evaluated<br />
as dispersants for CNTs. Dispersion via non-covalent<br />
functionalization is based on the direct contact between<br />
a CNT and a dispersant molecule (Liu et al., 1998; Grossiord<br />
et al., 2007). Such modification of the CNT surface facilitates<br />
the disaggregation (i.e. debundling) of CNT bundles into<br />
smaller diameter bundles (Liu et al., 1998) or even individual<br />
CNTs (O’Connell et al., 2002; Hyung et al., 2007) and leads to<br />
the stabilization of suspended CNTs via steric or electrostatic<br />
repulsion mechanisms or both. (See Supporting Documentation<br />
(SD), Section S.1.1, for a brief review of mechanisms of<br />
non-covalent functionalization of CNTs.)<br />
The dispersion of CNTs in water has been enhanced by<br />
mixing (Didenko et al., 2005; Hyung et al., 2007), sonication<br />
(Bandyopadhyaya et al., 2002; Liu et al., 2007; Salzmann et al.,<br />
2007), or mixing followed by sonication (O’Connell et al., 2001,<br />
water research 44 (2010) 505–520<br />
SWCNT suspensions on the toxicity of the dispersant and point to the potential of noncovalent<br />
functionalization with non-toxic dispersants as a method for the preparation of<br />
stable aqueous suspensions of biocompatible CNTs.<br />
ª 2009 Elsevier Ltd. All rights reserved.<br />
2002; Star et al., 2002; Islam et al., 2003; Moore et al., 2003;<br />
McDonald et al., 2006; Grossiord et al., 2007). These treatments<br />
were applied both in the absence (Salzmann et al., 2007) and in<br />
the presence of solubilizing agents: NOM (Hyung et al., 2007),<br />
Triton X-100 (Islam et al., 2003), Triton X-405 (Chappell et al.,<br />
2009), PVP-1300 (Didenko et al., 2005), GA (Bandyopadhyaya<br />
et al., 2002), and starch (Star et al., 2002). A three-step<br />
approach to solubilizing single-walled carbon nanotubes<br />
(SWCNTs) with amylose was suggested by Kim et al. (Kim<br />
et al., 2004): dispersion of SWCNT in water by sonication followed<br />
by treatment with amylose in dimethylsulfoxide<br />
(DMSO)–H2O mixture, followed by sonication allowing for<br />
molecularly controlled encapsulation of CNTs.<br />
1.2. Stability of CNT suspensions in water<br />
Previous studies of the long-term changes in suspensions of<br />
dispersed CNTs have focused on monitoring changes in the<br />
concentration of suspended CNTs (Jiang and Gao, 2003; Tseng<br />
et al., 2006; Lee et al., 2007; Marsh et al., 2007). By measuring<br />
UV–vis absorption at certain wavelength: 253 nm (Jiang and<br />
Gao, 2003), 300 nm (Sinani et al., 2005), 500 nm (Bahr et al.,<br />
2001; Lee et al., 2007), 530 nm (Marsh et al., 2007), and 800 nm<br />
(Hyung et al., 2007) the change in the concentration of suspended<br />
CNTs with time was determined.<br />
Aqueous suspensions of non-functionalized CNTs are<br />
known to be unstable. There are considerable quantitative<br />
differences, however, in the reported stability data for nonfunctionalized<br />
CNTs. The concentration of unmodified multiwalled<br />
carbon nanotubes (MWCNTs) suspended in deionized<br />
water was reported to decline 86 % over 2 h in one study<br />
(Marsh et al., 2007) and only 50% over 500 h in another study<br />
(Jiang and Gao, 2003). The suspensions of unmodified SWCNTs<br />
in deionized water were found to completely precipitate after<br />
only 4 h (Tseng et al., 2006).<br />
It was found that non-covalent modification of CNT<br />
surface drastically improved the stability of CNT suspensions<br />
(Jiang and Gao, 2003; Tseng et al., 2006; Hyung et al., 2007; Lee<br />
et al., 2007; Marsh et al., 2007). The stability of the suspension<br />
of non-covalently functionalized CNTs was found to be<br />
a function of the type (Hyung et al., 2007; Lee et al., 2007) and<br />
concentration (Chappell et al., 2009) of the dispersant, CNT<br />
length (Marsh et al., 2007) and the presence of inorganic salts<br />
in the solution (Saleh et al., 2009). SWCNTs stabilized with<br />
oligothiophene-terminated poly(ethylene glycol) produced<br />
suspensions that were significantly more stable than CNTs<br />
dispersed in water with the aid of sodium dodecyl sulfate<br />
(SDS) and Pluronic F127 (Lee et al., 2007). CNT suspensions in<br />
model Suwannee River NOM (SRNOM) solutions and in<br />
Suwannee River water were found to be considerably more<br />
stable than suspensions of CNTs dispersed in 1% aqueous<br />
solution of SDS (Hyung et al., 2007). Chappell et al. showed
that the stability of MWCNTs dispersed using two types of<br />
humic acids, Triton X-405, Brij 35, and SDS was enhanced in<br />
dose-dependent manner (Chappell et al., 2009). In a study of<br />
the aggregation kinetics of MWCNTs in aquatic systems,<br />
Suwannee River humic acid was shown to significantly<br />
enhance stability of MWCNTs suspensions in the presence of<br />
monovalent and divalent salts (Saleh et al., 2009). While<br />
studying the stability of MWCNTs in water, Marsh and coworkers<br />
(Marsh et al., 2007) observed that wrapping of the<br />
annealed CNTs with 1% SDS or charge doping increased their<br />
stability in the suspension; the authors demonstrated that<br />
shorter CNTs form more stable dispersions than longer CNTs<br />
of the same diameter. In summary, there is growing evidence<br />
that by the appropriate choice of a dispersant that modifies<br />
CNT surface non-covalently, highly stable CNT suspensions<br />
can be produced. However, to date there have been no systematic<br />
investigations that correlated long-term changes in size and charge<br />
of CNTs dispersed via non-covalent functionalization with the<br />
dispersion stability.<br />
1.3. Toxicity of CNTs<br />
1.3.1. Toxicity of CNTs towards eukaryotic cells: cytotoxicity<br />
Most nanomaterial toxicity studies have been performed with<br />
mammalian cells, in particular with lung and skin cell<br />
cultures reflecting the understanding that the most likely<br />
routes of an organism’s exposure to nanomaterials are<br />
respiratory and dermal contact. With the development of<br />
methods of CNT dispersion in aqueous media, the assessment<br />
of the toxicity of such dispersed CNTs becomes very important<br />
in view of their increased mobility and potential to enter<br />
water supplies. While the toxicity of CNTs was studied with<br />
respect to the type of the CNT (MWCNT versus SWCNT) (Ding<br />
et al., 2005; Jia et al., 2005; Kang et al., 2008a), surface functionalization<br />
(Sayes et al., 2006; Yu et al., 2007; Meng et al.,<br />
2009; Yun et al., 2009), CNT length (Muller et al., 2005; Kang<br />
et al., 2008b; Simon-Deckers et al., 2008), exposure dose<br />
(Cherukuri et al., 2006; Flahaut et al., 2006; Sayes et al., 2006;<br />
Kang et al., 2007; Pulskamp et al., 2007; Yang et al., 2008, 2009;<br />
Ye et al., 2009), degree of purification (Fiorito et al., 2006;<br />
Flahaut et al., 2006; Elias et al., 2007; Simon-Deckers et al.,<br />
2008), and degree of dispersion (Wick et al., 2007), only very<br />
limited information exists on the biocompatibility of CNTs as<br />
a function of surface coating.<br />
Several hypotheses have been put forth to explain the likely<br />
pathways of CNT toxicity: (i) oxidative stress induced by the<br />
formation of reactive oxygen species (ROS) generated at the<br />
surface of CNTs (Manna et al., 2005; Yang et al., 2008; Ye et al.,<br />
2009); (ii) the presence of residual catalyst, which is used during<br />
the CNT manufacturing process (Kang et al., 2008a);<br />
(iii) physical contact between a CNT and a cell (Kang et al., 2007);<br />
or (iv) a combination of these factors (Kang et al., 2008a). The<br />
dispersant used for the stabilization of CNTs in suspension was<br />
suggested as another possible cause of the observed nanotube<br />
toxicity (Sayes et al., 2006) (SeeSD, Section S.1.2 for a brief<br />
review of possible mechanisms of CNT toxicity).<br />
In most studies of the toxicity of non-covalently functionalized<br />
CNTs the dispersants were surfactants. In vivo<br />
assessment of the toxicity of SWCNTs modified with Pluronic<br />
F108 administered intravenously to rabbits showed the<br />
water research 44 (2010) 505–520 507<br />
absence of the acute toxicity (Cherukuri et al., 2006). Pluronic<br />
F-127-coated MWCNTs did not affect cell viability, apoptosis,<br />
and ROS formation in mouse and human neuroblastoma cells<br />
(Vittorio et al., 2009), and mouse cerebral cortex (Bardi et al.,<br />
2009). In contrast, MWCNTs dispersed using Pluronic F68<br />
caused cell death, changes in cell size and complexity, ROS<br />
production, interleukin-8 (IL-8) gene expression and nuclear<br />
factor (NF)-jB activation (Ye et al., 2009). CNTs dispersed using<br />
Tween 80 were toxic to mesothelioma cells (Wick et al., 2007),<br />
led to an inflammation of murine allergic airway with<br />
augmented humoral immunity (Inoue et al., 2009), and<br />
induced inflammatory and fibrotic responses when intracheally<br />
administrated to rats (Muller et al., 2005). More ROS<br />
were observed upon exposure of human lung epithelial or<br />
primary bronchial epithelial cells to SWCNTs dispersed using<br />
dipalmitoylphosphatidylcholine, a major component of lung<br />
surfactant (Herzog et al., 2009), as compared to dipalmitoylphosphatidylcholine-free<br />
samples. Low toxicity was observed<br />
in in vivo experiments when SWCNTs were dispersed using<br />
Tween 80 and intravenously injected into mice (Yang et al.,<br />
2008). In another study, SWCNTs dispersed in 1% SDS aqueous<br />
solution showed no cytotoxicity with respect to the human<br />
alveolar epithelial cells (Wörle-Knirsch et al., 2006). CNTs<br />
dispersed using Pluronic PF-127 solution did not affect<br />
viability, apoptosis and ROS generation in the human neuroblastoma<br />
cells after 3 days of incubation; however, cell<br />
proliferation decreased as incubation time increased to 2<br />
weeks (Vittorio et al., 2009). In the only paper that mentioned<br />
the potential toxicity effect of the dispersant, the authors<br />
suggested that excess surfactant was responsible for the<br />
observed increase in toxicity; in this work, the controlled<br />
exposure of cells to 1% Pluronic F108 produced a 10% decrease<br />
in the cells viability (Sayes et al., 2006).<br />
Very little is known about the effect of non-covalent wrapping<br />
with dispersants other than surfactants on the biocompatibility of<br />
CNTs. In vitro cytotoxicity of CNTs wrapped with poly(methyl<br />
vinylketone) decorated with a-N-acetylgalactosamine (Chen<br />
et al., 2006), nano-1 peptide (Chin et al., 2007) and cholesterolend-capped<br />
poly(2-methacryloyloxyethyl phosphorylcholine)<br />
(Xu et al., 2008) were examined after their contact with human<br />
lung epithelial-like cells, normal skin fibroblasts and human<br />
umbilical vein endothelial cell line. No impact on cell growth<br />
and proliferation was demonstrated in all three cases. Cytotoxicity<br />
of GA-stabilized MWCNTs upon exposure to A549 cells<br />
was observed by LDH, XTT and MTT bioassays in (Simon-<br />
Deckers et al., 2008). Two possible reasons were hypothesized<br />
to be responsible for this effect: (i) increased availability of GAstabilized<br />
MWCNTs and (ii) different intercellular accumulation<br />
pathway of GA-stabilized MWCNTs as compared to when<br />
carbon nanotubes were exposed in bundles.<br />
1.3.2. Toxicity of CNTs towards eukaryotic cells: genotoxicity<br />
and epigenetic toxicity<br />
Most toxicological studies of CNTs focused on the evaluation of<br />
cytotoxicity (Sayes et al., 2006; Elias et al., 2007; Gutierrez et al.,<br />
2007; Kang et al., 2007; Wick et al., 2007; Kang et al., 2008a,b;<br />
Yang et al., 2008; Bardi et al., 2009; Herzog et al., 2009; Inoue et al.,<br />
2009; Kang et al., 2009; Vittorio et al., 2009; Yang et al., 2009).<br />
However, genotoxicity (Kang et al., 2008a; Di Sotto et al., 2009;<br />
Lindberg et al., 2009; Wirnitzer et al., 2009; Yang et al., 2009)and
508<br />
epigenetic toxicity (see SD, Section S.1.3) (Upham et al., 1994);<br />
(Trosko et al., 1998) are other possible causes of cell damage or<br />
other adverse effects. The genotoxicity of SWCNTs dispersed<br />
using fetal bovine serum at (5–10) mg/mL concentration, with<br />
respect to primary mouse embryo fibroblasts was demonstrated<br />
using Comet assay (Yang et al., 2009). In another recent<br />
study, CNTs (dispersed in BEGM cell culture medium and subjected<br />
to ultrasonication) induced a dose-dependent increase in<br />
DNA damage as indicated by Comet assay and caused a significant<br />
increase in micronucleated cells (micronucleus assay) in<br />
human bronchial epithelial cells (Lindberg et al., 2009). Kang<br />
et al. observed high levels of stress-related gene products in<br />
Escherichia coli upon its exposure to CNTs, with the quantity and<br />
magnitude of expression being much higher in the presence of<br />
SWCNTs (Kang et al., 2008a); in this study CNTs were either<br />
deposited on a PVDF membrane surface or dispersed in saline<br />
solution. No mutagenic effect was observed in Salmonella<br />
microcosme test with baytubes Ò (high purity MWCNTs<br />
agglomerates, sonicated 10 min in deionized water) (Wirnitzer<br />
et al., 2009) and in bacterial reverse mutation assay (Ames test)<br />
with Salmonella typhimurium TA 98 and TA 100 strains and<br />
with E. coli WP2uvrA strain exposed to MWCNTs dispersed in<br />
DMSO (Di Sotto et al., 2009). There have been no reports to date on the<br />
epigenetic toxicity of carbon nanomaterials.<br />
1.3.3. Toxicity of CNTs in prokaryotic cells<br />
Most toxicity studies have focused on the effect of CNTs on<br />
mammalian cell lines. Only limited information exists on the<br />
cytotoxic effects of CNTs towards bacterial cells. Recently,<br />
antimicrobial activity of SWCNTs (Kang et al., 2007, 2008a,<br />
2009) and MWCNTs (Kang et al., 2008a,b, 2009) either suspended<br />
in aqueous solution (Kang et al., 2007, 2008a) or<br />
deposited on the surface of a PVDF microfilter (Kang et al.,<br />
2007, 2008a,b, 2009) towards gram-negative bacteria (E. coli<br />
and Pseudomonas aeruginosa (Kang et al., 2007, 2008a,b, 2009)<br />
and gram-positive (Staphylococcus epidermidis and Bacillus subtilis<br />
(Kang et al., 2009)) bacteria was reported. It was suggested<br />
that membrane damage to the cells was caused by the direct<br />
physical contact between the CNT and cell (Kang et al., 2007)<br />
or by a combination of direct physical contact and oxidative<br />
stress (Kang et al., 2008a). No effect on the percentage of E. coli<br />
inactivation was observed upon exposure of SRNOM-stabilized<br />
SWCNTs as compared of SWCNTs dispersed in the<br />
absence of SRNOM (Kang et al., 2009). Significant antimicrobial<br />
activity of CNTs composite films against Staphylococcus aureus<br />
and Staphylococcus warneri was reported in a separated study<br />
(Narayan et al., 2005). In contrast to the findings of above<br />
studies, no inhibition of E. coli growth and proliferation was<br />
reported in a study where microchannel-structured MWCNTs<br />
scaffolds were immersed into the culture medium with the<br />
cells (Gutierrez et al., 2007).<br />
1.4. Effect of growth media on the toxicity of<br />
nanoparticles<br />
The existing literature highlights the importance of the<br />
growth media in cytotoxicity testing and links the apparent<br />
cytotoxic effect of the nanomaterial to the salt and organic<br />
content of the culture and related physicochemical characteristics<br />
of the nanomaterial (Lyon et al., 2005; Tong et al.,<br />
water research 44 (2010) 505–520<br />
2007). Lyon et al. showed that in media with high salt<br />
content, the size of nC60 aggregates tends to increase as<br />
compared to that observed in low salt media (Lyon et al.,<br />
2005). In another study, MWCNTs were found to stimulate<br />
growth of unicellular protozoan Tetrahymena pyriformis in<br />
growth medium, which contained proteose peptone, yeast<br />
extract, and glucose; the increase in growth was attributed to<br />
the formation of peptone-MWCNTs conjugates, which were<br />
taken up by the microorganism (Zhu et al., 2006). While the<br />
aforementioned studies indicate that salt and organic<br />
composition of the medium, in which exposure studies are<br />
performed, may influence the interaction of CNTs with<br />
bacteria, there have been no studies that comparatively evaluated<br />
CNT toxicity in growth media with different organic loadings and<br />
salt contents.<br />
1.5. Objectives of this study<br />
This study addressed some of the knowledge gaps identified<br />
above. Aqueous suspensions of SWCNTs functionalized by<br />
a range of non-covalently bound dispersants of natural (NOM,<br />
GA, amylose) and synthetic (PVP, Triton X-100) origin, were<br />
prepared and evaluated in terms of their physicochemical and<br />
toxicity properties. The study pursued the following<br />
objectives:<br />
(i) To evaluate the long-term stability and its physicochemical<br />
determinants for non-covalently functionalized SWCNTs as<br />
a function of dispersant type. The evolution of the concentration,<br />
size, and charge of suspended SWCNTs was<br />
studied over the period of 28 days.<br />
(ii) To assess time-dependent cytotoxicity of non-covalently functionalized<br />
SWCNTs with respect to bacteria and mammalian<br />
cells as a function of dispersant type and growth media. The<br />
effect of SWCNTs on E. coli was studied by measuring cell<br />
viability after 3 h, 24 h, and 48 h of exposure in three<br />
types of growth media with different organic loadings<br />
and salt contents. The cytotoxicity of dispersed SWCNTs<br />
towards mammalian cells was assessed in NDU bioassay<br />
with rat liver epithelial cells after 30 min and 24 h of<br />
incubation.<br />
(iii) To assess time-dependent epigenetic toxicity of non-covalently<br />
functionalized SWCNTs as a function of dispersant type. In this<br />
study, we evaluated epigenetic toxicity of the SWCNTs to<br />
rat liver epithelial cells as a function of dispersant type in<br />
GJIC bioassay after 30 min and 24 h of incubation.<br />
2. Approach<br />
The set of chemically diverse dispersants was chosen based<br />
on their demonstrated effectiveness in solubilizing CNTs<br />
(Bandyopadhyaya et al., 2002; O’Connell et al., 2002; Star et al.,<br />
2002; Islam et al., 2003; Moore et al., 2003; Hyung et al., 2007;<br />
Liu et al., 2007) and potential adverse effects (Burnette, 1960;<br />
Chourasia and Jain, 2004; Dayeh et al., 2004; Schmitt et al.,<br />
2008). NOM, GA and PVP have LD50 doses of (54.8–58.5) mg<br />
(intravenous administration in mice) (EMEA, 1999), 2,000 mg<br />
(oral administration in rats) (Schmitt et al., 2008), and<br />
100,000 mg (oral administration in rats) (Burnette, 1960) per kg
of body weight, respectively. Amylose has been reported to be<br />
used in a colon-specific drug delivery due to its low toxicity<br />
and high biodegradability (Chourasia and Jain, 2004). Triton<br />
X-100 was reported to be toxic to protozoa, fish, and<br />
mammalian cells (Dayeh et al., 2004).<br />
Three batches of dispersed SWCNTs were prepared. The<br />
first batch was used to comprehensively evaluate the longterm<br />
stability of SWCNT suspensions over a period of 4 weeks<br />
in terms of concentration, effective hydrodynamic diameter,<br />
and z-potential of stabilized SWCNTs.<br />
The second batch was used to evaluate the viability of E.<br />
coli cells after their contact with dispersed SWCNTs by the<br />
quantification of colony forming units. To elucidate the<br />
effects of ionic and organic composition of the growth<br />
medium, the cytotoxicity of the SWCNTs suspended in three<br />
growth media of different organic and salt compositions<br />
was evaluated. First, in order to assess the acute cytotoxicity<br />
of dispersed SWCNTs, we conducted experiments in<br />
0.1 M NaCl. Then, in order to investigate the effect of<br />
SWCNTs on the ability of E. coli form colonies over time, we<br />
employed three types of growth media: (i) LB medium with<br />
higher organic chemical and salt content, (ii) MD medium<br />
with low salt content and organic load, and (iii) 0.1 M NaCl<br />
solution.<br />
The third batch was used to evaluate cyto- and epigenetic<br />
toxicity of the dispersed SWCNTs against rat liver epithelial<br />
cells by the (i) NDU and (ii) GJIC bioassays. The NDU assesses<br />
cell viability by measuring the accumulation of neutral red<br />
dye in lysosomes, which depends on the cell’s capacity to<br />
maintain pH gradients through the maintaining membrane<br />
integrity and production of adenosine triphosphate. The<br />
amount of dye incorporated by the cell is quantified spectrophotometrically<br />
(Borenfreund and Puerner, 1985). The NDU<br />
bioassay has been used to assess cytotoxicity of CNTs (Flahaut<br />
et al., 2006). The GJIC bioassay (Borenfreund and Puerner,<br />
1985; Weis et al., 1998; Herner et al., 2001; Satoh et al., 2003)<br />
utilizes the ability of epigenetic tumor promoters to alter level<br />
of GJIC (Yamasaki, 1990; Trosko et al., 1991). The degree of GJIC<br />
was quantified by measuring the distance (area) the fluorescent<br />
dye ( 1200 Da) travels (occupies) between the cells after<br />
a given time. Both NDU and GJIC bioassays were carried out<br />
with WB-F344 rat liver epithelial cells exposed to dispersed<br />
SWCNTs for different periods of time. The WB-F344 cell line<br />
was chosen because these normal diploid rat liver epithelial<br />
cells have already used in numerous studies of cytotoxic or<br />
epigenetic effects (e.g., Herner et al., 2001) thus allowing us to<br />
have a comparative basis.<br />
3. Materials and methods<br />
3.1. CNTs<br />
SWCNTs (purity > 90%), produced by catalytic chemical vapor<br />
deposition, were obtained from Cheap Tubes, Inc (Brattleboro,<br />
VT) and used as received. The SWCNTs were used as obtained,<br />
allowing us to mimic what might occur in the environment.<br />
As indicated by the manufacturer, the SWCNTs had an inside<br />
diameter in the range of 0.8 nm to 1.6 nm, an outer diameter in<br />
the range of 1 nm to 2 nm and were 5 mm to30mm in length.<br />
water research 44 (2010) 505–520 509<br />
3.2. Non-covalent functionalization: dispersants and<br />
dispersion procedures<br />
3.2.1. Dispersants<br />
Gum arabic (approx. 250 kDa; a complex mixture of saccharides<br />
and glycoproteins obtained from the acacia tree), PVP<br />
(approx. 29 kDa), Triton X-100 (approx. 625 Da; polyethylene<br />
glycol p-(1,1,3,3-tetramethylbutyl)-phenyl ether) and amylose<br />
(molecular weight not specified; a polymeric form of glucose<br />
and a constituent of potato starch) were purchased from<br />
Sigma–Aldrich (Milwaukee, WI). SRNOM reverse osmosis<br />
isolation was obtained from International Humic Substances<br />
Society (St Paul, MN). The molecular structure of dispersants<br />
used in this study is presented in Fig. 1.<br />
3.2.2. Dispersion procedure<br />
Aqueous solutions of PVP, Triton X-100 were prepared by<br />
dissolving PVP (40 mg) and Triton X-100 (0.4 mL) in 40 mL of<br />
water and adjusting the pH of both solutions to 7 with 0.1 M<br />
HCl. The aqueous solution of GA was prepared by mixing 1 g of<br />
GA in 50 mL of water and adjusting the pH of the mixture to 7<br />
with 0.1 M HCl; this mixture was left to settle for 24 h and then<br />
40 mL of supernatant was collected for use in the stability or<br />
toxicity studies. To prepare NOM solutions, two flasks of<br />
40 mL of water, each containing 8 mg of NOM, were stirred for<br />
24 h. Each solution was then filtered through a 0.22 mm filter<br />
under vacuum. The pH of one solution was kept at its original<br />
value (approx. 3.5) while the pH in another solution was<br />
adjusted to 7 with 0.1 M NaOH. SWCNTs were added to the<br />
above solutions to result in a 1 mg (SWCNT)/mL loading and<br />
were sonicated in Aquasonic 50 T water bath (VWR Scientific<br />
Products Corp, West Chester, PA) for 20 min.<br />
SWCNTs were dispersed with amylose based on modified<br />
version of the three-step approach reported by Kim et al. (Kim<br />
et al., 2004). SWCNTs (40 mg) were sonicated in 25 mL of water<br />
(pH 7) at approximately 75 W for 5 min using Sonicator 3000<br />
(Misonix, Inc., Farmingdale, NY) equipped with a microprobe.<br />
100 mg of amylose in 6.28 mL of DMSO were prepared and<br />
added to the sonicated suspension of SWCNT in water so that<br />
the DMSO/water ratio was 20% by volume. The resulting<br />
mixture was sonicated for another 5 min in order to remove<br />
the excess amylose and DMSO. The suspension was sonicated,<br />
centrifuged and refilled with water four times (See SD,<br />
Section S.2.3).<br />
Following the sonication, SWCNTs suspensions were<br />
divided into 12 mL aliquots and each aliquot was transferred<br />
into 15 mL centrifuge tube. After 24 h standing time, the top<br />
4 mL of each suspension was withdrawn in order to be used in<br />
the stability or toxicity studies. Overall, three batches of<br />
dispersed SWCNT suspensions were prepared and were used,<br />
correspondingly, for 1) the study of the long-term stability of<br />
SWCNTs suspensions, 2) the general viability bioassay with<br />
E. coli, and 3) NDU and GJIC bioassays.<br />
3.3. Physicochemical characterization of SWCNT<br />
suspensions<br />
In order to quantitatively assess the long-term stability<br />
SWCNT suspensions, a combination of characterization<br />
methods was employed.
510<br />
a<br />
GALP<br />
1<br />
3<br />
ARAF<br />
1<br />
3<br />
3<br />
ARAF<br />
ARAF<br />
1<br />
1<br />
3 6<br />
1<br />
GALP<br />
3 1<br />
GALP<br />
3 6 1<br />
GALP<br />
3 1<br />
GALP<br />
6<br />
6<br />
3 1<br />
3 1<br />
RHAP GALP<br />
RHAP GALP<br />
6 6<br />
1<br />
1<br />
GA<br />
GA<br />
4 4<br />
1<br />
ARAF<br />
GALP<br />
1<br />
3<br />
ARAF<br />
1<br />
1<br />
ARAF<br />
GALP = D-GALACTOPYRANOSE ARAF = L-ARABOFURANOSE<br />
GA = D-GLUCURONIC ACID RHAP = L-RHAMNOPYRANOSE<br />
b<br />
d<br />
C 8 H 17<br />
3.3.1. UV–vis spectrophotometry. Quantification of SWCNTs<br />
concentration in suspension<br />
The absorbance of SWCNT suspensions over the (200–800) nm<br />
wavelength range was measured over a period of 4 weeks<br />
(Multi-spec 1501, Shimadzu, Kyoto, Japan). For each type of<br />
SWCNT suspension prepared, the absorbance of SWCNT-free<br />
solution of the corresponding dispersant was used as<br />
a baseline.<br />
SWCNT suspensions used in our toxicity studies likely<br />
contained both individual and bundled SWCNTs. In fact, DLS<br />
measurements indirectly confirmed (see Section 4.2) the<br />
presence of dispersed SWCNT bundles and direct TEM<br />
imaging (see SD, Section S.3.3, Fig. S5) also showed the presence<br />
of SWCNTs bundles (note that TEM results need to be<br />
interpreted with caution as evaporation-induced aggregation<br />
could have contributed to bundling during TEM sample<br />
preparation). Incomplete exfoliation is the most likely and<br />
most environmentally relevant scenario; unfortunately, it also<br />
N<br />
O<br />
n<br />
O<br />
water research 44 (2010) 505–520<br />
OH<br />
n<br />
OH<br />
OH<br />
H 2 COH<br />
HO<br />
O<br />
N<br />
H 2<br />
(a-1)<br />
CH 2 OH<br />
OH<br />
O<br />
(a-3)<br />
O<br />
OH<br />
O OH<br />
HOOC<br />
HO<br />
OH<br />
HO<br />
COOH<br />
COH<br />
OH<br />
(a-2)<br />
CH 3<br />
OH OH<br />
HOOC<br />
(a-4)<br />
entails difficulties with quantifying the total concentration of<br />
suspended nanotubes as larger CNT bundles tend to separate<br />
from the suspension.<br />
In this study, we used UV–vis spectrophotometry to estimate<br />
the concentration of exfoliated SWCNTs in suspensions.<br />
It is known that only fully exfoliated SWCNTs absorb in the<br />
(200–1200) nm wavelength range. Bundled NTs do not absorb<br />
significantly in this range due to the tunnel coupling between<br />
metallic and semiconductive SWNTs (Lauret et al., 2004;<br />
Grossiord et al., 2007). Therefore, the UV absorbance by<br />
a SWCNT suspension can be used to selectively measure the<br />
concentration of individually dispersed (i.e. fully exfoliated or<br />
debundled) SWCNTs.<br />
To determine the concentration of suspended exfoliated<br />
SWCNTs, we first prepared a separate set of suspensions with<br />
ten-fold reduced SWCNT loading with respect to the CNT<br />
loading used in long-term stability and toxicity studies. Four<br />
suspensions (with PVP, NOM pH 3.5, NOM pH 7, and GA used<br />
O<br />
O<br />
OH<br />
OH<br />
OH OH<br />
O O<br />
OH<br />
O<br />
HO<br />
OH<br />
O O<br />
Fig. 1 – Molecular structure of solubilizers: (a) Gum arabic (a-1) galactose, (a-2) rhamnose, (a-3) arabinose, (a-4) glucuonic<br />
acid; (b) Poly(vinyl pyrrolidone); (c) A building block of humic acids, which has a compositional similarity to SRNOM; Dots<br />
represent chiral centers; (d) Triton X-100; (e) Amylose.<br />
c<br />
e<br />
n<br />
OH<br />
COOH<br />
OH<br />
OH
as dispersants) with 10-fold reduced SWCNT content (0.1 g/L)<br />
were prepared using the same procedures as described in<br />
Section 3.2 except that they were subjected to rigorous, prolonged<br />
sonication for 1 h at power of (70–80) W using horn<br />
sonicator (Sonicator 3000, Misonix, Inc., Farmingdale, NY). No<br />
settling of SWCNTs was observed over the short term<br />
following the application of this treatment indicating<br />
complete dispersion of SWCNTs. UV–vis absorbance spectra<br />
were recorded at different dilutions and a calibration curve for<br />
absorption at 500 nm (Bahr et al., 2001; Huang et al., 2002;<br />
Sinani et al., 2005; Lee et al., 2007; Salzmann et al., 2007) as<br />
a function of concentration was constructed, and coefficient<br />
of molecular extinction, 3, was determined. This coefficient<br />
was used to estimate the concentration of exfoliated SWCNTs<br />
in suspensions used in long-term stability and toxicity<br />
experiments. Note that by subjecting suspended SWCNTs to<br />
a very intense sonication treatment we aimed at maximizing<br />
the extent of exfoliation; however, it was not possible to<br />
ascertain that the exfoliation was complete. Thus coefficients<br />
of molecular extinction and exfoliated SWCNT concentrations<br />
determined using the recorded calibration curves were estimated<br />
values. SWCNT characteristics measured upon the<br />
preparation of CNT suspensions are given in Table S1 in SD.<br />
3.3.2. Size and charge of dispersed SWCNTs<br />
The effective hydrodynamic diameter and z-potential of suspended<br />
SWCNTs were determined after 1, 4, 7, 14, 21, and 28<br />
days of settling. The size and charge were measured by<br />
dynamic light scattering (DLS) and phase analysis light scattering<br />
techniques, respectively (ZetaPALS, BI_MAS Option,<br />
Brookhaven Instrument Corp., Holtsville, NY). The Smoluchowski<br />
equation was applied to convert the measured<br />
electrophoretic mobility of dispersed SWCNT to z-potential.<br />
Transmission electron microscopy (TEM) imaging was used as<br />
an auxiliary method to aid in the interpretation of dynamic<br />
light scattering data (see SD, Sections S.2.5 and S.3.4).<br />
3.4. Toxicity assessment: E. coli viability assay<br />
3.4.1. Media preparation<br />
Luria–Bertani (LB) growth medium was prepared according to<br />
the standard procedure (Atlas, 1993). Minimal Davis medium<br />
with 90% reduced potassium phosphate concentration<br />
(MD medium) was prepared (Fortner et al., 2005). 0.1 M NaCl<br />
solution was prepared by dissolving 9 g of NaCl in 1 L of water<br />
(pH 7) and autoclaving it for 15 min at 1 bar and 121 C. For the<br />
toxicity assessments, each component of LB, MD and NaCl<br />
media was prepared in 4-fold higher concentration as<br />
compared to original protocol and the aliquot part of the corresponding<br />
medium was added to SWCNTs suspension<br />
(as further described in ‘‘Quantification of cell viability’’<br />
Section). Luria–Bertani Petri plates were prepared according to<br />
the published method (Atlas, 1993).<br />
3.4.2. Preparation of E. coli culture<br />
E. coli K12 stock was prepared in glycerol and stored at 80 C.<br />
Prior to use, the stock was defrosted, and 30 mL of LB medium<br />
were inoculated with 5 mL of the stock. After overnight growth<br />
at 37 C, 5 mL of this suspension was spread onto LB agar plate<br />
and cultured at 37 C. Once distinct colonies were formed, the<br />
water research 44 (2010) 505–520 511<br />
agar plate was transferred to the refrigerator and kept at 4 C<br />
for up to one month. E. coli suspensions to be used in SWCNT<br />
cytotoxicity studies were prepared by scraping one colony<br />
from the surface of a Petri plate by aseptic loop and immersing<br />
the loop into 10 mL of LB or MD media in a 50 mL centrifuge<br />
tube. Tubes were placed on a shaker in an incubator 37 C for<br />
12 h. When 0.1 M NaCl was used as an exposure medium in<br />
colony forming units bioassay, E. coli were first grown in the LB<br />
medium, centrifuged for 5 min at 2250 rpm and washed with<br />
0.1 M NaCl as follows: the supernatant was decanted and<br />
replaced with an equal volume of 0.1 M NaCl, vortexed and<br />
resuspended by centrifugation. The washing procedure was<br />
repeated twice, presuming that after this treatment most of<br />
the remaining organic constituents of LB medium were<br />
removed from the E. coli suspension.<br />
3.4.3. Quantification of cell viability<br />
The SWCNTs suspension (1.425 mL), growth medium (475 mL)<br />
and 100 mL of the stock E. coli suspension were transferred into<br />
a 15 mL centrifuge tube and incubated under gentle shaking at<br />
37 C. Samples were taken after 3, 24, and 48 h and a series of<br />
dilutions (10 4 –10 6 ) was prepared for each sample. Five samples<br />
of 10 mL and 20 mL from each dilution were placed onto an agar<br />
plate and incubated at 37 C until distinct colonies developed.<br />
Colony forming units (CFU)/1 mL were calculated for each<br />
sample. Each experiment was run in triplicates with negative<br />
(bacterial suspension with the corresponding amount of<br />
ultrapure water) and vehicle (solution of the corresponding<br />
dispersant) controls, herein called vehicle control I and<br />
vehicle control II, respectively. The results are reported as<br />
a fraction of control (FOC) standard deviation (STD), calculated<br />
as the ratio of the average number of colonies grown<br />
after E. coli exposure to SWCNTs suspensions or vehicle<br />
control II to the average number of colonies grown in vehicle<br />
control I plates.<br />
3.5. Toxicity assessment: neutral red dye uptake<br />
bioassay<br />
For the NDU bioassay we adapted a published procedure<br />
(Borenfreund and Puerner, 1985; Weis et al., 1998; Satoh et al.,<br />
2003). A solution (0.015 w/v) of neutral red dye (3-amino-7-<br />
(dimethylamino)-2-methylphenazine hydrochloride) in<br />
D-medium was incubated at 37 C for 2 h and filtered through<br />
a 0.22 mm syringe filter (Millipore Corp., New Bedford, MA) to<br />
remove undissolved dye and ensure sterile conditions.<br />
Confluent WB-F344 cells (see SD, Section S.2.6 for details on the<br />
preparation of the cells) were exposed to 500 mL of each of<br />
SWCNTs suspension and incubated at 37 C for 30 min and 24 h<br />
under gentle shaking in a humidified atmosphere containing<br />
5% CO 2. After cells were exposed to SWCNTs, the exposure<br />
medium was removed by aspiration and the cells were washed<br />
with 1 mL of phosphate saline buffer (PBS). Following washing,<br />
2 mL of the neutral red dye solution per plate was added and<br />
the cells were incubated for 1 h at 37 C in the humidified<br />
atmosphere containing 5% CO2. Upon incubation, the cells<br />
were rinsed three times with PBS, and 2 mL of aqueous solution<br />
containing 1% acetic acid and 50% ethanol was added to each<br />
plate to lyse the cells. 1.5 mL of the lysate was transported into<br />
2 mL microcentrifuge tube and the optical density was
512<br />
recorded at 540 nm using a Beckman DU 7400 diode array<br />
detector (Beckman Instruments, Inc., Schaumburg, IL). The<br />
background absorbance was measured at 690 nm and then<br />
subtracted from the original absorbance. Each experiment was<br />
conducted in triplicates. The neutral red dye uptake was<br />
reported as the FOC (absorption of neutral red in the chemically<br />
treated sample divided by the absorption of neutral red in the<br />
nontreated control I sample). FOC values of 1.0 indicates noncytotoxic<br />
response while FOC values
a b<br />
Concentration (mg/L)<br />
c<br />
-potential (mV)<br />
200<br />
160<br />
120<br />
80<br />
40<br />
0<br />
0<br />
0 7 14 21 28 0 7 14 21 28<br />
Time (d)<br />
Time (d)<br />
0<br />
-15<br />
-30<br />
-45<br />
Triton X-100 stabilizes nanotubes by the formation of hemimicelles<br />
that cover nanotube surface with benzene rings<br />
providing p–p stacking between the surfactant molecule and<br />
nanotube core (Islam et al., 2003). The exfoliation (debundling)<br />
of SWCNTs in the presence of SRNOM was suggested to occur<br />
through the interaction of the aromatic moieties of natural<br />
organic matter and nanotube surface (Hyung et al., 2007; Liu<br />
et al., 2007; Hyung and Kim, 2008). Despite the differences in<br />
the physiosorption mechanisms responsible for the stabilization<br />
of CNTs in aqueous suspensions, all dispersants were<br />
effective, albeit to somewhat different extents.<br />
For all SWCNT suspensions, the calculated values of the<br />
concentration of exfoliated SWCNTs correlated well to the<br />
visual observations described above. As seen from Fig. 2a, the<br />
dispersion efficiency in terms of the concentration of<br />
dispersed SWCNTs was a function of the type of the dispersant,<br />
with GA producing suspensions having highest<br />
concentration of SWCNTs. The data were consistent among<br />
the three prepared batches of SWCNTs (see SD, Table S1).<br />
Didenko et al. (Didenko et al., 2005) suggested that after<br />
covering one single carbon nanotube with a long polymeric<br />
molecule, the remaining strands would react with other<br />
uncovered or partially covered SWCNTs thus bundling several<br />
nanotubes together. We speculate that this mechanism where<br />
water research 44 (2010) 505–520 513<br />
Effective diameter (nm)<br />
-60<br />
-45<br />
0 7 14 21 28 0 7 14 21 28<br />
Time (d)<br />
Time (d)<br />
Fig. 2 – The time-dependent characterization of (a) estimated concentration of the exfoliated SWCNTs in the dispersion,<br />
(b) effective diameter and (c) z-potential of dispersed SWCNTs, and (d) z-potential of dispersants alone (-B- for GA/SWCNTs,<br />
-,- for PVP/SWCNTs, -6- for NOM3.5/SWCNTs, -:- for NOM7/SWCNTs, --- for Triton X-100/SWCNTs, -C- for Amylose/<br />
SWCNTs). Notes: (1) The z-potential measurement was not conducted for amylose solution, because amylose is not watersoluble<br />
under room temperature. (2) Reported are results of only those measurements that were statistically significant, i.e.<br />
when sufficiently high photon count rates were recorded in dynamic light scattering measurements. (3) The absorption by<br />
suspensions of amylose-stabilized SWCNTs was not measured because the nanotube content in these suspensions could<br />
not be precisely determined; this was due to the incomplete separation at the centrifugation step of the suspension<br />
preparation process.<br />
d<br />
-potential (mV)<br />
1000<br />
800<br />
600<br />
400<br />
200<br />
0<br />
-15<br />
-30<br />
one polymer molecule links two or more nanotubes may be<br />
responsible for the higher dispersing efficiency of GA (by far<br />
the largest molecule among the target dispersants) towards<br />
SWCNTs compared to other dispersants. This hypothesis is<br />
supported by the measurements of the effective hydrodynamic<br />
diameter (Fig. 2b), and the estimation of the length of<br />
suspended particles with the size and the length of GA/<br />
SWCNTs being higher than those of PVP, Triton X-100 and<br />
SRNOM (both pHs). As the measured diameter of GA-stabilized<br />
SWCNTs aggregates (which could also be expected to be<br />
rather porous) was still relatively small, gravity settling did<br />
not lead to a significant removal of GA-stabilized SWCNTs<br />
from the dispersion.<br />
When comparing the solubilizing ability of SRNOM at<br />
different pH, one could see that SRNOM is a more efficient<br />
dispersant at pH 3.5 than at pH 7. This is consistent with the<br />
findings by Huyng et al. (Hyung and Kim, 2008) who reported<br />
that adsorption of SRNOM to MWCNTs increased when pH<br />
decreased due to more compact and coiled conformation of<br />
NOM at acidic pHs. When pH increases, carboxylic and<br />
phenolic groups of SRNOM deprotonate resulting in higher<br />
electrostatic repulsion between a CNT and a SRNOM molecule<br />
and, as a consequence, in a lower amount of organic matter<br />
adsorbed on the CNT surface. The better dispersion of
514<br />
SWCNTs at pH 3.5 could also be attributed, in part, to the steric<br />
hindrance imposed by SRNOM when a higher surface density<br />
of NOM on the surface of CNT bundles results in stronger<br />
repulsion between CNTs. This observation is supported by the<br />
long-term z-potential measurements (Fig. 2c), where the<br />
surface charge of stabilized SRNOM/SWCNTs at pH 3.5<br />
became less negative up to day 7, and then stabilized with<br />
increasing settling time. At the same time the surface charge<br />
of SWCNTs dispersed in SRNOM solution at pH 7 gradually<br />
increased after day 5. Indeed, had the electrostatic repulsion<br />
been the sole mechanism of SWCNTs stabilization in SRNOM<br />
solutions, the surface charge on the SWCNTs would have<br />
more rapidly become less negative at pH 3.5 than at pH 7. In<br />
summary, in terms of the effectiveness of SWCNT stabilization,<br />
the dispersants were ranked as follows: GA > Triton<br />
X-100 > PVP > NOM (pH 3.5) > NOM (pH 7).<br />
4.2. Hydrodynamic size of dispersed SWCNTs<br />
The hydrodynamic size of CNTs in a suspension is an<br />
important characteristic that affects the stability and,<br />
possibly, toxicity of dispersed CNTs. It should be noted that<br />
the effective hydrodynamic diameter of suspended SWCNTs<br />
measured using DLS can only be used as a very rough estimate.<br />
While the DLS data are interpreted with the assumption<br />
that the primary scatterers are spherical with an aspect ratio<br />
of 1, dispersed SWCNTs are long and tubular, with a very high<br />
aspect ratio (see SD, Fig. S3). To obtain a better approximation<br />
of the size distribution of dispersed SWCNT, the multimodal<br />
size distribution model was used. TEM imaging was employed<br />
to obtain auxiliary information on the size and morphology of<br />
stabilized SWCNT. Even though the measurement of effective<br />
hydrodynamic diameter cannot be relied on to compare the<br />
sizes of suspended SWCNTs in different dispersion media,<br />
these measurements can be used to compare how the average<br />
particle size for a given dispersant changes with settling time.<br />
The values of effective diameter as a function of time for<br />
different SWCNT suspensions are given in Fig. 2b.<br />
Generally, the effective size of SWCNTs dispersed using<br />
Triton X-100, GA and NOM did not change significantly with<br />
increasing settling time (Fig. 2b). In the case of amylosedispersed<br />
SWCNTs, a slight decrease in the effective size was<br />
observed due to settling of larger and unstable SWCNT<br />
aggregates, which left behind more uniformly sized smaller<br />
amylose-stabilized CNT clusters. The concentration of<br />
SWCNTs in the suspension was negatively correlated with the<br />
effective size of SWCNT. The notable apparent exception was<br />
the suspension of SWCNTs dispersed using GA. However, it<br />
should be noted that the aqueous suspension of GA<br />
contained GA colloids of the size that was 1) comparable to the<br />
size of dispersed SWCNTs and 2) decreased during the 4 weeks<br />
of settling (see SD, Fig S5). Thus, the effective size measurements<br />
for SWCNTs dispersed using GA should be interpreted<br />
with caution.<br />
4.3. Charge of dispersed SWCNTs<br />
For all suspensions, the z-potential of dispersed SWCNTs was<br />
negative and nearly constant over the entire duration of the<br />
experiment (Fig. 2c and SD, Table S1). Each z-potential<br />
water research 44 (2010) 505–520<br />
a<br />
Fraction of control (FOC)<br />
b<br />
Fraction of control (FOC)<br />
c<br />
Fraction of control (FOC)<br />
1.2<br />
1<br />
0.8<br />
0.6<br />
0.4<br />
0.2<br />
0<br />
1.2<br />
1<br />
0.8<br />
0.6<br />
0.4<br />
0.2<br />
0<br />
1.2<br />
1<br />
0.8<br />
0.6<br />
0.4<br />
0.2<br />
0<br />
NOM 3.5 NOM 7 GA Amylose PVP Triton<br />
NOM 3.5 NOM 7 GA Amylose PVP Triton<br />
NOM 3.5<br />
NOM 7 GA Amylose PVP Triton<br />
Fig. 3 – Viability of E. coli in (a) 0.1 M NaCl, (b) LB medium,<br />
and (c) MD medium. Exposure time: (a) 3 h, (b) 48 h, (c) 48 h.<br />
Open, hatched and cross-hatched bars correspond to<br />
control I, control II and dispersed SWCNTs. In the case of<br />
amylose ultrapure water was used as control.<br />
measurement for SWCNT suspensions was accompanied by<br />
a measurement of the z-potential of the dispersants in<br />
a SWCNT-free aqueous solution (Fig. 2d). After 1 d, only the GA<br />
solution scattered light sufficiently to enable a statistically<br />
meaningful z-potential measurement. After 4 d, solutions of<br />
NOM (pH 7), PVP, and Triton X-100 also scattered light
a 1.2<br />
b<br />
Fraction of control (FOC)<br />
1<br />
0.8<br />
0.6<br />
0.4<br />
0.2<br />
0<br />
NOM3.5 NOM7 GA Amylose PVP Triton<br />
sufficiently. After 7 d, all the dispersants except for NOM (pH<br />
3.5) achieved acceptable count rates during z-potential<br />
measurements. The gradual change in the scattering ability of<br />
the dispersant solutions may correspond to the aggregation of<br />
the dispersant molecules. The dispersants were ranked in<br />
order from the most to the least negative z-potential of<br />
dispersed SWCNTs: NOM (pH 7) > NOM (pH 3.5) > GA z Triton<br />
X-100 > PVP z amylose. The highly negative charge of NOMstabilized<br />
SWCNTs was likely due to the charge of NOM. The<br />
low stability of amylose-dispersed SWCNTs could be attributed<br />
to the combination of low surface charge and steric<br />
repulsion between stabilized SWCNTs in the dispersion.<br />
4.4. Assessment of cell toxicity in prokaryotic and<br />
eukaryotic systems<br />
4.4.1. General viability bioassay with E. coli (prokaryotic)<br />
Immediately after E. coli cells were exposed to SWCNTs suspended<br />
in growth medium as well as after 3 h of exposure, no<br />
SWCNT aggregation was visually observed (see SD, Table S2)<br />
in experiments with all three types of the media – 0.1 M NaCl,<br />
MD (medium with lower salt organics content) and LB<br />
(medium with higher salt and organics content). After 24 h of<br />
incubation, limited precipitation was observed for all SWCNT<br />
water research 44 (2010) 505–520 515<br />
Fraction of control (FOC)<br />
1.2<br />
1<br />
0.8<br />
0.6<br />
0.4<br />
0.2<br />
0<br />
NOM3.5 NOM7 GA Amylose PVP Triton<br />
Fig. 4 – Neutral red dye uptake by WB-F344 cells as a function of dispersant type (a – 30 min, b – 24 h of exposure). Open,<br />
hatched and cross-hatched bars correspond to control I, control II and dispersed SWCNTs. In the case of amylose control I<br />
and control II were ultrapure water.<br />
suspensions in LB and MD media. After 48 h of incubation<br />
more precipitation occurred in each type of media.<br />
No inhibition of E. coli colony forming ability was observed<br />
after 3 h of incubation with amylose-, NOM-, GA- and PVPstabilized<br />
SWCNTs in 0.1 M NaCl (Fig. 3a). The FOC values did<br />
not decline in any of these samples and no significant difference<br />
in the influence on CFU counts was found between<br />
dispersed SWCNTs and solutions of corresponding dispersants.<br />
When the bacteria suspension was brought in contact<br />
with Triton X-100-stabilized SWCNTs, approx. 25% loss of the<br />
cell viability was measured as compared to vehicle control I<br />
samples. However, there was no statistical difference between<br />
FOC of Triton-stabilized SWCNTs and the control (solution of<br />
Triton X-100 only). It remains unclear if the observed cytotoxicity<br />
of the these suspensions was due to i) SWCNTs<br />
stabilized by Triton X-100 or ii) the residual ‘‘free’’ (i.e. not<br />
associated with suspended SWCNTs) Triton X-100 potentially<br />
present in the solution or iii) the combined effect of both<br />
Triton X-100/SWCNT and dissolved Triton X-100. The<br />
complete separation of the Triton X-100/SWCNT and<br />
dissolved Triton X-100 could not be accomplished using<br />
centrifugation. Even for very long centrifugation times, the<br />
supernatant had grayish color indicating that some fraction of<br />
SWCNTs was not removed.<br />
Fig. 5 – Representative phase contrast images of WB-F344 cells incubated with 500 mLofH 2O (control I), 500 mL of Triton X-100<br />
(control II) and 500 mL of Triton X-100-stabilized SWCNTs. Black dots observed in Control II and Triton X-100-solubilized<br />
SWCNTs samples correspond to dead WB-F344 cells. All images were taken at 2003 magnification. Scale bar is 50 mm.
516<br />
Fig. 6 – Representative phase contrast (upper row) and UV epifluorescent (bottom row) images of WB-F344 cells incubated<br />
with 500 mLofH2O (control I), 500 mL of NOM pH 3.5 (control II) and 500 mL of NOM-stabilized SWCNTs (pH 4). All images were<br />
taken at 2003 magnification. Bright dots correspond to cells that absorb the dye; the absorption is indicative of cellular<br />
health. All images were taken at 2003 magnification. Scale bar is 50 mm.<br />
The ability of E. coli to grow and to form colonies in the<br />
presence of amylose-, GA-, PVP-, and NOM-stabilized SWCNTs<br />
in LB medium mimicked both control samples regardless of<br />
the contact time (see SD, Fig. 3b and Fig. S6). On the contrary,<br />
21 and 18 % mortality rates after 3 h of incubation were<br />
observed for E. coli in Triton X-100/SWCNTs suspension and<br />
Triton X-100 solution, respectively. After 24 h of contact, the<br />
number of colonies grown on the Petri plates decreased by<br />
30 % when bacteria were in contact with Triton X-100-stabilized<br />
SWCNTs and by 27 % when bacteria were in Triton X-100<br />
only solution (see SD, Fig. S6) as compared to vehicle control I<br />
a 1.2<br />
b<br />
Fraction of control (FOC)<br />
1<br />
0.8<br />
0.6<br />
0.4<br />
0.2<br />
0<br />
NOM 3.5 NOM 7 GA Amylose PVP<br />
water research 44 (2010) 505–520<br />
plates. As the exposure time increased to 48 h, E. coli resumed<br />
its growth; with FOC of both Triton X-100 suspensions<br />
approaching values for vehicle control I sample (Fig. 3b). When<br />
dispersed SWCNTs were tested in MD medium, no losses in<br />
cell viability for amylose-, GA-, PVP- and NOM-stabilized<br />
SWCNTs were measured (Fig. 3c; also see SD, Fig. S7). In the<br />
case of Triton X-100 suspensions, the reduction in E. coli<br />
survival was observed after 3 h of exposure; comparable<br />
losses of the E. coli viability were also observed for both Triton<br />
X-100-stabilized SWCNTs and the Triton X-100 only solution.<br />
However, after 24 h of exposure, fewer E. coli colonies were<br />
Fraction of control (FOC)<br />
1.2<br />
1<br />
0.8<br />
0.6<br />
0.4<br />
0.2<br />
0<br />
NOM 3.5 NOM 7 GA Amylose PVP<br />
Fig. 7 – GJIC in WB-F344 cells as a function of dispersant type (a – 30 min, b – 24 h of exposure). Open, hatched and<br />
cross-hatched bars correspond to control I, control II and dispersed SWCNTs. In the case of amylose control I and control II<br />
were ultrapure water.
observed on Petri plates from suspensions of Triton X-100stabilized<br />
SWCNTs and Triton X-100 solution only (vehicle<br />
control II) (33 and 44 %, respectively) (see SD, Fig. S7). As the<br />
incubation time increased to 48 h, bacterial CFU counts in the<br />
presence of both Triton X-100 samples increased and<br />
approached that of vehicle control I samples (Fig. 3c).<br />
As was previously mentioned, the exposure of Triton<br />
X-100-stabilized SWCNTs to MD medium for 24 h resulted in<br />
formation of large flake-like aggregates, which settled out of<br />
the suspension. It is possible that this low salt medium<br />
induced specific interactions between bacterial cells and<br />
Triton X-100-coated SWCNTs, which were suppressed in the<br />
medium having a higher ionic strength. This led to a larger<br />
decrease in E. coli viability in these samples in comparison<br />
with the losses in Triton X-100 solution only. Damaged or dead<br />
cell attached to the SWCNTs aggregates and formed debris at<br />
the bottom of the testing tubes. The simultaneous effect of<br />
bacterial and nanoparticles settling out of the solution has<br />
been previously described (Tang et al., 2007) where the<br />
response of Shewanella oneidensis to C 60–NH 2 nanoparticles<br />
was studied.<br />
The fact that we observed inhibition of E. coli viability after<br />
24 h of exposure in MD medium and did not observed this<br />
effect in LB medium highlights the importance of growth<br />
medium in cytotoxicity testing. Similarly, E. coli exposure to<br />
SWCNTs, stabilized with amylose, GA, PVP and NOM (both pH)<br />
did not affect CFU counts in any of the three media pointing to<br />
the importance of the dispersant in the assessments of<br />
biocompatibility of CNTs.<br />
4.4.2. NDU cytotoxicity bioassay on eukaryotic cells<br />
The assessment of cytotoxicity was determined by the<br />
viable uptake of neutral red into F344-WB rat liver epithelial<br />
cells. Nonviable cells lack the ability to absorb this dye.<br />
Results (Fig. 4) indicated that SWCNTs dispersed using NOM<br />
(pH 3.5), NOM (pH 7), GA, amylose and PVP were not<br />
cytotoxic. However, Triton X-100 detergent induced<br />
a significant cytotoxic effect independent of SWCNTs. This<br />
latter observation can also be seen under phase contrast<br />
microscopy of cells treated with Triton X-100 for 30 min<br />
(Fig. 5). The morphology of the cells drastically changed<br />
from that of the normal control cells, in which the cells<br />
became clearer due to loss of most cytosolic contents,<br />
except for the cell nuclei. After a 24 h treatment with Triton<br />
X-100, the cells completely detached or solubilized from the<br />
culture plates (data not shown).<br />
4.5. Assessment of epigenetic toxicity on eukaryotic cells:<br />
GJIC bioassay<br />
A representative fluorescent micrograph of the GJIC assay<br />
after treatment with (NOM þ SWCNTs) is shown in Fig. 6.<br />
The migration of the fluorescent dye through several cell<br />
layers is an indicator that the gap junction channels are<br />
open. After WB-F344 cells were incubated with dispersed<br />
SWCNTs suspensions for (1) 30 min and (2) 24 h, there was<br />
no significant affect on GJIC for all suspensions tested<br />
(Fig. 7). Due to Triton X-100-stabilized SWCNTs being<br />
cytotoxic, GJIC was not measured in cells exposed to this<br />
mixture. Also, when cells were examined under phase<br />
water research 44 (2010) 505–520 517<br />
contrast, no changes in size and shape of the individual<br />
cells were detected (Fig. 6). Hence, it can be concluded that<br />
under these experimental conditions WB-F344 cells exposed<br />
to dispersed SWCNTs retain normal intercellular communication<br />
irrespective of the applied treatment.<br />
5. Conclusions<br />
The stability of aqueous suspensions of SWCNTs non-covalently<br />
functionalized by a range of natural (gum arabic,<br />
amylose, Suwannee River natural organic matter) and<br />
synthetic (polyvinyl pyrrolidone, Triton X-100) dispersants<br />
that bind to SWCNT surface via different physiosorption<br />
mechanisms was evaluated. Despite the differences, all<br />
SWCNT suspensions remained relatively stable over a period<br />
of 4 weeks as indicated by the evolution of the concentration<br />
of suspended exfoliated SWCNT and by the fact that both the<br />
average effective hydrodynamic diameter and z-potential of<br />
suspended SWCNTs were relatively constant. The epigenetic<br />
toxicity of dispersed SWCNTs was evaluated for the first time<br />
using a gap junctional intercellular communication assay<br />
with WB-F344 rat liver epithelial cells resulting in no inhibition<br />
by SWCNTs non-covalently functionalized using GA, PVP,<br />
amylose, and SRNOM. There were no cytotoxic effects of these<br />
SWCNTs suspensions on either the prokaryotic, bacterial (E.<br />
coli), or eukaryotic (WB-F344 rat liver epithelial) cell types.<br />
Only when the dispersant itself was toxic, were losses of cell<br />
viability observed. Bacterial CFU counts or neutral red<br />
uptake was affected only by SWCNTs dispersed using Triton<br />
X-100, which showed cytotoxicity in the SWCNT-free<br />
solution (vehicle control II). The results suggest a strong<br />
dependence of the toxicity of SWCNT suspensions on the<br />
toxicity of the solubilizing agent and point to the potential of<br />
non-covalent functionalization with non-toxic dispersants as<br />
a method for the preparation of aqueous suspensions of<br />
biocompatible CNTs.<br />
Acknowledgements<br />
This work was supported by the National Science Foundation<br />
research grants CBET-0608320, CBET-056828 and OISE-<br />
0530174, by the National Institute of Environmental Health<br />
Sciences grants R01 ES013268-01A2 and Grant-in-Aid for<br />
Science Research. We thank Michael Housinger for his assistance<br />
with the preparation of SWCNT suspensions. ARR<br />
acknowledges the support by Michigan AWWA Fellowship for<br />
Water Quality and Treatment Study. The contents of this work<br />
are solely the responsibility of the authors and do not necessarily<br />
represent the official views of the National Institute of<br />
Environmental Health Sciences.<br />
Appendix.<br />
Supplementary data<br />
Supplementary data associated with this article can be found<br />
in the online version, at doi: doi:10.1016/j.watres.2009.09.042.
518<br />
references<br />
Atlas, R.M., 1993. Handbook of Microbiological Media. CRC Press,<br />
Boka Raton.<br />
Bahr, J.L., Mickelson, E.D., Bronikovski, M.J., Smalley, R.E., Tour, J.M.,<br />
2001. Dissolution of small diameter single-wall carbon<br />
nanotubes in organic solvents. Chemical Communications,<br />
193–194.<br />
Bandyopadhyaya, R., Nativ-Roth, E., Regev, O., Yerushalmi-<br />
Rozen, R., 2002. Stabilization of individual carbon nanotubes<br />
in aqueous solutions. Nano Letters 2 (1), 25–28.<br />
Bardi, G., Tognini, P., Ciofani, G., Raffa, V., Costa, M.,<br />
Pizzorusso, T., 2009. Pluronic-coated carbon nanotubes do<br />
not induce degeneration of cortical neurons in vivo and in<br />
vitro. Nanomedicine: Nanotechnology, Biology, and Medicine<br />
5, 96–104.<br />
Bonnet, P., Albertini, D., Bizot, H., Bernard, A., Chauvet, O., 2007.<br />
Amylose/SWNT composites: from solution to film – synthesis,<br />
characterization and properties. Composite Science and<br />
Technology 67, 817–821.<br />
Borenfreund, E., Puerner, J.A., 1985. Toxicology determined in<br />
vitro by morphological alterations and neutral red absorption.<br />
Toxicology Letters 24, 119–124.<br />
Burnette, L.W., 1960. A review of the physiological properties of<br />
PVP. Proceedings of the Scientific Section of the Toilet Goods<br />
Association 38, 1–4.<br />
Chappell, M.A., George, A.J., Dontsova, K.M., Porter, B.E., Price, C.L.,<br />
Zhou, P., Morikawa, E., Kennedy, A.J., Steevens, J.A., 2009.<br />
Surfactive stabilization of multi-walled carbon nanotube<br />
dispersions with dissolved humic substances. Environmental<br />
Pollution 157, 1081–1087.<br />
Chen, X., Tam, U.C., Czlapinski, J.L., Lee, G.S., Rabuka, D., Zettl, A.,<br />
Bertozzi, C.R., 2006. Interfacing carbon nanotubes with living<br />
cells. Journal of American Chemical Society 128, 6292–6293.<br />
Cherukuri, P., Gannon, C.J., Leeuw, T.K., Schmidt, H.K.,<br />
Smalley, R.E., Curley, S.A., Weisman, R.B., 2006. Mammalian<br />
pharmacokinetics of carbon nanotubes using intrinsic nearinfrared<br />
fluorescence. Proceedings of the National Academy of<br />
Sciences 103, 18882–18886.<br />
Chin, S.F., Baughman, R.H., Dalton, A.B., Diekmann, G.R.,<br />
Draper, R.K., Mikoryak, C., Musselman, I.H., Poenitzsch, V.Z.,<br />
Xie, H., Pantano, P., 2007. Amphiphilic helical<br />
peptide enhances the uptake of single-walled carbon<br />
nanotubes by living cells. Experimental Biological Medicine<br />
232 (9), 1236–1244.<br />
Chourasia, M.K., Jain, S.K., 2004. Polysaccharides for colon<br />
targeted drug delivery. Drug Delivery 11 (2), 129–148.<br />
Dayeh, V.R., Chow, S.L., Schirmer, K., Lynn, D.H., Bols, N.C.,<br />
2004. Evaluating the toxicity of Triton X-100 to protozoan,<br />
fish, and mammalian cells using fluorescent dyes as<br />
indicators of cell viability. Ecotoxicology and Environmental<br />
Safety 57 (3), 375–382.<br />
Di Sotto, A., Chiaretti, M., Carru, G.A., Bellucci, S., Mazzantia, G.,<br />
2009. Multi-walled carbon nanotubes: lack of mutagenic<br />
activity in the bacterial reverse mutation assay. Toxicology<br />
Letters 184, 192–197.<br />
Didenko, V.V., Moore, V.C., Baskin, D.S., Smalley, R.E., 2005.<br />
Visualization of individual single-walled carbon nanotubes by<br />
fluorescent polymer wrapping. Nano Letters 5 (8), 1563–1567.<br />
Ding, L.H., Stilwell, J., Zhang, T.T., Elboudware, J.O., Jiang, H.J.,<br />
Selegue, J.P., Cooke, P.A., Gray, J.W., Chen, F.F., 2005. Molecular<br />
characterization of the cytotoxic mechanism of multiwall<br />
carbon nanotubes and nanoonions on human skin fibroblast.<br />
Nano Letters 5, 2448–2464.<br />
Dodziuk, H., Ejchart, A., Anczewski, W., Ueda, H., Krinichnaya, E.,<br />
Dolgonosa, G., Kutner, W., 2003. Water solubilization,<br />
water research 44 (2010) 505–520<br />
determination of the number of different types of single-wall<br />
carbon nanotubes and their partial separation with respect to<br />
diameters by complexation with h-cyclodextrin. Chemical<br />
Communications, 986–987.<br />
El-Fouly, M.H., Trosko, J.E., Chang, C.C., 1987. Scrape-loading and<br />
dye transfer: a rapid and simple technique to study gap<br />
junctional intercellular communications. Experimental Cell<br />
Research 168, 422–430.<br />
Elias, A.L., Carrero-Sanchez, J.C., Terrones, H., Endo, M.,<br />
Laclette, H.P., Mauricio, M., 2007. Viability studies of pure<br />
carbon- and nitrogen-doped nanotubes with Entamoeba<br />
Histolytica: from amoebicidal to biocompatible structures.<br />
Small 3 (10), 1723–1729.<br />
Enyashin, A.N., Gemming, S., Seifert, G., 2007. DNA-wrapped<br />
carbon nanotubes. Nanotechnology 18, 245702.<br />
Fiorito, S., Serafino, A., Andreola, F., Bernier, P., 2006. Effects of<br />
fullerenes and single-wall carbone nanotubes on murine and<br />
human macrophages. Carbon 44, 1100–1105.<br />
Flahaut, E., Durrieu, M.C., Remy-Zolghadri, M., Bareille, R.,<br />
Baquey, C., 2006. Investigation of the cytotoxicity of CCVD<br />
carbon nanotubes towards human umbilical vein endothelial<br />
cells. Carbon 44 (6), 1093–1099.<br />
Fortner, J.D., Lyon, D.Y., Sayes, C.M., Boyd, A.M., Falkner, J.C.,<br />
Hotze, E.M., Alemany, L.B., Tao, Y.J., Guo, W., Ausman, K.D.,<br />
Colvin, V.L., Hughes, J.B., 2005. C 60 in water: nanocrystal<br />
formation and microbial response. Environmental Science<br />
and Technology 39 (11), 4307–4316.<br />
Georgakilas, V., Tagmatarchis, N., Pantarotto, D., Bianco, A.,<br />
Briand, J.-P., Prato, M., 2002. Amino acid functionalisation of<br />
water soluble carbon nanotubes. Chemical Communications,<br />
3050–3051.<br />
Grossiord, N., Schhoo, P., Meuldijk, J., Koning, C.E., 2007.<br />
Determination of the surface coverage of exfoliated carbon<br />
nanotubes by surfactant molecules in aqueous solution.<br />
Langmuir 23, 3646–3653.<br />
Gutierrez, M.C., Garcia-Carvajal, Z.Y., Hortiguela, M.J., Yuste, L.,<br />
Rojo, F., Ferrer, M.L., Monte, F., 2007. Biocompatible MWCNT<br />
scaffolds for immobilization and proliferation of E. coli. Journal<br />
of Materials Chemistry 17, 2992–2995.<br />
Herner, H.A., Trosko, J.E., Masten, S.J., 2001. The epigenetic<br />
toxicity of pyrene and related ozonation byproducts<br />
containing an aldehyde functional group. Environmental<br />
Science and Technology 35 (17), 3576–3583.<br />
Herzog, E., Byrne, H.J., Davoren, M., Casey, A., Duschl, A.,<br />
Oostingh, G.J., 2009. Dispersion medium modulates oxidative<br />
stress response of human lung epithelial cells upon exposure<br />
to carbon nanomaterial samples. Toxicology and Applied<br />
Pharmacology 236, 276–281.<br />
Huang, W.J., Taylor, S., Fu, K.F., Lin, Y., Zhang, D.H., Hanks, T.W.,<br />
Rao, A.M., Sun, Y.P., 2002. Attaching proteins to carbon<br />
nanotubes via diimide-activated amidation. Nano Letters 2 (4),<br />
311–314.<br />
Humic acids and their sodium salts. EMEA-The European agency<br />
for the evaluation of medicinal products. Available via. http://<br />
www.emea.europa.eu/pdfs/vet/mrls/055499en.pdf, 1999.<br />
Hyung, H., Fortner, J.D., Hughes, J.B., Kim, J.H., 2007. Natural organic<br />
matter stabilizes carbon nanotubes in the aqueous phase.<br />
Environmental Science and Technology 41 (1), 179–184.<br />
Hyung, H., Kim, J.-H., 2008. Natural organic matter (NOM)<br />
adsorption to multi-walled carbon nanotubes: effect of NOM<br />
characteristics and water quality parameters. Environmental<br />
Science and Technology 42 (12), 4416–4421.<br />
Iijima, S., 1991. Helical microtubules of graphitic carbon. Nature<br />
354 (6348), 56–58.<br />
Inoue, K.-i., Koike, E., Yanagisawa, R., Hirano, S., Nishikawa, M.,<br />
Takano, H., 2009. Effects of multi-walled carbon nanotubes on<br />
a murine allergic airway inflammation model. Toxicology and<br />
Applied Pharmacology 237 (3), 306–316.
Islam, M.F., Rojas, E., Bergey, D.M., Johnson, A.T., Yodh, A.G.,<br />
2003. High weight fraction surfactant solubilization of<br />
single-wall carbon nanotubes in water. Nano Letters 3 (2),<br />
269–273.<br />
Jia, G., Wang, H., Yan, L., Wang, X., Pei, R., Yan, T., Zhao, Y.,<br />
Guo, X., 2005. Cytotoxicity of carbon nanomaterials: singlewall<br />
nanotube, multi-wall nanotube, and fullerene.<br />
Environmental Science and Technology 39, 1378–1383.<br />
Jiang, J., Gao, L., 2003. Production of aqueous colloidal dispersions<br />
of carbon nanotubes. Journal of Colloid and Interface Science<br />
260, 89–94.<br />
Kang, S., Herzberg, M., Rodrigues, D.F., Elimelech, M., 2008a.<br />
Antibacterial effects of carbon nanotubes: size does matter!<br />
Langmuir 24 (13), 6409–6413.<br />
Kang, S., Mauter, M.S., Elimelech, M., 2008b. Physicochemical<br />
determinants of multiwalled carbon nanotube bacterial<br />
cytotoxicity. Environmental Science and Technology 42 (19),<br />
7528–7534.<br />
Kang, S., Mauter, M.S., Elimelech, M., 2009. Microbial cytotoxicity<br />
of carbon-based nanomaterials: implications for river water<br />
and wastewater effluent. Environmental Science and<br />
Technology 43 (7), 2648–2653.<br />
Kang, S., Pinault, M., Pfefferle, L.D., Elimelech, M., 2007. Singlewalled<br />
carbon nanotubes exhibit strong antimicrobial activity.<br />
Langmuir 23 (17), 8670–8673.<br />
Karajanagi, S.S., Yang, H.C., Asuri, P., Sellitto, E., Dordick, J.S.,<br />
Kane, R.S., 2006. Protein-assisted solubilization of singlewalled<br />
carbon nanotubes. Langmuir 22 (4), 1392–1395.<br />
Kim, O.-K., Je, J., Baldwin, J.W., Kooi, S., Pehrsson, P.E., Buckley, L.J.,<br />
2004. Solubilization of single-wall carbon nanotubes by<br />
supramolecular encapsulation of helical amylose. Journal of<br />
American Chemical Society 125, 4426–4427.<br />
Lauret, J.-S., Voisin, C., Cassabois, G., Roussignol, P., Delalande, C.,<br />
Capes, L.E., Valentin, E., Jost., O., 2004. Photocreated carrier<br />
dynamics in isolated carbon nanotubes. Semiconductor<br />
Science and Technology 19, S486–S488.<br />
Lee, J.U., Huh, J., Kim, K.H., Park, C., Jo, W.H., 2007. Aqueous<br />
suspension of carbon nanotubes via non-covalent<br />
functionalization with oligothiophene-terminated<br />
poly(ethylene glycol). Carbon 45, 1051–1057.<br />
Lin, Y., Taylor, S., Li, H.P., Fernando, K.A.S., Qu, L.W., Wang, W.,<br />
Gu, L.R., Zhou, B., Sun, Y.P., 2004. Advances toward<br />
bioapplications of carbon nanotubes. Journal of Materials<br />
Chemistry 14 (4), 527–541.<br />
Lindberg, H.K., Falck, G.C.-M., Suhonen, S., Vippola, M.,<br />
Vanhalab, E., Catalána, J., Savolainena, K., Norppa, H., 2009.<br />
Genotoxicity of nanomaterials: DNA damage and micronuclei<br />
induced by carbon nanotubes and graphite nanofibres in<br />
human bronchial epithelial cells in vitro. Toxicology Letters<br />
186, 166–173.<br />
Liu, J., Rinzler, A.G., Dai, H.J., Hafner, J.H., Bradley, R.K., Boul, P.J.,<br />
Lu, A., Iverson, T., Shelimov, K., Huffman, C.B., Rodriguez-<br />
Macias, F., Shon, Y.S., Lee, T.R., Colbert, D.T., Smalley, R.E.,<br />
1998. Fullerene pipes. Science 280 (5367), 1253–1256.<br />
Liu, Y.Q., Gao, L., Zheng, S., 2007. Debundling of single-walled<br />
carbon nanotubes by using natural polyelectrolytes.<br />
Nanotechnology 18 (36), 365702.<br />
Lyon, D.Y., Fortner, J.D., Sayes, C.M., Colvin, V.L., Hughes, J.B.,<br />
2005. Bacterial cell association and antimicrobial activity of<br />
a C60 water suspension. Environmental Toxicology and<br />
Chemistry 24, 2757–2762.<br />
Manna, S.K., Sarkar, S., Barr, J., Wise, K., Barrera, E.V., Jejelowo, O.,<br />
Rice-Ficht, A., Ramesh, G.T., 2005. Single-walled carbon<br />
nanotube induces oxidative stress and activates nuclear<br />
transcription factor-kB in human keratinocytes. Nano Letters<br />
5, 1676–1684.<br />
Marsh, D.H., Rance, G.A., Zaka, M.H., Whitby, R.J., Khlobystov, A.N.,<br />
2007. Comparison of the stability of multiwalled carbon<br />
water research 44 (2010) 505–520 519<br />
nanotube dispersions in water. Physical Chemistry Chemical<br />
Physics 9 (40), 5490–5496.<br />
McDonald, T.J., Engtrakul, C., Jones, M., Rumbles, G., Heben, M.J.,<br />
2006. Kinetics of PL quenching during single-walled<br />
carbon nanotube rebundling and diameter-dependent<br />
surfactant interactions. Journal of Physical Chemistry B 110,<br />
25339–25346.<br />
Meng, J., Yanga, M., Song, L., Kong, H., Wang, C.Y., Wang, R.,<br />
Wang, C., Xie, S.S., Xua, H.Y., 2009. Concentration control of<br />
carbon nanotubes in aqueous solution and its influence on the<br />
growth behavior of fibroblasts. Colloids and Surfaces B:<br />
Biointerfaces 71, 148–153.<br />
Moore, V.C., Strano, M.S., Haroz, E.H., Hauge, R.H., Smalley, R.E.,<br />
Schmidt, J., Talmon, Y., 2003. Individually suspended singlewalled<br />
carbon nanotubes in various surfactants. Nano Letters<br />
3 (10), 1379–1382.<br />
Muller, J., Huaux, F., Moreau, N., Misson, P., Heilier, J.-F., Delos, M.,<br />
Arras, M., Fonseca, A., Nagy, J.B., Lison, D., 2005. Respiratory<br />
toxicity of multi-wall carbon nanotubes. Toxicology and<br />
Applied Pharmacology 207 (31), 221–231.<br />
Narayan, R.J., Berry, C.J., Brigmon, R.L., 2005. Structural and<br />
biological properties of carbon nanotube composite films.<br />
Materials Science and Engineering B 123, 123–129.<br />
O’Connell, M.J., Bachilo, S.M., Huffman, C.B., Moore, V.C.,<br />
Strano, M.S., Haroz, E.H., Rialon, K.L., Boul, P.J., Noon, W.H.,<br />
Kittrell, C., Ma, J., Hauge, R.H., Weisman, R.B., Smalley, R.E.,<br />
2002. Band gap fluorescence from individual single-walled<br />
carbon nanotubes. Science 297, 593–596.<br />
O’Connell, M.J., Boul, P.J., Ericson, L.M., Huffman, C., Wang, Y.,<br />
Haroz, E., 2001. Reversible water-solubilization of singlewalled<br />
carbon nanotubes by polymer wrapping. Chemical<br />
Physics Letters 342, 265–271.<br />
Park, H.J., Heo, H.Y., Lee, S.C., Park, M., Lee, S.-S., Kim, J., Chang, J.Y.,<br />
2006. Dispersion of single-walled carbon nanotubes in water<br />
with polyphosphazene polyelectrolyte. Journal of Inorganic and<br />
Organometallic Polymers and Materials 16 (4) 359–364.<br />
Pulskamp, K., Worle-Knirsch, J.M., Hennrich, F., Kern, K., Krug, H.F.,<br />
2007. Human lung epithelial cells show biphasic oxidative burst<br />
after single-walled carbon nanotube contact. Carbon 45, 2241–<br />
2249.<br />
Saleh, N.D., Pfefferle, L.D., Elimelech, M., 2009. Aggregation<br />
kinetics of multiwalled carbon nanotubes in aquatic systems:<br />
measurements and environmental implications.<br />
Environmental Science and Technology 42 (21), 7963–7969.<br />
Salzmann, C.G., Chu, B.T.T., Tobias, G., Llewellyn, S.A.,<br />
Green, M.L.H., 2007. Quantitative assessment of carbon<br />
nanotube dispersions by Raman spectroscopy. Carbon 45 (5),<br />
907–912.<br />
Satoh, A.Y., Trosco, J., Masten, S.J., 2003. Epigenetic toxicity of<br />
hydroxylated biphenyls and hydroxylated polychlorinated<br />
biphenyls on normal rat liver epithelial cells. Environmental<br />
Science and Technology 37, 2727–2733.<br />
Sayes, C.M., Liang, F., Hudsona, J.L., Mendeza, J., Guob, W.,<br />
Beach, J.M., Moore, V.C., Doyle, C.D., West, J.L., Billups, W.E.,<br />
Ausmanb, K.D., Colvin, V.L., 2006. Functionalization density<br />
dependence of single-walled carbon nanotubes cytotoxicity in<br />
vitro. Toxicology Letters 161, 135–142.<br />
Schmitt, D., Tran, N., Riefler, S., Jacoby, J., Merkel, D., Marone, P.,<br />
Naouli, N., 2008. Toxicologic evaluation of modified gum<br />
acacia: mutagenicity, acute and subchronic toxicity. Food and<br />
Chemical Toxicology 46, 1048–1054.<br />
Simon-Deckers, A., Gouget, B., Mayne-L’Hermiteb, M., Herlin-<br />
Boimeb, N., Reynaudb, C., Carrièrea, M., 2008. In vitro<br />
investigation of oxide nanoparticle and carbon nanotube<br />
toxicity and intracellular accumulation in A549 human<br />
pneumocytes. Toxicology 253, 137–146.<br />
Sinani, V.A., Gheith, M.K., Yaroslavov, A.A., Rakhnyanskaya, A.A.,<br />
Sun, K., Mamedov, A.A., Wicksted, J.P., Kotov, N.A., 2005.
520<br />
Aqueous dispersions of single-wall and multiwall carbon<br />
nanotubes with designed amphiphilic polycations. Journal of<br />
American Chemical Society 127 (10), 3463–3472.<br />
Star, A., Steuerman, D.W., Heath, J.R., Stoddart, J.F., 2002.<br />
Starched carbon nanotubes. Angewandte Chemie<br />
International Edition 41 (14), 2508–2512.<br />
Tang, Y.J., Ashcroft, J.M., Chen, D., Min, G., Kim, C.-H.,<br />
Murkhejee, B., Larabell, C., Keasling, J.D., Chen, F.F., 2007.<br />
Charge-associated effects of fullerene derivatives on microbial<br />
structural integrity and central metabolism. Nano Letters 7 (3),<br />
754–760.<br />
Tasis, D., Tagmatarchis, N., Bianco, A. and Prato, M., 2006.<br />
Chemistry of carbon nanotubes. 106 (3), 1105–1136.<br />
Tong, Z., Bischoff, M., Nies, L., Applegate, B., Turco, R., 2007.<br />
Impact of fullerene (C60) on a soil microbial community.<br />
Environmental Science and Technology 41, 2985–2991.<br />
Trosko, J.E., Chang, B.V., Madhucar, J.E., Klauning, J.E., 1991.<br />
Chemical, oncogene, and growth factor inhibition of gap<br />
junctional intercellular communication: an alternative<br />
hypothesis to carcinogenesis. Pathobiology 58, 265–278.<br />
Trosko, J.E., Chang, C.-C., Upham, B., Wilson, M., 1998. Epigenetic<br />
toxicology as toxicant-induced changes in intracellular<br />
signalling leading to altered gap junctional intercellular<br />
communication. Toxicology Letters 102–103, 71–78.<br />
Tseng, C.-H., Wang, C.-C., Chen, C.-Y., 2006. Modification of<br />
multi-walled carbon nanotubes by plasma treatment and<br />
further use as templates for growth of CdS nanocrystals.<br />
Nanotechnology 17, 5602–5612.<br />
Upham, B.L., Masten, S.J., Lochwood, B.R., Trosko, J.E., 1994.<br />
Nongenotoxic effects of polycyclic aromatic hydrocarbons and<br />
their ozonation by-products on the intercellular<br />
communication of rat liver epithelial cells. Fundamental and<br />
Applied Toxicology 23, 470–475.<br />
Vaisman, L., Marom, G., Wagner, H.D., 2006. Dispersions of<br />
surface-modified carbon nanotubes in water-soluble and<br />
water-insoluble polymers. Advanced Functional Materials 16<br />
(3), 357–363.<br />
Vittorio, O., Raffa, V. and Cuschieri, A., Influence of purity and surface<br />
oxidation on cytotoxicity of multi-wall carbon nanotubes with<br />
human neuroblastoma cells. Nanomedicine: Nanotechnology,<br />
Biology, and Medicine, in press, doi:10.1016/j.nano.2009.02.006.<br />
Wang, Y., Iqbal, Z., Malhotra, S.V., 2005. Functionalization of<br />
carbon nanotubes with amines and enzymes. Chemical<br />
Physics Letters 402, 96–101.<br />
Weis, L.M., Rummel, A.M., Masten, S.J., Trosko, J.E., Upham, B.L.,<br />
1998. Bay or baylike regions of polycyclic aromatic<br />
hydrocarbons were potent inhibitors of gap junctional<br />
water research 44 (2010) 505–520<br />
intercellular communication. Environmental Health<br />
Perspectives 106 (1), 17–22.<br />
Wick, P., Manser, P., Limbach, L.K., Dettlaff-Weglikowskab, U.,<br />
Krumeich, F., Roth, S., Stark, W.J., Bruinink, A., 2007. The<br />
degree and kind of agglomeration affect carbon nanotube<br />
cytotoxicity. Toxicology Letters 168, 121–131.<br />
Wirnitzer, U., Herbold, B., Voetz, M., Ragot, J., 2009. Studies on the<br />
in vitro genotoxicity of baytubes Ò , agglomerates of engineered<br />
multi-walled carbon-nanotubes (MWCNT). Toxicology Letters<br />
186, 160–165.<br />
Wörle-Knirsch, J.M., Pulskamp, K., Krug, H.F., 2006. Oops they did<br />
it again! Carbon nanotubes hoax scientists in viability assays.<br />
Nano Letters 6 (6), 1261–1268.<br />
Xu, F.-M., Xu, J.-P., Ji, J., Shen, J.-C., 2008. A novel biomimetic<br />
polymer as amphiphilic surfactant for soluble and<br />
biocompatible carbon nanotubes (CNTs). Colloids and<br />
Surfaces B: Biointerfaces 67, 67–72.<br />
Yamasaki, Y., 1990. Gap junctional intercellular communication<br />
and carcinogenesis. Carcinogenesis 11, 1051–1058.<br />
Yang, H., Liu, C., Yang, D., Zhanga, H., Xia, Z., 2009. Comparative<br />
study of cytotoxicity, oxidative stress and genotoxicity induced<br />
by four typical nanomaterials: the role of particle size, shape<br />
and composition. Journal of Applied Toxicology 29, 69–78.<br />
Yang, S.-T., Wang, X., Jia, G., Guc, Y., Wang, T., Nie, H., Ge, C.,<br />
Wang, H., Liu, Y., 2008. Long-term accumulation and low<br />
toxicity of single-walled carbon nanotubes in intravenously<br />
exposed mice. Toxicology Letters 181, 182–189.<br />
Yang, Z., Chen, X.H., Chen, C.S., 2007. Noncovalent-wrapped<br />
sidewall multi-walled carbon nanotubes functionalization<br />
with polyimide. Polymer Composites 28 (1), 36–41.<br />
Ye, S.-F., Wu, Y.-H., Hou, Z.-Q., Zhang, Q.-Q., 2009. ROS and NF-jB<br />
are involved in upregulation of IL-8 in A549 cells exposed to<br />
multi-walled carbon nanotubes. Biochemical and Biophysical<br />
Research Communications 379, 643–648.<br />
Yu, B.Z., Yang, J.S., Li, W.X., 2007. In vitro capability of multi-walled<br />
carbon nanotubes modified with gonadotrophin releasing<br />
hormone on killing cancer cells. Carbon 45 (10), 1921–1927.<br />
Yun, Y., Dong, Z., Tan, Z., Schulz, M.J., Shanov, V., 2009. Fibroblast<br />
cell behavior on chemically functionalized carbon<br />
nanomaterials. Materials Science and Engineering C 29, 719–725.<br />
Zhu, Y., Ran, T., Li, Y., Guo, J., Li, W., 2006. Dependence of the<br />
cytotoxicity of multi-walled carbon nanotubes on the culture<br />
medium. Nanotechnology 16, 4668–4674.<br />
Zong, S., Cao, Y., Jua, H., 2007. Direct electron transfer of hemoglobin<br />
immobilized in multiwalled carbon nanotubes enhanced<br />
grafted collagen matrix for electrocatalytic detection of<br />
hydrogen peroxide. Electroanalysis 19 (7–8), 841–846.
Oxidation of atenolol, propranolol, carbamazepine and<br />
clofibric acid by a biological Fenton-like system mediated by<br />
the white-rot fungus Trametes versicolor<br />
Ernest Marco-Urrea a , Jelena Radjenović b , Gloria Caminal c , Mira Petrović b,d ,<br />
Teresa Vicent a, *, Damià Barceló b,e<br />
a<br />
Departament d’Enginyeria Química and Institut de Ciència i Tecnologia Ambiental, Universitat Autònoma de Barcelona (UAB),<br />
08193 Bellaterra, Spain<br />
b<br />
Department of Environmental Chemistry, IDAEA-CSIC, c/Jordi Girona 18–26, 08034 Barcelona, Spain<br />
c<br />
Unitat de Biocatàlisis Aplicada associada al IQAC (CSIC-UAB), Escola Tècnica Superior d’Enginyeria, UAB, 08193 Bellaterra, Spain<br />
d<br />
Institució Catalana de Recerca i Estudis Avançats (ICREA), Barcelona, Spain<br />
e<br />
Institut Català de Recerca de l’Aigua (ICRA), Parc Científic i Tecnológic de la Universitat de Girona, Pic de Peguera, 15, 17003 Girona, Spain<br />
article info<br />
Article history:<br />
Received 31 March 2009<br />
Received in revised form<br />
16 September 2009<br />
Accepted 21 September 2009<br />
Available online 22 September 2009<br />
Keywords:<br />
Trametes versicolor<br />
Pharmaceuticals<br />
Hydroxyl radical<br />
Carbamazepine<br />
Beta-blockers<br />
Clofibric acid<br />
1. Introduction<br />
abstract<br />
The presence of pharmaceuticals and personal care products<br />
(PPCPs) and their metabolites in waters has become an<br />
emerging environmental issue. To date, most attention has<br />
been focused on identification, fate and distribution of PPCPs<br />
water research 44 (2010) 521–532<br />
Available at www.sciencedirect.com<br />
journal homepage: www.elsevier.com/locate/watres<br />
Biological advanced oxidation of the pharmaceuticals clofibric acid (CA), carbamazepine<br />
(CBZP), atenolol (ATL) and propranolol (PPL) is reported for the first time. Extracellular<br />
oxidizing species were produced through a quinone redox cycling mechanism catalyzed by<br />
an intracellular quinone reductase and any of the ligninolytic enzymes of Trametes versicolor<br />
after addition of the lignin-derived quinone 2,6-dimethoxy-1,4-benzoquinone (DBQ)<br />
and Fe 3þ -oxalate in the medium. Time-course experiments with approximately 10 mg L 1<br />
of initial pharmaceutical concentration resulted in percent degradations above 80% after<br />
6 h of incubation. Oxidation of pharmaceuticals was only observed under DBQ redox<br />
cycling conditions. A similar degradation pattern was observed when CBZP was added at<br />
the environmentally relevant concentration of 50 mgL 1 . Depletion of DBQ due to the attack<br />
of oxidizing agents was assumed to be the main limiting factor of pharmaceutical degradation.<br />
The main degradation products, that resulted to be pharmaceutical hydroxylated<br />
derivatives, were structurally elucidated. The detected 4- and 7-hydroxycarbamazepine<br />
intermediates of CBZP degradation were not reported to date. Total disappearance of<br />
intermediates was observed in all the experiments at the end of the incubation period.<br />
ª 2009 Elsevier Ltd. All rights reserved.<br />
in municipal wastewater treatment plants (WWTPs), which<br />
are commonly found at very low concentrations (low ppb<br />
levels) (Radjenovic et al., 2007, 2009, Reemtsma et al., 2006). To<br />
avoid their potential adverse health effects on both humans<br />
and environment, research efforts are underway to develop<br />
efficient techniques for their removal. Among the alternatives<br />
* Corresponding author. Departament d’Enginyeria Química and Institut de Ciència i Tecnologia Ambiental. Universitat Autònoma de<br />
Barcelona (UAB). 08193 Bellaterra, Spain. Tel.: þ34 935812142; fax: þ34 935812013.<br />
E-mail address: Teresa.Vicent@uab.cat (T. Vicent).<br />
0043-1354/$ – see front matter ª 2009 Elsevier Ltd. All rights reserved.<br />
doi:10.1016/j.watres.2009.09.049
522<br />
to traditional water treatment processes, advanced oxidation<br />
processes (AOPs) have received great attention to degrade<br />
PPCPs (Klavarioti et al., 2009; Song et al., 2008). AOPs can be<br />
broadly described as aqueous phase oxidation methods based<br />
on the intemediacy of highly reactive species such as hydroxyl<br />
radicals ( , OH). This radical acts as a non-selective, very strong<br />
oxidant agent with the ability to react with chemicals giving<br />
dehydrogenated or hydroxylated derivates, until achieving<br />
their total mineralization. To date, the production of , OH<br />
radicals was based on chemical, photochemical, photocatalytical<br />
or electrochemical techniques, but little is known<br />
about the possibility to use biological systems to produce , OH<br />
radicals for pollutant degradation (Klavarioti et al., 2009). This<br />
biological approach was proposed recently for several whiterot<br />
fungi (WRF) including Trametes versicolor and Pleurotus<br />
eringyi through a simple quinone redox cycling mechanism<br />
that is referred to as advanced biooxidation (Gómez-Toribio<br />
et al., 2009a).<br />
WRF are basidiomycetes that are capable of extensive<br />
aerobic lignin depolymeration and mineralization due to the<br />
low substrate specificity and high reactivity of the enzymes<br />
they produce (principally laccase and peroxidases). Such<br />
combinations of enzymes with auxiliary oxidoreductive<br />
enzymes such as cytochrome P450 system are commonly<br />
involved in the degradation of several PPCPs as for example<br />
carbamazepine, clofibric acid, ibuprofen and several endocrine<br />
disrupting chemicals (Cabana et al., 2007; Marco-Urrea<br />
et al., 2009a).<br />
In this work, the anti-epileptic drug carbamazepine (CBZP),<br />
lipid regulator clofibric acid (CA), and b-blockers propranolol<br />
(PPL) and atenolol (ATL) were the selected pharmaceuticals due<br />
to their long-term use in Europe and North America and their<br />
subsequent occurrence in the aquatic environment (Alder<br />
et al., 2006). The metabolism of these pharmaceuticals in<br />
mammals has been well established. It is known that CBZP<br />
undergoes extensive hepatic metabolism by the cytochrome<br />
P450 system, with oxidation to 10,11-dihydro-10,11-epoxy<br />
carbamazepine, its further hydration to 11-dihydro-10,11epoxy<br />
carbamazepine and conjugation of the latter one with<br />
glucuronide being the main metabolic pathway of CBZP (Kerr<br />
et al., 1994). CA is the pharmacologically active derivative of<br />
clofibrate and several other fibrates, and it is metabolized to<br />
reactive acylating derivatives that have been shown to transacylate<br />
glutathione forming the corresponding glutathione<br />
conjugate (Grillo and Benet, 2001). The primary metabolic<br />
pathways of PPL are glucuronidation, side-chain oxidation and<br />
ring oxidation, with the main metabolites arising from Ndealkylation<br />
of the isopropanolamine side-chain, naphthalene<br />
ring hydroxylation, and side-chain O-glucuronidation (Luan<br />
et al., 2005). ATL is excreted eagerly via urine as an unchanged<br />
compound (90%) (Dollery, 1991), with a small percentage of<br />
atenolol-glucuronide (0.8–4.4%) and hydroxyatenolol (1.1–<br />
4.4%, hydroxylation of the benzilic position).<br />
Limited removal of ATL and CA and no removal of CBZP<br />
have been frequently observed in municipal WWTPs and<br />
subsequently they deserve more attention due to the high risk<br />
of passing through later barriers in partly closed water cycles<br />
(Radjenovic et al., 2007; Reemtsma et al., 2006; Ternes, 1998).<br />
Joss et al. (2003) found that CBZP was not expected to produce<br />
acute toxic effects in the aquatic biota, but chronic and<br />
water research 44 (2010) 521–532<br />
synergistic effects with other chemicals could not be<br />
excluded. According to their results and regarding the present<br />
European legislation on the classification and labelling of<br />
chemicals (92/32/EEC), they classified CBZP as ‘‘R52/53 Harmful<br />
to aquatic organisms and may cause long-term adverse<br />
effects in the aquatic environment’’. On the other side, CA has<br />
been reported to be a non-hazardous, yet environmentally<br />
persistent compound (estimated persistence of 21 years in the<br />
environment) (Buser et al., 1998; Ferrari et al., 2003). Nevertheless,<br />
the possibility of endocrine disruption activity of CA<br />
through interference with cholesterol synthesis was reported<br />
(Pfluger and Dietrich, 2001), which is particularly important<br />
regarding its long-term effects. Although specific environmental<br />
effects of b-blockers are low, in a mixture with other bblockers<br />
the effect of concentration addition can occur, as<br />
shown in tests with Daphnia magna and phototoxicity assays<br />
with green algae (Escher et al., 2006). Considering these facts,<br />
removal techniques need to be developed to effectively<br />
degrade the selected pharmaceuticals.<br />
The objective of this study was to demonstrate for the first<br />
time the degradation of the aforementioned pharmaceuticals<br />
by biologically induced oxidizing species, produced by the<br />
WRF T. versicolor. The advanced biooxidation strategy consists<br />
of the incubation of fungi with a lignin-derived quinone (2,6,dimethoxy-1,4-benzoquione,<br />
DBQ) and chelated ferric ion<br />
(Fe 3þ -oxalate). Under these conditions, fungi catalyze the<br />
conversion of the quinone into hydroquinone (DBQH2) byan<br />
intracellular quinone reductase, and subsequent oxidation of<br />
DBQH2 to semiquinone radicals (DBQ, ) is performed in the<br />
extracellular medium by any of the lignin modifying enzymes<br />
of the WRF (laccase and peroxidases). Then, Fenton’s reagent<br />
is formed by DBQ , autoxidation catalyzed by Fe 3þ , in which<br />
Fe 2þ ,<br />
and superoxide anion radical (O2 ) are generated<br />
(DBQ , þ Fe 3þ -oxalate / DBQ þ Fe 2þ -oxalate; and Fe 2þ -<br />
oxalate þ O2 # Fe 3þ -oxalate þ O2 , ), followed by O2 ,<br />
dismutation<br />
(O2 , þ HO2 , þ H þ / O2 þ H2O2). Production of , OH<br />
via quinone redox cycling described above was previously<br />
proposed by estimating production of 2-thiobarbituric acid<br />
reactive substances (TBARS) from 2-deoxyribose and hydroxylation<br />
of 4-hydroxybenzoic acid producing 3,4-dimethoxybenzoic<br />
acid, showing a high correlation between formation<br />
of Fenton’s reagent and TBARS production (Gómez-Toribio<br />
et al., 2009a). The inhibitory effect of , OH scavengers and<br />
catalase on TBARS production rate further strengthened , OH<br />
generation by this mechanism (Gómez-Toribio et al., 2009a;<br />
Guillén et al., 2000). However, besides production of , OH<br />
radicals, other oxidizing species such as ferryl ion might also<br />
be produced in typical Fenton reactions under certain conditions<br />
of pH and concentration of organic and inorganic ligands<br />
(Hug and Leupin, 2003).Therefore, more research is needed to<br />
ascertain the relative contribution of , OH produced under<br />
quinone redox cycling conditions.<br />
This quinone redox cycle mechanism applied to WRF<br />
increases the range of environmental pollutants susceptible to<br />
be degraded by these microorganisms due to the high oxidation<br />
power and low substrate specificity of the oxidizing species<br />
generated in comparison with their ligninolytic enzymes<br />
(Marco-Urrea et al., 2009b). These oxidizing species are<br />
produced rapidly in the extracellular medium and almost total<br />
destruction of pollutants can be achieved during the first hours
of incubations without the need of adaptation of fungi to<br />
pollutants (Gómez-Toribio et al., 2009a, b). Furthermore,<br />
degradation metabolites were elucidated, which were further<br />
oxidized either by the induced oxidizing agents or by the ligninolytic<br />
enzyme system of fungus. Studies concerning degradation<br />
of pharmaceuticals by fungi are very scarce, with 1- and<br />
2-hydroxy ibuprofen, and 1,2-dihydroxy ibuprofen being the<br />
only fungal metabolites of pharmaceutical reported up to date<br />
(Marco-Urrea et al., 2009a). Nevertheless, the knowledge of<br />
these metabolites is necessary for safe application of fungi<br />
biocatalyst as bioremediation technology, as well as for the<br />
elucidation of the key steps of the degradation process.<br />
2. Materials and methods<br />
2.1. Chemicals<br />
CA (CAS No. 882-09-7), CBZP (CAS No. 298-46-4), ATL (CAS No.<br />
29122-68-7), PPL (CAS No. 3506-09-0) and DBQ (CAS No. 530-55-2)<br />
were obtained from Sigma–Aldrich Co. All other chemicals used<br />
were of analytical grade.<br />
2.2. Fungus and culture conditions<br />
T. versicolor (ATCC#42530) was maintained by subculturing on<br />
2% malt extract agar slants (pH 4.5) at room temperature.<br />
Subcultures were routinely made every 30 days.<br />
Pellets of T. versicolor were produced by inoculating 1 mL of<br />
a mycelial suspension, prepared as described previously<br />
(Marco-Urrea et al., 2008), in 1 L Erlenmeyer flask containing<br />
250 mL of malt extract medium. This was shaken (135 rpm,<br />
r ¼ 25 mm) at 25 C for 5 days. Subsequent pellets formed by<br />
this process were transferred to another 1 L Erlenmeyer flask<br />
containing 250 mL of a defined medium described elsewhere<br />
(Marco-Urrea et al., 2008) and were also incubated for 2 days in<br />
shaking conditions.<br />
2.3. Degradation experiments<br />
In time-course experiments, induction of oxidizing agents in<br />
T. versicolor via quinone redox cycling was routinely performed<br />
as follows. Mycelial pellets from each sample flask<br />
were collected by filtration and washed three times with<br />
MilliQ water. Appropriate amounts of 2-day old washed<br />
mycelium pellets were incubated in the reaction mixture (see<br />
figure legends). Reaction mixture contained 500 mM DBQ, 100–<br />
300 mM Fe 3þ -oxalate, and 100 mM Mn 2þ in 25 mL 20 mM phosphate<br />
buffer, pH 5, based on previous optimization of , OH<br />
production in quinone redox cycling (Gómez-Toribio et al.,<br />
2009b). Twenty mL of a solution containing the corresponding<br />
pharmaceutical in acetonitrile was added into the flasks to<br />
give the desired final pharmaceutical concentration (approximately<br />
10 mg L 1 ). The flasks were incubated at 25 oC and<br />
130 rpm and samples were taken at each point for analysis.<br />
The samples were filtered through a Millex-GV (Millipore)<br />
0.22 mm filter and subsequently analyzed by HPLC.<br />
In order to investigate the degradation of pharmaceuticals<br />
present at environmentally relevant concentrations, CBZP<br />
was selected as a model compound and tests flasks were<br />
water research 44 (2010) 521–532 523<br />
amended with CBZP to a 50 mg L 1 concentration. Erlenmeyer<br />
flasks containing 50 mL of the corresponding reaction mixture<br />
were sacrificed at each experiment time and were filtered<br />
through 0.45 mm glass fiber filter from Whatman. The target<br />
compound was extracted in one step by solid phase extraction<br />
with Oasis HLB cartridges (60 mg adsorbent, Waters, Barcelona,<br />
Spain) as is described elsewhere (Radjenovic et al., 2007).<br />
Briefly, the cartridges were preconditioned sequentially with<br />
5 mL of methanol and 5 mL of deionized water at neutral pH.<br />
The cartridge was dried under vacuum and was eluted with<br />
two 2-mL portions of methanol and subsequently concentrated<br />
to dryness under a gentle nitrogen stream. The extracts<br />
were reconstituted with 0.5 mL 25:75 (v/v) acetonitrile-water.<br />
To obviate the possible influence of light on pharmaceuticals<br />
stability, all the experiments were carried out in the dark.<br />
Each experiment included control flasks (without Fe 3þ -<br />
oxalate). All the results included in the text and shown in<br />
figures are the mean and standard deviation of duplicate<br />
experiments.<br />
2.4. Analytical procedures<br />
2.4.1. Analysis of pharmaceuticals and DBQ<br />
Analysis of the pharmaceuticals and DBQ were performed<br />
using a Dionex 3000 Ultimate HPLC equipped with a UV<br />
detector at 230 nm. The column temperature was 30 C and<br />
a sample volume of 20 mL was injected from a Dionex autosampler.<br />
Chromatographic separation of CBZP, PPL, CA and<br />
DBQ was achieved on a GraceSmart RP 18 column<br />
(250 mm 4 mm, particle size 5 mm). The mobile phase consisted<br />
of 6.9 mmol L 1 acetic acid adjusted to pH 4 (by NaOH)<br />
with 35% v/v acetonitrile. It was delivered isocratically at 1 mL<br />
min 1 as was described elsewhere (Stafiej et al., 2007). ATL<br />
analysis was performed with a Ascentis C18 column<br />
(150 mm 4.6 mm, particle size 5 mm). Mobile phase of 0.01 M<br />
ammonium acetate (pH 7) and mobile phase B (acetonitrile)<br />
were delivered at flow rate of 1.2 mL min 1 and used for<br />
gradient elution of ATL (t ¼ 0 min A ¼ 95%, t ¼ 20 min A ¼ 80%).<br />
The detection limit was calculated to be
524<br />
Fig. 1 – Time-course of CBZP (a) and CA (b) degradation in T. versicolor cultures under quinone redox cycle conditions.<br />
Symbols: (B) CBZP and CA in the reaction mixture (25 mL 20 mM phosphate buffer containing 500 mM DBQ, 100–300 mM<br />
Fe 3D -oxalate, 100 mM Mn 2D , and 10 mg L L1 of the corresponding pharmaceutical); (C) CBZP and CA in the control treatment<br />
(non-containing DBQ and Fe 3D -oxalate); (-) DBQ(H2) in the reaction mixture; (:) metabolite formation expressed as relative<br />
area (A/A0) where A is the corresponding metabolite and A0 is the parent drug in the control treatment at time zero. In the<br />
case of CBZP where two hydroxylated isomers were found, black and white triangles refer to P254A and B, respectively.<br />
Incubations were carried out with 1.5 ± 0.1 and 1.7 ± 0.2 mg dry weight mL L1 for CBZ and CA, respectively.<br />
increased to 60% at 11 min, further increased to 90% of A in the<br />
next 3 min and held isocratically for 2 min. Together with<br />
returning to initial conditions, the total run time was 20 min.<br />
CA was analyzed in the negative ion (NI) mode with mobile<br />
phases consisting of (A) methanol; and (B) water. The elution<br />
started at 5% of A for the first min, which was raised to 70% in<br />
the next 7 min, to 90% in the next 2 min, and then returned to<br />
initial conditions. The total run time was 13 min. The injection<br />
volume of the sample was 10 mL.<br />
The mass spectrometry analysis on the QqToF instrument<br />
was performed in wide pass quadrupole mode, for MS<br />
experiments, with the ToF data being collected between m/z<br />
50–800. The capillary and cone voltages were set to 3000 and<br />
25–30 V, respectively. Data were collected in the centroid<br />
mode, with a scan accumulation time of 1 s. The instrument<br />
was operated at a resolution of 5000 (FWHM). The nebulisation<br />
gas was set to 500 L h 1 at a temperature of 350 C,<br />
water research 44 (2010) 521–532<br />
the cone gas was set to 50 L h 1 , and the source temperature<br />
to 120 C. All analyses were acquired using an independent<br />
reference spray via the LockSpray interference to ensure<br />
accuracy and reproducibility. Sulfaguanidine (C7H10N4O2S)<br />
was used as the internal lock mass in both PI mode<br />
([M þ H] þ ¼ m/z 215.0602) and NI mode ([M H] ¼ m/z<br />
213.0446). The LockSpray frequency was set at 11 s. Fragmentation<br />
of precursor ions was done by applying collision<br />
energies in the range of 10–35 eV, using argon as a collision<br />
gas at a pressure of ~20 psi.<br />
Elemental compositions of the molecular ions and their<br />
fragments were determined and exact masses were calculated<br />
with the help of MassLynx V4.1 software incorporated in the<br />
instrument. Since the software calculation of the accurate<br />
mass of cation is performed by adding a hydrogen atom<br />
instead of proton, mass of one electron (i.e., 0.0005) was subtracted<br />
from the calculated mass.
2.4.3. Mycelial dry weight<br />
Mycelial dry weights were determined by vacuum filtering the<br />
cultures through reweighed glass filters (Whatman GF/C,<br />
Maidstone, England). The filters containing the mycelial mass<br />
were placed in glass dishes and dried at 100 C to constant<br />
weight.<br />
3. Results and discussions<br />
3.1. Oxidation of CA, CBZP, PPL and ATL by T.<br />
versicolor under quinone redox cycling conditions<br />
The strategy proposed here to degrade the selected pharmaceuticals<br />
couples both AOP and biological strategies by<br />
inducing the production of oxidizing agents such as , OH in the<br />
white-rot fungus T. versicolor. In the experiments conducted,<br />
water research 44 (2010) 521–532 525<br />
Fig. 2 – Time-course of PPL (a) and ATL (b) degradation in T. versicolor cultures under quinone redox cycle conditions. Culture<br />
conditions were as described in the legend for Fig. 1. Symbols: (B) PPL and ATL in the reaction mixture; (C) PPL and ATL in<br />
the control treatment; (-) DBQ(H2) in the reaction mixture; (:) metabolite formation expressed as relative area (A/A0) where<br />
A is the corresponding metabolite and A0 is the parent drug in the control treatment at time zero. Incubations were carried<br />
out with 1.9 ± 0.1 mg dry weight mL L1 for both PPL and ATL.<br />
oxalate was used as chelant agent of Fe 3þ instead of EDTA and<br />
Mn 2þ was added into the incubation mixture, since the<br />
oxidizing power under these conditions was substantially<br />
improved as discussed previously (Gómez-Toribio et al.,<br />
2009b). Figs. 1 and 2 show the time-courses degradation of the<br />
selected pharmaceuticals in test flasks amended with<br />
approximately 10 mg L 1 of the target pharmaceuticals, when<br />
incubated under DBQ redox cycling conditions. Control<br />
treatments in the absence of DBQ and Fe 3þ but containing<br />
fungus were performed to demonstrate both the positive<br />
involvement of oxidizing agents formed from the quinone<br />
redox cycling mechanism and to discard the role of fungal<br />
sorption in the removal of pharmaceuticals. Furthermore,<br />
control treatments permit us to rule out the intracellular and<br />
extracellular enzymatic system of T. versicolor in the degradation<br />
of the pharmaceuticals. Previously CA and CBZP were<br />
reported to be degraded by the cyt P450 system of T. versicolor<br />
grown in a chemically define medium (Marco-Urrea et al.,
526<br />
Fig. 3 – Time-course of CBZP degradation added at low<br />
concentration of 50 mg L L1 in T. versicolor cultures under<br />
quinone redox cycle conditions. Symbols: (;) CBZP in the<br />
reaction mixture; (B) CBZP in the control treatment; (C)<br />
CBZP in uninoculated bottles. Incubations were carried out<br />
with 3.8 ± 0.1 mg dry weight mL L1 . Values plotted are<br />
means ± standard deviations for triplicate cultures.<br />
2009a) but as can be observed in Figs. 1 and 2 degradation did<br />
not occur in control treatments for the incubation period<br />
studied.<br />
Results in Figs. 1 and 2 show a steep decrease of the<br />
concentration of pharmaceuticals in the first hours of incubation<br />
when the fungus was incubated under quinone redox<br />
cycling conditions. Approximately 50% of CBZP, CA and ATL<br />
were degraded the first hour and a plateau was reached above<br />
80% after the sixth hour of incubation. However, the incubation<br />
was continued to 24 h to investigate the further degradation<br />
of possible metabolites formed during the degradation<br />
process (see Section 3.3).<br />
The consumption of DBQ was also investigated in the same<br />
samples used for the determination of pharmaceutical<br />
degradation. This analysis included the determination of DBQ<br />
and its counterpart (DBQH 2) levels and is shown in Figs. 1 and<br />
2 as the sum of both: DBQ(H 2). A high correlation between<br />
DBQ(H2) and pharmaceutical degradation was found, suggesting<br />
that DBQ(H2) consumption was due to the attack of the<br />
induced oxidizing agents. Furthermore, since quinone redox<br />
cycling and therefore the production of oxidizing species<br />
cannot be sustained without the presence of DBQ(H2), the<br />
plateau observed in the pharmaceutical degradation rate can<br />
be ascribed to DBQ(H 2) depletion.<br />
In an effort to discuss the relevance of this strategy in<br />
regards to the normal concentrations of pharmaceuticals<br />
found in aqueous environments (that are frequently detected<br />
at levels from ng L 1 to low mg L 1 ), one experiment was<br />
carried out at 50 mg L 1 . For this purpose CBZP was chosen as<br />
model compound since it is found to be recalcitrant in sewer<br />
treatment facilities and in the environment (Zhang et al.,<br />
2008). As is shown in Fig. 3 the concentration of CBZP was<br />
decreased to approximately 10 mg L 1 , due to the mechanisms<br />
of fungal sorption and oxidation-induced degradation. In this<br />
water research 44 (2010) 521–532<br />
experiment, adsorption and transformation of pharmaceutical<br />
by fungus cannot be clearly distinguished, since live<br />
cultures can also degrade adsorbed pollutants by intracellular<br />
mechanisms (Blánquez et al., 2004). In this case, adsorption<br />
was higher than that observed at 10 mg L 1 probably due to<br />
the highest concentration of fungal mycelium in the experimental<br />
flasks (see figure legend). Similarly to that described<br />
for higher concentrations, the plateau in CBZP degradation<br />
under quinone redox cycling conditions was observed at the<br />
fifth hour, when DBQ(H 2) was expected to be depleted under<br />
the tested conditions.<br />
3.2. QqToF-MS 2 fragmentation pathways of<br />
pharmaceuticals and their metabolites<br />
Table 1 summarizes the exact masses of molecular and fragment<br />
ions, together with recalculated mass errors and double<br />
bond equivalents (DBEs) given by the software (mass<br />
measurements accuracy threshold of 5 mg L 1 ). Since all<br />
fragment ions had an even number of electrons (i.e., DBE was<br />
half integer number in all cases), all neutral losses had likewise<br />
even electron configurations. These data were obtained under<br />
optimized conditions of collision energy and cone voltage in<br />
ESI(þ) and ESI( )MS 2 experiments on the QqToF instrument.<br />
In ESI(þ) full-scan experiments performed at UPLC-QqToF<br />
instrument, ATL (molecular ion [M þ H] þ 267) eluted at<br />
3.4 min. Another peak was observed at 3.2 min, with the<br />
molecular ion [M þ H] þ 283, denominated as metabolite P282.<br />
Considering that the mass of P282 was shifted 16 Da upwards<br />
relative to the parent compound, hydroxylation by , OH radical<br />
attack was assumed. In Table 1 are presented calculated and<br />
measured masses of fragment ions of ATL and its metabolite<br />
P282, determined in ESI(þ)-MS 2 experiments at the QqToF-MS<br />
instrument. The most abundant fragment ions in the MS 2<br />
spectrum of ATL were detected m/z 190 (loss of 77 Da, isopropylamine<br />
and water) and m/z 145 (further cleavage of CO<br />
and intermolecular cyclization and rearrangement) (see<br />
Fig. 4c). In the fragmentation pattern of the molecular ion<br />
[M þ H] þ 283, characteristic losses of isopropyl group (42 Da)<br />
and ammonia (17 Da) afforded the fragment ions at m/z 241<br />
and 224 (see Fig. 4d). The subsequent losses of water and CO<br />
from the fragment ion m/z 224 led to the formation of m/z 206<br />
and 178 fragment ions. The fragment ion m/z 161 was formed<br />
through cleavage of isopropylamine, water, CO and ammonia<br />
and intramolecular cyclization, as suggested in the insert in<br />
Fig. 4d. The fragment ion m/z 116 was also present in the MS 2<br />
spectrum of P282, indicating that the hydroxylation had to<br />
occur at the p-hydroxyphenylacetamide fraction of ATL. From<br />
the comparison of the two fragmentation patterns, it was<br />
deduced that the –OH group was attached to the alkyl sidechain,<br />
at the C-atom next to the ether oxygen.<br />
In the case of PPL, MS 2 fragmentation of the molecular ion<br />
[M þ H] þ 260 rendered intense signals at m/z 183 and 157,<br />
corresponding to the subsequent cleavages of aminoisopropyl<br />
moiety and water, and C2H2, respectively (see Fig. 4a). The less<br />
intense fragment ions at m/z 242 and 218 were formed through<br />
cleavages of water and isopropylamine, respectively. The<br />
fungal metabolite of PPL eluted 0.6 min earlier, with the<br />
molecular ion [M þ H] þ 276, and was denominated as P275.<br />
Besides the molecular ion with mass 16 Da greater than
Table 1 – Accurate mass measurement of product ions of carbamazepine (CBZP), clofibric acid (CA), propranolol (PPL), and atenolol (ATL) and their degradation products<br />
P254, P230, P275 and P282, respectively, as determined by UPLC-QqToF in ESI(D) MS 2 mode for CBZP, PPL and ATL, and ESI(L) MS 2 mode for CA.<br />
Comp. Precursor ion/product ion Elemental formula Mass (m/z) Error DBE a<br />
Exp. Theor. mDa ppm<br />
CBZP [M þ H] þ<br />
C15H13N2O 237.1019 237.1022 0.3 1.3 10.5 8.93<br />
[M þ Na] þ<br />
C15H12N2ONa 259.0844 259.0842 0.2 0.8 10.5<br />
[M þ H NH3] þ<br />
C15H10NO 220.0761 220.0757 0.4 1.8 11.5<br />
[M þ H CONH] þ<br />
C14H12N 194.0960 194.0964 0.4 2.1 9.5<br />
P254 [M þ H] þ<br />
C15H13N2O2 253.0976 253.0972 0.4 1.6 10.5 7.24; 7.81<br />
[M þ Na] þ<br />
C15H12N2O2Na 275.0799 275.0791 0.8 2.9 10.5<br />
[M þ H NH3] þ<br />
C15H10NO2 236.0710 236.0706 0.4 1.7 11.5<br />
[M þ H CONH] þ<br />
C14H12NO 210.0906 210.0913 0.7 3.3 9.5<br />
[M þ H CONH3] þ<br />
C14H10NO 208.0762 208.0762 0 0 10.5<br />
[M þ H CONH CO] þ<br />
C13H12N 182.0965 182.0964 0.1 0.5 8.5<br />
CA [M H] C10H10ClO3 213.0327 213.0324 0.3 1.4 5.5 5.90<br />
[M H C4H6O2] C6H4ClO 126.9944 126.9956 1.2 9.4 4.5<br />
[M H C6H4ClO] C4H6O2 85.0294 85.0290 0.4 4.7 2.5<br />
P230 [M H] C10H10ClO4 229.0254 229.0268 1.4 6.1 5.5 6.40<br />
[M H C4H6O2] C6H4ClO2 142.9909 142.9905 0.4 2.8 4.5<br />
PPL [M þ H] þ<br />
C16H22NO2 260.1653 260.1645 0.8 3.1 6.5 6.7<br />
[M þ H H2O] þ<br />
C16H20NO 242.1547 242.1539 0.8 3.3 7.5<br />
[M þ H C(CH3)2] þ<br />
C13H16NO2 218.1170 218.1176 0.6 2.7 6.5<br />
[M þ H H2O NH2CH(CH3) 2] þ<br />
C13H11O 183.0814 183.0804 1.0 5.5 8.5<br />
[M þ H H2O NH2CH(CH3) 2 C2H2] þ<br />
C11H9O 157.0648 157.0648 0 0 7.5<br />
[M þ H C10H8O] þ<br />
C6H14NO 116.1066 116.1070 0.4 3.4 0.5<br />
P275 [M þ H] þ<br />
C16H22NO3 276.1583 276.1594 1.1 4.0 6.5 6.1<br />
[M þ H H2O] þ<br />
C16H20NO2 258.1492 258.1489 0.3 1.2 7.5<br />
[M þ H H2O O NH2CH(CH3) 2] þ<br />
C13H11O 183.0795 183.0804 0.9 4.9 8.5<br />
[M þ H H2O O NH2CH(CH3)2 C2H2] þ<br />
C11H9O 157.0664 157.0648 1.6 10.1 7.5<br />
[M þ H C10H8O2] þ<br />
C6H14NO 116.1079 116.1070 0.9 7.7 0.5<br />
ATL [M þ H] þ<br />
C14H23N2O3 267.1696 267.1704 0.8 3.0 4.5 3.4<br />
[M þ H C(CH3)2] þ<br />
C11H17N2O3 225.1222 225.1234 5.3 1.2 4.5<br />
[M þ H C(CH3)2 NH3] þ<br />
C11H14NO3 208.0975 208.0969 0.6 2.9 5.5<br />
[M þ H C(CH3) 2 H2O] þ<br />
C11H12NO2 190.0874 190.0863 1.1 5.8 6.5<br />
[M þ H C(CH3) 2 NH3 H2O CO NH3] þ<br />
C10H9O 145.0650 145.0648 0.2 1.4 6.5<br />
[M þ H CHCONH2 H2O C(CH3)2 NH3] þ<br />
C9H9O 133.0658 133.0648 1.0 7.5 5.5<br />
[M þ H C8H9NO2] þ<br />
C6H14NO 116.1070 116.1071 0.1 0.9 0.5<br />
P282 [M þ H] þ<br />
C14H23N2O4 283.1663 283.1652 1.1 3.9 4.5 3.2<br />
[M þ H C(CH3) 2] þ<br />
C11H17N2O4 241.1186 241.1183 0.3 1.2 4.5<br />
[M þ H C(CH3)2 NH3] þ<br />
C11H14NO4 224.0901 224.0917 1.6 7.1 5.5<br />
[M þ H C(CH3)2 NH3 H2O] þ<br />
C11H12NO3 206.0807 206.0812 0.5 2.4 6.5<br />
[M þ H C(CH3) 2 NH3 H2O CO] þ<br />
C10H12NO2 178.0862 178.0863 0.1 0.6 5.5<br />
[M þ H C(CH3)2 NH3 CO H2O NH3] þ<br />
C10H9O2 161.0617 161.0597 2.0 12.4 6.5<br />
[M þ H C8H9NO3] þ<br />
C6H14NO 116.1055 116.1070 1.5 12.9 0.5<br />
Only product ions with abundances higher than 10% are taken into account.<br />
a DBE, double bond equivalent.<br />
b tr-UPLC retention time.<br />
tr b ,min<br />
water research 44 (2010) 521–532 527
528<br />
Fig. 4 – Spectra obtained in ESI(D)-MS/MS experiments at QqToF instrument (cone voltages 20–25 V, collision energies 15–<br />
25 eV) for: (a) standard mixture of PPL in methanol/water (v/v, 25/75) at 10 mg L L1 , (b) degradation product of PPL, P275, in<br />
the sample taken after 2 h, (c) standard mixture of ATL in methanol/water (v/v, 25/75) at 10 mg L L1 , and (d) degradation<br />
product of ATL, P282, in the sample taken after 4 h.<br />
[M þ H] þ 260, fragment ion at m/z 258 was observed, equivalent<br />
to the m/z 242 in the MS 2 spectrum of PPL (Fig. 4b). The rest of<br />
the observed fragment ions of the molecular ion [M þ H] þ 276<br />
were the same (i.e., m/z 116, 157 and 183), indicating that the<br />
naphthalene moiety was the site of , OH radical attack.<br />
In full-scan experiments CBZP eluted at 8.94 min, whereas<br />
two new peaks emerged at 7.24 and 7.81 min. The identical<br />
fragmentation patterns of these two peaks and their molecular<br />
weight (MW) 16 Da higher than the parent compound<br />
(see Figs. 5c,d) suggested that the formed products were two<br />
hydroxylated isomers of CBZP, denominated as P254A and B.<br />
In Table 1 exact and measured masses were presented for<br />
a more intense M254B. The collision-induced-dissociation<br />
experiments with CBZP (molecular ion [M þ H] þ 237) revealed<br />
the formation of only two fragment ions at m/z 220 and 194, as<br />
a consequence of the loss of ammonia and CONH group,<br />
respectively. Equivalent fragments were observed for P254A,B,<br />
at m/z 236 and 210. The strong signal observed at m/z 208<br />
suggested that the –OH group was probably attached at the C4<br />
and C7-atom for the two observed isomers, since the loss of 2<br />
H-atoms could be favoured by intramolecular cyclization as<br />
presented in the insert of Fig. 5d. Further cleavage of CO in the<br />
water research 44 (2010) 521–532<br />
fragment ion m/z 208 afforded the signal at m/z 182. Nevertheless,<br />
based on the retention times only it could not be<br />
deduced which peak belongs to which product. Probably the<br />
formation of intramolecular hydrogen bond between the<br />
attached oxygen and hydrogen atoms of the amide group<br />
caused one of the isomers to elute ~0.6 min later than the<br />
other.<br />
Finally, ESI( ) full-scan screenings of samples from the<br />
experiments with CA (molecular ion [M þ H] þ 213) revealed<br />
a new signal appearing 0.5 min after the CA peak, at 6.40 min.<br />
Similar to the other metabolites identified, its MW was<br />
deferring for 16 Da relative to the original drug, with the<br />
molecular ion [M þ H] þ 229 (i.e., P230). Very poor MS 2 fragmentation<br />
was observed for both parent compound and its<br />
fungal metabolite. The collision of [M þ H] þ 213 rendered two<br />
fragment ions at m/z 126 and 85, representing the cleavage of<br />
the ether bond (see insert in Fig. 5a), whereas in the spectrum<br />
of [M þ H] þ 229 only one fragment ion was detected at m/z 142,<br />
equivalent to the m/z 126 but with mass shifted upwards for<br />
16 Da. Nevertheless, this was enough evidence to assume<br />
hydroxylation of the benzene ring, although the exact position<br />
of the attached –OH group could not be deduced.
The degradation intermediates of each pharmaceutical<br />
detected in this section are found in Fig. 6.<br />
3.3. Discussion of the results on the fungal metabolites<br />
of pharmaceuticals<br />
The formation of metabolites was depicted in Figs. 1 and 2 and<br />
was expressed as relative area (A/A0) measured by integration<br />
of LC-MS peaks of the corresponding metabolite (A) and the<br />
parent drug in the control treatments at time zero (A0), since<br />
due to the lack of authentic analytical standards for the newly<br />
identified products their quantitative determination was not<br />
possible. Examination of the profiles show qualitatively<br />
similar formation pattern of metabolites across the incubation<br />
time, obtaining a maximum value of A/A 0 between the<br />
first and fourth hour and disappearing completely in all the<br />
cases after 24 h. The metabolites appeared to be formed at low<br />
concentration levels in comparison with the parent drug, with<br />
a maximum A/A0 values for each case in the range of 0.012<br />
(PPL, Fig. 2a) to 0.27 (CA, Fig. 1b).<br />
The same compound that was identified in our experiments<br />
as the fungal metabolite of ATL, P282, was previously<br />
reported as one of the intermediate products of solar photo-<br />
water research 44 (2010) 521–532 529<br />
Fig. 5 – Spectra obtained in MS/MS experiments at QqToF instrument (cone voltages 15–30 V, collision energies 15–25 eV)<br />
for: (a) standard mixture of CA in methanol/water (v/v, 25/75) at 10 mg L L1 , (b) degradation product of CA, P230, in the<br />
sample taken after 2 h, (c) standard mixture of CBZP in methanol/water (v/v, 25/75) at 10 mg L L1 , and (d) degradation product<br />
of CBZP, P254, in the sample taken after 2 h.<br />
Fenton and TiO2 photocatalytic treatment of ATL in pilotscale<br />
compound parabolic collectors (Radjenovic et al., 2009).<br />
In the case of these photocatalytic treatments, P282 was<br />
observed together with its keto tautomer (P280), and was<br />
formed through , OH-mediated reactions. Furthermore, Song<br />
et al. (2008) reported a product of ATL with MW 282, formed in<br />
the reaction of ATL with , OH radicals. This product was<br />
identified by 60 Co g-irradiation and LC-MS, although the , OH<br />
attacks was assumed to occur at the benzene ring. In the<br />
same study hydroxylated derivative of PPL was also detected<br />
with MW 275 and –OH group attached at the naphthalene<br />
ring, which coincides with the structure of P275 identified in<br />
the present study. Hydroxylation of naphthalene moiety is<br />
phase I reaction in the human metabolism of PPL, in which 4hydroxy<br />
PPL is formed (Luan et al., 2005). Cytochrome P450<br />
isozymes were found to be responsible for oxidative metabolic<br />
pathways of PPL in humans proceeding through naphthalene<br />
ring-hydroxylations at the 4-, 5-, and 7-positions and<br />
side-chain N-desisopropylation (Masubuchi et al., 1994).<br />
However, in our study no metabolites were found in fungal<br />
control treatments indicating that P275 was generated via<br />
induction of oxidizing agents instead of fungal cyt P450<br />
mechanism.
530<br />
Fig. 6 – Selected pharmaceuticals and proposed structures of their degradation products.<br />
Sirés et al. (2007) proposed the formation of a hydroxylated<br />
intermediate as the observed in our study after the<br />
hydroxylation of CA on its C 2-position by electro-Fenton and<br />
photoelectro-Fenton. This proposed metabolite was not<br />
identified by pure standards in the mentioned study but it<br />
was assigned to a dehydrated species of 2-(4-chloro-2hydroxyphenoxy)-2-methylpropionic<br />
acid on the basis of its<br />
mass fragmentation spectra. On the other hand, by using an<br />
HPLC-DAD and FLD, Doll and Frimmel (2004) proposed TiO2<br />
photocatalytic degradation pathway of CA, which did not<br />
include the hydroxylated product as the one observed in our<br />
study. During photocatalysis CA underwent substitution of<br />
–Cl by –OH group in the para-position, whereas products such<br />
as 2-(4-hydroxyphenoxy)-isobutyric acid and hydroquinone<br />
were formed. On the other side, cleavage of isobutyric acid<br />
from the side-chain afforded 4-chlorophenol and 4-chlorocatechol.<br />
Interestingly, in the abovementioned reports,<br />
substrates of laccase such as phenol and 4-chlorophenol<br />
have been described as typical metabolites of CA degradation<br />
by AOP and once produced they could therefore be mineralized<br />
by the ligninolytic enzymatic system of WRF. But<br />
perhaps even more interesting is the production of different<br />
water research 44 (2010) 521–532<br />
quinone metabolites (also described for CA degradation by<br />
, OH attack) that could theoretically be incorporated into the<br />
quinone redox cycling enhancing the degradation rates of<br />
the target pollutant.<br />
As far as CBZP is concerned, there are few studies<br />
regarding identification of products in , OH-induced degradation<br />
pathway. Vogna et al. (2004) reported the formation of<br />
toxic acridine intermediates in UV/H2O2 treatment of CBZP,<br />
formed in , OH and , HO2-mediated reactions. Chiron et al.<br />
(2006) also found acridine as major photodegradation intermediate<br />
of CBZP after direct photolysis, but in the presence of<br />
Fe þ3<br />
they identified a hydroxycarbamazepine structure<br />
derived from a , OH attack on the C 6 aromatic ring. The<br />
formation of 2- and 3-hydroxylated derivatives of CBZP has<br />
been reported in the metabolism of CBZP by the cytochrome<br />
P450 in mammals (Mandrioli et al., 2001; Pearce et al., 2002) as<br />
well as in the two model fungi Cunninghamella elegans and<br />
Umbelopsis ramanniana used to study mammalian metabolism<br />
for many drugs (Kang et al., 2008). As far as we know, the two<br />
hydroxylated metabolites of CBZP in the 4- and 7-position<br />
(P254A and B) detected in this study are reported for the first<br />
time.
4. Conclusions<br />
Biological advanced oxidation of ATL, CA, CBZP and PPL was<br />
demonstrated for the first time by inducing extracellular<br />
oxidizing species in T. versicolor. Under quinone redox cycling<br />
conditions, pharmaceuticals degradation (added at 10 mg L 1 )<br />
reached a plateau after 6 h of incubation achieving degradation<br />
levels up to 80%. The plateau observed was due to DBQ<br />
depletion, which was shown to be concomitant to pharmaceutical<br />
degradation. The feasibility of this novel technology<br />
to degrade pharmaceuticals at low levels commonly found in<br />
the environment (50 mg L 1 ) was also demonstrated. Hydroxylated<br />
intermediates of all pharmaceuticals were identified<br />
and they disappeared after the incubation period, suggesting<br />
the total mineralization of the target compound.<br />
Acknowledgements<br />
This work was supported by the Spanish MICINN (project<br />
CTM2007-60971/TECNO) and MMAMRM (project 010/PC08/3-04).<br />
The Department of Chemical Engineering of the UAB is the<br />
Unit of Biochemical Engineering of the XRB de la Generalitat de<br />
Catalunya. E. Marco-Urrea, T. Vicent and G. Caminal are<br />
members of a Consolidated Research Group of Catalonia<br />
(2005SGR 00220). Jelena Radjenović gratefully acknowledges<br />
the JAE Program (CSIC-European Social Funds).<br />
references<br />
Alder, A.C., Bruchet, A., Carballa, M., Clara, M., Joss, A., Löffler, D.,<br />
McArdell, C.S., Miksch, K., Omil, F., Tuhkanen, T., Ternes, T.A.,<br />
2006. Consumption and occurrence. In: Ternes, T.A., Joss, A.<br />
(Eds.), Human Pharmaceuticals, Hormones and Fragrances.<br />
IWA Publishing, London, pp. 15–54.<br />
Blánquez, P., Casas, N., Font, X., Gabarrell, X., Sarrà, M.,<br />
Caminal, G., Vicent, T., 2004. Mechanism of textile metal dye<br />
biotransformation by Trametes versicolor. Water Research 38<br />
(8), 2166–2172.<br />
Buser, H.-R., Müller, M.D., Theobald, N., 1998. Occurrence of the<br />
pharmaceutical drug clofibric acid and the herbicide<br />
mecoprop in various Swiss Lakes and in the North Sea.<br />
Environmental Science and Technology 32 (1), 188–192.<br />
Cabana, H., Jones, J.P., Agathos, S.N., 2007. Elimination of<br />
endocrine disrupting chemicals using white-rot fungi and<br />
their lignin modifying enzymes: a review. Engineering in Life<br />
Sciences 7 (5), 429–456.<br />
Chiron, S., Minero, C., Vione, D., 2006. Photodegradation<br />
processes of the antiepileptic drug carbamazepine, relevant to<br />
estuarine waters. Environmental Science and Technology 40<br />
(19), 5977–5983.<br />
Doll, T.E., Frimmel, F.H., 2004. Kinetic study of photocatalytic<br />
degradation of carbamazepine, clofibric acid, iomeprol and<br />
iopromide assisted by different TiO 2 materials – determination<br />
of intermediates and reaction pathways. Water Research 38 (4),<br />
955–964.<br />
Dollery, C., 1991. Therapeutic Drugs. Churchill Livingstone,<br />
Edinburgh.<br />
Escher, B.I., Bramaz, N., Richter, M., Lienert, J., 2006. Comparative<br />
ecotoxicological hazard assessment of beta-blockers and their<br />
water research 44 (2010) 521–532 531<br />
human metabolites using a mode-of-action-based test battery<br />
and a QSAR approach. Environmental Science and Technology<br />
40 (23), 7402–7408.<br />
Ferrari, B., Paxéus, N., Giudice, R.L., Pollio, A., Garric, J., 2003.<br />
Ecotoxicological impact of pharmaceuticals found in treated<br />
wastewaters: study of carbamazepine, clofibric acid, and<br />
diclofenac. Ecotoxicology and Environmental Safety 55 (3),<br />
359–370.<br />
Gómez-Toribio, V., García-Martín, A.B., Martínez, M.J.,<br />
Martínez, A.T., Guillén, F., 2009a. Induction of extracellular<br />
hydroxyl radical production by white-rot fungi through<br />
quinone redox cycling. Applied and Environmental<br />
Microbiology 75 (12), 3944–3953.<br />
Gómez-Toribio, V., García-Martín, A.B., Martínez, M.J.,<br />
Martínez, A.T., Guillén, F., 2009b. Enhancing the production of<br />
hydroxyl radicals by Pleurotus eryngii via quinone redox cycling<br />
for pollutant degradation. Applied and Environmental<br />
Microbiology 75 (12), 3954–3962.<br />
Grillo, M.P., Benet, L.Z., 2001. Interaction of gglutamyltranspeptidase<br />
with clofibryl- S-acyl-glutathione in<br />
vitro and in vivo in rat. Chemical Research in Toxicology 14 (8),<br />
1033–1040.<br />
Guillén, F., Gómez-Toribio, V., Martínez, M.J., Martínez, A.T., 2000.<br />
Production of hydroxyl radical by the synergistic action of<br />
fungal laccase and aryl alcohol oxidase. Archives of<br />
Biochemistry and Biophysics 383 (1), 142–147.<br />
Hug, S.J., Leupin, O., 2003. Iron-catalyzed oxidation of arsenic(III)<br />
by oxygen and by hydrogen peroxide: pH-dependent<br />
formation of oxidants in the Fenton reaction. Environmental<br />
Science and Technology 37 (12), 2734–2742.<br />
Joss, A., Repetto, G., Rios, J.C., Hazen, M.J., Molero, M.L., del<br />
Peso, A., Salguero, M., Fernandez-Freire, P., Perez-Martin, J.M.,<br />
Camean, A., 2003. Ecotoxicological evaluation of<br />
carbamazepine using six different model systems with<br />
eighteen endpoints. Toxicology in Vitro 17 (5-6), 525–532.<br />
Kang, S.-I., Kang, S.-Y., Hur, H.-G., 2008. Identification of fungal<br />
metabolites of anticonvulsant drug carbamazepine. Applied<br />
Microbiology and Biotechnology 79 (4), 663–669.<br />
Kerr, B.M., Thummel, K.E., Wurden, C.J., Klein, S.M., Kroetz, D.L.,<br />
Gonzalez, F.J., Levy, R.H., 1994. Human liver carbamazepine<br />
metabolism: role of CYP3A4 and CYP2C8 in 10,11-epoxide<br />
formation. Biochemical Pharmacology 47 (11), 1969–1979.<br />
Klavarioti, M., Mantzavinos, D., Casinos, D., 2009. Removal of<br />
residual pharmaceuticals from aqueous systems by advanced<br />
oxidation processes. Environment International 35 (2),<br />
402–417.<br />
Luan, L.-J., Shao, Q., Ma, J.-Y., Zeng, S., 2005. Stereoselective<br />
urinary excretion of S-( )- and R-(þ)-propranolol glucuronide<br />
following oral administration of RS-propranolol in Chinese<br />
Han subjects. World Journal of Gastroenterology 11 (12),<br />
1822–1824.<br />
Mandrioli, R., Albani, F., Casamenti, G., Sabbioni, C., Raggi, M.A.,<br />
2001. Simultaneous high-performance liquid chromatography<br />
determination of carbamazepine and five of its metabolites in<br />
plasma of epileptic patients. Journal of Chromatography B 762<br />
(2), 109–116.<br />
Marco-Urrea, E., Parella, T., Gabarrell, X., Caminal, G., Vicent, T.,<br />
Reddy, C.A., 2008. Mechanistics of trichloroethylene<br />
mineralization by the white-rot fungus Trametes versicolor.<br />
Chemosphere 70 (3), 404–410.<br />
Marco-Urrea, E., Pérez-Trujillo, M., Vicent, T., Caminal, G., 2009a.<br />
Ability of white-rot fungi to remove selected pharmaceuticals<br />
and identification of degradation products of ibuprofen by<br />
Trametes versicolor. Chemosphere 74 (6), 765–772.<br />
Marco-Urrea, E., Aranda, E., Caminal, G., Guillén, F., 2009b.<br />
Induction of hydroxyl radical production in Trametes versicolor<br />
to degrade recalcitrant chlorinated hydrocarbons. Bioresource<br />
Technology 100 (23), 5757–5762.
532<br />
Masubuchi, Y., Hosokawa, S., Horie, T., Suzuki, T., Ohmori, S.,<br />
Kitada, M., Narimatsu, S., 1994. Cytochrome P450 isozymes<br />
involved in propranolol metabolism in human liver<br />
microsomes. The role of CYP2D6 as ring-hydroxylase and<br />
CYP1A2 as N-desisopropylase. Drug Metabolism and<br />
Disposition 22 (6), 909–915.<br />
Pearce, R.E., Vakkalagadda, G.R., Steven Leeder, J., 2002. Pathways<br />
of carbamazepine bioactivation in vitro I. Characterization of<br />
human cytochromes P450 responsible for the formation of<br />
2- and 3-hydroxylated metabolites. Drug Metabolism and<br />
Disposition 30 (11), 1170–1179.<br />
Pfluger, P., Dietrich, D.R., 2001. Effects on pharmaceuticals in the<br />
environmentdan overview and principle considerations. In:<br />
Kümmerer, K. (Ed.), Pharmaceuticals in the<br />
EnvironmentdSources, Fate, Effects and Risks. Springer, Berlin,<br />
pp. 11–17.<br />
Radjenovic, J., Petrovic, M., Barceló, D., 2007. Analysis of<br />
pharmaceuticals in wastewater and removal using<br />
a membrane bioreactor. Analytical and Bioanalytical<br />
Chemistry 387 (4), 1365–1377.<br />
Radjenovic, J., Sirtori, C., Petrovic, M., Barceló, D., Sixto, M., 2009.<br />
Solar photocatalytic degradation of persistent<br />
pharmaceuticals at pilot-scale: Kinetics and characterization<br />
of major intermediate products, Applied catalysis B:<br />
Environmental 89 (1–2), 255–264.<br />
water research 44 (2010) 521–532<br />
Reemtsma, T., Weiss, S., Mueller, J., Petrovic, M., González, S.,<br />
Barceló, D., Ventura, F., Knepper, T.P., 2006. Polar pollutants<br />
entry into the water cycle by municipal wastewater:<br />
a European perspective. Environmental Science and<br />
Technology 40 (17), 5451–5458.<br />
Sirés, I., Arias, C., Cabot, P.Ll., Centellas, F., Garrido, J.A.,<br />
Rodríguez, R.M., Brillas, E., 2007. Degradation of clofibric acid<br />
in acidic aqueous medium by electro-Fenton and<br />
photoelectro-Fenton. Chemosphere 66 (9), 1660–1669.<br />
Song, W., Cooper, W.J., Mezyk, S.P., Greaves, J., Peake, B.M., 2008.<br />
Free radical destruction of b-blockers in aqueous solution.<br />
Environmental Science and Technology 42 (4), 1256–1261.<br />
Stafiej, A., Pyrzynska, K., Regan, F., 2007. Determination of antiinflammatory<br />
drugs and estrogens in water by HPLC with UV<br />
detection. Journal of Separation Science 30 (7), 985–991.<br />
Ternes, T.A., 1998. Occurrence of drugs in German sewage<br />
treatment plants and rivers. Water Research 32 (11), 3245–<br />
3260.<br />
Vogna, D., Marotta, R., Andreozzi, R., Napolitano, A., D’Ischia, M.,<br />
2004. Kinetic and chemical assessment of the UV/H 2O 2<br />
treatment of antiepileptic drug carbamazepine. Chemosphere<br />
54 (4), 497–505.<br />
Zhang, Y., Geissen, S.U., Gal, C., 2008. Carbamazepine and<br />
diclofenac: removal in wastewater treatment plants and<br />
occurrence in water bodies. Chemosphere 73 (8), 1151–1161.
Loadings, trends, comparisons, and fate of achiral and<br />
chiral pharmaceuticals in wastewaters from urban tertiary<br />
and rural aerated lagoon treatments<br />
Sherri L. MacLeod a,1 , Charles S. Wong a,b, *<br />
a<br />
Department of Chemistry, University of Alberta, Edmonton, Alberta, T6G 2G2, Canada<br />
b<br />
Environmental Studies Program and Department of Chemistry, Richardson College for the Environment, University of Winnipeg,<br />
Winnipeg, Manitoba, R3B 2E9, Canada<br />
article info<br />
Article history:<br />
Received 1 June 2009<br />
Received in revised form<br />
14 September 2009<br />
Accepted 23 September 2009<br />
Available online 27 September 2009<br />
Keywords:<br />
Pharmaceuticals<br />
Passive samplers<br />
Wastewater<br />
Loadings<br />
Trends<br />
Chiral drugs<br />
1. Introduction<br />
abstract<br />
Pharmaceutical occurrence, fate, and toxicity in aquatic<br />
ecosystems has been studied for more than 10 years (Daughton<br />
and Ternes, 1999; Halling-Sorensen et al., 1998; Kolpin<br />
water research 44 (2010) 533–544<br />
Available at www.sciencedirect.com<br />
journal homepage: www.elsevier.com/locate/watres<br />
A comparison of time-weighted average pharmaceutical concentrations, loadings and<br />
enantiomer fractions (EFs) was made among treated wastewater from one rural aerated<br />
lagoon and from two urban tertiary wastewater treatment plants (WWTPs) in Alberta,<br />
Canada. Passive samplers were deployed directly in treated effluent for nearly continuous<br />
monitoring of temporal trends between July 2007 and April 2008. In aerated lagoon effluent,<br />
concentrations of some drugs changed over time, with some higher concentrations in<br />
winter likely due to reduced attenuation from lower temperatures (e.g., less microbially<br />
mediated biotransformation) and reduced photolysis from ice cover over lagoons; however,<br />
concentrations of some drugs (e.g. antibiotics) may also be influenced by changing use<br />
patterns over the year. Winter loadings to receiving waters for the sum of all drugs were<br />
700 and 400 g/day from the two urban plants, compared with 4 g/day from the rural plant.<br />
Per capita loadings were similar amongst all plants. This result indicates that measured<br />
loadings, weighted by population served by WWTPs, are a good predictor of other effluent<br />
concentrations, even among different treatment types. Temporal changes in chiral drug<br />
EFs were observed in the effluent of aerated lagoons, and some differences in EF were<br />
found among WWTPs. This result suggests that there may be some variation of microbial<br />
biotransformation of drugs in WWTPs among plants and treatment types, and that the<br />
latter may be a good predictor of EF for some, but not all drugs.<br />
ª 2009 Elsevier Ltd. All rights reserved.<br />
et al., 2002; Stamm et al., 2008). While drugs have been<br />
consistently detected in surface water, the effects of these<br />
complex mixtures of bioactive pollutants on non-target<br />
aquatic organisms are not fully understood (Fent et al., 2006).<br />
Accurate exposure scenarios are essential to assess toxicity<br />
Abbreviations: CR, Capital Region; DL, method detection limit; EF, enantiomer fraction; GB, Gold Bar; LLB, Lac La Biche; PEC, predicted<br />
environmental concentration; POCIS, polar organic chemical integrative sampler; SD, standard deviation; TWA, time-weighted average;<br />
WWTP, wastewater treatment plant.<br />
* Corresponding author. Richardson College for the Environment, University of Winnipeg, Winnipeg, Manitoba, R3B 2E9, Canada.<br />
E-mail addresses: smacleod@ualberta.ca (S.L. MacLeod), wong.charles.shiu@alum.mit.edu (C.S. Wong).<br />
1 Department of Chemistry, Université de Montréal, Montréal, Quebec, H3C 3J7, Canada.<br />
0043-1354/$ – see front matter ª 2009 Elsevier Ltd. All rights reserved.<br />
doi:10.1016/j.watres.2009.09.056
534<br />
and assign risk. Exposure, in turn, can be determined by either<br />
prediction or by measurement of wastewater or surface water<br />
concentrations. Prescription and sales information are not<br />
readily available, and may not be ideal for predicting environmental<br />
concentrations of some drugs (Coetsier et al., 2009)<br />
because of the uncertainty associated with use and metabolism<br />
(Letzel et al., 2009). In addition, removal of drugs from<br />
wastewater may range from 0 to 100% depending on the drug<br />
and the treatment process (Batt et al., 2007; Letzel et al., 2009,<br />
Lishman et al., 2006), even for the same compound such as<br />
diclofenac (Letzel et al., 2009), further complicating prediction<br />
of environmental concentrations. In Europe, where a tiered<br />
approach is taken for requirements of fate and effects studies,<br />
predicted environmental concentrations (PECs) are calculated<br />
as a worst-case scenario, without accounting for any metabolism,<br />
biodegradation, or retention by treatment (EMEA, 2006).<br />
For PEC values above 10 ng/L, higher tiered assessment is<br />
required (EMEA, 2006), and while in a French study, PECs were<br />
accurate for some drugs (carbamazepine, diclofenac,<br />
propranolol), local measured environmental concentrations<br />
of others were found to be much lower than those predicted,<br />
even very close (10 m) to effluent outfalls (Coetsier et al., 2009).<br />
Wastewater and surface water concentrations may be<br />
heavily affected by abiotic transformation processes such as<br />
photolysis, which depends on light intensity over the seasons<br />
(Vieno et al., 2005); as well as biotransformation, which would<br />
depend on microbial activity and on temperature. The latter is<br />
particularly significant for the numerous chiral drugs. As with<br />
other chiral pollutants (Wong, 2006), the enantiomers of<br />
pharmaceuticals are affected identically by abiotic processes,<br />
but may be affected differently by biologically mediated<br />
processes in wastewater treatment (Fono and Sedlak, 2005;<br />
MacLeod, 2009; MacLeod et al., 2007b; Nikolai et al., 2006). Drug<br />
enantiomers may also exhibit differential toxicity to aquatic<br />
life (Stanley et al., 2006). Thus, a robust means by which to<br />
predict environmental loadings and fate of drugs from wastewater<br />
effluents is valuable for exposure and risk assessment.<br />
Direct measurement of drug loadings from wastewater<br />
treatment plants (WWTPs) has proven accurate for prediction<br />
of drug concentrations in receiving waters (Letzel et al., 2009;<br />
MacLeod, 2009). Such measurements can be accomplished via<br />
repeated grab sampling (Sacher et al., 2008), which is tedious<br />
and time-consuming, or continuous flow-proportional effluent<br />
collection (Letzel et al., 2009). However, passive sampling<br />
devices such as the Polar Organic Chemical Integrative Sampler<br />
(POCIS) (Alvarez et al., 2004) can provide a simple means of<br />
obtaining time-weighted average (TWA) loadings of polar<br />
contaminants, such as drugs, over longer time periods with<br />
lower detection limits (Jones-Lepp et al., 2004; MacLeod et al.,<br />
2007a; Zhang et al., 2008). While a high-frequency, discrete<br />
sampling approach may capture the real-time heterogeneity of<br />
contaminant loadings, passive samplers have the advantage of<br />
providing integrated, continuous sample collection, and are<br />
thus useful as an alternative or complementary sampling<br />
approach for dissolved phase chemicals.<br />
Most research on drugs in the environment has, thus far,<br />
focused on measurement of drugs in the waste or receiving<br />
waters from larger (10,000 to millions) population centers<br />
(Carballa et al., 2007; Chen et al., 2006; Hua et al., 2006; Kimura<br />
et al., 2007; Lajeunesse et al., 2008; Letzel et al., 2009; Lissemore<br />
water research 44 (2010) 533–544<br />
et al., 2006; Loraine and Pettigrove, 2006; Managaki et al., 2007;<br />
Sacher et al., 2008; Vieno et al., 2005), with little attention paid<br />
to the drug output from the wastewater of smaller (
communities. Treated effluent from LLB is discharged<br />
continuously to Field Lake which discharges into Lac La Biche<br />
(the lake) via Red Deer Brook, while GB and CR both discharge<br />
treated effluent to the North Saskatchewan River and we<br />
have previously measured drugs in both of these receiving<br />
waters (MacLeod, 2009; MacLeod et al., 2007a). To our<br />
knowledge, this is the first study to use POCIS to undertake<br />
long-term, nearly continuous, monitoring of drugs, particularly<br />
chiral drugs, in municipal effluents from a small population<br />
center.<br />
2. Materials and methods<br />
2.1. Materials<br />
We used POCIS (Environmental Sampling Technologies,<br />
St. Joseph, MO) in the ‘pharmaceutical’ configuration: 200 mg<br />
Oasis HLB sorbent (Waters, Milford, MA) between two polyethersulfone<br />
membranes (45.8 cm 2 total surface area) held<br />
together by stainless steel rings and attached to a spindle<br />
holder. This configuration was used as POCIS sampling rates<br />
are available for the target analytes (Tables 1 and S2), all<br />
commonly used drugs. Chemical analytes (>98% purity) and<br />
used as received (sources in Table S2). Solvents and reagents<br />
for chromatography (Fisher Scientific, Ottawa, Canada) were<br />
of HPLC grade; nanopure water (18 MU cm) was supplied by<br />
a Nanopure Ultrapure system (Barnsteam/Thermolyne,<br />
Dubuque, IA). Details are published elsewhere (MacLeod, 2009;<br />
MacLeod et al., 2007a, b).<br />
Table 1 – Pooled winter time-weighted average<br />
concentrations for drugs detected at all three WWTPs in<br />
each of the three deployment periods.<br />
Analyte Gold<br />
Bar<br />
(ng/L) SD<br />
Capital<br />
Region<br />
(ng/L) SD<br />
Lac la<br />
Biche<br />
(ng/L) SD<br />
Atenolol 72 270 120 440 36 160<br />
Carbamazepine 200 130 290 190 63 39<br />
Celecoxib 17 8 28 13 22 10<br />
Citalopram 23 5 67 8 5.7 1.7<br />
Clarithromycin 220 140 420 270 120 90<br />
Codeine 450 380 920 710 990 710<br />
Diclofenac 1200 680 2,000 1300 460 280<br />
Erythromycin 13 10 19 15 3.1 2.8<br />
Gemfibrozil 150 59 190 72 72 34<br />
Metoprolol 58 47 65 52 29 23<br />
Naproxen 66 58 530 510 280 250<br />
Paroxetine 0.7 0.9 1.4 1.8 0.2 0.2<br />
Propranolol 4.7 3.0 14 8.5 1.4 1.1<br />
Sotalol 27 17 66 22 16 11<br />
Temazepam 130 71 160 76 120 60<br />
Triclosan 1.5 2.3 53 69 6.9 8.7<br />
Trimethoprim 38 39 65 70 10 12<br />
Sum of drugs<br />
(mg/L)<br />
2.2 5.0 2.2<br />
Pooled values calculated from n ¼ 6 for Gold Bar and Capital Region<br />
WWTPs, n ¼ 9 for Lac la Biche WWTP.<br />
water research 44 (2010) 533–544 535<br />
2.2. Passive sampling in WWTP effluent<br />
Sampling of WWTP effluent took place between July 2007 and<br />
April 2008 (Table S1).<br />
During each sampling period at the CR and LLB WWTPs,<br />
two and three POCIS, respectively, were protected in perforated<br />
stainless steel cages secured with nylon rope, and<br />
deployed in final effluent. During each sampling period at GB<br />
WWTP, two caged samplers were deployed in a tank into<br />
which treated effluent was continuously pumped. Field blanks<br />
were brought to all sites at deployment and retrieval, at which<br />
time POCIS were individually put on ice until return to the<br />
laboratory, where they were stored at 20 C until chemical<br />
extraction.<br />
2.3. Chemical extraction and instrumental analysis<br />
Details on chemical extraction and instrumental analysis are<br />
provided elsewhere (MacLeod, 2009; MacLeod et al., 2007a, b)<br />
and in Supplemental Information. In brief, POCIS sorbent was<br />
washed into a glass column, and methanol used to elute target<br />
analytes. The extract was filtered, reduced to 5–10 mL via<br />
rotary evaporation, filtered through 0.22 mm Acrodisc polytetrafluoroethylene<br />
syringe filters (Pall Life Sciences, Ann<br />
Arbor, MI, USA), and nitrogen-evaporated to dryness. Internal<br />
standards (Table S2) were added, and extracts were reconstituted<br />
to 1 mL with methanol. Achiral chromatography for<br />
determining concentrations (MacLeod et al., 2007a) was<br />
performed with an Ultra C18 reversed-phase column<br />
(150 mm 4.6 mm internal diameter (i.d.) 5 mm particle size<br />
(dp), Restek, Bellefonte, PA, USA), while enantioselective<br />
chromatography for determining enantiomer compositions<br />
was performed with a Chirobiotic V column (250 mm 4.6 mm<br />
i.d. 5 mm d p, Advanced Separation Technologies, Whippany,<br />
NJ, USA) for all chiral analytes except temazepam, for which<br />
a Chiralpak AD-RH column (MacLeod, 2009) was used<br />
(150 mm 4.6 mm i.d. 5 mm d p, Daicel Chemical Industries,<br />
West Chester, PA). Analytes were detected by electrospray<br />
ionization-tandem mass spectrometry using an Applied Biosystems<br />
QTrap triple-quadruple instrument (Foster City, CA) in<br />
multiple reaction monitoring mode.<br />
2.4. Quality assurance/quality control and<br />
data handling<br />
Detection limits (DLs, Table S2) were determined as the<br />
analyte concentration in 1 mL POCIS extract providing<br />
a signal-to-noise ratio of 3, and ranged from 0.001 to 1.6 ng/<br />
POCIS. Drugs were not detected in any blanks. Calculated<br />
time-weighted average concentrations and loadings (hereafter<br />
called ‘‘concentrations’’ and ‘‘loadings’’, respectively) are<br />
presented as mean standard deviation (SD). Data analysis<br />
was carried out as previously described elsewhere (MacLeod,<br />
2009; MacLeod et al., 2007a, b). Statistical comparisons<br />
between two groups were carried out by t-test, while three or<br />
more groups were compared by ANOVA, with either Tukey’s<br />
test for multiple comparisons or with Dunnett’s test for<br />
comparison with a control, as indicated. Differences noted are<br />
statistically significant (all p < 0.05). As expected, analytes for<br />
which the POCIS sampling rate was known with higher
536<br />
precision (Table S2) were generally more conducive to statistical<br />
significance testing.<br />
For chiral drugs atenolol, citalopram, fluoxetine, metoprolol,<br />
nadolol, pindolol, propranolol, salbutamol, sotalol, and<br />
temazepam, enantiomer fractions (EFs) described enantiomer<br />
composition (Harner et al., 2000):<br />
EF ¼<br />
E1<br />
E1 þ E2<br />
ðþÞ<br />
¼<br />
ðþÞþð Þ<br />
where E 1 and E 2 are the first and second eluted enantiomers,<br />
respectively, when the enantiomer optical rotation was<br />
unknown (i.e., all analytes except atenolol, propranolol, and<br />
fluoxetine). Further information on data quality and handling<br />
is available in Supplementary Data.<br />
3. Results and discussion<br />
3.1. Time-weighted average pharmaceutical<br />
concentrations in treated effluents<br />
3.1.1. Aerated lagoons: Lac la Biche<br />
Most studies to date have relied on grab or short-term<br />
composite sampling to assess temporal trends in pharmaceutical<br />
concentrations in effluents (Batt et al., 2007; Chen et al,<br />
2006; Hua et al., 2006; Loraine and Pettigrove 2006, Stülten et al.,<br />
2008; Vieno et al., 2005). However, temporal trends assumed<br />
from grab sampling may not be representative of longer time<br />
periods. Because POCIS can provide longer and continuous<br />
temporal integration, it may provide more accurate assessment<br />
of temporal trends. However, passive samplers cannot<br />
provide reliable information on concentration fluctuations<br />
over time scales shorter than sample integrative periods,<br />
which can result in very different conclusions about temporal<br />
changes in concentration over time (Shaw and Mueller, 2009).<br />
Drug concentrations at LLB (Fig. 2) measured over six<br />
sampling periods of 35–51 days (Table S1) ranged from 0.07 ng/L<br />
(paroxetine, propranolol) to 1100 ng/L (codeine). Fenoprofen,<br />
fluoxetine, ketoprofen, omeprazole, roxithromycin, sildenafil,<br />
sulfadimethoxine, sulfamethazine, sulfapyridine, sulfisoxazole,<br />
tadalafil and vardenafil were not detected in LLB treated<br />
effluent above their respective DLs in any sampling period<br />
(Table S2). Concentrations of atenolol and propranolol were<br />
similar to effluent grab sampled from LLB in September 2005<br />
(Nikolai et al., 2006); however, metoprolol was higher (200 ng/L).<br />
Diclofenac and naproxen concentrations at LLB were also<br />
similar to that in effluents of the activated sludge WWTP of<br />
Aura, Finland, which has a similar population (4000) to LLB<br />
(Vieno et al., 2005). Concentrations were also generally similar<br />
to those in an aerated lagoon WWTP in Louisiana, USA (Conkle<br />
et al., 2008) for drugs common to both studies: atenolol, carbamazepine,<br />
gemfibrozil, metoprolol, naproxen, and sotalol.<br />
Thus, our TWA concentrations are likely representative of<br />
those expected in small community WWTP effluents.<br />
There were statistically significant temporal concentration<br />
changes (Fig. 2) for celecoxib, citalopram, clarithromycin,<br />
codeine, diclofenac, erythromycin, gemfibrozil, naproxen,<br />
propranolol, sotalol, and temazepam. Temporal concentration<br />
trends were similar for clarithromycin, codeine, diclofenac,<br />
water research 44 (2010) 533–544<br />
(1)<br />
erythromycin, naproxen and propranolol, with higher<br />
concentrations during winter sampling periods (December to<br />
April) than at other times (July to December, Fig. 2). For atenolol,<br />
carbamazepine, metoprolol, triclosan and trimethoprim,<br />
the effluent concentrations did not change significantly<br />
over the sampling periods (Fig. 2). For drugs with temporal<br />
differences in concentration, the changes in concentration<br />
ranged from 2 (propranolol) to 960 ng/L (codeine), whereas<br />
drugs without statistically significant temporal differences<br />
had smaller concentration changes over time, from 8 (triclosan)<br />
to 85 ng/L (atenolol).<br />
While the average daily volume of treated effluent<br />
discharged decreased over the six sampling periods (Table S1),<br />
this decrease was not statistically significant. Thus, TWA<br />
loadings are reflective of concentrations. Of the drugs studied,<br />
photodegradation likely plays a major role in the dissipation<br />
of diclofenac (Andreozzi et al., 2003), naproxen (Lin et al., 2006;<br />
Lin and Reinhard, 2005), and propranolol (Andreozzi et al.,<br />
2003) in aquatic systems. The higher effluent concentrations<br />
of those drugs in winter (Fig. 2) may be at least partially the<br />
result of decreased photolysis, due to ice cover on lagoons<br />
from November to March (Siebold, 2008) blocking sunlight<br />
from the water column (Vieno et al., 2005), and fewer hours of<br />
sunlight during those months (MacLeod, 2009; Vieno et al.,<br />
2005). It is impossible to ascertain the amount of photodegradation<br />
in the lagoon during ice-free periods with our<br />
data. While the LLB lagoon wastewater was not clear in color,<br />
it is likely that photodegradation of drugs took place at least at<br />
the lagoon surface during ice-free periods, and continuous<br />
aeration of lagoon waters for oxygenation purposes would<br />
bring water at depth to the surface to provide some photodegradation<br />
treatment. Lower winter temperatures also likely<br />
influenced elimination of pharmaceuticals, through lower<br />
reaction rate constants for abiotic transformation, and<br />
reduced microbial activity for biotransformation (Kim and<br />
Carlson, 2007). Effluent pH was 8 0.5 over the six sampling<br />
periods and effluent temperatures ( C) were 21 3, 16 2,<br />
4 3, 1 0, 1.4 0.5, 4 2, chronologically for the six<br />
sampling periods (Siebold, 2008). The concentrations of those<br />
three drugs decreased again in March/April, as temperatures<br />
and number of daylight hours increased, and ice cover was<br />
reduced or absent on the lagoons (Fig. 2). Since naproxen can<br />
be readily biodegraded in sewage (Yu et al., 2006), decreased<br />
biodegradation may also contribute to the higher effluent<br />
concentrations for naproxen in winter.<br />
It is possible that changes in temperature may affect<br />
calculated TWA concentrations, as sampling rates used in<br />
this study were determined at room temperature (22–28 C,<br />
MacLeod et al., 2007a), and WWTP temperatures during<br />
colder months were considerably less. However, we cannot<br />
presently quantify the compound-dependent effect of such<br />
changes on our observed trends. Lower temperatures would<br />
result in lower analyte aqueous diffusion coefficients and<br />
hence lower sampling rates, estimated to be roughly 75%<br />
over a 20 C ambient temperature change (MacLeod et al.,<br />
2007a). When applied to water colder than under the calibration<br />
condition, sampling rates would then provide an<br />
underestimate of the calculated TWA concentrations, the<br />
effect of which would be compound-dependent. Such an<br />
effect may be pronounced in the case of drugs with higher
concentrations in winter than in other times (e.g., antibiotics<br />
such as clarithromycin and erythromycin, Fig. 2). Temporal<br />
effects on concentrations were less for other analytes, and<br />
may either be insignificant (e.g., triclosan and gemfibrozil) or<br />
of little practical significance even if statistically significant,<br />
given the relatively narrow range (e.g., from a factor of two,<br />
to an order of magnitude) in concentrations over the seasons<br />
studied (Fig. 2).<br />
water research 44 (2010) 533–544 537<br />
Fig. 2 – Time-weighted average concentrations (ng/L ± SD; ) and enantiomer fractions (EF) of chiral drugs (,) in LLB treated<br />
wastewater during six sampling periods from July 2007 to August 2008 (Table S1). Hash marks on x-axis indicate break in<br />
continuous sampling between September and October 2008. Significant concentration differences are indicated by lowercase<br />
letters while EF differences are indicated by upper-case letters. Plots without letters do not have concentrations or EFs,<br />
as appropriate, statistically different from one another over sampling periods. Non-racemic EFs are indicated by an asterisk,<br />
and EFs that were corrected for enantiomer-specific matrix effects are indicated by ^. RS, racemic standards. Error bars (SD)<br />
represent propagated error from replicate POCIS, analytical processing and instrumental uncertainty, and error associated<br />
with POCIS sampling rates.<br />
Without measurements of influent concentration, we<br />
cannot discern whether temporal concentrations in effluent<br />
are more influenced by changes in removal efficiency, as<br />
suggested as likely by many studies (Hua et al., 2006; Kim and<br />
Carlson, 2007; Loraine and Pettigrove, 2006; Sacher et al., 2008;<br />
Vieno et al., 2005) or in use patterns over time, which may be<br />
the case for antibiotics. In a Swiss study (McArdell et al., 2003),<br />
higher macrolide concentrations were found in winter than in
538<br />
other periods, consistent with our observations for the macrolides<br />
clarithromycin and erythromycin (Fig. 2). In the Swiss<br />
study, as with this study, raw wastewater concentrations<br />
were not measured so removal efficiency could not be<br />
assessed. However, the Swiss authors cited a personal<br />
communication which indicated that macrolide sales were<br />
higher in winter, indicating that increased use may contribute<br />
to higher environmental concentrations.<br />
The winter concentrations in LLB treated effluent, which<br />
were sampled mostly at the same time as those at GB and CR<br />
(see Section 3.1.3), changed over time for some but not all of<br />
our target analytes (Fig. 2). Concentration changes were less<br />
prevalent among the three winter sampling months,<br />
compared with other periods. Among the sampling periods<br />
between December and April (Tukey’s test), there were<br />
statistically significant changes in effluent concentration for<br />
only three drugs. The celecoxib concentration in December/<br />
January was three times that in January/March and twice that<br />
in March/April with magnitudes of difference of 20 and 15 ng/L,<br />
respectively. Sotalol had December/January concentrations<br />
were eight times that in both January/March and March/April<br />
and magnitudes of 33 ng/L for both comparisons. While the<br />
concentration of citalopram in January/March was 50% higher<br />
than that in March/April, the magnitude of this difference was<br />
only 2 ng/L. The general lack of effluent concentration change<br />
over the winter indicates that there is no change in drug use<br />
patterns and/or in the ability of aerated lagoons to remove<br />
drugs effectively over this season.<br />
It is important to note that concentration changes in<br />
WWTPs serving small communities may well be sensitive to<br />
changes in use patterns of the population served. If a drug was<br />
completely excreted from the body in unchanged form and<br />
did not undergo any removal during wastewater treatment,<br />
then the addition of a single additional individual’s twice-aday,<br />
100 mg active ingredient dose at LLB would result in an<br />
effluent concentration change of 100 ng/L, which would rival<br />
existing concentrations at LLB (Fig. 2)! In actuality, only<br />
a fraction of most drugs are excreted as parent compounds,<br />
and WWTP treatment provides some elimination. For<br />
example, 3% of carbamazepine is excreted unmetabolized<br />
from humans, with 30% removal based on a survey of Canadian<br />
WWTPs (Miao et al., 2005). Assuming these conditions<br />
hold for LLB, then a 2 ng/L change in effluent concentration, or<br />
a 3–10% increase (Fig. 2), would result from the same dosage.<br />
However, without specific usage information, excretion data,<br />
and WWTP-specific removal efficiencies, we cannot make<br />
more specific predictions of the sensitivity of drug concentrations<br />
in wastewater on usage in small communities.<br />
3.1.2. Tertiary treatment: Gold Bar and Capital Region<br />
The concentration of detectable drugs (Table 1, Table S3) in GB<br />
effluent ranged from 0.03 (omeprazole) to 1350 ng/L (diclofenac)<br />
while that in CR ranged from 0.09 (roxithromycin) to<br />
2520 ng/L (diclofenac). These concentrations are similar to<br />
grab sample measurements for our analytes in other Canadian<br />
WWTP effluents (Hua et al., 2006; Lishman et al., 2006;<br />
Miao et al., 2005), indicating that our passive sampler derived<br />
TWA concentrations were comparable with previous work.<br />
However, concentrations of chiral drugs in this study were<br />
generally lower by a factor of at least 10 compared with a grab<br />
water research 44 (2010) 533–544<br />
sample from GB in April 2007 (Macleod et al., 2007b), suggesting<br />
possible annual or other temporal changes in<br />
concentrations as discussed below.<br />
Since only two samplers were deployed in each of GB and<br />
CR during each sampling period (Table S1, Table S3), temporal<br />
trends could not be statistically assessed. However, few<br />
temporal trends were evident. In GB effluent, the temazepam<br />
concentration in March was double (range >100 ng/L) that in<br />
December/January, and in CR effluent, the naproxen concentration<br />
increased seven-fold over the winter (range >700 ng/L).<br />
Also in CR effluent, the roxithromycin concentration in<br />
December/January was nine times higher than that in both<br />
February/March and March (range 0.8 ng/L), whereas the<br />
opposite trend was seen for sulfamethazine (increased three<br />
times over time, range 3 ng/L). Fenoprofen, ketoprofen, sulfadimethoxine,<br />
sulfisoxazole, and tadalafil were not detected<br />
in either urban WWTP above their respective DLs.<br />
When the TWA concentrations for each drug were pooled<br />
for each WWTP (n ¼ 6 for each WWTP), statistical comparisons<br />
revealed no significant differences between urban<br />
effluent concentrations for any drug (Table 1). As with the<br />
aerated lagoon effluent, there was little evidence of changes in<br />
drug concentration over the monthly integrated sampling<br />
periods. Seasonality in removal efficiency and/or or drug use<br />
patterns could not be assessed for the tertiary plants as only<br />
winter data was collected. The concentrations of each drug<br />
were similar between the two tertiary plants (Table 1), likely<br />
a reflection of similar per capita drug and water use between<br />
the populations since treatment processes are nearly identical<br />
in the two plants (Letzel et al., 2009). We do not expect<br />
concentrations in these large-community WWTPs to be<br />
influenced significantly by use changes in a small number of<br />
individuals; using the same assumptions as in the previous<br />
section, the same individual’s carbamazepine dose change<br />
would result in a 0.02 ng/L increase in GB concentrations, an<br />
insignificant amount.<br />
3.1.3. Comparison of concentrations in aerated lagoon and<br />
tertiary treated effluents in winter<br />
Drug concentrations in effluents depend on a number of<br />
factors. Removal efficiency during treatment is known to vary<br />
by drug and by WWTP (Batt et al., 2007; Letzel et al., 2009).<br />
Differences in sewage retention time (Batt et al., 2007; Göbel<br />
et al., 2007; Kimura et al., 2007), and/or the co-metabolic<br />
activity of microbes (Joss et al., 2006; Kim and Carlson, 2007;<br />
Kimura et al., 2007; Yu et al., 2006) can produce differences in<br />
the efficacy of wastewater treatment for removal of some<br />
drugs. For other chemicals, sorption capacity, rather than<br />
retention time or biodegradation, can be most important for<br />
removal during treatment (Carballa et al., 2007; Joss et al.,<br />
2006). Differences in sunlight attenuation while wastewater is<br />
in outdoor tanks (tertiary) and lagoons may also play a role for<br />
some drugs, as the turbidity of the wastewater may vary, thus<br />
varying the amount of sunlight attenuation and thus the<br />
amount available for photolysis. Drugs in GB and CR were<br />
exposed to sunlight for less than 24 h, and received at least 8 s<br />
of low power UV exposure (City of Edmonton, 2009) for<br />
disinfection of effluent before release. The drugs in LLB are<br />
exposed to sunlight for as long as the daylight hours of 90 days<br />
(Duigou, 2006), providing more time for photolysis.
Temperature can also play a role, as previously noted. In<br />
addition, both GB and CR use secondary activated sludge<br />
treatment. These tanks are warmed in the winter to maintain<br />
efficiency, especially in cold climates such as that in Alberta.<br />
Thus, the water temperatures in those tanks may be higher<br />
than that in the LLB lagoon pond under ice (w4 C). As a result,<br />
biological activity capable of biodegrading drugs may be more<br />
active in GB and CR compared with LLB.<br />
Notwithstanding these potential differences in incidental<br />
removal during treatment, the concentrations of several drugs<br />
detected at all three plants were generally comparable, with<br />
CR having the highest total concentration (Table 1). Most<br />
drugs (17 out of 24 detectable) were found in all three WWTPs<br />
in winter (Table 1). No differences in winter pooled TWA<br />
concentrations (three winter sampling periods with n ¼ 9 for<br />
LLB, and n ¼ 6 for GB and for CR) among the three effluents for<br />
atenolol, celecoxib, codeine, metoprolol, naproxen, paroxetine,<br />
temazepam, triclosan and trimethoprim. For eight other<br />
drugs, the concentrations in urban effluents were from 2 to 12<br />
times higher than that in rural effluent, with concentration<br />
differences that ranged from 9 to 1600 ng/L (Table 1). For citalopram<br />
and gemfibrozil, the concentrations at GB were<br />
greater than those at LLB. For carbamazepine, citalopram,<br />
clarithromycin, diclofenac, erythromycin, gemfibrozil,<br />
propranolol, and sotalol, the concentrations in CR effluent<br />
were greater than those in LLB effluent. Between the two<br />
urban WWTPs, the concentrations of citalopram, propranolol<br />
and sotalol at CR were greater than those at GB. Several drugs<br />
(fluoxetine, paroxetine, roxithromycin, sildenafil, sulfamethazine,<br />
sulfapyridine, vardenafil), infrequently or not detected<br />
in the rural WWTP effluent, were consistently detected in<br />
the urban WWTP effluents (Tables S2–S3). These drugs were<br />
generally present at low concentrations (100 ng/L). The discrepancy with sulfapyridine may be<br />
explained by differences in drug use among the communities,<br />
but is more likely caused by treatment differences. Sulfapyridine<br />
is generally administered via sulfasalazine, which is<br />
cleaved in the colon to release sulfapyridine (10–35%), while<br />
another 30–50% is released as sulfasalazine and a metabolite<br />
(Neumann, 1989), both of which may be cleaved in biological<br />
wastewater treatment to release sulfapyridine (Göbel et al.,<br />
2005). It is possible that such cleavage does not occur in the<br />
aerated lagoons, leading to a lower concentration of sulfapyridine<br />
in LLB effluent. For the other drugs, differences<br />
reflect differences between the communities in terms of water<br />
and/or drug use, and/or a difference in the fate of drugs in the<br />
WWTPs (Hua et al., 2006; Kim and Carlson, 2007; Loraine and<br />
Pettigrove, 2006; Sacher et al., 2008; Vieno et al., 2005).<br />
3.2. Loadings of pharmaceuticals to surface waters from<br />
aerated lagoon and tertiary treated effluents<br />
Average daily loadings (Table S4) in winter were calculated for<br />
each plant by multiplying the average daily water discharge by<br />
the TWA concentration of each drug during each sampling<br />
period. Total winter loadings for GB, CR, and LLB were 700, 400<br />
and 4 g/day, respectively. For the winter sampling periods,<br />
average daily loadings from GB were generally twice that<br />
from CR. Average daily loadings from LLB were on the order<br />
water research 44 (2010) 533–544 539<br />
of mg/day. This was at least an order of magnitude lower than<br />
loadings from GB and CR, as expected based on the populations<br />
served (Letzel et al., 2009). The above result suggests<br />
that population-weighted per capita TWA daily loadings<br />
would be similar among the three WWTPs. Indeed, this was<br />
the case for most drugs, for which there were no statistically<br />
significant differences among population-weighted average<br />
daily loadings (Fig. 3), e.g., atenolol, celecoxib, clarithromycin,<br />
codeine, gemfibrozil, metoprolol, naproxen, temazepam,<br />
triclosan, and trimethoprim. Urban per capita average daily<br />
loadings were greater than those from the rural WWTP in<br />
a few cases: CR was greater than LLB (carbamazepine,<br />
citalopram, diclofenac, propranolol, erythromycin, sotalol),<br />
GB was greater than LLB (citalopram), while the per capita<br />
loading from CR was greater than GB for three drugs<br />
(citalopram, propranolol, sotalol). These per capita loading<br />
differences, although statistically significant, may not be<br />
practically significant, as all were all less than an order of<br />
magnitude (two to eight times) and the absolute differences<br />
ranged from 2 (propranolol, CR > GB) to 390 mg/person per day<br />
(diclofenac, CR > LLB). All of these differences in loadings<br />
between WWTPs were also noted as differences in concentration.<br />
Loading differences were not found for clarithromycin<br />
and gemfibrozil, for which differences in concentration were<br />
found among WWTPs. Thus, we conclude that population is<br />
a good predictor of human-use drug release into receiving<br />
waters, at least for over-the-counter or long-term prescription<br />
drugs such as those in this study.<br />
Such predictions are likely to be more accurate when based<br />
on direct measurements, particularly from demographically<br />
similar regions (Stamm et al., 2008), and most accurate when<br />
also based on comparable wastewater treatment. In a ten-year<br />
study of drugs in the Rhine, concentrations of diclofenac and<br />
carbamazepine were found to be relatively constant and given<br />
consistency in use patterns, and although raw wastewater<br />
concentrations were not measured, the authors attributed any<br />
temporal variations to differences in WWTP removal<br />
efficiency (Sacher et al., 2008). Their conclusion was that any<br />
measures that had been implemented to reduce loadings were<br />
thus far ineffective. Because our estimates are based on<br />
continuous measurements of pharmaceutical concentrations<br />
(Letzel et al., 2009; Miao et al., 2005; Stamm et al., 2008), rather<br />
than instantaneous grab sampling, they are likely to be more<br />
robust as TWA concentrations from passive samplers are less<br />
susceptible to short-term fluctuations in concentration over<br />
periods shorter than the sampler deployment periods. Efforts<br />
have been undertaken to find a suitable mechanism for<br />
removal of drugs from the waste stream (Ikehata et al., 2008),<br />
and such efforts should continue, because as our results<br />
further highlight, even sophisticated WWTPs like GB and CR<br />
produce wastewater with drug concentrations that are<br />
comparable with water treated by aerated lagoons.<br />
As previously noted, loadings at WWTPs serving small<br />
communities are potentially more sensitive to small changes<br />
in use patterns than larger communities. The twice-a-day,<br />
100 mg/dose from one individual at LLB would result in<br />
a loading increase of 50 mg/person per day at 100% excretion<br />
and no WWTP removal, or 1 mg/person per day (4% increase,<br />
Fig. 3) using the carbamazepine assumptions previously<br />
given. At GB, the equivalent increase in loadings would
540<br />
water research 44 (2010) 533–544<br />
Fig. 3 – Time-weighted average per capita daily loadings (:, mg/person per day) and enantiomer fractions (,, EF), both ± SD,<br />
of drugs at three WWTPs: GB, Gold Bar; CR, Capital Region (n [ 6), LLB, Lac La Biche. Differences in daily loadings are<br />
indicated by lower-case letters while EF Statistically significant differences are indicated by upper-case letters. Plots without<br />
letters do not have loadings or EFs, statistically different from one another over sampling periods. Asterisks indicate EFs<br />
that are significantly non-racemic compared with racemic standards. Error bars (SD) represent propagated error from<br />
replicate POCIS, analytical processing and instrumental uncertainty, and error associated with POCIS sampling rates.
generally be insignificant (Fig. 3), at 0.27 and 0.006 mg/person<br />
per day, respectively.<br />
3.3. Enantiomer fractions of chiral drugs in<br />
treated effluents<br />
3.3.1. Temporal trends in enantiomer fractions of chiral drugs<br />
in aerated lagoon effluent<br />
We measured EFs in LLB to discern whether temporal changes<br />
existed in enantiomer composition of chiral drugs released by<br />
WWTPs. Changes in EF may arise from differential interaction<br />
of chiral drug enantiomers with other chiral chemicals (e.g.,<br />
enzymes). These may result from temporal differences in<br />
enantiomer-specific metabolism of patients (Mehvar and<br />
Brocks, 2001; Schmekel et al., 1999), and/or from enantiomerspecific,<br />
microbial enzyme mediated biotransformations<br />
during wastewater treatment, as previously noted for atenolol<br />
(MacLeod et al., 2007b; Nikolai et al., 2006), citalopram<br />
(MacLeod et al., 2007b) and propranolol (Fono and Sedlak,<br />
2005; MacLeod et al., 2007b; Nikolai et al., 2006), but not<br />
metoprolol (Fono and Sedlak, 2005; MacLeod et al., 2007b;<br />
Nikolai et al., 2006), salbutamol (MacLeod et al., 2007b), or<br />
sotalol (MacLeod et al., 2007b). Because of enantiomer-specific<br />
metabolism by the target organism, chiral pharmaceuticals<br />
may not necessarily enter the environment with the same<br />
enantiomeric composition as the dose. This is in contrast to<br />
chiral persistent organic pollutants, which generally enter the<br />
environment as a racemate, or sometimes as a single enantiomer.<br />
Thus, changes in enantiomeric composition from the<br />
excreted EF, not just from a racemic value of 0.5, would be<br />
environmentally significant as such a change would indicate<br />
post-release biochemical weathering. However, given that<br />
few environmental measurements of drug EFs exist to date,<br />
our discussion here focuses on changes in EF from the racemate,<br />
the most likely enantiomeric composition of formulated<br />
chiral drugs not sold as single enantiomers (e.g., naproxen).<br />
With the exception of tempazepam, chiral drugs were<br />
generally non-racemic in LLB effluent (Fig. 2) compared with<br />
racemic standards (Dunnett’s test). Atenolol was enriched in<br />
( )-atenolol during some sampling periods, with changes over<br />
time, consistent with previous observations in 2005 at LLB<br />
(Nikolai et al., 2006). Citalopram was mainly enriched in the<br />
first-eluted enantiomer except during August/September.<br />
Metoprolol was enriched in the E2-enantiomer when nonracemic,<br />
which was not previously observed at LLB in individual<br />
grab samples (Nikolai et al., 2006). Salbutamol was<br />
always enriched in the second eluted enantiomer, except<br />
during January/March wherein it was E1-salbutamol that was<br />
enriched. Sotalol was always non-racemic, with no change in<br />
EF over time. Temazepam was racemic with one exception,<br />
wherein E 2-temazepam was enriched. Propranolol EFs could<br />
not be measured (supplemental data).<br />
Temporal changes in EF were observed (Tukey’s test) for all<br />
drugs except sotalol in LLB effluent (Fig. 2, Table S5). Without<br />
direct measurement of the EF in influent and during the<br />
treatment process, we cannot conclusively determine<br />
whether EFs in effluent arose from metabolism prior to bodily<br />
excretion or from processes during treatment. The latter may<br />
be plausible, given changes in activity and composition of<br />
microbial consortia during biological sewage treatment from<br />
water research 44 (2010) 533–544 541<br />
temperature changes over the seasons. For drugs such as<br />
citalopram and salbutamol, changes in EF may also reflect<br />
differences in availability or use of single-enantiomer<br />
formulations (Maier et al., 2001). Because of the temporal<br />
changes in EF observed, our results show that an EF measured<br />
at one time point may not necessarily be characteristic of the<br />
enantiomer composition of chiral drugs discharged from<br />
a WWTP, and hence insufficient to characterize exposures in<br />
receiving waters.<br />
3.3.2. Comparison of chiral drug EFs from aerated lagoons<br />
and tertiary treated wastewater<br />
Winter EFs from LLB and from the urban tertiary WWTPs were<br />
compared to gain insight on whether EFs could be predicted by<br />
treatment type, and may possibly be caused by similar human<br />
metabolism and release and/or microbial degradation during<br />
WWTP treatment. Due to low sample sizes, statistical tests and<br />
trend analyses were not carried out on the temporal EF data<br />
from the urban tertiary WWTPs. Instead, winter EFs were pooled<br />
for each drug from GB and CR and compared with EF standard<br />
(Dunnett’s test), to each other (t-test, with Welch’s correction for<br />
unequal variances as needed), and to the pooled EFs from the<br />
three winter sampling periods at LLB (Tukey’s test). Standard<br />
deviations in pooled data reflect propagated variability from<br />
both temporal changes in EFsample and from replicates.<br />
The two tertiary WWTP effluents had similar EFs during<br />
winter for all chiral drugs except sotalol (Fig. 3), suggesting<br />
that both treatment plants were processing those drugs in the<br />
same stereoselective manner during microbial treatment.<br />
At all three WWTPs (Fig. 3), metoprolol EFs were racemic in<br />
effluents, while sotalol was enriched in the E 2-enantiomer and<br />
differed in EF between the tertiary effluents. Racemic metoprolol<br />
was previously observed in both GB and LLB effluents<br />
(MacLeod, 2009; MacLeod et al., 2007b; Nikolai et al., 2006) and<br />
in the effluent of a WWTP in Dallas TX (Fono et al., 2006), with<br />
no change in EF from raw to treated wastewater (MacLeod<br />
et al., 2007b). After correction for enantiomer-specific matrix<br />
effects (MacLeod et al., 2007b), sotalol was previously found to<br />
be racemic in GB effluent (MacLeod et al., 2007b) via grab<br />
samples. This observation suggests that changes in sotalol EFs<br />
might have occurred over time in GB effluent; however, we<br />
cannot unequivocally conclude that this must be the case<br />
given our earlier caveats that a single EF cannot necessarily<br />
characterize the enantiomer composition. No statistically<br />
significant differences were observed for salbutamol EFs in<br />
the three effluents, despite enrichment of E 2-salbutamol in<br />
both tertiary effluents. This observation is largely a result of<br />
the high variability in salbutamol EFs in LLB and CR. In turn,<br />
the variability may be a function of the drug’s input to<br />
WWTPs, given that salbutamol is available as an R-salbutamol-only<br />
formulation, while S-salbutamol is metabolized<br />
slower than its antipode (Schmekel et al., 1999). Citalopram<br />
(Fig. 3) was enriched in the E 1-enantiomer in all effluents, with<br />
differences in EF between tertiary and aerated lagoon output.<br />
As with salbutamol, this difference could be influenced<br />
by differences in prescription patterns, given the availability<br />
of escitalopram, the S-enantiomer formulation. Citalopram<br />
was previously found with similar EFs (0.55–0.65) in GB<br />
effluent (MacLeod et al., 2007b). Temazepam was enriched in<br />
the E2-enantiomer at both GB and LLB, but racemic in CR,
542<br />
wherein its EF was different from that in LLB, but not from that<br />
in GB. No information regarding temazepam EFs in treated<br />
effluent is available, except our previous results (MacLeod,<br />
2009). For atenolol, all effluents had EFs that were different<br />
from one another, with GB containing racemic atenolol, CR<br />
containing more (þ)-atenolol and LLB containing more<br />
( )-atenolol. The EF of atenolol from healthy renal excretion is<br />
0.476 (Mehvar and Brocks, 2001), and 90% of atenolol is<br />
excreted unchanged in urine (Bourne, 1981). Thus, a process<br />
other than healthy renal excretion, and potentially specific to<br />
microbial communities of WWTPs, was influencing atenolol<br />
EFs in effluent. Likewise, the differences in EFs observed<br />
among WWTP effluents is likely due, at least in part, to variations<br />
in enantioselective microbial biotransformation of<br />
drugs during wastewater treatment.<br />
As we have noted, it is not clear what the enantiomeric<br />
composition of drugs may be upon excretion from humans,<br />
but non-racemic release is likely. As such, we must delineate<br />
between environmental biotransformation and in changes in<br />
enantiomeric composition of the source material in assessing<br />
the environmental fate of drug enantiomers, as has been done<br />
for chiral pesticides (e.g., metolachlor) also released in nonracemic<br />
amounts (Kurt-Karakus et al., 2008).<br />
4. Conclusions<br />
Although other studies have used POCIS to monitor drugs in<br />
treated wastewater, none have done so for more than 30 days<br />
(Jones-Lepp et al., 2004; Zhang et al., 2008). Herein, we monitored<br />
LLB effluent for 271 days and found that the concentration<br />
of drugs in effluent varied over seasonal time scales. Other<br />
studies have mainly focused on larger urban centers. By<br />
comparing drugs in both rural and urban wastewater effluents,<br />
we find that pharmaceutical loadings to aquatic ecosystems<br />
scale with population. However, we also found that the<br />
concentrations of drugs in effluent from WWTPs with biological<br />
activated sludge treatment and UV disinfection were<br />
comparable with concentrations in effluent from simpler<br />
treatment methods like aerated lagoons. Although absolute<br />
loadings were generally lower, drugs were still detectable in<br />
effluent from small communities. Thus, if pharmaceutical<br />
contamination of surface water presents a risk to aquatic life,<br />
that risk is also present near smaller centers, especially those<br />
which are landlocked and not discharging to a large surface<br />
water body to provide effluent dilution.<br />
Overall, temporal changes in EF were found for all chiral<br />
drugs except sotalol, with temporal changes in concentration<br />
for all chiral drugs except atenolol and metoprolol. The<br />
concentration of atenolol and metoprolol in LLB effluent did<br />
not change over time, but the EFs of these drugs did change.<br />
Conversely, although the concentration of sotalol in LLB<br />
effluent changed over time, the EF remained constant.<br />
Although temporal changes in concentration may not be<br />
occurring, the relative proportions of enantiomers being<br />
released to the environment may be changing over time. Also,<br />
for some drugs the EF may remain constant, but the concentration<br />
of the drug may change.<br />
Similar conclusions can be drawn from the comparison<br />
between EFs and per capita daily loadings among the three<br />
water research 44 (2010) 533–544<br />
WWTPs. Loading and EF differences were found among<br />
WWTPs for citalopram and sotalol, while for atenolol and<br />
temazepam, there were no differences in loading but there<br />
were changes in EF, and for metoprolol there were no differences<br />
in loading or EF among the three WWTPs. Simply put,<br />
although concentrations and loadings may be accurately<br />
predicted based on population-weighted measured effluent<br />
concentrations in other treatment plants, the EFs of chiral<br />
drugs released by one WWTP may not necessarily reflect that<br />
released by another, probably from site-specific biochemical<br />
activity. Thus, enantiomer-specific monitoring of chiral drugs<br />
may be useful, particularly for drugs with enantiomer-specific<br />
toxicity to aquatic life (Stanley et al., 2006).<br />
Acknowledgements<br />
We thank D. Bleackley, V. Cooper, M. Ross, B. Asher and<br />
E. McClure (University of Alberta), R. Litwinow and D. Seehagel<br />
(GB WWTP), D. Cikaluk (CR WWTP) and G. Siebold (LLB WWTP)<br />
for sampling assistance, and H. Li (Trent University) for<br />
providing unpublished POCIS sampling rates for citalopram and<br />
sotalol. Funding was provided to C.S.W. from the Canada<br />
Research Chairs Program, Canada’s Natural Sciences and Engineering<br />
Research Council (NSERC), and the Society of Environmental<br />
Toxicology and Chemistry (SETAC) Early Career Award<br />
for Applied Ecological Research, cosponsored by the American<br />
Chemistry Council; and to S.L.M. as fellowships from NSERC,<br />
Alberta Ingenuity Fund, and the American Chemical Society<br />
Division of Analytical Chemistry, sponsored by DuPont.<br />
Appendix.<br />
Supplementary data<br />
Information about sampling locations and times, analytical<br />
and data treatment procedures, concentrations, loadings, and<br />
EFs are available in the online supplementary data to<br />
Supplementary data associated with this article can be found<br />
in the online version at doi:10.1016/j.watres.2009.09.056.<br />
references<br />
Alberta Capital Region Wastewater Commission, 2009. http://<br />
www.acrwc.ab.ca accessed February 25.<br />
Alvarez, D.A., Petty, J.D., Huckins, J.N., Jones-Lepp, T.L.,<br />
Getting, D.T., Goddard, J.P., Manahan, S.E., 2004. Development<br />
of a passive, in situ, integrative sampler for hydrophilic organic<br />
contaminants in aquatic environments. Environ. Toxicol.<br />
Chem. 23, 1640–1648.<br />
Andreozzi, R., Raffaele, M., Nicklas, P., 2003. Pharmaceuticals in<br />
STP effluents and their solar photodegradation in the aquatic<br />
environment. Chemosphere 50, 1319–1330.<br />
Batt, A.L., Kim, S., Aga, D.S., 2007. Comparison of the occurrence of<br />
antibiotics in four full-scale wastewater treatment plants with<br />
varying designs and operations. Chemosphere 68, 428–435.<br />
Bourne, G.R., 1981. The metabolism of b-adrenoreceptor blocking<br />
drugs. In: Bridges, J.W., Chasseaud, L.F. (Eds.), Progress in Drug<br />
Metabolism 6. John Wiley & Sons Ltd, pp. 77–110.
Carballa, M., Omil, F., Ternes, T., Lema, J.M., 2007. Fate of<br />
pharmaceutical and personal care products (PPCPs) during<br />
anaerobic digestion of sewage sludge. Water Res. 41, 2139–2150.<br />
Chen, M., Ohman, K., Metcalfe, C., Ikonomou, M.G., Amatya, P.L.,<br />
Wilson, J., 2006. Pharmaceuticals and endocrine disruptors in<br />
wastewater treatment effluents and in the water supply<br />
system of Calgary, Alberta, Canada. Water Qual. Res. J. Can.<br />
41, 351–364.<br />
City of Edmonton, 2009. http://www.edmonton.ca. accessed<br />
February 25.<br />
Coetsier, C.M., Spinelli, S., Lin, L., Roig, B., Touraud, E., 2009.<br />
Discharge of pharmaceutical products (PPs) through<br />
a conventional biological sewage treatment plant: MECs vs<br />
PECs? Environ. Int 35, 787–792.<br />
Conkle, J.L., White, J.R., Metcalfe, C.D., 2008. Reduction of<br />
pharmaceutically active compounds by a lagoon wetland<br />
wastewater treatment system in Southeast Louisiana.<br />
Chemosphere 73, 1741–1748.<br />
Daughton, C.G., Ternes, T.A., 1999. Pharmaceuticals and personal<br />
care products in the environment: agents of subtle change?<br />
Environ. Health. Perspect 107, 907–938.<br />
Duigou, E., 2006. Town of Lac La Biche, Personal communication.<br />
EMEA, 2006. Note for guidance on environmental risk assessment of<br />
medicinal products for human use. The European Agency for the<br />
Evaluation of Medicinal Products (EMEA) CMPC/SWP/4447/00.<br />
Environment Canada, 2009. http://www.ec.gc.ca/soer-ree/<br />
English/Indicator_series/new_issues.cfm?issue_id¼7&tech_<br />
id¼28#bio_pic. accessed February 27.<br />
Fent, K., Weston, A.A., Caminada, D., 2006. Ecotoxicology of<br />
human pharmaceuticals. Aquat. Toxicol 76, 122–159.<br />
Fono, L.J., Sedlak, D.L., 2005. Use of the chiral pharmaceutical<br />
propranolol to identify sewage discharges into surface waters.<br />
Environ. Sci. Technol 39, 9244–9252.<br />
Fono, L.J., Kolodziej, E.P., Sedlak, D.L., 2006. Attenuation of<br />
wastewater-derived contaminants in a effluent-dominated<br />
river. Environ. Sci. Technol 40, 7257–7262.<br />
Göbel, A., McArdell, C.S., Joss, A., Siegrist, H., Giger, W., 2007. Fate<br />
of sulfonamides, macrolides, and trimethoprim in different<br />
wastewater treatment methodologies. Sci. Total Environ 372,<br />
361–371.<br />
Göbel, A., Thomsen, A., McArdell, C.S., Joss, A., Giger, W., 2005.<br />
Occurrence and sorption behavior of sulfonamides,<br />
macrolides, and trimethoprim in activated sludge treatment.<br />
Environ. Sci. Technol 39, 3981–3989.<br />
Halling-SBrensen, B., Nielsen, S.N., Lanzky, P.F., Ingerslev, F.,<br />
Holten-LützhBft, H.C., JBrgensen, S.E., 1998. Occurrence, fate<br />
and effects of pharmaceutical substances in the environment<br />
– a review. Chemosphere 36, 357–393.<br />
Harner, T., Wiberg, K., Norstrom, R., 2000. Enantiomer fractions<br />
are preferred to enantiomer ratios for describing chiral<br />
signatures in environmental analysis. Environ. Sci. Technol<br />
34, 218–220.<br />
Hua, W.Y., Bennett, E.R., Miao, X.-S., Metcalfe, C.D., Letcher, R.J.,<br />
2006. Seasonality effects on pharmaceuticals and s-triazine<br />
herbicides in wastewater effluent and surface water from the<br />
Canadian side of the Upper Detroit River. Environ. Toxicol.<br />
Chem. 25, 2356–2365.<br />
Ikehata, K., El-Din, M.G., Snyder, S.A., 2008. Ozonation and<br />
advanced oxidation treatment of emerging organic pollutants<br />
in water and wastewater. Ozone Sci. Eng 30, 21–26.<br />
Jones-Lepp, T.L., Alvarez, D.A., Petty, J.D., Huckins, J.N., 2004.<br />
Polar organic chemical integrative sampling and liquid<br />
chromatography-electrospray/ion-trap mass spectrometry for<br />
assessing selected prescription and illicit drugs in treated<br />
sewage effluents. Arch. Environ. Contam. Toxicol 47, 427–439.<br />
Joss, A., Zabcynski, S., Göbel, A., Hoffmann, B., Löffler, D.,<br />
McArdell, C.S., Ternes, T.A., Thomsen, A., Siegrist, H., 2006.<br />
Biological degradation of pharmaceuticals in municipal<br />
water research 44 (2010) 533–544 543<br />
wastewater treatment: proposing a classification scheme.<br />
Water Res. 40, 1686–1696.<br />
Kim, S.-C., Carlson, K., 2007. Temporal and spatial trends in<br />
the occurrence of human and veterinary antibiotics in<br />
aqueous and river sediment matrices. Environ. Sci. Technol<br />
41, 50–57.<br />
Kimura, K., Hara, H., Watanabe, Y., 2007. Elimination of selected<br />
acidic pharmaceuticals from municipal wastewater by an<br />
activated sludge system and membrane bioreactors. Environ.<br />
Sci. Technol 41, 3708–3714.<br />
Kolpin, D.W., Furlong, E.T., Meyer, M.T., Thurman, E.M., Zaugg, S.D.,<br />
Barber, L.B., Buxton, H.T., 2002. Pharmaceuticals, hormones,<br />
and other organic wastewater contaminants in U.S. streams,<br />
1999–2000: a national reconnaissance. Environ. Sci. Technol 36,<br />
1202–1211.<br />
Kurt-Karakus, P.B., Bidleman, T.F., Muir, D.C.G., Cagampan, S.J.,<br />
Struger, J., Sverko, E., Small, J.M., Jantunen, L.M., 2008. Chiral<br />
current-use herbicides in Ontario streams. Environ. Sci.<br />
Technol 42, 8452–8458.<br />
Lajeunesse, A., Gagnon, C., Sauvé, S., 2008. Determination of basic<br />
antidepressants and their N-desmethyl metabolites in raw<br />
sewage and wastewater using solid-phase extraction and<br />
liquid chromatography-tandem mass spectrometry. Anal.<br />
Chem. 80, 5325–5333.<br />
Letzel, M., Metzner, G., Letzel, T., 2009. Exposure assessment of<br />
the pharmaceutical diclofenac based on long-term<br />
measurements of the aquatic input. Environ. Int 35, 363–368.<br />
Lin, A.Y.-C., Reinhard, M., 2005. Photodegradation of common<br />
environmental pharmaceuticals and estrogens in river water.<br />
Environ. Toxicol. Chem. 24, 1303–1309.<br />
Lin, A.Y.-C., Plumlee, M.H., Reinhard, M., 2006. Natural<br />
attenuation of pharmaceuticals and alkylphenol<br />
polyethoxylate metabolites during river transport:<br />
photochemical and biological transformation. Environ.<br />
Toxicol. Chem. 25, 1458–1464.<br />
Lishman, L., Smyth, S.A., Sarafin, K., Kleywegt, S., Toito, J.,<br />
Peart, T., Lee, B., Servos, M., Beland, M., Seto, P., 2006.<br />
Occurrence and reductions of pharmaceuticals and personal<br />
care products and estrogens by municipal wastewater<br />
treatment. Sci. Total Environ 367, 544–588.<br />
Lissemore, L., Hao, C., Yang, P., Sibley, P.K., Mabury, S.,<br />
Solomon, K.R., 2006. An exposure assessment for selected<br />
pharmaceuticals within a watershed in Southern Ontario.<br />
Chemosphere 64, 717–729.<br />
Loraine, G.A., Pettigrove, M.E., 2006. Seasonal variations in<br />
concentrations of pharmaceuticals and personal care<br />
products in drinking water and reclaimed wastewater in<br />
Southern California. Environ. Sci. Technol 40, 687–695.<br />
MacLeod, S.L., 2009. Pharmaceutical occurrence and fate in<br />
wastewater and receiving surface waters in two Alberta<br />
watersheds. PhD dissertation. University of Alberta,<br />
Edmonton, Alberta.<br />
MacLeod, S.L., McClure, E.L., Wong, C.S., 2007a. Laboratory<br />
calibration and field deployment of the Polar Organic<br />
Chemical Integrative Sampler for pharmaceuticals and<br />
personal care products in wastewater and surface water.<br />
Environ. Toxicol. Chem. 26, 2517–2529.<br />
MacLeod, S.L., Sudhir, P., Wong, C.S., 2007b. Stereoisomer<br />
analysis of wastewater-derived beta-blockers, selective<br />
serotonin re-uptake inhibitors, and salbutamol by highperformance<br />
liquid chromatography-tandem mass<br />
spectrometry. J. Chromatogr. A 1170, 23–33.<br />
Maier, N.M., Franco, P., Lindner, W., 2001. Separation of<br />
enantiomers: needs, challenges, perspectives. J. Chromatogr.<br />
A 906, 3–33.<br />
Managaki, S., Murata, A., Takada, H., Tuyen, B.C., Chiem, N.H.,<br />
2007. Distribution of macrolides, sulfonamides, and<br />
trimethoprim in tropical waters: ubiquitous occurrence of
544<br />
veterinary antibiotics in the Mekong Delta. Environ. Sci.<br />
Technol 41, 8004–8010.<br />
McArdell, C.S., Molnar, E., Suter, M.J.-F., Giger, W., 2003.<br />
Occurrence and fate of macrolide antibiotics in wastewater<br />
treatment plants and in the Glatt Valley Watershed.<br />
Switzerland. Environ. Sci. Technol 37, 5479–5486.<br />
Mehvar, R., Brocks, D.R., 2001. Stereospecific pharmacokinetics<br />
and pharmacodynamics of beta-adrenergic blockers in<br />
humans. J. Pharm. Pharm. Sci. 4, 185–200.<br />
Miao, X.-S., Yang, J.-J., Metcalfe, C.D., 2005. Carbamazepine and its<br />
metabolites in wastewater and in biosolids in a municipal<br />
wastewater treatment plant. Environ. Sci. Technol 39, 7469–7475.<br />
Neumann, J., 1989. Untersuchungen zur Bioverfügbarkeit und<br />
Pharmakokinetik von Sulfasalazin und seinen Metaboliten.<br />
PhD thesis. Free University of Berlin.<br />
Nikolai, L.N., McClure, E.L., MacLeod, S.L., Wong, C.S., 2006.<br />
Stereoisomer quantification of the b-blocker drugs atenolol,<br />
metoprolol, and propranolol in wastewaters by chiral highperformance<br />
liquid chromatography-tandem mass<br />
spectrometry. J. Chromatogr. A 1131, 103–109.<br />
Sacher, F., Ehmann, M., Gabriel, S., Graf, C., Brauch, H.-J., 2008.<br />
Pharmaceutical residues in the river Rhine-results of a onedecade<br />
monitoring programme. J. Environ. Monit. 10, 664–670.<br />
Schmekel, B., Rydberg, I., Norlander, B., Sjöswärd, K.N., Ahlner, J.,<br />
Anderson, R.G.G., 1999. Stereoselective pharmacokinetics of<br />
S-salbutamol after administration of the racemate in healthy<br />
volunteers. Eur. Respir. J 13, 1230–1235.<br />
Shaw, M., Mueller, J.F., 2009. Time integrative passive sampling:<br />
how well do Chemcatchers integrate fluctuating pollutant<br />
concentrations? Environ. Sci. Technol 43, 1443–1448.<br />
Siebold, G., 2008. Lac La Biche County, Personal communication.<br />
water research 44 (2010) 533–544<br />
Stamm, C., Alder, A.C., Fenner, K., Hollender, J., Kraus, M.,<br />
McArdell, C.S., Ort, C., Schneider, M.K., 2008. Spatial and<br />
temporal patterns of pharmaceuticals in the aquatic<br />
environment. Geog. Compass 2/3, 920–955.<br />
Stanley, J.K., Ramirez, A.J., Mottaleb, M., Chambliss, C.K.,<br />
Brooks, B.W., 2006. Enantiospecific toxicity of the b-blocker<br />
propranolol to Daphnia magna and Pimephales promelas.<br />
Environ. Toxicol. Chem. 25, 1780–1786.<br />
Statistics Canada, 2009. http://www12.statcan.ca. accessed<br />
February 25.<br />
Stülten, D., Zühlke, S., Lamshöft, M., Spiteller, M., 2008.<br />
Occurrence of diclofenac and selected metabolites in sewage<br />
effluents. Sci. Total Environ 405, 310–316.<br />
Vieno, N.M., Tuhkanen, T., Kronberg, L., 2005. Seasonal variation<br />
in the occurrence of pharmaceuticals in effluents from<br />
a sewage treatment plant and in the recipient water. Environ.<br />
Sci. Technol 39, 8220–8226.<br />
Vasskog, T., Anderssen, T., Pedersen-Bjergaard, S., Kallenborn, R.,<br />
Jensen, E., 2008. Occurrence of selective serotonin reuptake<br />
inhibitors in sewage and receiving waters at Spitsbergen and<br />
in Norway. J. Chromatogr. A 1185, 194–205.<br />
Wong, C.S., 2006. Environmental fate processes and biochemical<br />
transformations of chiral emerging organic pollutants. Anal.<br />
Bioanal. Chem. 386, 544–558.<br />
Yu, J.T., Bouwer, E.J., Coelhan, M., 2006. Occurrence and<br />
biodegradability studies of selected pharmaceuticals and personal<br />
care products in sewage effluent. Agric. Water Mgmt 86, 72–80.<br />
Zhang, Z., Hibberd, A., Zhou, J.L., 2008. Analysis of emerging<br />
contaminants in sewage effluent and river water:<br />
comparison between spot and passive sampling. Anal. Chim.<br />
Acta 607, 37–44.
Degradation of fifteen emerging contaminants at mgL L1 initial<br />
concentrations by mild solar photo-Fenton in MWTP effluents<br />
N. Klamerth a,b , L. Rizzo c , S. Malato a, *, Manuel I. Maldonado a ,<br />
A. Agüera b , A.R. Fernández-Alba b<br />
a<br />
Plataforma Solar de Almería-CIEMAT. Carretera Senés km 4, 04200 Tabernas (Almería), Spain<br />
b<br />
Pesticide Residue Research Group, University of Almería, 04120 Almería, Spain<br />
c<br />
Department of Civil Engineering, University of Salerno, via Ponte don Melillo, 84084 Fisciano (SA), Italy<br />
article info<br />
Article history:<br />
Received 9 March 2009<br />
Received in revised form<br />
23 September 2009<br />
Accepted 28 September 2009<br />
Available online 4 October 2009<br />
Keywords:<br />
Emerging contaminants<br />
Pharmaceuticals treatment<br />
Photo-Fenton<br />
Solar photocatalysis<br />
1. Introduction<br />
abstract<br />
Due to their growing use, pharmaceuticals, like the antiinflammatory<br />
ibuprofen, the antibiotic flumequine and the<br />
antiepileptic carbamazepine, endocrine disruptors like<br />
bisphenol A and atrazine, personal care products like oxybenzone<br />
and parabens (PHBA), synthetic musks and<br />
fragrances like musk xylene and galaxolide, pesticides like<br />
isoproturon and endosulphan and illicit drugs like THC aznd<br />
cocaine, to name just a few, and other xenobiotic substances,<br />
are found in increasing quantities in wastewater, surface<br />
water, and even in drinking water (Kasprzyk-Hordern et al.,<br />
2009; Kim et al., 2007; Mitch et al., 2003; Esplugas et al., 2007;<br />
Ternes, 1998). Since the use of these substances cannot be<br />
controlled or eliminated as they are ever present in our daily<br />
water research 44 (2010) 545–554<br />
Available at www.sciencedirect.com<br />
journal homepage: www.elsevier.com/locate/watres<br />
* Corresponding author. Tel.: þ34 950387940; fax: þ34 950365015.<br />
E-mail address: sixto.malato@psa.es (S. Malato).<br />
0043-1354/$ – see front matter ª 2009 Elsevier Ltd. All rights reserved.<br />
doi:10.1016/j.watres.2009.09.059<br />
The degradation of 15 emerging contaminants (ECs) at low concentrations in simulated and<br />
real effluent of municipal wastewater treatment plant with photo-Fenton at unchanged pH<br />
and Fe ¼ 5mgL 1 in a pilot-scale solar CPC reactor was studied. The degradation of those<br />
15 compounds (Acetaminophen, Antipyrine, Atrazine, Caffeine, Carbamazepine, Diclofenac,<br />
Flumequine, Hydroxybiphenyl, Ibuprofen, Isoproturon, Ketorolac, Ofloxacin, Progesterone,<br />
Sulfamethoxazole and Triclosan), each with an initial concentration of 100 mgL 1 ,<br />
was found to depend on the presence of CO3 2 and HCO3 (hydroxyl radicals scavengers) and<br />
on the type of water (simulated water, simulated effluent wastewater and real effluent<br />
wastewater), but is relatively independent of pH, the type of acid used for release of<br />
hydroxyl radicals scavengers and the initial H2O2 concentration used. Toxicity tests with<br />
Vibrio fisheri showed that degradation of the compounds in real effluent wastewater led to<br />
toxicity increase.<br />
ª 2009 Elsevier Ltd. All rights reserved.<br />
lives, their release into the environment has to be optimized<br />
and restricted, as they pose risks to the environment, public<br />
health and aquatic systems and they are responsible for<br />
building up microbiological resistance, feminisation of higher<br />
organisms and ecotoxicological issues (Laville et al., 2004).<br />
Particularly relevant examples of such emerging contaminants<br />
(ECs), such as those mentioned above, which are ubiquitously<br />
present in influents and effluents of MWTPs in the<br />
high ng L 1 to low mgL 1 range, do not need to be persistent to<br />
be hazardous, because they are introduced continuously into<br />
the environment (Fono et al., 2006; Jackson and Sutton, 2008;<br />
Nakada et al., 2008; Petrovic et al., 2003).<br />
Conventional MWTPs, typically based on biological<br />
processes, are capable of removing some substances, but nonbiodegradable<br />
compounds may escape the treatment and be
546<br />
released into the environment (Göbel et al., 2007; Carballa<br />
et al., 2004; Ternes et al., 2007). ECs have been found in the<br />
MWTP effluents at mean concentrations ranging from 0.1 to<br />
20 mgL 1 (Castiglioni et al., 2006; Martínez Bueno et al., 2007;<br />
Radjenović et al., 2007; Richardson, 2007; Zhao et al., 2009).<br />
Concern about the growing problem of the continuously rising<br />
concentrations of these compounds must be emphasized, and<br />
therefore, the application of more thorough wastewater<br />
treatment protocols, including the use of new and improved<br />
technologies, is a necessary task.<br />
Reusable water should be free of these persistent, toxic,<br />
endocrine-disrupting or non-biodegradable substances,<br />
(Radjenović et al., 2007; Teske and Arnold, 2008), and therefore,<br />
an effective tertiary treatment is required to remove<br />
these substances completely.<br />
Among the advanced technologies that may be used to<br />
remove these pollutants, (Gogate and Pandit, 2004; Saritha<br />
et al., 2007; Huber et al., 2005) advanced oxidation processes<br />
(AOPs), through the generation of hydroxyl radicals which are<br />
able to mineralise most organic molecules yielding CO 2 and<br />
inorganic ions as final products, are a particularly attractive<br />
option. (Farré et al., 2005; Gebhardt and Schröder, 2007; Gültekin<br />
and Ince, 2007; Ning et al., 2007; Rosenfeldt and Linden,<br />
2004; Rosenfeldt et al., 2007; Ternes et al., 2003) The generation<br />
of the OH radicals can be achieved: electrochemically (Cañizares<br />
et al., 1999; Pelegrini et al., 2001; Zhou et al., 2005),<br />
sonochemically (Mantzavinos et al., 2004; Papadaki et al.,<br />
2004; Lesko et al., 2006), photochemically (Esplugas et al., 2005;<br />
Bali et al., 2003; Bremner et al., 2006), and by homogeneous or<br />
heterogenous catalysis (Zepp et al., 1992; Martinez et al., 2005)<br />
in acid or basic media (Glaze et al., 1987; Hislop and Bolton,<br />
1999; Neyens and Baeyens, 2003). Most of the AOPs make use<br />
of a combination of either oxidants and irradiation (O 3/H 2O 2/<br />
UV), or a catalyst and irradiation (Fe 2þ /H 2O 2; UV/TiO 2). The<br />
drawbacks which make them economically disadvantageous<br />
depend on the specific AOP: (i) High electricity demand (e.g.<br />
ozone and UV-based AOPs), (ii) the relatively large amounts of<br />
oxidants and/or catalysts (e.g. ozone, hydrogen peroxide and<br />
iron-based AOPs), and (iii) the pH operating conditions (e.g.<br />
Fenton and photo-Fenton). This is why, although AOPs are<br />
well known for their capacity for oxidising and mineralising<br />
almost any organic contaminant, commercial applications are<br />
still scarce. Processes like photo-Fenton may be applied to<br />
commercial applications by using solar energy as a light<br />
source, optimizing the pH range and the amounts of oxidant/<br />
catalyst required.<br />
AOP efficiency in the removal of ECs has typically been<br />
studied in demineralised water and bench scale, at initial<br />
concentrations in the milligram-to-gram range, which is not<br />
realistic compared to the concentrations detected in real<br />
water and wastewater (Farré et al., 2005; Lapertot et al., 2007;<br />
Kassinos et al., 2009; Malato et al., 2007). This work focused on<br />
solar photo-Fenton degradation of the ECs typically found in<br />
the effluents of MWTPs, leaving the treated wastewater suitable<br />
for reuse. Moreover, to make the process of interest for<br />
practical applications high iron concentrations (mM range),<br />
excessive amounts of H 2O 2 and a pH under 3 must be avoided<br />
(Pignatello et al., 2006). A new approach aimed at finding<br />
a very mild photo-Fenton treatment (low iron concentration<br />
and H2O2 dose at neutral pH), has been proposed (Moncayo-<br />
water research 44 (2010) 545–554<br />
Lasso et al., 2008; Klamerth et al., 2009). In this paper, a pilotscale<br />
solar photo-Fenton treatment was run at a pH between 6<br />
and 7, with starting concentrations of 5 mg L 1 Fe, and<br />
50 mg L 1 H2O2. Synthetic water (SW), simulated effluent<br />
wastewater (SE) and real effluent wastewater (RE) were tested<br />
in this study to which a mixture of 15 ECs, consisting of<br />
pharmaceuticals, pesticides and personal care products,<br />
selected from a list of 80 compounds found in MWTP effluents<br />
in previous studies (Martínez Bueno et al., 2007), were added at<br />
low concentrations (100 mgL 1 each).<br />
2. Material and methods<br />
2.1. Reagents<br />
All reagents used for chromatographic analyses, acetonitrile,<br />
methanol, and ultrapure (MilliQ) water, were HPLC grade.<br />
Analytical standards for chromatography analyses were<br />
purchased from Sigma-Aldrich. Table 1 and Scheme 1 list the<br />
15 compounds (pharmaceuticals, pesticides and personal care<br />
products) used. Photo-Fenton experiments were performed<br />
using iron sulphate (FeSO4$7H2O), reagent-grade hydrogen<br />
peroxide (30% w/v), sulphuric acid and hydrochloric acid for<br />
carbonate stripping, all provided by Panreac. The filters used<br />
were syringe driven 0.2 mm Millex nylon membrane filters<br />
from Millipore. The reagents used in the preparation of the<br />
synthetic water (NaHCO 3, CaSO 4$2H 2O, MgSO 4, and KCl), and<br />
simulated effluent wastewater (Peptone, Meat extract, Urea,<br />
K 2HPO 4, CaCl 2$2H 2O, MgSO 4$7H 2O and NaCl) were provided by<br />
Panreac.<br />
2.2. Polluted waters<br />
Taking into consideration that typical EC concentrations in the<br />
effluent are in the 0.1–20.0 mgL 1 range, it was decided to work<br />
at 100 mgL 1 , which is a compromise between (i) a high enough<br />
concentration to characterise kinetics using conventional<br />
Table 1 – LOD, LOQ and absorption l of the selected<br />
compounds.<br />
Name LOD<br />
(mgL 1 )<br />
LOQ<br />
(mgL 1 )<br />
Absorption<br />
l [nm]<br />
Acetaminophen 1.4 2.8 245<br />
Antipyrine 2.1 4.2 205<br />
Atrazine 0.6 1.2 223<br />
Caffeine 0.7 1.5 205<br />
Carbamazepine 0.8 1.6 211<br />
Diclofenac 2.7 5.5 277<br />
Flumequine 2.1 4.3 248<br />
Hydroxybiphenyl 1.6 3.2 243<br />
Ibuprofen 2.5 5.0 222<br />
Isoproturon 1.5 3.0 205<br />
Ketorolac 2.1 4.2 321<br />
Ofloxacin 1.3 2.6 295<br />
Progesterone 2.0 4.0 248<br />
Sulfamethoxazole 1.4 2.7 267<br />
Triclosan 5.0 10.0 280
Name Structure Name Structure<br />
Acetaminophen<br />
Ibuprofen<br />
analgesic /<br />
antipyretic<br />
Antipyrine<br />
analgesic<br />
Atrazine<br />
herbicide<br />
Caffeine<br />
stimulant<br />
Carbamazepine<br />
anticonvulsant<br />
Diclofenac<br />
antiinflammatory<br />
Flumequine<br />
broadspectrum<br />
antibiotic<br />
Hydroxybiphenyl<br />
biocide<br />
HO NH<br />
O<br />
N<br />
Cl<br />
N<br />
O<br />
N<br />
N N<br />
O<br />
N<br />
NH 2<br />
O<br />
nonsteroidal<br />
antiinflammatory<br />
Isoproturon<br />
phenylurea<br />
herbicide<br />
H<br />
H<br />
N N N Ketorolac<br />
O<br />
N<br />
H<br />
N<br />
Cl<br />
Cl<br />
N<br />
N<br />
O<br />
OH<br />
antiinflammatory<br />
Ofloxacin<br />
gramnegative<br />
antibiotic<br />
Progesterone<br />
steroid<br />
hormone<br />
Sulfamethoxaz<br />
ole<br />
bacteriostatic<br />
antibiotic<br />
F COOH Triclosan<br />
OH<br />
water research 44 (2010) 545–554 547<br />
O<br />
N<br />
anti-bacterial/<br />
fungal agent<br />
N<br />
N<br />
O<br />
O<br />
F<br />
N<br />
O<br />
H 2 N S<br />
Scheme 1 – Name and structure of the 15 selected compounds.<br />
Cl<br />
O<br />
Cl<br />
O<br />
O<br />
N<br />
H<br />
N<br />
H<br />
N<br />
O<br />
O<br />
O<br />
OH<br />
OH<br />
N<br />
N<br />
O<br />
O<br />
O<br />
O<br />
OH<br />
Cl<br />
OH
548<br />
C/C 0<br />
1.0<br />
0.5<br />
0.0<br />
1.0<br />
0.5<br />
0.0<br />
1.0<br />
0.5<br />
0.0<br />
Fenton<br />
-40 -20 0 20 40 60 80 100<br />
t EXP , min<br />
illumination<br />
t 30W , min<br />
Acetaminophen<br />
Caffeine<br />
Ofloxacine<br />
Antipyrine<br />
Sulfamethoxazole<br />
Carbamazepine<br />
Flumequine<br />
Ketorolac<br />
Atrazine<br />
Isoproturon<br />
Hydroxybenzone<br />
Diclofenac<br />
Ibuprofen<br />
Progesterone<br />
Triclosan<br />
Fig. 1 – Degradation of the 15 ECs (0.1 mg L L1 each) by<br />
photo-Fenton with 5 mg L L1 Fe without pH adjustment in<br />
SW acidified for release of carbonates.<br />
analytical techniques, and (ii) low enough to simulate real<br />
conditions.<br />
Synthetic water (SW), simulated effluent wastewater (SE) and<br />
real effluent wastewater (RE) to which a mixture of the 15 ECs<br />
previously listed had been added at low concentrations<br />
(100 mgL 1 ) were tested. The experimental protocol was designed<br />
to study solar photo-Fenton with a relatively uncomplicated<br />
aqueous matrix (SW) first, before gradually increasing<br />
complexity by studying SE and finally RE, in order to acquire<br />
information on process operation.<br />
Demineralised water used for preparing SW and SE in the<br />
pilot plant was supplied by the Plataforma Solar de Almería<br />
C/C 0<br />
1.0<br />
0.5<br />
0.0<br />
1.0<br />
0.5<br />
0.0<br />
1.0<br />
0.5<br />
0.0<br />
Fenton<br />
t EXP , min<br />
illumination<br />
water research 44 (2010) 545–554<br />
(PSA) distillation plant (conductivity < 10 mScm 1 ,Cl ¼ 0.7–<br />
0.8 mg L 1 , NO3 ¼ 0.5 mg L 1 , organic carbon < 0.5 mg L 1 ).<br />
The preparation of standard moderately-hard freshwater<br />
(SW) was carried out taking into account the characteristics of<br />
ground water in the province of Almería (southern Spain). SW<br />
was prepared by mixing 96 mg L 1 of NaHCO3, 60mgL 1 of<br />
CaSO4$2H2O, 60 mg L 1 of MgSO4, and 4 mg L 1 of KCl (Standard<br />
Methods, 1998). The simulated effluent wastewater (SE)<br />
consists of SW and Peptone (32 mg L 1 ), Meat extract<br />
(22 mg L 1 ), Urea (6 mg L 1 ), K 2HPO 4 (28 mg L 1 ), CaCl 2$2H 2O<br />
(4 mg L 1 ), MgSO 4$7H 2O(2mgL 1 ) and NaCl (7 mg L 1 ) which<br />
derives to an initial DOC (dissolved organic carbon) of<br />
25 mg L 1 (OECD, 1999). RE was taken downstream of the<br />
Almería MWTP secondary biological treatment and used as<br />
received within the next 2 days. Initial COD and DOC were at<br />
60 mg L 1 and 25 mg L 1 , respectively.<br />
2.3. Solar photo-Fenton pilot plant<br />
Photo-Fenton experiments were performed at the Plataforma<br />
Solar de Almería in a pilot compound parabolic collector (CPC)<br />
solar plant designed for solar photocatalytic applications. This<br />
reactor is composed of two 11dL modules with twelve Pyrex<br />
glass tubes (30 mm O.D.) mounted on a fixed platform tilted 37<br />
(local latitude). The water flows (20 L min 1 ) directly from one<br />
module to the other and finally to a 10dL reservoir. The piping<br />
and valves (3 L) between the reactor and the tank are black<br />
HDPE, which is highly resistant to chemicals, weatherproof<br />
and opaque, preventing any photochemical effect from outside<br />
the collectors. The total illuminated area is 3 m 2 , the total<br />
volume (two modules þ reservoir tank þ piping and valves) is<br />
35 L (VT) and the irradiated volume is 22 L (Vi). Solar ultraviolet<br />
radiation (UV) was measured by a global UV radiometer<br />
(KIPP&ZONEN, model CUV 3) mounted on a platform tilted 37<br />
consumed H 2O 2<br />
0 60 120 180 240 300<br />
t 30W , min<br />
Acetaminophen<br />
Caffeine<br />
Ofloxacine<br />
Antipyrine<br />
Sulfamethoxazole<br />
Carbamazepine<br />
Flumequine<br />
Ketorolac<br />
Atrazine<br />
Isoproturon<br />
Hydroxybenzone<br />
Diclofenac<br />
Ibuprofen<br />
Progesterone<br />
Triclosan<br />
Fig. 2 – Degradation of the 15 ECs (0.1 mg L L1 each) by photo-Fenton (5 mg L L1 Fe, no pH adjustment) and hydrogen peroxide<br />
consumption in SE acidified for carbonate release.<br />
120<br />
90<br />
60<br />
30<br />
0<br />
consumed H 2 O 2 , mg L -1
Table 2 – Final concentration of ECs in two different experiments with SE after similar irradiation times, with<br />
[Fe]0 [ 5mgL L1 : 300 min with [H2O2]0 [ 50 mg L L1 and addition of 50 mg L L1 when the initial dose was almost consumed<br />
and so on; 336 min with [H2O2]0 [ 5mgL L1 and addition of H2O2 every 30–60 min until the end of the experiment. Fenton<br />
degradation (%) in the dark and first order kinetic constants (k) for photo-Fenton are also included.<br />
[H2O2]0 ¼ 50 mg L 1<br />
[H2O2]0 ¼ 5mgL 1<br />
(the same as the CPCs). The temperature inside the reactor was<br />
continuously recorded by a temperature probe (Crioterm<br />
PT-100 3H) inserted in the piping. A plant diagram has been<br />
published elsewhere (Kositzi et al., 2004). With Eq. (1), combination<br />
of the data from several days’ experiments and their<br />
comparison with other photocatalytic experiments is possible,<br />
UV Vi<br />
t30W;n ¼ t30W;n 1 þ Dtn ; Dtn ¼ tn tn 1; t0 ¼ 0ðn ¼ 1Þ (1)<br />
30 VT<br />
where tn is the experimental time for each sample, UV is the<br />
average solar ultraviolet radiation (l < 400 nm) measured<br />
between tn 1 and tn, and t30W is a ‘‘normalized illumination<br />
time’’. In this case, time refers to a constant solar UV power of<br />
30 Wm 2 (typical solar UV power on a perfectly sunny day<br />
around noon).<br />
2.4. Experimental setup<br />
t 30W C/C 0 Fenton<br />
deg. X [%]<br />
The mixture of the 15 compounds (100 mgL 1 each) dissolved in<br />
Methanol at 2.5 g L 1 (mother solution) was added directly<br />
(1.4 mL) into the pilot plant and well homogenized by turbulent<br />
recirculation for 15 min. Methanol was used to dissolve the<br />
contaminants (TOC from methanol was 12 mg L 1 ). The pH in<br />
SW was 7.6, in SE 7.8 and in RE 8.0, which favoured bicarbonate<br />
ions, so at the beginning of the process, when the collectors<br />
were still covered, enough acid to remove CO 3 2 and HCO3 ,<br />
known OH radical scavengers, was added. Recirculation time<br />
for this process was between 15 and 60 min. When the total<br />
inorganic carbon (TIC) concentration was below the desired<br />
level (10 mg L 1 ), 50 mg L 1 of peroxide was added and homogenised<br />
by recirculation for 15 min. Finally, Fe 2þ as FeSO4$7H20<br />
was added (5 mg L 1 ). After 15 min of recirculation, in which<br />
the Fenton reaction started, the collectors were uncovered and<br />
photo-Fenton began. Hydrogen peroxide and iron were<br />
measured in every sample taken. The experiments normally<br />
water research 44 (2010) 545–554 549<br />
k [min 1 ] t 30W C/C 0 Fenton<br />
deg. X [%]<br />
lasted as long as there was H 2O 2 to be consumed (usually 3–4 h),<br />
in the case of experiments which lasted longer than that, the<br />
experiment was covered at the end of the day and the next day,<br />
after recirculation for 15 min, a sample was taken, peroxide<br />
was added (50 mg L 1 ) and the covers again removed.<br />
2.5. Analytical measurements<br />
k [min 1 ]<br />
Acetaminophen 261
550<br />
Table 3 – Final EC concentration in RE with<br />
[Fe]0 [ 5mgL L1 ,[H2O2]0 [ 50 mg L L1 after an irradiation<br />
time of t30W [ 276 min. Fenton degradation (%) in the<br />
dark and first order kinetic constant (k) for photo-Fenton<br />
are also included.<br />
t30W C/C0 Fenton k [min<br />
deg. X [%]<br />
1 ]<br />
Acetaminophen 109
eactor (t ExP ¼ 15 min). Removal of all compounds but atrazine<br />
(18%) was in the range between 30% (caffeine) and 60%<br />
(ibuprofen). Fenton degradation was very quick after adding just<br />
5mgL 1 of Fe 2þ but it did not proceed further until the reactor<br />
was uncovered, as Fe 3þ reduction to Fe 2þ was inefficient without<br />
illumination. All the pollutants but triclosan (87%) and atrazine<br />
(77%) were efficiently degraded (removal over 94%) under photo-<br />
Fenton conditions, at t30W ¼ 90 min. Moreover, some pollutants<br />
(ofloxacin and diclofenac) were degraded to below the LOD as<br />
early as t 30W ¼ 32 min. The total amount of H 2O 2 consumed was<br />
37 mg L 1 and the iron concentration decreased due to precipitation<br />
from 5 mg L 1 to 3.1 mg L 1 , probably due to colloidal<br />
aggregation, while the pH varied from an original 7.6–5.3 (after<br />
adding the FeSO4$7H2O) to a final pH ¼ 3.8 at t30W ¼ 90 min.<br />
Similar results were found by adding H2SO4 (56 mg L 1 )<br />
instead of HCl. H2SO4 has certain advantages (handling,<br />
volatility, etc.) over HCl, and Cl ions may act as radical<br />
scavengers. Experiments did not show any significant difference<br />
between using HCl or H2SO4. Indeed, the total concentration<br />
of chloride and sulphate measured in SW was<br />
43 mg L 1 Cl when HCl was added and 145 mg L 1 SO 4 2 when<br />
H 2SO 4 was added respectively. It is well known that inorganic<br />
species (chloride, sulphate, phosphates, etc) are usually<br />
detrimental to the photo-Fenton reaction rate (Pignatello<br />
et al., 2006) due to complexation of these ions with Fe 2þ or<br />
Fe 3þ , or to the scavenging of hydroxyl radicals and formation<br />
of less reactive inorganic ions (De Laat et al., 2004), but it has<br />
also been observed that they significantly reduce the photo-<br />
Fenton reaction rate only if they occur at concentrations over<br />
500 mg L 1 , usually very far from the concentration found in<br />
natural water (Bacardit et al., 2007; Zapata et al., 2009).<br />
3.2. Photo-Fenton tests using SE<br />
The following experiment dealt with the photo-Fenton treatment<br />
of SE, which was characterised by original TOC and COD<br />
of 37 mg L 1 and 80 mg L 1 , respectively. Two different<br />
approaches to the H2O2 dose were studied: (i) 50 mg L 1 were<br />
added at the beginning and again when the initial dose was<br />
almost consumed and so on; (ii) 5 mg L 1 were added every<br />
30–60 min (also depending on consumption) until the end of<br />
the experiment. In this case, 56 mg L 1 H 2SO 4 acid were added.<br />
A recirculation time of 15 min was necessary for carbonates to<br />
be released (TIC < 5mgL 1 ). As can be seen in Fig. 2, degradation<br />
is much slower with both Fenton (t < 0) and photo-<br />
Fenton (t30W > 0) than in the experiment with SW shown in<br />
Fig. 1. The residual concentration of the contaminants at the<br />
end of the experiments (t30W ¼ 300 min and 336 min, respectively)<br />
can be seen in Table 2. The pH varied during the<br />
experiment with high H 2O 2 concentrations from the original<br />
7.6–4.9 at the end, while the iron concentration varied from<br />
5mgL 1 to 3.2 mg L 1 . During the experiment with low H 2O 2<br />
concentrations the pH varied from 8.1 to 4.9, while iron<br />
concentration went from 4.8 to 3.1 mg L 1 . The EC degradation<br />
behaviour did not change significantly, no matter which<br />
hydrogen peroxide dose approach was used, as can be seen in<br />
Table 2, although the overall amount of peroxide consumed<br />
was much lower (110 mg in 50-mg-L 1 additions compared to<br />
52 mg with the 5-mg-L 1 additions). It should be mentioned,<br />
although not relevant to the purpose of the experiments,<br />
water research 44 (2010) 545–554 551<br />
because the main purpose was to degrade ECs, that TOC<br />
mineralisation was rather low in both cases (around 25%).<br />
3.3. Photo – Fenton treatment of RE<br />
The initial DOC, TIC and COD were 36 mg L 1 , 106 mg L 1 and<br />
60 mg L 1 , respectively. In this case, 406 mg L 1 H2SO4 were<br />
added to reach < 20 mg L 1 TIC and recirculation time in the pilot<br />
plant necessary to remove carbonate species was 30 min.<br />
According to previous tests, photo-Fenton treatment at a TIC over<br />
20 mg L 1 leads to very slow degradation. Interestingly, although<br />
the degradation rate of all ECs was faster than in the previous<br />
experiments in SE, as observed in Table 3, DOC degradation was<br />
similar to SE (around 20%). This can be explained by the presence<br />
of humic acids in RE which produce solvated electrons and<br />
hydroxyl radicals upon irradiation (Fukushima and Tatsumi,<br />
2001; Prosen and Zupančič-Kralj, 2005; Auger et al., 2002).<br />
3.4. The effect of photo-Fenton treatment on<br />
V. fisheri toxicity<br />
In Fig. 3,theresultsofV. fisheri toxicity tests with SE and RE after<br />
photo-Fenton experiments are compared. Three different steps<br />
in toxicity behaviour (marked 1, 2 and 3 in Fig. 3) can be<br />
described. Before photo-Fenton starts, SE toxicity (20–30% inhibition)<br />
may be due to the mixture of pollutants as well as to the<br />
formation of intermediates during Fenton in the dark, particularly<br />
from the fast oxidation of some of the ECs. On the contrary,<br />
simultaneous activation (negative inhibition) is also observable<br />
in the RE experiment. This completely different behaviour may<br />
be explained by the RE – nutrients and salt contents, which in<br />
someway favoured the growth of V. fisheri until the toxic effect of<br />
pollutants and their oxidation intermediates was neutralized.<br />
In both SE and RE, when photo-Fenton starts up, from an of<br />
H2O2 dose ¼ 0mgL 1 up to H2O2 dose ¼ 50 mg L 1 , the degradation<br />
of parent pollutants begins forming more toxic intermediates,<br />
which drastically increase the inhibition rate in SE<br />
(Fig. 3, Step 2). The previously favourable conditions of RE<br />
probably can no longer efficiently balance the increasingly<br />
toxic effect, so inhibition start to gradually increase in RE too.<br />
In the last step (3) inhibition with RE is quite constant, probably<br />
because toxic organic intermediates take longer to<br />
mineralise; on the other hand, toxicity in SE almost totally<br />
disappeared. If we compare these results with residual<br />
concentration of the four parent ECs selected for comparison<br />
(selected for their representativeness of the behaviour of the<br />
15 different compounds) it may be observed that photo-Fenton<br />
treatment was more efficient in the RE experiment than in<br />
the SE. So in terms of toxicity, the higher removal rate in RE<br />
experiment may result in a correspondingly larger amount of<br />
more toxic intermediates than in the SE experiment or the<br />
generation of other intermediates from RE organics. Therefore,<br />
toxicity assessment be should taken into account as well<br />
as EC degradation during AOP wastewater treatment.<br />
4. Conclusions<br />
- The experiments showed that ECs at low concentrations<br />
(mgL 1 range) can be successfully degraded to negligible
552<br />
concentrations with solar photo-Fenton at low iron<br />
concentrations (5 mg L 1 ) and low initial H2O2 (50 mg L 1 )<br />
concentrations without adjusting the pH.<br />
- One limiting factor for the degradation rate is the presence of<br />
CO3 2 and HCO3 - which are very efficient OH radical scavengers<br />
and which have to be removed prior to the photo-<br />
Fenton reaction. This can be done either with HCl or H2SO4,<br />
as the range of the final concentration of Cl or SO4 2 does not<br />
influence the reaction. Therefore, either one may be used as<br />
economic/operating reasons recommend, but not for any<br />
kinetic effect. It should be taken into account that around<br />
400 mg L 1 of H 2SO 4 (0.05 V kg 1 ) were necessary for CO 2<br />
stripping in RE, which means around 0.025 V m 3 . However,<br />
these costs should be evaluated on a case by case basis as it<br />
strongly depends on the particular RE.<br />
- Changing the operating conditions by lowering H2O2<br />
concentration (5 mg L 1 ) and adjusted by properly<br />
controlled dosing could lead to comparable degradation of<br />
the compounds with lower peroxide consumption.<br />
- Degradation is faster in RE due to humic acids producing<br />
solvated electrons and hydroxyl radicals upon irradiation,<br />
and these contribute to the radicals produced by photo-<br />
Fenton.<br />
Toxicity from the formation of oxidation intermediates<br />
should be taken into account in the treatment of effluents<br />
from wastewater treatment plants by AOPs, because they<br />
could increase final toxicity.<br />
Acknowledgements<br />
Funding for this work was provided by the Spanish Ministry of<br />
Science and Innovation under the Consolider-Ingenio 2010<br />
programme (Project CSD2006–00044 TRAGUA; http://www.<br />
consolider-tragua.com) and by the Andalusia Regional Government<br />
(Project no. P06-TEP-02329). Nick Klamerth would to thank<br />
the University of Almería and CIEMAT for his Ph. D. research<br />
grant. Luigi Rizzo wishes to thank the University of Salerno<br />
(Italy) and the Province of Salerno for the research grant which<br />
allowed him to work in the above mentioned project at the<br />
Plataforma Solar de Almería. The authors also wish to thank<br />
Mrs. Deborah Fuldauer for English language correction.<br />
references<br />
Auger, J.P., Richard, C., Trubetskaya, O., Trubetskoj, O., Lévèque, J.,<br />
Andreux, F., 2002. Photoinductive efficiency of soil extracted<br />
humic and fulvic acids. Chemosphere 49, 259–262.<br />
Bacardit, J., Stötzner, J., Chamarro, E., Esplugas, S., 2007. Effects of<br />
salinity on the photo-Fenton process. Industrial and<br />
Engineering Chemistry Research 46, 7615–7619.<br />
Bali, U., Catalkaya, E.C., Sengul, F., 2003. Photochemical<br />
degradation and mineralisation of phenol: a comparative<br />
study. Journal of Environmental Science and Health – Part A<br />
Toxic/Hazardous Substances and Environmental Engineering<br />
38, 2259–2275.<br />
Bremner, D.H., Burgess, A.E., Houllemare, D., Namkung, K., 2006.<br />
Phenol degradation using hydroxyl radicals generated from<br />
zero-valent iron and hydrogen peroxide. Applied Catalysis B:<br />
Environmental 63, 15–19.<br />
water research 44 (2010) 545–554<br />
Cañizares, P., Domínguez, J.A., Rodrigo, M.A., Villaseñor, J.,<br />
Rodríguez, F., 1999. Effect of the current intensity in the<br />
electrochemical oxidation of aqueous phenol wastes at an<br />
activated carbon and steel anode. Industrial and Engineering<br />
Chemistry Research 38, 3779–3785.<br />
Carballa, M., Omil, F., Lema, J.M., Llompart, M., Garcia-Jares, C.,<br />
Rodriguez, I., Gomez, M., Ternes, T., 2004. Behaviour of<br />
pharmaceuticals, cosmetics and hormones in a sewage<br />
treatment plant. Water Research 38, 2918–2926.<br />
Castiglioni, S., Bagnati, R., Fanelli, R., Pomati, F., Calamari, D.,<br />
Zuccato, E., 2006. Removal of pharmaceuticals in sewage<br />
treatment plants in Italy. Environmental Science and<br />
Technology 40, 357–363.<br />
De Laat, J., Truong Le, G., Legube, B., 2004. A comparative study of<br />
the effects of chloride, sulfate and nitrate ions on the rates of<br />
decomposition of H 2O 2 and organic compounds by Fe(II)/H 2O 2<br />
and Fe(III)/H 2O 2. Chemosphere 55, 715–723.<br />
Esplugas, S., Rodríguez, M., Malato, S., Pulgarin, C., Contreras, S.,<br />
Curcó, D., Giménez, J., 2005. Optimizing the solar photo Fenton<br />
process in the treatment of contaminated water.<br />
Determination of intrinsic kinetic constants for scale-up. Solar<br />
Energy 79, 360–368.<br />
Esplugas, S., Bila, D., Krause, L., Dezotti, M., 2007. Ozonation and<br />
advanced oxidation technologies to remove endocrine<br />
disrupting chemicals (EDCs) and pharmaceuticals and<br />
personal care products (PPCPs) in water effluents. A review.<br />
Journal of Hazardous Materials 149, 631–642.<br />
Farré, M.J., Franch, M.I., Malato, S., Ayllon, J., Peral, J.,<br />
Doménech, X., 2005. Degradation of some biorecalcitrant<br />
pesticides by homogeneous and heterogeneous photocatalytic<br />
ozonation. Chemosphere 58, 1127–1133.<br />
Fono, L.J., Kolodziej, E.P., Sedlak, D.L., 2006. Attenuation of<br />
wastewater-derived contaminants in an effluent<br />
dominated river. Environmental Science and Technology 40,<br />
7257–7262.<br />
Fukushima, M., Tatsumi, K., 2001. Degradation pathways of<br />
pentachlorophenol by photo-Fenton systems in the presence<br />
of iron(III), humic acids and hydrogen peroxide.<br />
Environmental Science and Technology 35, 1771–1778.<br />
Gebhardt, W., Schröder, H., 2007. Liquid chromatography-<br />
(tandem) mass spectrometry for the follow-up of the<br />
elimination of persistent pharmaceuticals during wastewater<br />
treatment applying biological wastewater treatment and<br />
advanced oxidation. Journal of Chromatography A 1160, 34–43.<br />
Glaze, W., Kang, W., Chapin, H., 1987. Chemistry of water<br />
treatment processes involving ozone, hydrogen peroxide<br />
and ultraviolet radiation. Ozone Science & Engineering 9,<br />
335–352.<br />
Göbel, A., McArdell, C., Joss, A., Siegrist, H., Giger, W., 2007. Fate of<br />
sulfonamides, macrolides and trimethoprim in different<br />
wastewater treatment technologies. Science of the Total<br />
Environment 372, 361–371.<br />
Gogate, P.R., Pandit, A.B., 2004. A review of imperative<br />
technologies for wastewater treatment. I: oxidation<br />
technologies at ambient conditions. Advances in<br />
Environmental Research 8, 501–551.<br />
Gültekin, I., Ince, N.H., 2007. Synthetic endocrine disruptors in<br />
the environment and water remediation by advanced<br />
oxidation processes. Journal of Environmental Management<br />
85, 816–832.<br />
Hislop, K.A., Bolton, J.R., 1999. The photochemical generation of<br />
OH radicals in the UV–Vis/ferrioxalate/H2O2 system.<br />
Environmental Science and Technology 33, 3119–3126.<br />
Huber, M., Coronen, S., Ternes, T., von Gunten, U., 2005.<br />
Oxidation of pharmaceuticals during water treatment with<br />
chlorine dioxide. Water Research 39, 3607–3617.<br />
Jackson, J., Sutton, R., 2008. Sources of endocrine disrupting<br />
chemicals in urban wastewater, Oakland, CA. Science of the<br />
Total Environment 405, 153–160.
Kassinos, D., Varanva, N., Michael, C., Piera, P., 2009.<br />
Homogenous oxidation of aqueous solutions of atrazine and<br />
fenitrothion through dark and photo-Fenton reactions.<br />
Chemosphere 74, 866–872.<br />
Kasprzyk-Hordern, B., Dinsdale, R.M., Guwy, A.J., 2009. The<br />
removal of pharmaceuticals, personal care products,<br />
endocrine disruptors and illicit drugs during wastewater<br />
treatment and its impact on the quality of receiving waters.<br />
Water Research 43, 363–380.<br />
Kim, S.D., Cho, J., Kim, I.S., Vanderford, B.J., Snyder, S.A., 2007.<br />
Occurrence and removal of pharmaceuticals and endocrine<br />
disruptors in South Korean surface, drinking, and waste<br />
waters. Water Research 41, 1013–1021.<br />
Klamerth, N., Miranda, N., Malato, S., Agüera, A., Fernández-<br />
Alba, A.R., 2009. Degradation of emerging contaminants at low<br />
concentrations in MWTPs effluents with mild solar photo-<br />
Fenton and TiO 2. Catalysis Today 144, 124–130.<br />
Kositzi, M., Poulios, I., Malato, S., Cáceres, J., Campos, A., 2004.<br />
Solar photocatalytic treatment of synthetic municipal<br />
wastewater. Water Research 38, 1147–1154.<br />
Lapertot, M., Ebrahimi, S., Dazio, S., Rubinelli, A., Pulgarin, C.,<br />
2007. Photo-Fenton and biological integrated process for<br />
degradation of a mixture of pesticides. Journal of<br />
Photochemistry and Photobiology A: Chemistry 186, 34–40.<br />
Laville, N., Ait-Aissa, S., Gomez, E., Casellas, C., Porcher, J.M.,<br />
2004. Effects of human pharmaceuticals on cytotoxicity, EROD<br />
activity and ROS production in fish hepatocytes. Toxicology<br />
196, 41–55.<br />
Lesko, T., Colussi, A., Hoffmann, M., 2006. Sonochemical<br />
decomposition of phenol: evidence for a synergistic effect of<br />
ozone and ultrasound for the elimination of total organic<br />
carbon from water. Environmental Science and Technology<br />
40, 6818–6823.<br />
Malato, S., Blanco, J., Alarcón, D.C., Maldonado, M.I., Fernández-<br />
Ibáñez, P., Gernjak, W., 2007. Photocatalytic decontamination<br />
and disinfection of water with solar collectors. Catalysis<br />
Today 122, 137–149.<br />
Mantzavinos, D., Vassilakis, C., Pantidou, A., Psillakis, E.,<br />
Kalogerakis, N., 2004. Sonolysis of natural phenolic<br />
compounds in aqueous solutions: degradation pathways and<br />
biodegradability. Water Research 38, 3110–3118.<br />
Martinez, F., Calleja, G., Melero, J.A., Molina, R., 2005.<br />
Heterogeneous photo-Fenton degradation of phenolic<br />
aqueous solutions over iron containing SBA-15 catalyst.<br />
Applied Catalysis B: Environmental 60, 181–190.<br />
Martínez Bueno, M.J., Agüera, A., Gómez, M.J., Hernando, M.D.,<br />
García-Reyes, J.F., Fernández-Alba, A.R., 2007. Application of<br />
liquid chromatography/quadrupole-linear ion trap mass<br />
spectrometry and time-of-flight mass spectrometry to the<br />
determination of pharmaceuticals and related<br />
contaminants in wastewater. Analytical Chemistry 79,<br />
9372–9384.<br />
Mitch, W.A., Sharp, J.O., Trussel, R.R., Valentine, R.L., Alvarez-<br />
Cohen, L., Seldak, D.L., 2003. N-Nitrosodimethylamine (NDMA)<br />
as a drinking water contaminant: A review. Environmental<br />
Engineering and Science 20, 389–404.<br />
Moncayo-Lasso, A., Pulgarin, C., Benítez, N., 2008. Degradation of<br />
DBPs’ precursors in river water before and after slow sand<br />
filtration by photo-Fenton process at pH 5 in a solar CPC<br />
reactor. Water Research 42, 4125–4132.<br />
Nakada, N., Kiri, K., Shinohara, H., Harada, A., Kuroda, K.,<br />
Takizawa, S., Takada, H., 2008. Evaluation of pharmaceuticals<br />
and personal care products as water-soluble molecular<br />
markers of sewage. Environmental Science and Technology<br />
42, 6347–6353.<br />
Ning, B., Graham, N., Zhang, Y., Nakonechny, M., El-Din, M.G.,<br />
2007. Degradation of endocrine disrupting chemicals by<br />
ozone/AOPs. Ozone Science & Engineering 29, 153–176.<br />
water research 44 (2010) 545–554 553<br />
Neyens, E., Baeyens, J., 2003. A review of classic fenton’s<br />
peroxidation as an advanced oxidation technique. Journal of<br />
Hazardous Materials 98, 33–50.<br />
Nogueira, R.F.P., Oliveira, M.C., Paterlini, W.C., 2005. Simple and<br />
fast spectrophotometric determination of H 2O 2 in photo-<br />
Fenton reactions using metavanadate. Talanta 66, 86–91.<br />
OECD Guidelines for testing of Chemicals, Simulation Test-<br />
Aerobic Sewage Treatment 303A, 1999.<br />
Papadaki, M., Emery, R.J., Abu-Hassan, M.A., Diaz-Bustos, A.,<br />
Metcalfe, I.S., Mantzavinos, D., 2004. Sonocatalytic oxidation<br />
processes for the removal of contaminants containing<br />
aromatic rings from aqueous effluents. Separation and<br />
Purification Technology 34, 35–42.<br />
Pelegrini, R.T., Freire, R.S., Duran, N., Bertazzoli, R., 2001.<br />
Photoassisted electrochemical degradation of organic<br />
pollutants on a DSA type oxide electrode: process test for<br />
a phenol synthetic solution and its application for the E1<br />
bleach kraft mill effluent. Environmental Science and<br />
Technology 35, 2849–2853.<br />
Petrovic, M., Gonzalez, S., Barceló, D., 2003. Analysis and removal<br />
of emerging contaminants in wastewater and drinking water.<br />
Trends in Analytical Chemistry 22, 685.<br />
Pignatello, E., Oliveros, A., McKay, 2006. Advanced oxidation<br />
processes for organic contaminant destruction based on the<br />
Fenton reaction and related chemistry. Critical Reviews in<br />
Environmental Science and Technology 36, 1–84.<br />
Prosen, H., Zupančič-Kralj, L., 2005. Evaluation of photolysis and<br />
hydrolysis of atrazine and its first degradation product in the<br />
presence of humic acids. Environmental Pollution 133, 517–529.<br />
Radjenović, J., Petrović, M., Barceloć, D., Petrović, M., 2007.<br />
Advanced mass spectrometric methods applied to the study of<br />
fate and removal of pharmaceuticals in wastewater<br />
treatment. Trends in Analytical Chemistry 26, 1132–1144.<br />
Richardson, S., 2007. Water analysis: emerging contaminants and<br />
current issues. Reviews in Analytical Chemistry 79, 4295–4324.<br />
Rosenfeldt, E.J., Linden, K.G., 2004. Degradation of endocrine<br />
disrupting chemicals bisphenol A, ethinyl estradiol and estradiol<br />
during UV photolysis and advanced oxidation processes.<br />
Environmental Science and Technology 38, 5476–5483.<br />
Rosenfeldt, E.J., Chen, P.J., Kullman, S., Linden, K.G., 2007.<br />
Destruction of estrogenic activity in water using UV advanced<br />
oxidation. Science of the Total Environment 377, 105–113.<br />
Saritha, P., Aparna, C., Himabindu, V., Anjaneyulu, Y., 2007.<br />
Comparison of various advanced oxidation processes for the<br />
degradation of 4-chloro-2-nitrophenol. Journal of Hazardous<br />
Materials 149, 609–614.<br />
United Book Press Inc.. In: Clesceri, L.S., Greenberg, A.E., Eaton, A.<br />
D. (Eds.), Standard Methods for the Examination of Water and<br />
Wastewater, 20th ed. American Public Health Association,<br />
American Water Works Association, Water Environment<br />
Federation, Baltimore, Maryland.<br />
Ternes, T., 1998. Occurrence of drugs in German sewage<br />
treatment plants and rivers. Water Research 32, 3245–3260.<br />
Ternes, T., Struber, J., Herrmann, N., McDowell, D., Ried, A.,<br />
Kampmann, M., Teiser, B., 2003. Ozonation: a tool for removal<br />
of pharmaceuticals contrast media and musk fragrances from<br />
wastewater. Water Research 37, 1976–1982.<br />
Ternes, T., Bonerz, M., Herrmann, N., Teiser, B., Andersen, R.,<br />
2007. Irrigation of treated wastewaters in Braunschweig,<br />
Germany: an option to remove pharmaceuticals and musk<br />
fragrances. Chemosphere 66, 894–904.<br />
Teske, S.S., Arnold, R.G., 2008. Removal of natural and xenoestrogens<br />
during conventional wastewater treatment. Reviews<br />
in Environmental Science and Biotechnology 7, 107–124.<br />
Zapata, A., Oller, I., Bizani, E., Sánchez-Pérez, J.A., Maldonado, I.,<br />
Malato, S., 2009. Evaluation of operational parameters<br />
involved in solar photo-Fenton degradation of a commercial<br />
pesticide mixture. Catalysis Today 144, 94–99.
554<br />
Zepp, R.G., Faust, B.C., Hoigné, J., 1992. Hydroxyl radical<br />
formation in aqueous reactions pH(3–8) of iron(II) with<br />
hydrogen peroxide: the photo-fenton reaction. Environmental<br />
Science and Technology 26, 313–319.<br />
Zhao, J.-L., Ying, G.-G., Wang, L., Yang, J.-F., Yang, X.-B., Yang, L.-H.,<br />
Li, X., 2009. Determination of phenolic endocrine disrupting<br />
chemicals and acidic pharmaceuticals in surface water of the<br />
water research 44 (2010) 545–554<br />
Pearl River in South China by GC-MS. Science of the Total<br />
Environment 407, 962–974.<br />
Zhou, M., Dai, Q., Lei, L., Ma, C., Wang, D., 2005. Long life<br />
modified lead dioxide anode for organic wastewater<br />
treatment: electrochemical characteristics and degradation<br />
mechanism. Environmental Science and Technology 39,<br />
363–370.
Oxidative transformation of micropollutants during municipal<br />
wastewater treatment: Comparison of kinetic aspects of<br />
selective (chlorine, chlorine dioxide, ferrate VI , and ozone)<br />
and non-selective oxidants (hydroxyl radical)<br />
Yunho Lee a , Urs von Gunten a,b, *<br />
a<br />
Eawag, Swiss Federal Institute of Aquatic Science and Technology, Ueberlandstrasse 133, P.O. Box 611, 8600 Duebendorf, Switzerland<br />
b<br />
Institute of Biogeochemistry and Pollutant Dynamics, ETH Zurich, 8092 Zurich, Switzerland<br />
article info<br />
Article history:<br />
Received 1 October 2009<br />
Received in revised form<br />
13 November 2009<br />
Accepted 23 November 2009<br />
Available online 27 November 2009<br />
Keywords:<br />
Oxidation processes<br />
Ozone<br />
Chlorine<br />
Chlorine dioxide<br />
Ferrate VI<br />
Hydroxyl radical<br />
water research 44 (2010) 555–566<br />
Available at www.sciencedirect.com<br />
journal homepage: www.elsevier.com/locate/watres<br />
abstract<br />
Chemical oxidation processes have been widely applied to water treatment and may serve<br />
as a tool to minimize the release of micropollutants (e.g. pharmaceuticals and endocrine<br />
disruptors) from municipal wastewater effluents into the aquatic environment. The<br />
potential of several oxidants for the transformation of selected micropollutants such as<br />
atenolol, carbamazepine, 17a-ethinylestradiol (EE2), ibuprofen, and sulfamethoxazole was<br />
assessed and compared. The oxidants include chlorine, chlorine dioxide, ferrate VI , and<br />
ozone as selective oxidants versus hydroxyl radicals as non-selective oxidant. Secondorder<br />
rate constants (k) for the reaction of each oxidant show that the selective oxidants<br />
react only with some electron-rich organic moieties (ERMs), such as phenols, anilines,<br />
olefins, and deprotonated-amines. In contrast, hydroxyl radicals show a nearly diffusioncontrolled<br />
reactivity with almost all organic moieties (k 10 9 M 1 s 1 ). Due to a competition<br />
for oxidants between a target micropollutant and wastewater matrix (i.e. effluent<br />
organic matter, EfOM), a higher reaction rate with a target micropollutant does not<br />
necessarily translate into more efficient transformation. For example, transformation<br />
efficiencies of EE2, a phenolic micropollutant, in a selected wastewater effluent at pH 8<br />
varied only within a factor of 7 among the selective oxidants, even though the corresponding<br />
k for the reaction of each selective oxidant with EE2 varied over four orders of<br />
magnitude. In addition, for the selective oxidants, the competition disappears rapidly after<br />
the ERMs present in EfOM are consumed. In contrast, for hydroxyl radicals, the competition<br />
remains practically the same during the entire oxidation. Therefore, for a given oxidant<br />
dose, the selective oxidants were more efficient than hydroxyl radicals for transforming<br />
ERMs-containing micropollutants, while hydroxyl radicals are capable of transforming<br />
micropollutants even without ERMs. Besides EfOM, ammonia, nitrite, and bromide were<br />
found to affect the micropollutant transformation efficiency during chlorine or ozone<br />
treatment.<br />
ª 2009 Elsevier Ltd. All rights reserved.<br />
* Corresponding author at: Eawag, Swiss Federal Institute of Aquatic Science and Technology, Ueberlandstrasse 133, P.O. Box 611,<br />
8600 Duebendorf, Switzerland. Tel.: þ41 44 823 5270; fax: þ41 44 823 5028.<br />
E-mail address: vongunten@eawag.ch (U. von Gunten).<br />
0043-1354/$ – see front matter ª 2009 Elsevier Ltd. All rights reserved.<br />
doi:10.1016/j.watres.2009.11.045
556<br />
1. Introduction<br />
Effluents of municipal wastewater treatment plants (WWTPs)<br />
have been identified as a major source of micropollutants,<br />
such as hormones, pharmaceuticals, and personal care<br />
products (Ternes and Joss, 2006). To minimize a discharge of<br />
micropollutants from WWTPs to the receiving waters and<br />
thus to prevent adverse ecological effects, there have been<br />
discussions on upgrading WWTPs with additional treatment<br />
steps such as activated carbon and ozonation (Nowotny et al.,<br />
2007; Hollender et al., 2009). This is also desirable for the<br />
quality of drinking water if the wastewater-impacted<br />
receiving waters are used as a drinking water source via<br />
indirect potable reuse.<br />
The focus of the current study is on the chemical oxidation.<br />
Chemical oxidants, such as chlorine, chlorine dioxide, ozone,<br />
and hydroxyl radicals, have been widely applied for disinfection<br />
of drinking waters and wastewaters and often for the<br />
transformation/elimination of undesired micropollutants<br />
from drinking water (Letterman, 1999). Chlorine is worldwide<br />
one of the most commonly used disinfectant for drinking<br />
water and wastewater treatment (Letterman, 1999). Chlorine<br />
dioxide has been used as an alternative disinfectant for<br />
chlorine, often to minimize the formation of chlorine-based<br />
disinfection by-products (e.g. trihalomethane) (Gates, 1998;<br />
Letterman, 1999). The application of ozone in drinking water<br />
treatment is also widespread and the main reasons for its use<br />
are disinfection and oxidation (e.g. micropollutants transformation)<br />
or a combination of both (Hoigné, 1998; von<br />
Gunten, 2003a). Even though chlorine, chlorine dioxide, and<br />
ozone have been widely used for disinfection (Letterman,<br />
1999), their potentials for transforming micropollutants in<br />
wastewaters have received attention only in recent years<br />
(Ternes and Joss, 2006). In addition, ferrate VI is an emerging<br />
water treatment chemical due to its dual functions as an<br />
oxidant and a subsequent coagulant/precipitant as ferric<br />
hydroxide (Lee et al., 2009). Finally, the hydroxyl radical is<br />
a primary oxidant in advanced oxidation processes (AOPs)<br />
such as UV/H2O2 and O3/H2O2 and their application are<br />
considered for transforming various micropollutants in<br />
drinking waters as well as wastewaters for water reuse<br />
(Letterman, 1999; Asano et al., 2007). The hydroxyl radical is<br />
quite different from other oxidants with respect to its high<br />
reactivity and thus low selectivity (Buxton et al., 1988). In this<br />
paper, chlorine, chlorine dioxide, ferrate VI and ozone will be<br />
referred as selective oxidants as opposed to non-selective<br />
hydroxyl radicals.<br />
The oxidants mentioned above can achieve a transformation<br />
of various micropollutants during wastewater<br />
treatment. The transformation efficiency mainly depends on:<br />
(i) the reactivity of the oxidant to target micropollutants; and<br />
(ii) towards matrix components present in water (e.g. dissolved<br />
organic matter, DOM) which determines the stability of<br />
the oxidant. In general, the reactions of the oxidants follow<br />
second-order kinetics, i.e. the reaction rate is proportional to<br />
the concentrations of both reactants. The corresponding<br />
second-order reaction rate constants (k) range from 1to<br />
10 10 M 1 s 1 in aqueous solution (Buxton et al., 1988; Deborde<br />
and von Gunten, 2008; Lee et al., 2009; Neta et al., 1988; von<br />
water research 44 (2010) 555–566<br />
Gunten, 2003a). Currently, k values are available for a wide<br />
range of organic compounds including many micropollutants<br />
(Fig. 1 and Table 1 as examples). For ozone reactions more<br />
than 500 k values have been measured (Neta et al., 1988 von<br />
Gunten, 2003a), and a few thousand k values are available for<br />
hydroxyl radical reactions (Buxton et al., 1988). There are also<br />
a few hundred k values available for chlorine, chlorine<br />
dioxide, and ferrate VI (Deborde and von Gunten, 2008; Lee<br />
et al., 2009; Neta et al., 1988). These second order rate<br />
constants show that the selective oxidants react preferentially<br />
with electron-rich organic moieties (ERMs), such as activated<br />
aromatic compounds (i.e. phenol, aniline, and polycyclic<br />
aromatics), organosulfur compounds, and deprotonated<br />
amines. Additionally, ozone and ferrate VI react with olefins.<br />
Contrary to the selective oxidants, hydroxyl radicals react<br />
with almost all organic moieties (e.g. aliphatic CeH bond) with<br />
nearly diffusion-controlled rates (i.e. k 10 9 M 1 s 1 ).<br />
Therefore, micropollutants containing the ERMs<br />
mentioned above can be potentially transformed by the<br />
selective oxidants. On the contrary, hydroxyl radicals can be<br />
effective for transforming any type of micropollutant.<br />
However, a high reactivity of the oxidants to the ERMs is also<br />
responsible for a rapid oxidant consumption by the ERMs<br />
included in the matrix components, such as DOM. In addition<br />
to DOM, some inorganic species such as ammonia, nitrite,<br />
sulfide, iron(II), and manganese(II) can be potential<br />
consumers of the selective oxidants. In the case of hydroxyl<br />
radicals, their rapid consumption by almost all types of<br />
matrix components can lead to a significantly lower transformation<br />
efficiency of micropollutants than the selective<br />
oxidants. Therefore, the kinetic information (i.e. k values) of<br />
the oxidants with the ERMs and some other moieties is<br />
crucial to assess the transformation efficiency of structurallydiverse<br />
micropollutants by a given oxidant in a real wastewater<br />
matrix.<br />
The aim of this study was to compare each of the selective<br />
oxidants and non-selective hydroxyl radicals with respect to<br />
their efficiency for transforming micropollutants during the<br />
treatment of the wastewater effluents. This comparison is<br />
important to provide criteria for choosing an appropriate<br />
oxidation process for a specific class of micropollutants based<br />
on the chemical structure (i.e. presence of ERM). For this<br />
purpose, the following issues were addressed:<br />
The k of each oxidant with selected organic model<br />
compounds and micropollutants (containing various ERMs)<br />
available in literature were summarized and compared.<br />
The oxidant consumption kinetics in a wastewater effluent<br />
were determined.<br />
The oxidative transformations of selected micropollutants<br />
containing various ERMs were conducted in a wastewater<br />
effluent for varying oxidant doses (0–150 mM). The transformation<br />
efficiency was then compared for the oxidants<br />
based on the required oxidant dose to achieve a certain<br />
degree of a micropollutant transformation.<br />
The effect of ammonia, nitrite, and bromide was investigated<br />
in respect to the transformation efficiency of selected<br />
micropollutants during the treatment of a wastewater<br />
effluent by the selected oxidants and hydroxyl radicals.
-1 -1<br />
s<br />
a b c<br />
k,<br />
M<br />
10<br />
10<br />
10 8<br />
10 6<br />
10 4<br />
10 2<br />
10 0<br />
HO.<br />
d e f<br />
-1 -1<br />
s<br />
k,<br />
M<br />
10 10<br />
10 8<br />
10 6<br />
10 4<br />
10 2<br />
10 0<br />
HO.<br />
HFeO 4 –<br />
2. Materials and methods<br />
2.1. Reagents and wastewater<br />
Atenolol (ATL), carbamazepine (CBZ), 17a-ethinylestradiol<br />
(EE2), ibuprofen (IBP), and sulfamethoxazole (SMX) were<br />
obtained from Sigma–Aldrich with a purity higher than 98%.<br />
Stock solutions of these pharmaceuticals were prepared with<br />
Milli-Q purified water (Barenstad). Other chemicals and<br />
solvents were purchased from various commercial suppliers<br />
and used as received.<br />
A 10-L grab sample of secondary wastewater effluent<br />
(RDWW) was obtained from the Regensdorf wastewater<br />
treatment plant near Zurich, Switzerland. The sample was<br />
transported to the laboratory within several hours of sampling<br />
and vacuum-filtered with a 0.45 mm cellulose nitrate<br />
membrane prior to storage at 4 C. RDWW has a DOC of 5.0 mg<br />
CL 1 , 1.2 mM ammonia (17 mg NL 1 ), 6 mM nitrite (84 mg NL 1 ),<br />
5.3 mM carbonate alkalinity, 0.35 mM (28 mg L 1 )ofBr and<br />
apHofw8. The concentrations of selected micropollutants<br />
(ATL, CBZ, EE2, IBP, and SMX) originally present in RDWW<br />
O 3<br />
phenol aniline butenol (olefin)<br />
ClO 2<br />
HOCl<br />
glycine<br />
o (1 amine)<br />
HOCl<br />
O 3<br />
HFeO 4 –<br />
water research 44 (2010) 555–566 557<br />
HO. HO.<br />
HO.<br />
HOCl<br />
HFeO 4 –<br />
HFeO 4 –<br />
O 3<br />
ClO 2<br />
HOCl<br />
dimethylamine<br />
o (2 amine)<br />
5 6 7 8 9 10 5 6 7 8 9 10 5 6 7 8 9 10<br />
pH pH pH<br />
O 3<br />
ClO 2<br />
were negligible relative to the concentrations spiked for<br />
oxidation experiments, which were 0.2–1 mM for each<br />
substrate.<br />
2.2. Oxidants<br />
HO.<br />
HFeO 4 –<br />
Stock solutions of chlorine (5–20 mM) were prepared by<br />
diluting a commercial solution of sodium hypochlorite (10%<br />
active chlorine, Riedel-deHaen, Germany). Chlorine dioxide<br />
(w5 mM) was produced by mixing potassium peroxodisulfate<br />
(K 2S 2O 8) with sodium chlorite (NaClO 2) according to a method<br />
described by Gates (1998). Ozone was produced with a Fischer<br />
502 ozone generator and its stock solutions (w1.5 mM) were<br />
produced by sparging ozone-containing oxygen through Milli-<br />
Q water that was cooled in an ice bath (Bader and Hoigné,<br />
1981). Potassium ferrate VI of high purity (w90%) was prepared<br />
by the method of Thompson et al. (1951). Stock solutions of<br />
ferrate VI (2 mM) were freshly prepared by dissolving solid<br />
samples of potassium ferrate VI (K2FeO4) in Milli-Q water and<br />
subsequent filtration (pH z 9.2). Hydroxyl radicals were in<br />
situ generated by photolysis of H2O2 (0.2–5 mM). Irradiations<br />
O 3<br />
trimethylamine<br />
o (3 amine)<br />
Fig. 1 – pH dependent second-order rate constants (k) for the reaction of the oxidants, chlorine (HOCl), chlorine dioxide (ClO2),<br />
ferrate VI (HFeO4 L ), hydroxyl radicals (HO ), and ozone (O3) with (a) phenol, (b) aniline, (c) butenol, (d) glycine, (e)<br />
dimethylamine, and (f) trimethylamine. k values for chlorine: Deborde and von Gunten, (2008); for chlorine dioxide and<br />
ozone: Neta et al., (1988); for ferrate VI : Lee et al. (2005a, 2008, 2009); and for hydroxyl radicals: Buxton et al., (1988). Speciesspecific<br />
second-order rate constants and the pKa values of the compounds used to generate the data are provided in<br />
Supplementary Information, SI-excel.<br />
O 3<br />
ClO 2<br />
HOCl
558<br />
Table 1 – Second-order rate constants for the reaction of oxidants with selected pharmaceuticals and inorganic species<br />
(ammonia, nitrite, and bromide). a<br />
17a-Ethinylestradiol (EE2),<br />
synthetic steroid estrogen<br />
Sulfamethoxazole (SMX),<br />
sulfonamide antibiotic<br />
Carbamazepine (CBZ),<br />
antiepileptic drug<br />
Atenolol (ATL), b-blocker<br />
Ibuprofen (IBP), antiphlogistic<br />
water research 44 (2010) 555–566<br />
Compounds pK a Oxidant k (species<br />
of compound),<br />
M 1 s 1<br />
k at pH 8,<br />
M 1 s 1<br />
10.4 HO 9.8 10 9<br />
9.8 10 9<br />
O3 1.8 10 5 (neutral),<br />
3.7 10 9 1.5 10<br />
(anion)<br />
7<br />
ClO2 4.6 10 8 (anion) 1.8 10 6<br />
HFeO4 9.4 10 2 (neutral),<br />
5.4 10 5 4.2 10<br />
(anion)<br />
2<br />
HOCl 3.5 10 5 (anion) 3.4 10 2<br />
Ref.<br />
Huber et al. (2003)<br />
Deborde et al. (2005)<br />
Huber et al. (2005)<br />
Lee et al. (2005a)<br />
Deborde et al. (2004)<br />
5.7 HO 5.5 10 9<br />
5.5 10 9<br />
Huber et al. (2003)<br />
O3 4.7 10 4 (neutral),<br />
5.7 10 5 5.7 10<br />
(anion)<br />
5<br />
Dodd et al. (2006)<br />
ClO2 7.9 10 3 (anion) 7.9 10 3<br />
Huber et al. (2005)<br />
HFeO4 9.2 10 3b<br />
95 Lee et al. (2009)<br />
HOCl 1.1 10 3 (neutral),<br />
2.4 10 3 5.7 10<br />
(anion)<br />
2<br />
Dodd and Huang<br />
(2004)<br />
– HO 8.8 10 9<br />
O3 3.0 10 5<br />
ClO 2<br />
Table 1 (continued)<br />
Compounds pKa Oxidant k (species<br />
of compound),<br />
M 1 s 1<br />
were performed in quartz tubes using a merry-go-round<br />
photoreactor (Hans Mangels GmbH, Bornheim-Roisdor,<br />
Germany), equipped with a low pressure mercury lamp (Heraues<br />
Noblelight model TNN 15/32, nominal power 15 W)<br />
emitting essentially monochromatic light at l ¼ 254 nm. In the<br />
case of SMX, a medium pressure mercury lamp (Heraues<br />
Noblelight model TQ 718, nominal power 500 W) with a 0.25 M<br />
sodium nitrate filter solution producing light at l > 320 nm<br />
was used to minimize the direct photo-transformation of SMX<br />
(less than 10%). Fluence and fluence rate were determined by<br />
chemical actinometry at low optical density using 5 mM<br />
aqueous atrazine as an actinometer described by Meunier<br />
et al. (2006). Stock solutions of each oxidant were standardized<br />
spectrophotometrically based on their molar absorption<br />
coefficient: 3 ¼ 350 M 1 cm 1 at 292 nm for OCl (Johnson and<br />
Margerum, 1991), 3 ¼ 1230 M 1 cm 1 at 359 nm for ClO2 (Gates,<br />
1998), 3 ¼ 3000 M 1 cm 1 at 260 nm for O3 (Liu et al., 2001),<br />
3 ¼ 1150 M 1 cm 1 at 510 nm for FeO4 2<br />
(Lee et al., 2005b), and<br />
3 ¼ 40 M 1 cm 1 at 240 nm for H2O2 (Bader et al., 1988).<br />
2.3. Oxidant consumption kinetics and transformation<br />
of selected micropollutants in wastewater<br />
Oxidant consumption experiments for the selective oxidants<br />
were performed in a secondary wastewater effluent (RDWW)<br />
at pH 8 (20 mM borate buffer) in 50 mL reaction volume. Forty<br />
to 45 mM of each oxidant was applied to RDWW and 0.2–1 mL<br />
of the reaction solution was sampled for the measurements of<br />
residual oxidants over a 1 h reaction time.<br />
Transformation of the selected micropollutants was tested<br />
in RDWW spiked with each micropollutant and treated by<br />
each oxidant at various oxidant doses (0–150 mM) at 24 1 C<br />
and pH 8. The micropollutants, ATL (amine), CBZ (olefin), EE2<br />
(phenol), and SMX (aniline), were selected as representative<br />
substances with different ERMs that belong to different usage<br />
classes and have an environmental relevance (Ternes and<br />
Joss, 2006). IBP was chosen as a model micropollutant without<br />
the ERMs. Each micropollutant was spiked separately at<br />
concentrations of 0.2–1 mM into 20 mL of the wastewater<br />
buffered at pH 8. For the selective oxidants, stock solutions of<br />
water research 44 (2010) 555–566 559<br />
k at pH 8,<br />
M 1 s 1<br />
Br bromide – HO w1.1 10 9<br />
w1.1 10 9<br />
von Gunten (2003b)<br />
O3 2.6 10 2<br />
2.6 10 2<br />
Liu et al. (2001)<br />
ClO2
560<br />
irradiation time. For details of the kinetic method, see<br />
Supplementary Information, SI-word.<br />
2.5. Analytical methods<br />
Oxidant concentrations in wastewater effluents were<br />
measured by several colorimetric methods. Chlorine and<br />
chlorine dioxide were determined by the ABTS method<br />
developed by Pinkernell et al. (2000); ozone by the indigo<br />
method (Bader and Hoigné, 1981); ferrate VI by the ABTS<br />
method described by Lee et al. (2005b); H2O2 by the DPD/<br />
peroxidase method (Bader et al., 1988). Micropollutant (or<br />
pCBA) concentrations at sub to low mM range were determined<br />
with an Agilent 1100 series HPLC system equipped with<br />
a Nucleosil 100-5 C18 column (125 4 or 250 4 mm) and a UV<br />
(diode array) and/or fluorescence detector. Separation was<br />
achieved by different gradient programs with the aqueous<br />
phase consisting either of 10 mM phosphoric acid or 10 mM<br />
citric acid (pH 5) and the solvent phase either of acetonitrile or<br />
methanol. The injection volume was 100 mL and the flow rate<br />
was set to 1 mL min 1 . The repeatability was determined to be<br />
4% while limit of quantifications (LOQ) were inferred from the<br />
lowest standard concentrations used. The limits of quantification<br />
were in all cases sufficient to determine a transformation<br />
of micropollutants of more than 96% ( 1.4 logtransformation).<br />
3. Results and discussion<br />
3.1. Rate constants of oxidants with selected organic<br />
compounds<br />
Fig. 1 summarizes the pH-dependent k for the reaction of the five<br />
oxidants (chlorine, chlorine dioxide, ferrate VI , ozone, and<br />
hydroxyl radical) with the selected ERM-containing model<br />
compounds, such as phenol and aniline (activated aromatic<br />
compounds), butenol (olefin), glycine (primary (1 ) amine),<br />
dimethylamine (secondary (2 ) amine), and trimethylamine<br />
(tertiary (3 ) amine). If the k value is lower than 1 M 1 s 1 ,itisnot<br />
shown in Fig. 1. The species-specific k values which were used to<br />
generate Fig. 1 are provided in Supplementary Information, SIexcel.<br />
For more details on the kinetics of each oxidant, refer the<br />
corresponding references provided in SI-excel.<br />
The k values for phenol and aniline, as two aromatic<br />
compounds, decrease in the order of hydroxyl radical -<br />
> ozone > chlorine dioxide > chlorine z ferrate VI (Figs. 1a,b).<br />
The k values for the other compounds show a slightly different<br />
reactivity order. Chlorine and chlorine dioxide are not reactive<br />
to butenol, therefore the reactivity trend for butenol is<br />
hydroxyl radical > ozone > ferrate VI (Fig. 1c). For three amine<br />
compounds, chlorine shows a particular high reactivity<br />
whereas chlorine dioxide has a low reactivity. For glycine as 1<br />
amine, the k values decrease in the order of hydroxyl radical<br />
> chlorine > ozone > ferrate VI > chlorine dioxide (Fig. 1d).<br />
For dimethylamine as 2 amine, the reactivity order is<br />
hydroxyl radical > ozone z chlorine > ferrate VI > chlorine<br />
dioxide (Fig. 1e). Finally, for trimethylamine as 3 amine, the<br />
reactivity order is hydroxyl radical > ozone > chlorine dioxide<br />
chlorine > ferrate VI (Fig. 1f).<br />
water research 44 (2010) 555–566<br />
In the pH range 5–10, hydroxyl radicals react with phenol,<br />
aniline, and butenol with nearly diffusion-controlled rates.<br />
However, the k values for the reaction with three amine<br />
compounds are one or two orders of magnitude lower (10 7 –<br />
10 8 M 1 s 1 at pH 6–8). This is due to the relative low reactivity of<br />
hydroxyl radicals with the protonated-amine moieties. Nevertheless,<br />
the k of hydroxyl radicals with the amine compounds<br />
with longer hydrocarbon-chains can be higher than 10 9 M 1 s 1 .<br />
This is due to the high reactivity of hydroxyl radical with CeH<br />
bonds (k [ 10 8 M 1 s 1 ) (Buxton et al., 1988). In addition,<br />
hydroxyl radicals react with benzenes (i.e. non-activated<br />
aromatics) with k > 10 9 M 1 s 1 (Buxton et al., 1988). Ozone<br />
reacts fast with all ERMs shown in Fig. 1:thek values decrease in<br />
the order of aniline phenol > butenol dimethylamine z<br />
trimethylamine > glycine in the pH range 6–8. Chlorine dioxide<br />
shows a high reactivity only with phenol, aniline, and trimethylamine.<br />
Chlorine is reactive to all ERMs except butenol. The k<br />
values decrease in the order of glycine > dimethylamine<br />
aniline > phenol z trimethylamine in the pH<br />
range 6–8. Finally, ferrate VI reacts with all ERMs except trimethylamine.<br />
The k values decrease in the order of aniline > phenol<br />
z glycine > butenol dimethylamine in the pH range 6–8.<br />
For the selective oxidants, ozone shows the widest range of k<br />
values (i.e. w7 orders of magnitude) while ferrate VI shows the<br />
narrowest range of k values (i.e. w4 orders of magnitude) toward<br />
the selected ERMs in the pH range 6–8.<br />
3.2. Consumption kinetics of the selective oxidants in<br />
a wastewater effluent<br />
Fig. 2 shows the consumption kinetics of the selective<br />
oxidants (ozone, ferrate VI , chlorine, and chlorine dioxide) in<br />
a secondary wastewater effluent from Regensdorf,<br />
Switzerland (RDWW) at pH 8 for an oxidant dose of 40–45 mM.<br />
The selected RDWW has a low concentration of inorganic<br />
nitrogen species ([ammonia] ¼ 1.2 mM and [nitrite] ¼ 6 mM),<br />
therefore the consumption of each oxidant is mainly caused<br />
by EfOM ([DOC] ¼ 5mg C L 1 ). Except ferrate VI , all oxidants<br />
showed fast consumptions within 2 min after the oxidant<br />
addition. The stability of the oxidants follows the order of<br />
ferrate VI z chlorine dioxide chlorine [ ozone. Ozone was<br />
depleted in less than 2 min (Fig. 2a, inset), whereas the other<br />
oxidants were slowly consumed within 60 min. The<br />
observed low stability of ozone is consistent with its high<br />
reactivity to all ERMs shown in Fig. 1. Chlorine and chlorine<br />
dioxide showed a similar decay pattern: a rapid decrease<br />
within 2 min after the oxidant addition (termed ‘initial<br />
phase’), followed by a slow decrease over 60 min of reaction<br />
time (termed ‘second phase’). Based on the k values shown in<br />
Fig. 1, the chlorine consumption in the initial phase is<br />
expected to be mainly caused by aniline- and 1 &2 aminemoieties<br />
while in the second phase phenolic- and 3 aminemoieties<br />
become important. In the case of chlorine dioxide,<br />
the consumption in the initial phase might be mainly caused<br />
by phenolic- and aniline-moieties while in the second phase<br />
by 2 &3 amine-moieties. Contrary to chlorine and chlorine<br />
dioxide, ferrate VI decreased smoothly along the reaction time,<br />
which is consistent with its narrow distribution of k values<br />
with the ERMs shown in Fig. 1. The achieved oxidant exposures<br />
for the conditions shown in Fig. 2 were 4.5 10 2 M 1
a<br />
C oncentration,<br />
µM<br />
c<br />
C oncentration,<br />
µM<br />
50<br />
40<br />
30<br />
20<br />
10<br />
0<br />
50<br />
40<br />
30<br />
20<br />
10<br />
0<br />
0<br />
0.0 0.5 1.0 1.5 2.0 2.5 3.0<br />
Time, min<br />
0 10 20 30 40 50 60<br />
0 10 20 30 40 50 60<br />
Time, min<br />
s 1 for ferrate VI and chlorine dioxide, 3.6 10 2 M 1 s 1 for<br />
chlorine, and 6.0 10 4 M 1 s 1 for ozone.<br />
3.3. Transformation of selected micropollutants in<br />
a wastewater effluent: comparison between selective<br />
oxidants and non-selective hydroxyl radicals<br />
C oncentration,<br />
µM<br />
A comparison was made for the transformation efficiency of<br />
selected micropollutants in a secondary wastewater effluent<br />
(RDWW) by the selective oxidants (chlorine, chlorine dioxide,<br />
ferrate VI , and ozone) and the non-selective hydroxyl radicals.<br />
The selected micropollutants include compounds with ERMs,<br />
such as 17a-ethinylestradiol (EE2, phenol), sulfamethoxazole<br />
(SMX, aniline), carbamazepine (CBZ, olefin), and atenolol (ATL,<br />
2 amine), and ibuprofen (IBP, no ERM). For the selective<br />
oxidants, the reaction time of 1 h was given to simulate realistic<br />
treatment conditions. Ozone was completely consumed<br />
after 1 h whereas for chlorine, chlorine dioxide, and ferrate VI ,<br />
more than 10% of the oxidant concentration remained after<br />
1 h if the oxidant doses were higher than 40 mM (Fig. 2). For<br />
hydroxyl radicals, the oxidant doses were quantified by the<br />
kinetic method described in the Supplementary Information,<br />
SI-word. Fig. 3 shows the logarithm of the relative residual<br />
concentration of each selected micropollutant during treatment<br />
of RDWW at pH 8 as a function of the oxidant doses. The<br />
transformation efficiency of micropollutants varied significantly<br />
depending on the types of the selective oxidants and<br />
50<br />
40<br />
30<br />
20<br />
10<br />
water research 44 (2010) 555–566 561<br />
b<br />
ozone ferrateVI 0 10 20 30 40 50 60<br />
d<br />
chlorine chlorine dioxide<br />
0 10 20 30 40 50 60<br />
Time, min<br />
Fig. 2 – Consumption kinetics of the selective oxidants, (a) ozone, (b) ferrate VI , (c) chlorine, and (d) chlorine dioxide, in<br />
a secondary wastewater effluent (RDWW) at pH 8. Symbols represent measured data and lines connect each data point to<br />
show the trend.<br />
micropollutants, and will be discussed in the following<br />
sections.<br />
3.3.1. EE2 and SMX<br />
For these two micropollutants containing activated aromatic<br />
moieties, all four selective oxidants were more efficient than<br />
hydroxyl radicals in terms of the oxidant doses required to<br />
achieve more than one-log (90%) transformation (Figs. 3a,b).<br />
For the selective oxidants, especially in the case of chlorine<br />
and ferrate VI , there was a lag-phase at lower oxidant doses<br />
where the micropollutant concentration decreased little with<br />
increasing oxidant doses. The low transformation efficiency<br />
in the lag-phase is attributed to the high competition for the<br />
selective oxidants between EE2/SMX and the ERMs of EfOM.<br />
With an increase of the selective oxidant doses higher than<br />
the initial consumption by ERMs, the residual concentration of<br />
EE2 and SMX started to decrease significantly. This can be<br />
understood by the depletion of the most reactive ERMs of<br />
EfOM and the corresponding smaller competition for EE2 and<br />
SMX toward the selective oxidants. For hydroxyl radicals, on<br />
the contrary, the log-relative residual concentration of EE2<br />
and SMX was linearly proportional to the applied hydroxyl<br />
radical dose. This indicates that the magnitude of the<br />
competition for hydroxyl radicals between micropollutants<br />
and wastewater effluent matrix remained constant during the<br />
entire oxidation. Therefore, it can be concluded that the initial<br />
oxidation products of EfOM by hydroxyl radicals have similar<br />
k values for their reaction with hydroxyl radicals compared to
562<br />
l 0<br />
og<br />
C/<br />
C<br />
log C/<br />
C0<br />
a<br />
0<br />
-1<br />
17a-ethinylestradiol (EE2)<br />
0<br />
Sulfamethoxazole (SMX)<br />
0<br />
Carbamazepine (CBZ)<br />
ClO2 HOCl<br />
O3 -<br />
HFeO4 ClO2 -2<br />
HOCl<br />
ClO2 -2<br />
O3 HOCl<br />
-2<br />
O3 -<br />
HFeO4 0 20 40 60 80 100 0 30 60 90 120 150 0 30 60 90 120 150<br />
Oxidant dose, µM Oxidant dose, µM Oxidant dose, µM<br />
0<br />
-1<br />
O 3<br />
HO.<br />
Atenolol (ATL) Ibuprofen (IBP)<br />
0<br />
ClO2 HOCl<br />
ClO2HOCl O3 + t-BuOH<br />
HO.<br />
-<br />
HFeO4 -2<br />
0 30 60 90 120 150<br />
-2<br />
0 30 60 90 120 150<br />
-2<br />
0 30 60 90 120 150<br />
Oxidant dose, µM Oxidant dose, µM Oxidant dose, µM<br />
the parent EfOM. This leads to a constant consumption rate of<br />
hydroxyl radicals. Hence, selective oxidants can be much<br />
more efficient than hydroxyl radicals especially when a higher<br />
degree of transformation is required (e.g. >90%) for EE2 or<br />
SMX. The characteristics of transformations of EE2 and SMX<br />
by hydroxyl radicals (i.e. a linear log-transformations vs.<br />
hydroxyl radical doses) were also observed for CBZ, ATL, and<br />
IBP (Figs. 3c–e, respectively).<br />
For EE2 (Fig. 3a), the transformation efficiency to achieve an<br />
one-log transformation for the oxidants was in the order of<br />
chlorine dioxide (3 mM) > ozone z ferrate VI (10 mM) > chlorine<br />
(20 mM) > hydroxyl radical (50 mM) where the value in the<br />
parenthesis shows the required oxidant dose. The very high<br />
efficiency of chlorine dioxide for EE2 transformation can be<br />
understood by its high reactivity to phenolic moieties (Fig. 1a),<br />
but relative low reactivity to amine moieties (Figs. 1d–f). Since<br />
phenolic and amine moieties are expected to be the two main<br />
ERMs present in EfOM, chlorine dioxide can be selective for<br />
the transformation of the micropollutants containing<br />
phenolic moieties in presence of EfOM. In contrast, the relatively<br />
low efficiency of chlorine for EE2 transformation can be<br />
understood by its high reactivity to amine moieties (Figs. 1d–f),<br />
but low reactivity to phenolic moieties (Fig. 1a).<br />
The one-log transformation efficiency in the case of SMX<br />
(Fig. 3b) was in the order of chlorine dioxide (15 mM) z ozone<br />
(20 mM) > ferrate VI<br />
(45 mM) z chlorine (48 mM) > hydroxyl<br />
water research 44 (2010) 555–566<br />
b c<br />
-1<br />
-1<br />
HO.<br />
HO. HO.<br />
-1<br />
-<br />
HFeO4 d e f<br />
O 3<br />
-<br />
HFeO4 0<br />
-1<br />
p-chlorobenzoate (pCBA)<br />
Fig. 3 – Logarithm of the residual concentrations (log(c/c 0)) of selected micropollutants as a function of oxidant doses in<br />
a secondary wastewater effluent (RDWW) at pH 8: (a) EE2, (b) SMX, (c) CBZ, (d) ATL, and (e) IBP. Symbols represent measured<br />
data and lines connect each data point to show the trend. The lines for hydroxyl radicals represent the linear regression of<br />
data. For the selective oxidants, the reaction time of 1 h was given to simulate realistic treatment conditions. Hydroxyl<br />
radicals were in situ produced by UV254 nm/H2O2 photolysis and their doses were quantified by the kinetic method described<br />
in the Supplementary Information, SI-word. For (e) IBP, ‘O3 D tBuOH’ indicates that the ozonation was conducted in<br />
presence of 60 mM tBuOH to exclude the contribution of hydroxyl radicals to the IBP transformation.<br />
HO.<br />
radical (80 mM). The relative lower transformation of SMX<br />
compared to EE2 at the same oxidant dose (Figs. 3a vs. b) are<br />
consistent with the lower k values of SMX than EE2 for the<br />
reaction with the oxidants at pH 8 except the case of chlorine<br />
(Table 1). The sulfonyl-moiety deactivates the aniline-moiety<br />
of SMX toward oxidation as an electron-withdrawing group.<br />
Therefore, the k values for the reaction of SMX with the<br />
selective oxidants are lower than those for aniline (SI-excel).<br />
In contrast, the cyclic ring-moiety activates the phenolicmoiety<br />
of EE2 toward oxidation as an electron-donating group.<br />
Therefore, the k values for the reaction of EE2 with the<br />
selective oxidants are higher than those for phenol (SI-excel).<br />
The smaller transformation of SMX than EE2 at the same<br />
chlorine doses (Figs. 3a vs. b) is not consistent with the 1.7-fold<br />
higher k for the reaction of chlorine with SMX at pH 8 (Table 1).<br />
Dodd et al. found that the reaction of chlorine with SMX<br />
produced the N-chlorinated SMX (i.e. chlorine-adduct to the<br />
aniline-moiety of SMX) as an initial product (Dodd and Huang,<br />
2004). The k values reported in Table 1 correspond to the rate<br />
constant of this initial reaction step. The N-chlorinated SMX<br />
can be reduced back to the parent, SMX, upon its reaction with<br />
reductants such as thiosulfate, which was employed as<br />
a quenching reagent in this study. Therefore, the actual extent<br />
of SMX transformation during chlorination could be smaller<br />
than the predictions based on the k values corresponding to<br />
the initial chlorine addition to SMX. Finally, even though the k
values of ozone or chlorine dioxide for its reaction with EE2<br />
and SMX are several orders of magnitude higher than those of<br />
chlorine or ferrate VI (Table 1), the transformation efficiency in<br />
RDWW among the selective oxidants differed only within<br />
a factor of 20. This is caused by their competitive consumption<br />
by micropollutants and EfOM.<br />
3.3.2. CBZ<br />
The one-log transformation efficiency in the case of CBZ<br />
(Fig. 3c) was in the order of ozone (30 mM) > ferrate VI<br />
(43 mM) > hydroxyl radical (100 mM) [ chlorine z chlorine<br />
dioxide. Chlorine and chlorine dioxide achieved little transformations<br />
of CBZ (
564<br />
3.5.2. Ammonia and nitrite<br />
In addition to EfOMs, inorganic nitrogen compounds, such as<br />
ammonia and nitrite, commonly present in poorly-nitrified/<br />
denitrified wastewater effluents can rapidly consume some<br />
oxidants and affect the transformation efficiency of micropollutants.<br />
The effect of ammonia and nitrite was investigated<br />
using EE2 as an example. From preliminary experiments,<br />
oxidant doses to achieve w80% transformation of EE2 in<br />
RDWW were determined. As a next step, the RDWW spiked<br />
with additional ammonia or nitrite (up to 100 mM) was treated<br />
with the pre-determined oxidant dose. In the case of the<br />
selective oxidants, the residual EE2 concentration was determined<br />
after 1 h of reaction time.<br />
Fig. 4a shows that in the case of chlorine, ammonia<br />
significantly decreased the degree of transformation of EE2. At<br />
an ammonia concentration higher than 20 mM ([NH4 þ ]0/<br />
[HOCl]0 1), the EE2 transformation was less than 10%. This is<br />
because chlorine reacts rapidly with ammonia, forming<br />
chloramines (Deborde and von Gunten, 2008). When the<br />
þ<br />
[NH4 ]0/[HOCl] 0 1, monochloramine (NH2Cl) is the predominant<br />
chloramine species which has a low reactivity to EE2<br />
(k ¼ 0.24 M 1 s 1 ) (Lee et al., 2008). For the other selective<br />
oxidants and hydroxyl radicals, the EE2 transformation was<br />
not affected by the presence of ammonia within the tested<br />
concentration. This is consistent with the low reactivity of<br />
these oxidants with ammonia (Table 1).<br />
Fig. 4b shows that nitrite significantly reduced the transformation<br />
of EE2 in the case of chlorine and ozone, which is<br />
consistent with the high k values of these two oxidants with<br />
nitrite (Table 1). For chlorine, the EE2 transformation decreased<br />
from 80 to 40% with an increase of the nitrite concentration<br />
from 6 to 30 mM ([NO2 ] 0/[HOCl] 0 ¼ from 0.3 to 1.5) and remained<br />
rather constant at w40% with a further increase of the nitrite<br />
concentration even up to 100 mM ([NO2 ]0/[HOCl]0 ¼ 5). The<br />
oxidation of nitrite by chlorine to nitrate proceeds through<br />
some reactive intermediates such as NO2Cl and NO2 þ (Johnson<br />
% transformation<br />
+<br />
[NH4 ]0 /[HOCl] 0<br />
0 1 2 3 4 5<br />
100<br />
80<br />
60<br />
40<br />
20<br />
HOCl<br />
O 3 ClO 2 Fe(VI) HO.<br />
0<br />
0 20 40 60 80 100<br />
+<br />
[NH4 ]0 , µM<br />
water research 44 (2010) 555–566<br />
a b<br />
100<br />
80<br />
60<br />
40<br />
20<br />
and Margerum, 1991). Reactions of these intermediates with<br />
EE2 might be responsible for the incomplete inhibition of EE2<br />
transformation by chlorine at the high nitrite concentration.<br />
For ozone, the EE2 transformation decreased gradually from 80<br />
to 20% with an increase of the nitrite concentration from 6 to<br />
100 mM ([NO2 ]0/[O3]0 ¼ from 0.8 to 12.5). Nitrite had little effect<br />
on the EE2 transformation in the case of chlorine dioxide and<br />
ferrate VI , which is consistent with the relatively low k values of<br />
these two oxidants with nitrite (Table 1). For hydroxyl radical,<br />
nitrite has a slightly positive effect on EE2 transformation even<br />
though it reacts rapidly with nitrite ( OH þ NO 2 / OH þ NO 2 ,<br />
k ¼ 10 10 M 1 s 1 , Table 1). In RDWW spiked with nitrite at the<br />
concentration of 100 mM, w80% of the produced hydroxyl<br />
radicals is scavenged by nitrite during the initial phase of the<br />
UV/H2O2 treatment. This is based on the relative reaction rate<br />
of hydroxyl radicals with nitrite vs. the remaining matrix<br />
components (i.e. EfOM, H2O2, and carbonate). Therefore, the<br />
slightly enhanced transformation of EE2 in presence of 100 mM<br />
nitrite might be explained by the reaction of EE2 with some<br />
intermediates from the reaction of hydroxyl radical with<br />
nitrite such as NO 2 (i.e. EE2 þ NO 2 / EE2 oxid þ NO 2 ). The k<br />
value for the reaction of NO 2 with EE2 is estimated to be<br />
w10 5 M 1 s 1 at pH 8 from the k value for the reaction of NO 2<br />
with 4-methylphenol (Neta et al., 1988).<br />
3.5.3. Bromide<br />
Bromide is present in waters and wastewaters at concentrations<br />
of 10 to several hundred mgL 1 (von Gunten, 2003b). Since<br />
the oxidation of bromide typically produces bromine, which is<br />
highly reactive to phenol- and amine-moieties, the presence of<br />
bromide during oxidative wastewater treatment can affect the<br />
transformation efficiency of micropollutants. Among<br />
the selective oxidants, only chlorine and ozone react with<br />
bromide (Table 1), generating bromine. Since bromine is about<br />
three orders of magnitude more reactive toward phenols than<br />
chlorine (Lee and von Gunten, 2009), transformation of<br />
−<br />
[NO2 ]0 /[O3 ] 0<br />
0 2 4 6 8 10 12<br />
−<br />
[NO2 ]0 /[HOCl] 0<br />
0 1 2 3 4 5<br />
0<br />
0 20 40 60 80 100<br />
−<br />
[NO2 ]0 , µM<br />
Fig. 4 – Effect of (a) ammonia (NH 4 D ) and (b) nitrite (NO2 L ) on the transformations of EE2 during treatment of a secondary<br />
wastewater effluent (RDWW) by different oxidants at pH 8. Preliminary experiments were conducted to determine the<br />
oxidant dose for each oxidant to achieve a 80% transformation of EE2 in RDWW without additionally spiked ammonia and<br />
nitrite. They were 20 mM for chlorine, 3 mM for chlorine dioxide, 8 mM for ozone, 8 mM for ferrate VI , and 37 mM for hydroxyl<br />
radicals. Symbols represent measured data and lines connect each data point to show the trend.<br />
HO.<br />
ClO 2<br />
Fe(VI)<br />
HOCl<br />
O 3
phenolic micropollutants can be significantly enhanced during<br />
chlorination of bromide-containing waters. A recent study<br />
showed that bromine produced from bromide was mainly<br />
responsible for 17a-ethinylestradiol transformation during<br />
chlorination of a wastewater effluent (Lee and von Gunten,<br />
2009). In the case of ozone, most phenolic- and amine-moieties<br />
are already oxidized by ozone before the significant formation<br />
of bromine from bromide. Hence, the presence of bromide<br />
affects little the transformation efficiency of micropollutants<br />
during ozonation. However, an enhanced transformation of 1<br />
amine-containing micropollutants is expected due to the<br />
relatively low reactivity of ozone versus high reactivity of<br />
bromine to 1 amines. For example, the k values at pH 7 for the<br />
reaction with glycine are 1.6 10 2 and 4.7 10 5 M 1 s 1 for<br />
ozone and bromine, respectively (SI-excel). In addition, like to<br />
the enhanced ammonia oxidation by ozone in the presence of<br />
bromide as a catalyst (Haag et al., 1984), N-brominated 1<br />
amines can be oxidized by ozone with a faster rate than the<br />
parent 1 amines and the corresponding reaction regenerates<br />
bromide. Finally, bromide has only a small effect on the<br />
transformation efficiency of micropollutants during advanced<br />
oxidation processes. This is due to (i) the low reactivity of<br />
hydroxyl radicals with bromide ( OH þ Br / OH þ Br ,<br />
k ¼ 1.1 10 9 M 1 s 1 , Table 1) which results in the formation of<br />
HOBr and in the case of H2O2-based AOPs (ii) the fast reaction of<br />
HOBr with H2O2 leading to bromide (von Gunten, 2003b).<br />
Formation of potentially carcinogenic bromate during<br />
ozonation of bromide-containing waters is one of drawbacks<br />
of ozonation. In recent two studies, only low concentration of<br />
bromate formation (i.e.
566<br />
Buxton, G.V., Greenstock, C.L., Helman, W.P., Ross, W.P., 1988.<br />
Critical review of rate constants for reactions of hydrated<br />
electrons, hydrogen atoms and hydroxyl radicals ( OH/ O )in<br />
aqueous solution. J. Phys. Chem. Ref. Data 17, 513–886.<br />
Deborde, M., von Gunten, U., 2008. Reactions of chlorine with<br />
inorganic and organic compounds during water<br />
treatmentdkinetics and mechanisms: a critical review. Water<br />
Res. 42, 13–51.<br />
Deborde, M., Rabouan, S., Gallard, H., Legube, B., 2004. Aqueous<br />
chlorination kinetics of some endocrine disruptors. Environ.<br />
Sci. Technol 38, 5577–5583.<br />
Deborde, M., Rabouan, S., Duguet, J.P., Legube, B., 2005. Kinetics of<br />
aqueous ozone-induced oxidation of some endocrine<br />
disruptors. Environ. Sci. Technol 39, 6086–6092.<br />
Dodd, M.C., Huang, C.H., 2004. Transformation of the<br />
antibacterial agent sulfamethoxazole in reactions with<br />
chlorine: kinetics, mechanisms, and pathways. Environ. Sci.<br />
Technol 38, 5607–5615.<br />
Dodd, M.C., Buffle, M.O., von Gunten, U., 2006. Oxidation of<br />
antibacterial molecules by aqueous ozone: moiety-specific<br />
reaction kinetics and application to ozone-based wastewater<br />
treatment. Environ. Sci. Technol 40, 1969–1977.<br />
Gates, D., 1998. The Chlorine Dioxide Handbook. American Water<br />
Works Association, Denver.<br />
Haag, W.R., Hoigné, J., Bader, H., 1984. Improved ammonia<br />
oxidation by ozone in the presence of bromide ion during<br />
water treatment. Water Res. 18, 1125–1128.<br />
Hoigné, J., 1998. Chemistry of aqueous ozone and transformation<br />
of pollutants by ozonation and advanced oxidation processes.<br />
In: Hrubec, J. (Ed.), The Handbook of Environmental<br />
Chemistry. Springer-Verlag, Berlin Heidelberg, pp. 83–141.<br />
Hollender, J., Zimmermann, S.G., Koepke, S., Krauss, M.,<br />
McArdell, C.S., Ort, C., Singer, H., von Gunten, U., Siegrist, H.,<br />
2009. Elimination of organic micropollutants in a municipal<br />
wastewater treatment plant upgraded with a full-scale postozonation<br />
followed by sand filtration. Environ. Sci. Technol 43,<br />
7862–7869.<br />
Huber, M.M., Canonica, S., Park, G.Y., von Gunten, U., 2003.<br />
Oxidation of pharmaceuticals during ozonation and advanced<br />
oxidation processes. Environ. Sci. Technol 37, 1016–1024.<br />
Huber, M.M., Korhonen, S., Ternes, T.A., von Gunten, U., 2005.<br />
Oxidation of pharmaceuticals during water treatment with<br />
chlorine dioxide. Water Res. 39, 3607–3617.<br />
Johnson, D.W., Margerum, D.W., 1991. Non-metal redox kinetics:<br />
a reexamination of the mechanism of the reaction between<br />
hypochlorite and nitrite ions. Inorg. Chem. 30, 4845–4851.<br />
Kumar, K., Margerum, D.W., 1987. Kinetics and mechanisms of<br />
general-acid-assisted oxidation of bromide by hypochlorite<br />
and hypochlorous acid. Inorg. Chem. 26, 2706–2711.<br />
Lee, Y., Yoon, J., von Gunten, U., 2005a. Kinetics of the oxidation<br />
of phenols and phenolic endocrine disruptors during water<br />
treatment with ferrate (Fe(VI)). Environ. Sci. Technol 39,<br />
8978–8984.<br />
Lee, Y., Yoon, J., von Gunten, U., 2005b. Spectrophotometric<br />
determination of ferrate (Fe(VI)) in water by ABTS. Water Res.<br />
39, 1946–1953.<br />
Lee, Y., Escher, B.I., von Gunten, U., 2008. Efficient removal of<br />
estrogenic activity during oxidative treatment of waters<br />
containing steroid estrogens. Environ. Sci. Technol 42, 6333–6339.<br />
water research 44 (2010) 555–566<br />
Lee, Y., von Gunten, U., 2009. Transformation of 17aethinylestradiol<br />
during water chlorination: effects of bromide<br />
on kinetics, products, and transformation pathways. Environ.<br />
Sci. Technol 43, 480–487.<br />
Lee, Y., von Gunten, U. Kinetics of ferrate VI reactions with<br />
inorganic compounds in water. in preparation.<br />
Lee, Y., Zimmermann, S.G., Kieu, A.T., von Gunten, U., 2009. Ferrate<br />
(Fe(VI)) application for municipal wastewater treatment:<br />
a novel process. Environ. Sci. Technol 43, 3831–3838.<br />
Letterman, R.D., 1999. Water Quality and Treatment, fifth ed.<br />
McGraw-Hill, New York.<br />
Liu, Q., Schurter, L.M., Muller, C.E., Aloisio, S., Francisco, J.S.,<br />
Margerum, D.W., 2001. Kinetics and mechanisms of aqueous<br />
ozone reactions with bromide, sulfite, hydrogen sulfite, iodide,<br />
and nitrite ions. Inorg. Chem. 40, 4436–4442.<br />
Meunier, L., Canonica, S., von Gunten, U., 2006. Implications of<br />
sequential use of UV and ozone for drinking water quality.<br />
Water Res. 40, 1864–1876.<br />
Neta, P., Huie, E., Ross, A.B., 1988. Rate constants for reactions of<br />
inorganic radicals in aqueous solution. J. Phys. Chem. Ref.<br />
Data 17, 1027–1264.<br />
Nowotny, N., Epp, B., von Sonntag, C., Fahlenkamp, H., 2007.<br />
Quantification and modelling of the elimination behavior of<br />
ecologically problematic wastewater micropollutants by<br />
adsorption on powdered and granulated activated carbon.<br />
Environ. Sci. Technol 41, 2050–2055.<br />
Pinkernell, U., Nowack, B., Gallard, H., von Gunten, U., 2000. Methods<br />
for the photometric determination of reactive bromine and<br />
chlorine species with ABTS. Water Res. 34, 4343–4350.<br />
Pinkston, K.E., Sedlak, D.L., 2004. Transformation of aromatic<br />
ether- and amine-containing pharmaceuticals during chlorine<br />
disinfection. Environ. Sci. Technol 38, 4019–4025.<br />
Rosario-Ortiz, F.L., Mezyk, S.P., Doud, D.F.R., Snyder, S.A., 2008.<br />
Quantitative correlation of absolute hydroxyl radical rate<br />
constants with non-isolated effluent organic matter bulk<br />
properties in water. Environ. Sci. Technol 42, 5924–5930.<br />
Sharma, V.K., Bloom, J.T., Joshi, V.N., 1998. Oxidation of ammonia<br />
by ferrate(VI). J. Environ. Sci. Health A33, 635–650.<br />
Ternes, T.A., Joss, A., 2006. Human Pharmaceuticals, Hormones<br />
and Fragrances. The Challenge of Micropollutants in Urban<br />
Water Management. IWA Publishing, London, New York.<br />
Thompson, G.W., Ockerman, G.W., Schreyer, J.M., 1951.<br />
Preparation and purification of potassium ferrate. VI. J. Am.<br />
Chem. Soc. 73, 1379–1381.<br />
von Gunten, U., 2003a. Ozonation of drinking water: Part I. Oxidation<br />
kinetics and product formation. Water Res. 37, 1443–1467.<br />
von Gunten, U., 2003b. Ozonation of drinking water: Part II.<br />
Disinfection and by-products formation in presence of<br />
bromide, iodide or chlorine. Water Res. 37, 1469–1487.<br />
Wert, E.C., Rosario-Ortiz, F.L., Drury, D.D., Snyder, S.A., 2007.<br />
Formation of oxidation byproducts from ozonation of<br />
wastewater. Water Res. 41, 1481–1490.<br />
Westerhoff, P., Mezyk, S.P., Cooper, W.J., Minakata, D., 2007.<br />
Electron pulse radiolysis determination of hydroxyl radical rate<br />
constants with Suwannee river fulvic acid and other dissolved<br />
organic matter isolates. Environ. Sci. Technol 41, 4640–4646.<br />
Zepp, R.G., Hoigné, J., Bader, H., 1987. Nitrate-induced<br />
photooxidation of trace organic chemicals in water. Environ.<br />
Sci. Technol 21, 443–450.
A review of phytoestrogens: Their occurrence<br />
and fate in the environment<br />
Ze-hua Liu a,b, *, Yoshinori Kanjo a , Satoshi Mizutani a<br />
a<br />
Department of Urban Engineering, Graduate School of Engineering, Osaka City University,<br />
3-3-138 Sugimoto, Sumiyoshi-ku, Osaka 558-8585, Japan<br />
b<br />
Department of Chemistry and Chemical Engineering, Shanghai University of Engineering Science,<br />
333 Longteng Road, Songjiang, Shanghai 201620, PR China<br />
article info<br />
Article history:<br />
Received 11 December 2008<br />
Received in revised form<br />
12 March 2009<br />
Accepted 16 March 2009<br />
Available online 24 March 2009<br />
Keywords:<br />
Phytoestrogens<br />
Endocrine disrupting compounds<br />
Wastewater<br />
Urinary excretion rate<br />
Monitoring<br />
1. Introduction<br />
abstract<br />
Endocrine disrupting compounds (EDCs) are chemicals with<br />
the potential to elicit negative effects on the endocrine<br />
systems of humans and wildlife. They include a broad class of<br />
chemicals such as natural estrogens/androgens, synthetic<br />
estrogens/androgens, industrial chemicals as well as phytoestrogens<br />
(Liu et al., 2009). As EDCs exist in extremely low<br />
concentrations (mg/L or ng/L) in the environment, one of the<br />
key points for the research is sensitive and accurate<br />
measurement of EDCs. Nowadays, two categories of methods<br />
are available; chemical analyses such as LC–MS/MS, GC–MS as<br />
well as HPLC and bioassays (in vitro and in vivo). Chemical<br />
water research 44 (2010) 567–577<br />
Available at www.sciencedirect.com<br />
journal homepage: www.elsevier.com/locate/watres<br />
Phytoestrogens are plant compounds with estrogenic activities. Many edible plants, some<br />
of which are common in the human diet, are rich in phytoestrogens. Almost all phytoestrogens<br />
eaten daily by people were reported partly recovered in urine or feces, which can<br />
be regarded as one of the main sources of their occurrence in municipal wastewaters. As<br />
they may act as one part of the endocrine disrupting compounds (EDCs) in water systems,<br />
some phytoestrogens have been monitored and detected in wastewater and other various<br />
environments. It is very difficult to monitor numerous unknown EDCs in complex wastewater<br />
samples, and it is helpful if some estimation of target EDCs can be done before<br />
monitoring. With this in mind, this review will: (1) summarize estrogenic activities or<br />
estrogenic potencies of phytoestrogens by different bioassays; (2) summarize daily urinary<br />
excretion rates of phytoestrogens by humans, and compare their urinary excretion rates to<br />
that of estrone, which suggests that most phytoestrogens may occur in municipal wastewaters;<br />
(3) collect and summarize published data on the occurrence and fate of phytoestrogens<br />
in various environments.<br />
ª 2009 Elsevier Ltd. All rights reserved.<br />
analyses can measure precise concentrations of known target<br />
EDCs in environmental samples, while bioassays provide<br />
direct information of the estrogenic activity of complex<br />
mixtures of EDCs that are likely to occur in water irrespective<br />
of the causative compounds. For large-scale monitoring<br />
research, bioassays (in vitro) seem most appropriate due to<br />
their low cost and high throughput capability. However, for<br />
the complexity of environmental samples (usually wastewater<br />
samples), values measured by different bioassays on<br />
the same samples differed greatly, which have lead to questions<br />
over the validity and effectiveness of bioassays for<br />
environmental samples. To find widely acceptable standard<br />
bioassays for environmental samples, the chemically derived<br />
* Corresponding author: Department of Urban Engineering, Graduate School of Engineering, Osaka City University, 3-3-138 Sugimoto,<br />
Sumiyoshi-ku, Osaka 558-8585, Japan. Tel./fax: þ81 6 6605 3048.<br />
E-mail address: liuzehua78@yahoo.co.jp (Z.-h. Liu).<br />
0043-1354/$ – see front matter ª 2009 Elsevier Ltd. All rights reserved.<br />
doi:10.1016/j.watres.2009.03.025
568<br />
values are often compared to those of bioassays, which<br />
accelerates monitoring a multitude of endocrine-like chemicals<br />
in environmental samples (Furuichi et al., 2004; Nakada<br />
et al., 2004; Tan, 2006; Nelson et al., 2007; Salste et al., 2007). In<br />
Fernandez et al. (2007), as many as 30 EDCs have been monitored<br />
in 9 wastewater treatment plants in Canada and the<br />
results of chemically derived estrogen equivalents (EEQ) were<br />
also compared with those measured by recombinant yeast<br />
assay (RYA). The monitoring of numerous EDCs in wastewater<br />
is difficult, and would be preferable if some guidelines were<br />
applicable to wastewater in which EDCs may exist in high<br />
values, while other EDCs may not.<br />
Phytoestrogens are defined as any plant compounds<br />
structurally and/or functionally similar to ovarian and<br />
placental estrogens and their active metabolites (Whitten and<br />
Patisaul, 2001). Phytoestrogens with rather high concentrations<br />
widely exist in numerous plants such as soybeans,<br />
fruits, cabbages and so on (Liggin et al., 2000; US Department<br />
of Agriculture, USDA, 2003). Some of these are widelyconsumed<br />
foods, especially in Asia, where high amounts of<br />
soy are consumed in direct or indirect ways (Adlercreutz et al.,<br />
1991; Nagata et al., 2007). Although most research on phytoestrogens<br />
has suggested that they yield positive effects on<br />
human health (Setchell and Cassidy, 1999; Lissin and Cook,<br />
2000; Kris-Etherton et al., 2002), some research denoted that<br />
most phytoestrogens also acted as endocrine disruptors,<br />
which could cause intersex in aquatic organisms (Kiparissis<br />
et al., 2003; Tzchori et al., 2004). In addition, genotoxicity of<br />
phytoestrogens was reviewed by Stopper et al. (2005), who<br />
pointed out some possible adverse effects of phytoestrogens<br />
on humans.<br />
Due to the possible negative effects mentioned above,<br />
much attention has been paid to the occurrence and fate of<br />
phytoestrogens by environmental researchers in municipal/<br />
industrial wastewater or surface waters, in which they were<br />
detected with high frequency (Kiparissis et al., 2001; Lagana<br />
et al., 2004; Bacaloni et al., 2005; Erbs et al., 2007). Liu et al. (in<br />
press) detected 15 EDCs at two municipal wastewater treatment<br />
plants, in which the concentrations of daidzein in<br />
influents were from 3900 to 12,000 ng/L and its chemically<br />
derived estrogenic equivalents (EEQ) were 0.9–5.3% of those<br />
measured by the ER-binding assay (bisphenol A (BPA) and<br />
nonylphenol (NP) totally contributed 0.7–4.1% in the same<br />
samples), indicating a relatively high proportion in wastewater<br />
which was not negligible. Like natural estrogens or<br />
androgens, the main source of phytoestrogens in municipal<br />
wastewaters can also be attributed to daily excretions of urine<br />
or feces by humans. Therefore, a thorough understanding of<br />
phytoestrogens from daily excretions by humans is helpful for<br />
estimating their possible concentrations in municipal wastewater.<br />
This estimation has been applied to natural estrogens<br />
(Johnson et al., 2000; Johnson and Williams, 2004), and it is<br />
also helpful for suggesting possible priority phytoestrogens<br />
when monitoring for numerous EDCs in wastewater samples<br />
is carried out. Although there is much data on the excretion<br />
rates of phytoestrogens from human urine, few conclusions<br />
have been drawn for the purpose above. In addition, in terms<br />
of convenience to the comparison of chemical analyses with<br />
bioassays, relative binding affinities (RBA) or estrogenic/<br />
androgenic potencies are also important.<br />
water research 44 (2010) 567–577<br />
The objectives of this review are: (1) to summarize reported<br />
estrogenic/androgenic activities of phytoestrogens by<br />
different bioassays; (2) to summarize the daily urinary excretion<br />
rates of phytoestrogens by humans; and (3) to summarize<br />
reported data on the occurrence and fate of phytoestrogens in<br />
the various environments.<br />
2. Estrogenic/androgenic activities of<br />
phytoestrogens<br />
Phytoestrogens are a large family with two major classes,<br />
flavonoids and lignans. The number of flavonoids is greater<br />
than the number of lignans. Flavonoids also possess higher<br />
relative estrogenic activities than those in lignans (Whitten<br />
and Patisaul, 2001). One interest for researchers of environmental<br />
engineering relating to EDCs is the estrogenic/androgenic<br />
activities of phytoestrogens. There is much available<br />
data on their estrogenic/androgenic activities by different<br />
bioassays. To give an overview of their estrogenic/androgenic<br />
activities, some published data are outlined in Table 1, in<br />
which the estrogen standard substance is 17b-estradiol (E2)<br />
and the androgen standard substance is testosterone (Te) or<br />
methyltrienolone (R1881). From Table 1, it is easily seen that<br />
almost all phytoestrogens are far weaker estrogenic<br />
substances, except in the transient gene expression assay<br />
(TGEA), in which the estrogenic potencies of some phytoestrogens<br />
are at the same level, or even stronger than that of E2.<br />
Compared to the estrogenic activities of industrial chemicals<br />
BPA or NP (Blair et al., 2000; Gutendorf and Westendorf, 2001;<br />
Liu et al., 2009), most phytoestrogens are in the same order of<br />
magnitude, which denote that the estrogenic activities<br />
derived from phytoestrogens can not be neglected when they<br />
remain in high concentrations in water samples. In addition,<br />
some phytoestrogens also possess androgenic activities<br />
(Table 1). With the RBA or estrogenic potency in Table 1, we<br />
can calculate the contribution ratio of phytoestrogens (Ratio)<br />
to values measured by bioassays using Eq. (1) and the contribution<br />
ratio of phytoestrogens (Ratioc) to the chemically<br />
derived EEQ using Eq. (2):<br />
Ratio ¼ EEQ P<br />
c RBAiðEPiÞ Ci<br />
¼<br />
(1)<br />
EEQb EEQb where EEQc and EEQb are the chemically calculated EEQ and<br />
the measured EEQ by bioassay, respectively; Ci represents the<br />
concentration of an individual EDC in wastewater sample or<br />
the urinary excretion rate of an individual phytoestrogen from<br />
human and RBAi(EP)i is its corresponding estrogenic activity.<br />
Ratioc ¼ EEQi ¼<br />
EEQc RBAiðEPiÞ Ci<br />
P<br />
RBAiðEPiÞ Ci<br />
where EEQ i is the chemically derived EEQ of an individual<br />
phytoestrogen in urine or an individual EDC in wastewater.<br />
To monitor phytoestrogens in complex wastewater or<br />
other water samples, it is very important to estimate their<br />
concentrations possibly existing in environmental samples.<br />
For phytoestrogens in municipal wastewater, one of the main<br />
sources can be regarded as human urine or feces. Therefore,<br />
their daily urinary excretion rates were concluded as shown<br />
below.<br />
(2)
Table 1 – Estrogenic/androgenic activities of phytoestrogens by different bioassays * .<br />
Class Subclass Phytoestrogen CAS RBA (ER) RBA (AR) Estrogenic potency (EP)<br />
hER(a) a;b<br />
hER(b) b<br />
hAR(a) a<br />
rAR(a) c<br />
RGA a (ER(a)) YES d<br />
(ER(a))<br />
TGEA b<br />
ER(a) ER(b)<br />
Flavonoids Isoflavone Daidzein 486-66-8 1.8e-3; 1e-3 5e-3 N.B – 6.67e-5 3e-6 0.97 0.8<br />
Dihydrodaidzein 17238-05-0 –; – – – – – – – –<br />
Formononetin 485-72-3 2.65e-4;
570<br />
Table 2 – Daily excretion rates of phytoestrogens from human urine.<br />
Phytoestrogens Gender condition<br />
(number)<br />
water research 44 (2010) 567–577<br />
Diet Daily excretion rates<br />
(average)<br />
mg/d mgE2/d c<br />
Mean EEQ Reference<br />
Genistein Men (267) Normal a<br />
n.d.–7579.4 (132.4) 0.159 Low et al., 2005<br />
Women (250) Normal – b (2172.7) 2.607 Dai et al., 2002 d<br />
Women (19) Soy protein diets 1810.6–2837.5 (2357.8) 2.829 Lampe et al., 2001<br />
Women (268) Normal 31.5–1251.3 (219.8) 0.264 Tonkelaar et al., 2001 d<br />
Women (18) Normal 183.5–280.8 (227) 0.272 Xu et al., 2000<br />
7.1 mg total isoflavones/d 362.4–492.9 (422.7) 0.507<br />
65 mg total isoflavones/d 1521.1–2096.2 (1785.7) 2.143<br />
132 mg isoflavones/d 3045.8–4144.6 (3553.1) 4.264<br />
Men (49) Normal – (78.4) 0.094 Lampe et al., 1999<br />
Women (49) Normal – (45.9) 0.055<br />
Men (2) Normal 368.6–587.2 (477.9) 0.573 Adlercreutz et al., 1995 e<br />
Women (3) Normal (2 vegetarians) 482.6–3653.3 (1749.9) 2.100 Adlercreutz et al., 1995<br />
Women (18) Normal 10.8–281.0 (48.6) 0.058 Lampe et al., 1994<br />
Daidzein Men (267) Normal n.d.–4309.8 (227.0) 0.409 Low et al., 2005<br />
Women (250) Normal – (4390.7) 7.903 Dai et al., 2002<br />
Women (19) Soy protein diets 2517.0–5008.5 (3686.4) 6.636 Lampe et al., 2001<br />
Women (18) Normal 200.8–323.4 (254.7) 0.458 Xu et al., 2000<br />
7.1 mg total isoflavones/d 303.8–523.5 (398.9) 0.718<br />
65 mg total isoflavones/d 1281.6–2255.8 (1700.3) 3.061<br />
132 mg isoflavones/d 2722.6–4692.5 (3574.3) 6.434<br />
Men (49) Normal – (211.0) 0.380 Lampe et al., 1999<br />
Women (49) Normal – (94.1) 0.169<br />
Men (2) Normal 1074.2–1113.6 (1093.9) 1.969 Adlercreutz et al., 1995<br />
Women (3) Normal (2 vegetarians) 1240.4–5865.3 (3624.1) 6.523 Adlercreutz et al., 1995<br />
Women (18) Normal 28.0–2333.9 (180.5) 0.325 Lampe et al., 1994<br />
Women (41) Normal 56.1–322.9 (124.2) 0.224 Adlercreutz et al., 1986<br />
Dihydrodaidzein Women (250) Normal – (1340.2) – Dai et al., 2002<br />
Women (18) Normal 151.2–222.7 (183.5) – Xu et al., 2000<br />
7.1 mg total isoflavones/d 221.7–339.5 (274.4) –<br />
65 mg total isoflavones/d 853.1–1334.1 (1066.8) –<br />
132 mg isoflavones/d 1528.8–2341.4 (1891.9) –<br />
ODMA Men (267) Normal n.d.–381.9 (48.1) – Low et al., 2005<br />
Women (250) Normal – (1299.1) – Dai et al., 2002<br />
Women (19) Soy protein diets 206.6–1859.5 (707.0) – Lampe et al., 2001<br />
women (18) Normal 142.6–214.4 (174.8) – Xu et al., 2000<br />
7.1 mg total isoflavones/d 206.4–297.8 (247.9) –<br />
65 mg total isoflavones/d 1045.2–1344.5 (1265.5) –<br />
132 mg isoflavones/d 2236.9–3227.8 (2687.3) –<br />
Men (2) Normal 22.8–335.2 (179.0) – Adlercreutz et al., 1995<br />
Women (3) Normal (2 vegetarians) 171.0–627.1 (379.6) – Adlercreutz et al., 1995<br />
Women (41) Normal 6.5–27.4 (12.7) – Adlercreutz et al., 1986<br />
Equol Men (267) Normal n.d.–1280.5 (18.2) 0.157 Low et al., 2005<br />
Women (19) Soy protein diet n.d.–581.4 (271.0) 2.331 Lampe et al., 2001<br />
Women (18) Normal 45.1–64.4 (53.8) 0.447 Xu et al., 2000<br />
7.1 mg total isoflavones/d 61.3–120.4 (85.8) 0.738<br />
65 mg total isoflavones/d 126.9–249.5 (178.1) 1.532<br />
132 mg isoflavones/d 189.5–372.4 (265.8) 2.286<br />
Men (49) Normal – (21.8) 0.187 Lampe et al., 1999<br />
Women (49) Normal – (17.0) 0.146<br />
Men (2) Normal 26.6–117.5 (72.1) 0.620 Adlercreutz et al., 1995<br />
Women (3) Normal (2 vegetarians) 16.9–60.8 (37.7) 0.324 Adlercreutz et al., 1995<br />
Women (18) Normal 14.5–43.6 (26.6) 0.229 Lampe et al., 1994<br />
Women (41) Normal 15.5–24.7 (20.4) 0.175 Adlercreutz et al., 1986
Table 2 (continued)<br />
Phytoestrogens Gender condition<br />
(number)<br />
3. Occurrence and fate of phytoestrogens in<br />
the environment<br />
3.1. Phytoestrogens in human urine<br />
Phytoestrogens in human urine are from the diet or dietary<br />
metabolites, and their concentrations in urine have been<br />
found dose dependent on their consumption (Karr et al.,<br />
1997; Zhang et al., 1999; Xu et al., 2000; Peeters et al., 2003).<br />
For disease related items, there has been much research on<br />
the daily excretion rates of urinary phytoestrogens, in which<br />
the excretion rates varied greatly with race and dietary<br />
customs (Horn-Ross et al., 1997; Seow et al., 1998). To give an<br />
overview, published data on urinary excretion rates are<br />
outlined in Table 2. In order to compare the estrogenic<br />
potencies of phytoestrogens from human urine, the authors<br />
water research 44 (2010) 567–577 571<br />
Diet Daily excretion rates<br />
(average)<br />
mg/d mgE2/d c<br />
Mean EEQ Reference<br />
Glycitein Men (267) Normal n.d.–556.2 (51.3) 0.014 Low et al., 2005<br />
Women (250) Normal – (690.8) 0.193 Dai et al., 2002<br />
Women (18) Normal 73.9–94.1 (83.3) 0.023 Xu et al., 2000<br />
7.1 mg total isoflavones/d 109.4–141.3 (120.8) 0.034<br />
65 mg total isoflavones/d 149.0–292.8 (209.9) 0.059<br />
132 mg isoflavones/d 222.3–436.9 (311.8) 0.087<br />
Coumestrol Women (18) Normal 8.3–9.5 (8.9) 0.023 Xu et al., 2000<br />
7.1 mg total isoflavones/d 7.9–9.2 (8.6) 0.023<br />
65 mg total isoflavones/d 8.4–9.7 (9.0) 0.024<br />
132 mg isoflavones/d 8.8–10.2 (9.5) 0.025<br />
Enterolactone Men (267) Normal n.d.–13218.5 (1020.4) 0.033 Low et al., 2005<br />
Women (250) Normal – (1873.5) 0.060 Dai et al., 2002<br />
Women (19) Soy protein diets 477.3–2386.7 (1555.1) 0.050 Lampe et al., 2001<br />
Women (268) Normal 19.0–6157.9 (1276.2) 0.041 Tonkelaar et al., 2001<br />
Women (18) Normal 364.3–417.1 (389.6) 0.012 Xu et al., 2000<br />
7.1 mg total isoflavones/d 347–489.6 (412.3) 0.013<br />
65 mg total isoflavones/d 599.6–858.3 (717.2) 0.023<br />
132 mg isoflavones/d 960.6–1352.0 (1139.6) 0.036<br />
Men (49) Normal – (683.2) 0.022 Lampe et al., 1999<br />
Women (49) Normal – (584.7) 0.019<br />
Men (2) Normal 426.9–1592.8 (1009.9) 0.032 Adlercreutz et al., 1995<br />
Women (3) Normal (2 vegetarians) 1305.2–2488.7 (1851.2) 0.059 Adlercreutz et al., 1995<br />
Women (41) Normal 611.6–1244.0 (914.3) 0.029 Adlercreutz et al., 1986<br />
Naringenin Women (250) Normal – (1217.0) 0.065 Dai et al., 2002<br />
Men and women (77) Normal – (653.8) 0.035 Nielsen et al., 2002<br />
Quercetin Men and women (77) Normal – (23.4) – Nielsen et al., 2002<br />
Phloretin Men and women (77) Normal – (84.8) 0.013 Nielsen et al., 2002<br />
Kaempferol Men and women (77) Normal – (51.3) 0.015 Nielsen et al., 2002<br />
a Normal diet.<br />
b Not available.<br />
c Calculated as EEQi in Eq. (2), values of RBA based on ER-binding assay (NITE, 2008).<br />
d Transferred from nmol/mg creatinine or mmol/mol creatinine to mg/d, and creatinine excretion rate of 1 g/d for women was chosen (Jackson,<br />
1966).<br />
e Calculated from their free phytoestrogens and their conjugates.<br />
converted the urinary excretion rate of the individual phytoestrogens<br />
into a corresponding value, calculated as EEQi in<br />
the numerators of Eq. (2); we label it ‘‘excretion rate of EEQ’’.<br />
As is shown in Table 2, out of 12 phytoestrogens, genistein<br />
and daidzein are the two most important phytoestrogens<br />
evaluated by both urinary excretion rate and excretion rate<br />
of EEQ; followed by equol, for which its urinary excretion<br />
rates by normal diets are low but with a relatively high<br />
corresponding excretion rate of EEQ. Enterolactone and<br />
naringenin are the phytoestrogens with high urinary excretion<br />
rates, but very low estrogenic potencies (low RBA, Table<br />
1). Therefore, the corresponding excretion rates of EEQ by<br />
enterolactone and naringenin seem relatively low. Data of<br />
urinary excretion rates of some phytoestrogens, especially<br />
for quercetin, phloretin as well as kaempferol, is very<br />
limited. For better comparison, it would be helpful if more<br />
data were available.
572<br />
In addition, biochanin A and formononetin have been<br />
found existing in high concentrations in red clovers of<br />
Canada, and have also been detected in many fruits, beans<br />
and vegetables (Mazur et al., 1996; Sivesind and Seguin, 2005;<br />
Kuhnle et al., 2007, 2008). Few data on urinary excretion rates<br />
are available. As a reference, one published paper showed the<br />
average fecal excretion rate was about 1.5 mg/d for biochannin<br />
A and 11.9 mg/d for formononetin (Jenner et al.,<br />
2005). A phytosterol b-sitosterol with estrogenic activity<br />
(Table 1) was found rich in a wide range of fruits, vegetables,<br />
nuts, and many kinds of seed oils (Grosso et al., 2000; Beveridge<br />
et al., 2002, 2005; Iwatsuki et al., 2003; Holser et al.,<br />
2004; Phillips et al., 2005; Jimenez-Escrig et al., 2006; Zhang<br />
et al., 2006). Although there is no available data on the<br />
urinary excretion rate of b-sitosterol, very high human fecal<br />
excretion rates of 52–322 mg/d and a high human fecal<br />
concentration of 121 mg/g in dry feces (average) were reported<br />
(Eneroth et al., 1964; Leeming et al., 1996), which suggested<br />
that b-sitosterol might also exist in wastewater with a very<br />
high concentration.<br />
From the view of statistics, a rough estimation on the<br />
arithmetic average of urinary excretion rates was done on<br />
the data of normal diets in Table 2, which was also applied<br />
on human urinary excretion rates of natural estrogens and<br />
androgens (Liu et al., submitted for publication). If the loss of<br />
phytoestrogens during the course of flow to wastewater<br />
treatment plants is low, their concentrations in wastewater<br />
can be compared to those of natural estrogens. Although it is<br />
well known that 17b-estradiol (E2) is the strongest natural<br />
estrogen, estrone (E1) is the one which was detected<br />
frequently with high concentrations in effluent of WWTPs<br />
(Baronti et al., 2000; Joss et al., 2004; Johnson et al., 2007; Liu<br />
et al., 2009). Therefore, the average urinary excretion rates of<br />
phytoestrogens were compared to that of E1 as shown in<br />
Fig. 1, and the urinary excretion rate of E1 was modified from<br />
Liu et al. (submitted for publication), for which the urinary<br />
excretion rates of two pregnant women out of 177 persons<br />
(11.3&) were also included. From Fig. 1 it is known that the<br />
urinary excretion rates of phytoestrogens are 1–86 times that<br />
of E1, except for coumestrol (0.45 times); Among phytoestrogens<br />
in Table 2, genistein and daidzein, frequently detected<br />
in wastewaters, are 35 and 86 times that of E1,<br />
urinary excretion rates(µg/d)<br />
10000<br />
1000<br />
100<br />
10<br />
1<br />
E1<br />
Genistein<br />
Daidzein<br />
Dihydrodaidzein<br />
ODMA<br />
Glycitein<br />
Equol<br />
Coumestrol<br />
Enterlactone<br />
Naringerin<br />
Quercetin<br />
Phloretin<br />
Fig. 1 – Comparison of urinary excretion rates between<br />
phytoestrogens and E1.<br />
water research 44 (2010) 567–577<br />
Kaemperol<br />
respectively. According to published results, the average<br />
concentrations of genistein and daidzein in influent of one<br />
WWTP in Italy were about 3 and 9 times that of the detected<br />
E1 (Lagana et al., 2004), while the concentrations of daidzein<br />
in 12 influent samples of two municipal WWTPs in Japan<br />
showed about 8–46 times (22 times on average) that of the<br />
detected E1 (Liu et al., in press). Compared to detected<br />
concentrations for genistein and daidzein, their estimated<br />
concentrations in Fig. 1 are about 10 times and 2–10 times<br />
higher. The large differences may result from the different<br />
loss rates before they reach WWTPs, different excretion<br />
ratios of urine and feces for phytoestrogens and E1, as well as<br />
regional dietary customs. Although coumestrol is the leastpresent<br />
urinary excretion phytoestrogen in Fig 1, it was also<br />
reported existing in municipal WWTPs (Bacaloni et al., 2005).<br />
Based on this fact, the other 11 phytoestrogens in Fig. 1 are<br />
very likely to be present in municipal wastewater. Based on<br />
EEQ, phytoestrogens such as daidzein, genistein, equol and<br />
glycitein should not be neglected, for which the excretion<br />
rates of EEQ are about 35%, 10%, 2% and 1% that of human<br />
urinary E1. As enterlactone and naringerin are much weaker<br />
estrogens than daidzein (Table 1), although their urinary<br />
excretion rates are on the same level as daidzein, the corresponding<br />
excretion rates of EEQ are just about 0.5% and<br />
0.7% that of human urinary E1. However, the urinary excretion<br />
rates of enterlactone in some regions may be much<br />
higher than that in Fig. 1 (Lampe et al., 1994; Wang, 2002;<br />
Milder et al., 2005; Valentin-Blasini et al., 2005; Penalvo et al.,<br />
2008), which may yield a higher contribution ratio in wastewater.<br />
In addition, the urinary excretion rates of dihydrodaidzein<br />
and ODMA are almost the same as that of<br />
daidzein, but the data on their estrogenic activities is not<br />
available; therefore, their corresponding excretion rates of<br />
EEQ can not be applied here.<br />
3.2. Phytoestrogens in wastewaters and the other<br />
water bodies<br />
As predicted above that most phytoestrogens may exist in<br />
wastewater, some phytoestrogens have been proven to not<br />
only exist in wastewater but also in surface waters (Kawanishi<br />
et al., 2004; Lagana et al., 2004; Bacaloni et al., 2005; Erbs et al.,<br />
2007). Mycoestrogens (Table 1) are zearalenone and its derivatives.<br />
They are macrocyclic toxins produced by several<br />
Fusarium species, colonizing maize, sorghum, wheat, barley<br />
oats and other cereal grains, which were also reported<br />
occurring in wastewaters and surface waters (Lagana et al.,<br />
2004). Published data on the occurrence and fate of phytoestrogens<br />
in the course of WWTPs are included in Table 3.<br />
As stated in Section 1, to give a whole profile of EDCs in the<br />
wastewater of WWTPs, a combination of chemical analyses<br />
and bioassays has been used, and estrogenic contribution<br />
ratios of detected EDCs to those of chemical derived EEQ have<br />
also been compared. As shown in Table 3, it is known that<br />
some phytoestrogens have been detected in both influent and<br />
effluent, and their removal efficiencies varied greatly as<br />
shown for natural estrogens and androgens (Liu et al., 2009).<br />
Data in Table 3 also suggested that the estrogenic contribution<br />
ratios of phytoestrogens could not be ignored in wastewaters<br />
when present in relatively high concentrations. For example,
Table 3 – Occurrence and fate of phytoestrogens in wastewater.<br />
Phytoestrogens Country WWTP<br />
type (n)<br />
the authors calculated that six phytoestrogens in influents of<br />
a municipal WWTP contributed over 2.4% of total chemically<br />
calculated EEQ including E1, E2, E3 and EE2 (RBA values of<br />
EDCs are chosen from NITE, 2008), and increased to over 4.5%<br />
for the corresponding effluents (Lagana et al., 2004); In 12<br />
influent samples of two municipal WWTPs, the contribution<br />
ratios of daidzein accounted for 3.0–14.8% of total chemically<br />
calculated EEQ (Liu et al., in press). In the influents of a swine<br />
wastewater treatment plant, the contribution ratios of phytoestrogen<br />
equol accounted for as high as about 64% of total<br />
chemically derived EEQ (Furuichi et al., 2006). In addition, for<br />
the extremely high concentrations of b-sitosterol in wastewaters<br />
(Table 3), its estrogenic contribution ratio in wastewater<br />
seems relatively high and it also can not be neglected.<br />
Similar to natural estrogens and androgens, about 97–<br />
100% of phytoestrogens excreted from human urine are<br />
phytoestrogen conjugates, in which 62–97% of phytoestrogens<br />
are glucuronide conjugates (Adlercreutz et al., 1995).<br />
Although there is no information on the occurrence and fate<br />
of phytoestrogen conjugates in WWTPs, the glycoside<br />
conjugates of daidzein and genistein (daidzin and genistin),<br />
which have structures similar to those of their glucuronide<br />
conjugates, have been proven hydrolyzable to daidzein and<br />
water research 44 (2010) 567–577 573<br />
Methods LOD Influent Effluent Average<br />
removal (%)<br />
Reference<br />
Daidzein Germany AS (2) – 10 – n.d.–10 – Pawlowski et al., 2003<br />
Italy AS (1) LC–MS/MS 2.5–3.5 75–120 7–16 88 Lagana et al., 2004<br />
Italy AS (2) LC–MS/MS 3.5 97–1685 5–81 88 or 98 Bacaloni et al., 2005<br />
Japan AS (2) LC–MS/MS 5–60 b<br />
3900–12000 5.7–23 99.8 Liu et al., in press<br />
Daidzin Italy AS (2) LC–MS/MS 9–15 18–74 n.d.–22 69 or 100 Bacaloni et al., 2005<br />
Genistein Canada – (1) LC–MS/MS – 13100 10100 23 Kiparissis et al., 2001<br />
Italy As (1) LC–MS/MS 1.3–4.3 195–384 15–83 81 Lagana et al., 2004<br />
Italy AS (2) LC–MS/MS 5–6 160–954 7–22 97 Bacaloni et al., 2005<br />
Genistin Italy AS (2) LC–MS/MS 9–13 11–62 n.d.–19 66 or 100 Bacaloni et al., 2005<br />
Glycitein Italy AS (2) LC–MS/MS 6–8 6–25 n.d.–16 62 or 88 Bacaloni et al., 2005<br />
Formononetin Italy AS (2) LC–MS/MS 5–8 n.d.–10 n.d. – Bacaloni et al., 2005<br />
Coumestrol Italy AS (2) LC–MS/MS 2–5 9–23 n.d.–19 66 or 100 Bacaloni et al., 2005<br />
Biochanin-A Italy As (1) LC–MS/MS 0.9–5.4 8–18 3–5 73 Lagana et al., 2004<br />
Italy AS (2) LC–MS/MS 5–9 14–76 n.d.–21 67 or 90 Bacaloni et al., 2005<br />
Equol Japan – (1) LC–MS/MS – 9.4e5–1.1e6 170–410 99.99 Furuichi et al., 2006<br />
Zearlenone Italy AS (1) LC–MS/MS 0.6–1.7 3–18 3–10 53 Lagana et al., 2004<br />
a-Zearalanol Italy AS (1) LC–MS/MS 5.4–8.3 n.d.–10 n.d.–7 0 Lagana et al., 2004<br />
b-Zearalanol Italy AS (1) LC–MS/MS 3.7–8.4 n.d.–8 n.d.–5 33 Lagana et al., 2004<br />
a-Zearalenol Germany AS (2) – 10 – 20 or 30 – Pawlowski et al., 2003<br />
b-Sitosterol Switzerland – (1) GC–MS 125 – 400–1300 – Baig et al., 2008<br />
Canada AS (3);<br />
TF/SC (1);<br />
L (1)<br />
GC–HRMS 12–39 7303–35033 350–42110 – Fernandez et al., 2007<br />
USA – (2) GC–MS – 32000 a<br />
– – Cain et al., 2008<br />
Germany – (2) GC–MS – 14000or 16000 – – Radke, 2005<br />
Germany AS (2) – 50 – 60 or 90 – Pawlowski et al., 2003<br />
Germany – (7) GC–MS/MS – 15900 1900 88 Heberer, 2002<br />
France AS (2) GC–MS – 4194–49920 1100–50165 – Quemeneur and Marty,<br />
1994<br />
Enterolactone Australian – (1) LC–MS 1 600 – – Kang et al., 2006<br />
Finland AS (1) LC–MS/MS – 1223 659 46 Smeds et al., 2007<br />
AS, activated sludge; TF/SC, trickling filter/solid contact; L, lagoon; –, not available; n.d., not detected; LOD, limit of detection.<br />
a Maximum concentration.<br />
b Limit of quantification.<br />
genistein by the glucosidase of human intestinal bacteria<br />
(Hur et al., 2000). The detected daidzin and genistin in some<br />
effluents of WWTPs suggested their incomplete hydrolysis<br />
during the whole activated sludge process (Table 3). In addition,<br />
based on the fact that natural estrogen conjugates<br />
occurred both the influent and effluent of WWTPs (D’Ascenzo<br />
et al., 2003; Isobe et al., 2003; Komori et al., 2004; Reddy et al.,<br />
2005; Nakada et al., 2006), phytoestrogen conjugates may also<br />
be present in influent and effluent of WWTPs with high<br />
frequency.<br />
Published data on the occurrence of phytoestrogens in<br />
other various environments are collected in Table 4. From<br />
Table 4, it is known that most phytoestrogens including<br />
mycoestrogens were found in rivers, and in some occasions<br />
the concentrations of phytoestrogens in drainage water were<br />
much higher than those in river water, which indicated that<br />
the drainage water was also a source of phytoestrogens in<br />
river waters. In Table 4 b-sitosterol showed very high<br />
concentrations in rivers and it was also detected in the<br />
atmosphere. However, it has not been detected in drinking<br />
water (Quemeneur and Marty, 1994; Heberer, 2002; Martins<br />
et al., 2007; Stackelberg et al., 2007). In addition, enterlactone<br />
was seldom monitored in wastewater as a micro-pollutant,
574<br />
Table 4 – Phytoestrogens in the other various environments.<br />
Phytoestrogens River water<br />
(ng/L)<br />
whereas the recent investigation denoted that it not only<br />
existed in wastewater, but also in river water, lake water, sea<br />
water and drinking water (Table 4).<br />
4. Conclusions<br />
Drainage water<br />
(ng/L)<br />
Urinary excretion rates of phytoestrogens from humans and<br />
their occurrence and fate in the environment were summarized<br />
here. It was concluded that most phytoestrogens may be<br />
present in municipal wastewaters as can be estimated from<br />
their urinary excretion rates. Although phytoestrogens are<br />
very weak estrogens when compared to natural estrogens,<br />
their possible high concentrations in wastewater may lead<br />
them to contribute a non-negligible part of EEQ. Compared to<br />
natural estrogens or industrial chemicals such as BPA, NP and<br />
Octylphenol (OP), the research of phytoestrogens in wastewaters<br />
is just beginning. However, with the progress of<br />
analytic methods, knowledge of phytoestrogens in wastewater,<br />
such as removal mechanisms of phytoestrogens at<br />
water research 44 (2010) 567–577<br />
River sediments<br />
(ng/g)<br />
WWTPs, will grow. On some occasions, chemical advanced<br />
oxidation should also be adopted to degrade phytoestrogens<br />
when purer water is required.<br />
references<br />
Sea water<br />
(ng/L)<br />
Drinking<br />
water<br />
Air<br />
(pg/m 3 )<br />
Daidzein 2–3 [1]; 5–30 [3] – a<br />
2–4 [2];<br />
42900 [4];<br />
Bacaloni, A., Cavaliere, C., Faberi, A., Foglia, P., Samperi, R.,<br />
Lagana, A., 2005. Determination of isoflavones and coumestrol<br />
in river water and domestic wastewater sewage treatment<br />
plants. Analytica. Chimica. Acta 531, 229–237.<br />
Baig, S., Hansmann, G., Paolini, B., 2008. Ozone oxidation of<br />
estrogenic active substances in wastewater and drinking<br />
water. Wat. Sci. Technol 58, 451–458.<br />
Baronti, C., Curini, R., D’Ascenzo, G., Corcia, A.D., Gentili, A.,<br />
Samperi, R., 2000. Monitoring natural and synthetic estrogens<br />
at activated sludge sewage treatment plants and in a receiving<br />
river water. Environ. Sci. Technol 34, 5059–5066.<br />
Beck, M., Radke, M., 2006. Determination of sterols, estrogens and<br />
inorganic ions in waste water and size-segregated aerosol<br />
particles emitted from waste water treatment. Chemosphere<br />
64, 1134–1140.<br />
Beck, I.C., Bruhn, R., Gandrass, J., Ruck, W., 2005. Liquid<br />
chromatography-tandem mass spectrometry analysis of<br />
estrogenic compounds in coastal surface water of the Baltic<br />
Sea. J. Chromatogr. A 1090, 98–106.<br />
Beveridge, T.H.J., Li, T.S.C., Drover, J.C.G., 2002. Phytosterol content<br />
in American ginseng seed oil. J. Agric. Food Chem. 50, 744–750.<br />
Beveridge, T.H.J., Girard, B., Kopp, T., Drover, J.C.G., 2005. Yield<br />
and composition of grape seed oils extracted by supercritical<br />
carbon dioxide and petroleum ether: varietal effects. J. Agric.<br />
Food Chem. 53, 1799–1804.<br />
Blair, R.M., Fang, H., Branham, W.S., Hass, B.S., Dial, S.L.,<br />
Moland, C.L., Tong, W., 2000. The estrogen receptor relative<br />
binding affinities of 188 natural and xenochemicals: structural<br />
diversity of ligands. Toxicol. Sci. 54, 138–153.<br />
Bovee, T.F.H., Helsdingen, R.J.R., Rietjens, I.M.C.M., Keijer, J.,<br />
Hoogenboom, R.L.A.P., 2004. Rapid yeast estrogen bioassays<br />
stably expressing human estrogen receptor a and b, and green<br />
fluorescent protein: a comparison of different compounds with<br />
both receptor types. J. Steroid Biochem. Mol. Biol. 91, 99–109.<br />
Cain, T.G., Kolpin, D.W., Vargo, J.D., Wichman, M.D., 2008.<br />
Occurrence of antibiotics, pharmaceuticals and sterols at<br />
select surface and wastewater sites in Iowa. http://info.ngwa.<br />
org/gwol/pdf/042379985.pdf. 2008.11.24.<br />
Chou, C.C., Liu, Y.P., 2004. Determination of fecal sterols in the<br />
sediments of different wastewater outputs by GC-MS. Intern. J.<br />
Environ. Anal. Chem. 84, 379–388.<br />
D’Ascenzo, G., Corcia, A.D., Mancini, A.G.R., Mastropasqua, R.,<br />
Nazzari, M., Samperi, R., 2003. Fate of natural estrogen<br />
conjugates in municipal sewage transport and treatment<br />
facilities. Sci. Total Environ 302, 199–209.<br />
Dai, Q., Franke, A.A., Jin, F., Shu, X.O., Hebert, J.R., Custer, L.J.,<br />
Cheng, J.R., Gao, Y.T., Zheng, W., 2002. Urinary excretion of<br />
phytoestrogens and risk of breast cancer among Chinese<br />
women in Shanghai. Cancer Epidem. Biomarkers 11, 815–821.<br />
Devane, M., Saunders, D., Gilpin, B., 2006. Faecal sterols and<br />
fluorescene whiteners as indicators of the source of faecal<br />
contamination. Chem. New Zealand 70, 74–77.<br />
Eneroth, P., Hellstrom, K., Ryhage, R., 1964. Identification and<br />
quantification of neutral fecal steroids by gas-liquid<br />
chromatography and mass spectrometry: studies of human<br />
excretion during two dietary regimens. J. Lipid Res. 5, 245–262.<br />
Erbs, M., Hoerger, C.C., Hartmann, N., Bucheli, T.D., 2007.<br />
Quantification of six phytoestrogens at the nanogram per liter<br />
level in aqueous environmental samples using 13 C 3-labeled<br />
internal standards. J. Agric. Food Chem. 55, 8339–8345.<br />
Fang, H., Tong, W., Branham, W.S., Moland, C.L., Dial, S.L.D.,<br />
Hong, H., Xie, Q., Perkins, R., Owens, W., Sheehan, D.M., 2003.<br />
Study of 202 natural, synthetic, and environmental chemicals<br />
for binding to the androgen receptor. Chem. Res. Toxicol 16,<br />
1338–1358.<br />
Fattore, E., Benfenati, E., Marelli, R., Cools, E., Fanelli, R., 1996.<br />
Sterols in sediment samples from Venice lagoon. Italy.<br />
Chemosphere 33, 2383–2393.<br />
water research 44 (2010) 567–577 575<br />
Fernandez, M.P., Ikonomou, M.G., Buchanan, I., 2007. An<br />
assessment of estrogenic organic contaminants in Canadian<br />
wastewaters. Sci. Total Environ 373, 250–269.<br />
Furuichi, T., Kannan, K., Giesy, J.P., Masunaga, S., 2004.<br />
Contribution of known endocrine disrupting substances to the<br />
estrogenic activity in Tama River water samples from Japan<br />
using instrumental analysis and in vitro reporter gene assay.<br />
Wat. Res. 38, 4491–4501.<br />
Furuichi, T., Kannan, K., Suzuki, K., Tanaka, S., Giesy, J.P.,<br />
Masunaga, S., 2006. Occurrence of estrogenic compounds in<br />
and removal by a swine farm waste treatment plant. Environ.<br />
Sci. Technol 40, 7896–7902.<br />
Grosso, N.R., Nepote, V., Guzman, C.A., 2000. Chemical<br />
composition of some wild peanut species (Arachis L.) seeds. J.<br />
Agric. Food Chem. 48, 806–809.<br />
Gutendorf, B., Westendorf, J., 2001. Comparison of an array of in<br />
vitro assays for the assessment of the estrogenic potential of<br />
natural and synthetic estrogens, phytoestrogens and<br />
xenoestrogens. Toxicology 166, 79–89.<br />
Hartmann, N., Erbs, M., Wettstein, F.E., Schwarzenbach, R.P.,<br />
Bucheli, T.D., 2007. Quantification of estrogenic mycotoxins at<br />
the ng/L level in aqueous environmental samples using<br />
deuterated internal standards. J. Chromatogr. A 1138, 132–140.<br />
Heberer, T., 2002. Tracking persistent pharmaceutical residues<br />
from municipal sewage to drinking water. J. Hydrol 266,<br />
175–189.<br />
Holser, R.A., Bost, G., Boven, M.V., 2004. Phytosterol composition<br />
of hybrid Hibiscus seed oils. J. Agric. Food Chem. 52, 2546–2548.<br />
Horn-Ross, P.L., Barnes, S., Kirk, M., Coward, L., Parsonnet, J.,<br />
Hiatt, R.A., 1997. Urinary phytoestrogen levels in young<br />
women from a multiethnic population. Cancer Epidem.<br />
Biomar 6, 339–345.<br />
Hur, H.G., Lay Jr., J.O., Beger, R.D., Freeman, J.P., Rafii, F., 2000.<br />
Isolation of human intestinal bacteria metabolizing the<br />
natural isoflavone glycosides daidzin and genistin. Arch.<br />
Microbiol. 174, 422–428.<br />
Isobe, T., Shiraishi, H., Yasuda, M., Shinoda, A., Suzuki, H.,<br />
Morita, M., 2003. Determination of estrogens and their<br />
conjugates in water using solid-phase extraction followed by<br />
liquid chromatography-tandem mass spectrometry. J.<br />
Chromatogr. A 984, 195–202.<br />
Iwatsuki, K., Akihisa, T., Tokuda, H., Ukiya, M., Higashihara, H.,<br />
Mukainaka, T., Izuka, M., Hayashi, Y., Kimura, Y., Nishino, H.,<br />
2003. Sterol ferulates, sterols, and 5-alk(en)yl resorcinols from<br />
wheat, rye and corn bran oils and their inhibitory effects on<br />
Epstein-Barr virus activation. J. Agric. Food Chem. 51, 6683–<br />
6688.<br />
Jackson, S., 1966. Creatinine in urine as an index of urinary<br />
excretion rate. Health Phys. 12, 843–850.<br />
Jenner, A.M., Rafter, J., Halliwell, B., 2005. Human fecal water<br />
content of phenolics: the extent of colonic exposure to<br />
aromatic compounds. Free Radic. Biol. Med 38, 763–772.<br />
Jimenez-Escrig, A., Santos-Hidalgo, A.B., Saura-Calixto, F., 2006.<br />
Common sources and estimated intake of plant sterols in the<br />
Spanish diet. J. Agric. Food Chem 54, 3462–3471.<br />
Johnson, A.C., Williams, R.J., 2004. A model to estimate influent<br />
and effluent concentrations of estradiol. estrone, and<br />
ethinylestradiol at sewage treatment works. Environ. Sci.<br />
Technol 38, 3649–3658.<br />
Johnson, A.C., Belfroid, A., Di Corcia, A., 2000. Estimating steroid<br />
oestrogen inputs into activated sludge treatment works and<br />
observations on their removal from the effluent. Sci. Total<br />
Environ 256, 163–173.<br />
Johnson, A.C., Williams, R.J., Simpson, P., Kanda, R., 2007. What<br />
difference might sewage treatment performance make to<br />
endocrine disruption in rivers. Environ. Pollut 147, 194–202.<br />
Joss, A., Andersen, H., Ternes, T., 2004. Removal of estrogens in<br />
municipal wastewater treatment under aerobic and anaerobic
576<br />
conditions: consequences for plant optimization. Environ. Sci.<br />
Technol 38, 3047–3055.<br />
Kang, J.G., Price, W.E., Hick, L.A., 2006. Simultaneous<br />
determination of isoflavones and lignans at trace levels in<br />
natural waters and wastewater samples using liquid<br />
chromatography/electrospray ionization ion trap mass<br />
spectrometry. Rapid Commun. Mass Spectrom 20, 2411–2418.<br />
Karr, S.C., Lampe, J.W., Hutchins, A.M., Slavin, J.L., 1997. Urinary<br />
isoflavonoid excretion in humans is dose dependent at low to<br />
moderate levels of soy-protein comsumption. Am. J. Clin. Nutr<br />
66, 46–51.<br />
Kawanishi, M., Takamura-Enya, T., Ermawati, R., Shimohara, C.,<br />
Sakamoto, M., Matsukawa, K., Matsuda, T., Murahashi, T.,<br />
Matsushi, T., Wakabayashi, K., Watanabe, T., Tashiro, Y.,<br />
Yagi, T., 2004. Detection of genistein as an estrogenic<br />
contaminant of river water in Osaka. Environ. Sci. Technol 38,<br />
6424–6429.<br />
Kiparissis, Y., Hughes, R., Metcalfe, C., Ternes, T., 2001.<br />
Identification of the isoflavonoid genistein in bleached kraft<br />
mill effluent. Environ. Sci. Technol 35, 2423–2427.<br />
Kiparissis, Y., Balch, G.C., Metcalfe, T.L., Metcalfe, C.D., 2003.<br />
Effects of the isoflavones genistein and equol on the gonadal<br />
development of Japanese medaka (Oryzias latipes). Environ.<br />
Health Persp 111, 1158–1163.<br />
Komori, K., Tanaka, H., Okayasu, Y., Yasojima, M., Sato, C., 2004.<br />
Analysis and occurrence of estrogen in wastewater in Japan.<br />
Wat. Sci. Technol 50, 93–100.<br />
Kostelac, D., Rechkemmer, G., Briviba, K., 2003. Phytoestrogens<br />
modulate binding response of estrogen receptor a and b to the<br />
estrogen response element. J. Agric. Food Chem. 51, 7632–<br />
7635.<br />
Kris-Etherton, P.M., Hecker, K.D., Bonanome, A., Coval, S.M.,<br />
Binkoski, A.E., Hilpert, K.F., Griel, A.E., Etherton, T.D., 2002.<br />
Bioactive compounds in foods: their role in the prevention of<br />
cardiovascular disease and cancer. Am.. J. Med 113 (Suppl), 71–88.<br />
Kuhnle, G.G.C., Dell’Aquila, C., Kussmaul, M., Bingham, S.A., 2007.<br />
Extraction and quantification of phytoestrogens in foods using<br />
automated solid-phase extraction and LC/MS/MS. Anal. Chem.<br />
79, 9234–9239.<br />
Kuhnle, G.G.C., Dell’aquila, C., Aspinall, S.M., Runswick, S.A.,<br />
Mulligan, A.A., Bingham, S.A., 2008. Phytoestrogen content of<br />
beverages, nuts, seeds, and oils. J. Agric. Food Chem. 56, 7311–<br />
7315.<br />
Kuiper, G., Lemmen, J.G., Carlsson, B., Corton, J.C., Safe, S.H., van<br />
der Saag, P.T., van der Burg, B., Gustafsson, J.A., 1998.<br />
Interaction of estrogenic chemicals and phytoestrogens with<br />
estrogen receptor b. Endocrinology 139, 4252–4263.<br />
Lagana, A., Bacaloni, A., De Leva, I., Faberi, A., Fago, G., Marino, A.,<br />
2004. Analytical methodologies for determining the<br />
occurrence of endocrine disrupting chemicals in sewage<br />
treatment plants and natural waters. Analytica. Chimica. Acta<br />
501, 79–88.<br />
Lampe, J.W., Wartini, M.C., Kurzer, M.S., Adlercreutz, H., Slavin, J.<br />
L., 1994. Urinary lignan and isoflavonoid excretion in<br />
premenopausal women consuming flaxseed powder. J. Clin.<br />
Nutr 60, 122–128.<br />
Lampe, J.W., Gustafson, D.R., Hutchins, A.M., Martini, M.L., Li, S.,<br />
Wahala, K., Grandits, G.A., Potter, J.D., Slavin, J.L., 1999.<br />
Urinary isoflavonoid and lignan excretion on a western diet:<br />
relation to soy, vegetable, and fruit intake. Cancer Epidem.<br />
Biomarkers 8, 699–707.<br />
Lampe, J.W., Skor, H.E., Li, S., Wahala, K., Howald, W.N., Chen, C.,<br />
2001. Wheat bran and soy protein feeding do not alter urinary<br />
excretion of the isoflavan equol in premenopausal women. J.<br />
Nutr 131, 740–744.<br />
Leeming, R., Ball, A., Ashbolt, N., Nichols, P., 1996. Using faecal<br />
sterols from humans and animals to distinguish faecal<br />
pollution in receiving waters. Wat. Res. 30, 2893–2900.<br />
water research 44 (2010) 567–577<br />
Liggin, J., Bluck, L.J.C., Runswick, S., Atkinson, C., Coward, W.A.,<br />
Bingham, S.A., 2000. Daidzein and genistein content of fruits<br />
and nuts. J. Nutr. Biochem 11, 326–331.<br />
Lissin, L.W., Cook, J.P., 2000. Phytoestrogens and cardiovascular<br />
health. J. Am. Coll. Cardiol 35, 1403–1410.<br />
Liu, Z.H., Kanjo, Y., Mizutani, S., 2009. Removal mechanisms for<br />
endocrine disrupting compounds (EDcs) in wastewater<br />
treatment-physical means, biodegradation, and chemical<br />
advanced oxidation: A review. Sci. Total Environ 407, 731–748.<br />
Liu, Z.H., Kanjo, Y., Mizutani, S. Excretions of natural estrogens<br />
and androgens from humans, and their occurrence and fate in<br />
the environments. Sci. Total Environ, submitted for<br />
publication.<br />
Liu, Z.H., Mamoru, I., Kanjo, Y., Yamamoto, A. Profile and removal<br />
of endocrine disrupting chemicals (EDCs) in two municipal<br />
wastewater treatment plants by using an ER (AR) competitive<br />
ligand binding assay and chemical analyses. J. Environ. Sci. in<br />
press. doi:10.1016/S1001-0742(08)62356-6.<br />
Low, Y.L., Taylor, J.I., Grace, P.B., Dowsett, M., Folkerd, E.,<br />
Doody, D., Dunning, A.M., Scollen, S., Mulligan, A.A., Welch, A.<br />
A., Luben, R.N., Khaw, K.T., Day, N.E., Wareham, N.J.,<br />
Bingham, S.A., 2005. Polymorphisms in the CYP19 gene may<br />
affect the positive correlations between serum and urine<br />
phytoestrogen metabolites and plasma androgen<br />
concentrations in men. J. Nutr 135, 2680–2686.<br />
Markiewicz, L., Garey, J., Adlercreutz, H., Gurpide, E., 1993. In vitro<br />
bioassays of non-steroidal phytoestrogens. J. Steroid Biochem.<br />
Mol. Biol. 45, 399–405.<br />
Martins, C., Fillmann, G., Montone, R.C., 2007. Natural and<br />
anthropogenic sterols inputs in surface sediments of Patos<br />
Lagoon. Brazil. J. Braz. Chem. Soc. 18, 106–115.<br />
Matthews, J., Celius, T., Halgren, R., Zacharewski, T., 2000.<br />
Differential estrogen receptor binding of estrogenic<br />
substances: a species comparison. J. Steroid Biochem. Mol.<br />
Biol. 74, 223–234.<br />
Mazur, W., Fotsis, T., Wahala, K., Ojalas, S., Salakka, A.,<br />
Adlercreutz, H., 1996. Isotope dilution gas chromatographicmass<br />
spectrometric method for the determination of<br />
isoflavonoids, coumestrol, and lignans in food samples. Anal.<br />
Biochem 233, 169–180.<br />
Milder, I.E.J., Feskens, E.J.M., Arts, I.C.W., de Mesquita, H.B.B.,<br />
Hollman, P.C.H., Kromhout, D., 2005. Intake of the plant<br />
lignans secoisolariciresinol, matairesinol, lericiresinol, and<br />
pinoresinol in Dutch men and women. J. Nutr 135, 1202–1207.<br />
Mueller, S., Simon, S., Chae, K., Metzler, M., Korach, K., 2004.<br />
Phytoestrogens and their human metabolites show distinct<br />
agonistic and antagonistic properties on estrogen<br />
receptora(ERa) and ER(b) in human cells. Toxicol. Sci. 80,<br />
14–25.<br />
Nagata, Y., Miyanaga, N., Okumura, K., Goto, K., Naito, S.,<br />
Fujimoto, K., Hirao, Y., Takahashi, A., Tsukamoto, T.,<br />
Akaza, H., 2007. Dietary isoflavones may protect against<br />
prostate cancer in Japanese men. J. Nutr 137, 1974–1979.<br />
Nakada, N., Nyunoya, H., Nakamura, M., Hara, A., Iguchi, T.,<br />
Takada, H., 2004. Identification of estrogenic compounds in<br />
wastewater effluent. Environ. Toxicol. Chem. 23, 2807–2815.<br />
Nakada, N., Higashitani, T., Miyajima, K., Komori, K., Suzuki, Y.,<br />
2006. Development of method for identification of major<br />
substances inducing estrogenic activity contained in sewage<br />
and river waters. J. Environ. Chem. 16, 389–401 (in Japanese<br />
with English abstract).<br />
Nelson, J., Bishay, F., van Roodselaar, A., Ikonomou, M., Law, F.C.<br />
P., 2007. The use of in vitro bioassays to quantify endocrine<br />
disrupting chemicals in municipal wastewater treatment<br />
plants effluents. Sci. Total Environ 374, 80–90.<br />
Nielsen, S.E., Freese, R., Kleemola, P., Mutanen, M., 2002.<br />
Flavonoids in human urine as biomarkers for intake of fruits<br />
and vegetables. Cancer Epidem. Biomarkers 11, 459–466.
Nishihara, T., Nishikawa, J.I., Kanayama, T., Dakeyama, F.,<br />
Saito, K., Imagawa, M., Takatori, S., Kitagawa, Y., Hori, S.,<br />
Utsumi, S., 2000. Estrogenic activities of 517 chemicals by<br />
yeast two-hybrid assay. J. Health Sci. 46, 282–298.<br />
NITE, 2008. http://www.safe.nite.go.jp/english/index.html.<br />
2008.11.<br />
Pawlowski, S., Ternes, T., Bonerz, M., Kluczka, T., van der Burg, B.,<br />
Nau, H., Erdinger, L., Braunbeck, T., 2003. Combined in situ and<br />
in vitro assessment of the estrogenic activity of sewage and<br />
surface water samples. Toxicol. Sci. 75, 57–65.<br />
Peeters, P.H.M., Keinan-Boker, L., van der Schouw, Y.T.,<br />
Grobbee, D.E., 2003. Phytoestrogens and breast cancer risk.<br />
Review of the epidemiological evidence. Breast Cancer Res.<br />
Treat 77, 177–183.<br />
Penalvo, J.L., Adlercreutz, H., Uehara, M., Ristimaki, A.,<br />
Watanabe, S., 2008. Lignan content of selected foods from<br />
Japan. J. Agric. Food Chem. 56, 401–407.<br />
Phillips, K.M., Ruggio, D.M., Ashraf-Khorassan, M., 2005.<br />
Phytosterol composition of nuts and seeds commonly<br />
consumed in the United States. J. Agric. Food Chem. 53, 9436–<br />
9445.<br />
Quemeneur, M., Marty, Y., 1994. Fatty acids and sterols in<br />
domestic wastewaters. Wat. Res. 28, 1217–1226.<br />
Radke, M., 2005. Sterols and anionic surfactants in urban aerosol:<br />
emissions from wastewater treatment plants in relation to<br />
background concentrations. Environ. Sci. Technol 39, 4391–<br />
4397.<br />
Reddy, S., Iden, C.R., Brownawell, B., 2005. Analysis of steroid<br />
conjugates in sewage influent and effluent by liquid<br />
chromatography-tandem mass spectrometry. Anal. Chem. 77,<br />
7032–7038.<br />
Salste, L., Leskinen, P., Virta, M., Kronberg, L., 2007.<br />
Determination of estrogens and estrogenic activity in<br />
wastewater effluent by chemical analysis and bioluminescent<br />
yeast assay. Sci. Total Environ 378, 343–351.<br />
Seow, A., Shi, C.Y., Franke, A.A., Hankin, J.H., Lee, H.P., Yu, M.C.,<br />
1998. Isoflavonoid levels in spot urine are associated with<br />
frequency dietary soy intake in a population-based sample of<br />
middle-aged and older Chinese in Singapore. Cancer Epidem.<br />
Biomar 7, 135–140.<br />
Setchell, K.D.R., Cassidy, A., 1999. Dietary isoflavones: biological<br />
effects and relevance to human health. J. Nutr 129 (Suppl),<br />
758–767.<br />
Sivesind, E., Seguin, P., 2005. Effects of the environment, cultivar,<br />
maturity, and preservation method on red clover isoflavone<br />
concentration. J. Agric. Food Chem. 53, 6397–6402.<br />
Smeds, A.I., Willfor, S.M., Pietarinen, S.P., Peltonen-Sainio, P.,<br />
Reunanen, M.H.T., 2007. Occurrence of ‘‘mammalian’’ lignans<br />
in plant and water sources. Planta 226, 639–646.<br />
water research 44 (2010) 567–577 577<br />
Stackelberg, P.E., Furlong, E.T., Meyer, M.T., Zaugg, S.D.,<br />
Hendersen, A.K., Reissman, D.B., 2004. Persistence of<br />
pharmaceutical compounds and other organic wastewater<br />
contaminants in a conventional drinking-water-treatment<br />
plant. Sci. Total Environ 329, 99–113.<br />
Stackelberg, P.E., Gibs, J., Furlong, E.T., Meyer, M.T., Zaugg, S.D.,<br />
Lippincott, R.L., 2007. Efficiency of conventional drinkingwater-treatment<br />
processes in removal of pharmaceuticals and<br />
other organic compounds. Sci. Total Environ 377, 255–272.<br />
Song, T.T., Hendrich, S., Murphy, P.A., 1999. Estrogenic activity of<br />
glycitein, a soy isoflavone. J. Agric. Food Chem. 47, 1607–1610.<br />
Stopper, H., Schmitt, E., Kobras, K., 2005. Genotoxicity of<br />
phytoestrogens. Mutat. Res. 574, 139–155.<br />
Tan, B.L.L., 2006. Chemical and biological analyses of selected<br />
endocrine disruptors in wastewater treatment plants in South<br />
East Queensland, Australia. Doctorate dissertation. Griffith<br />
University, Australia.<br />
Tonkelaar, I.D., Keinan-Boker, L., Veer, P.V., Arts, C.J.M.,<br />
Adlercreutz, H., Thijssen, J.H.H., Peeters, H.M., 2001. Urinary<br />
phytoestrogens and postmenopausal breast cancer risk.<br />
Cancer Epidem. Biomarkers 10, 223–228.<br />
Tzchori, I., Degani, G., Elisha, R., Eliyahu, R., Hurvitz, A., Vaya, J.,<br />
Moav, B., 2004. The influence of phytoestrogens and<br />
oestradiol-17b on growth and sex determination in the<br />
European eel (Anguilla anguilla). Aquat. Res. 35, 1213–1219.<br />
USDA (US Department of Agriculture), 2003. USDA database for<br />
the flavonoid content of selected foods. http: www.nal.usda.<br />
gov/fnic/foodcomp.<br />
Valentin-Blasini, L., Sadowski, M.A., Walden, D., Caltabiano, L.,<br />
Needham, L.L., Barr, D.B., 2005. Urinary phytoestrogen<br />
concentrations in the U.S. population (1999–2000). J. Expo.<br />
Anal. Environ. Epidemiol 15, 509–523.<br />
Wang, L.Q., 2002. Mammalian phytoestrogens: enterodiol and<br />
enterolactone. J. Chromatogr. B 777, 289–309.<br />
Whitten, P.L., Patisaul, H.B., 2001. Cross-species and interassay<br />
comparisons of phytoestrogen action. Environ. Health Persp<br />
109 (Suppl), 5–20.<br />
Xu, X., Duncan, A.M., Wangen, K.E., Kurzer, M.S., 2000. Soy<br />
consumption alters endogenous estrogen metabolism in<br />
postmenopausal women. Cancer Epidem. Biomarkers 9, 781–786.<br />
Zhang, Y., Wang, G.J., Song, T.T., Murphy, P.A., Hendrich, S., 1999.<br />
Urinary disposition of the soybean isoflavones daidzein,<br />
genistein and glycitein differs among humans with moderate<br />
fecal isoflavone degradation activity. J. Nutr 129, 957–962.<br />
Zhang, X., Cambrai, A., Miesch, M., Roussi, S., Raul, F., Aoude-<br />
Werner, D., Marchioni, E., 2006. Separation of D5-and D7phytoesterols<br />
by adsorption chromatography liquid<br />
chromatography for quantitative analysis of phytosterols in<br />
foods. J. Agric. Food Chem. 54, 1196–1202.
Occurrence of emerging pollutants in urban wastewater<br />
and their removal through biological treatment followed<br />
by ozonation<br />
Roberto Rosal a, *, Antonio Rodríguez a , José Antonio Perdigón-Melón a , Alice Petre a ,<br />
Eloy García-Calvo a , María José Gómez b , Ana Agüera b , Amadeo R. Fernández-Alba b<br />
a<br />
Department of Chemical Engineering, University of Alcalá, 28771 Alcalá de Henares, Spain<br />
b<br />
Department of Analytical Chemistry, University of Almería, 04010 Almería, Spain<br />
article info<br />
Article history:<br />
Received 24 February 2009<br />
Received in revised form<br />
26 June 2009<br />
Accepted 3 July 2009<br />
Available online 9 July 2009<br />
Keywords:<br />
Emerging pollutants<br />
Pharmaceuticals<br />
Personal care products<br />
Wastewater treatment<br />
Ozonation<br />
1. Introduction<br />
abstract<br />
The presence of a wide variety of pharmaceutical and<br />
personal care products (PPCP) in water and wastewater has<br />
been frequently reported after the findings of Ternes (1998)<br />
and Daughton and Ternes (1999). These compounds are<br />
a source of concern because they are used and released in<br />
large quantities and their physical and chemical properties<br />
water research 44 (2010) 578–588<br />
Available at www.sciencedirect.com<br />
journal homepage: www.elsevier.com/locate/watres<br />
This work reports a systematic survey of over seventy individual pollutants in a Sewage<br />
Treatment Plant (STP) receiving urban wastewater. The compounds include mainly pharmaceuticals<br />
and personal care products, as well as some metabolites. The quantification in<br />
the ng/L range was performed by Liquid Chromatography–QTRAP–Mass Spectrometry and<br />
Gas Chromatography coupled to Mass Spectrometry. The results showed that paraxanthine,<br />
caffeine and acetaminophen were the main individual pollutants usually found<br />
in concentrations over 20 ppb. N-formyl-4-amino-antipiryne and galaxolide were also<br />
detected in the ppb level. A group of compounds including the beta-blockers atenolol,<br />
metoprolol and propanolol; the lipid regulators bezafibrate and fenofibric acid; the antibiotics<br />
erythromycin, sulfamethoxazole and trimethoprim, the antiinflammatories diclofenac,<br />
indomethacin, ketoprofen and mefenamic acid, the antiepileptic carbamazepine<br />
and the antiacid omeprazole exhibited removal efficiencies below 20% in the STP treatment.<br />
Ozonation with doses lower than 90 mM allowed the removal of many individual<br />
pollutants including some of those more refractory to biological treatment. A kinetic model<br />
allowed the determination of second order kinetic constants for the ozonation of bezafibrate,<br />
cotinine, diuron and metronidazole. The results show that the hydroxyl radical<br />
reaction was the major pathway for the oxidative transformation of these compounds.<br />
ª 2009 Elsevier Ltd. All rights reserved.<br />
contribute to their widespread distribution into the environment.<br />
The presence of small concentration of PPCP has been<br />
associated to chronic toxicity, endocrine disruption and the<br />
development of pathogen resistance. The consequences are<br />
particularly worrying in aquatic organisms as they are subjected<br />
to multigenerational exposure (Halling-Sørensen et al.,<br />
1998). The presence of micropollutants also endangers the<br />
reuse of treated wastewater, a generally proposed solution to<br />
* Corresponding author. Department of Chemical Engineering, Facultad de Quamica, University of Alcalá, 28771 Alcalá de Henares,<br />
Madrid, Spain. Tel.: þ34 91 885 5099; fax: þ34 91 885 5088.<br />
E-mail address: roberto.rosal@uah.es (R. Rosal).<br />
0043-1354/$ – see front matter ª 2009 Elsevier Ltd. All rights reserved.<br />
doi:10.1016/j.watres.2009.07.004
Notation<br />
Ci concentration of a given organic compound (M)<br />
CHO$ concentration of hydroxyl radical (M)<br />
CO3 concentration of dissolved ozone (M)<br />
CO3 equilibrium concentration of dissolved<br />
ozone (M)<br />
DO3 diffusivity of ozone in water, m 2 s 1<br />
Dow apparent octanol–water partition coefficient<br />
(dimensionless)<br />
fOH fraction of a given compound degraded by<br />
hydroxyl radicals<br />
Ha Hatta number, defined in Eq. (5) (dimensionless)<br />
achieve a sustainable water cycle management (Muñoz et al.,<br />
2009). PPCP represent a rising part of the trace organic<br />
micropollutants found in urban and domestic wastewaters<br />
that reach sewage treatment plants (STP), either metabolised<br />
or not (Castiglioni et al., 2006). Many of these substances<br />
escape to conventional activated sludge wastewater treatments<br />
allowing them to reach surface water streams and<br />
distribute in the environment (Tauxe-Wuersch et al., 2005).<br />
The need for treatment technologies that can provide safe<br />
treated effluents led to the proposals for upgrading STP and to<br />
implement new competing technologies for biological degradation<br />
of organic matter like membrane bioreactor (Radjenovic<br />
et al., 2009). In addition to these strategies, effective<br />
tertiary treatment technologies are also needed in order to<br />
ensure a safe use for reclaimed wastewater. The available<br />
technologies include oxidation processes alone or combined<br />
with nanofiltration or reverse osmosis (Ernst and Jekel, 1999).<br />
Many oxidation processes have been described for the<br />
removal of organic compounds in wastewater. Ozone-based<br />
and Advanced Oxidation Processes (AOP) using hydrogen<br />
peroxide or radiation have been repeatedly proposed for this<br />
task (Gogate and Pandit, 2004a; Ikehata et al., 2006). Proper<br />
combinations of AOP can also be considered in order to treat<br />
the more refractory pollutants. Fenton and Fenton-based<br />
systems, heterogeneous photocatalysis, and ultraviolet or<br />
ozone-based oxidation processes have been described (Gogate<br />
and Pandit, 2004b; Comninellis et al., 2008). The choice of the<br />
most suitable technology or combination lies on the quality<br />
required for the reclaimed water.<br />
Ozone has been largely used as oxidant in drinking water<br />
treatment and repeatedly proposed to remove organics in<br />
wastewater treatment (Raknes, 2005; Beltrán, 2004). The ozone<br />
molecule can react with many organic compounds, particularly<br />
those unsaturated or containing aromatic rings or<br />
heteroatoms also being able to decompose in water to form<br />
hydroxyl radicals. In a previous work (Rosal et al., 2008a) we<br />
studied the ozonation of wastewater from the secondary<br />
clarifier of urban and domestic STP by using ozone (pH w 8) and<br />
ozone–hydrogen peroxide (O3/H2O2). The presence of hydrogen<br />
peroxide improved the mineralization of dissolved carbon<br />
from 15% to over 90% after one hour, from which most part took<br />
place during the first five minutes. The disappearance of<br />
a selected group of over thirty pharmaceuticals revealed<br />
removal efficiencies were over >99% for most compounds after<br />
water research 44 (2010) 578–588 579<br />
kL liquid-phase individual mass transfer<br />
coefficient, m s –1<br />
kO3 second order ozone-based rate constant for<br />
a direct ozonation reaction (M 1 s 1 )<br />
kHO$ second order rate constant for a reaction with<br />
hydroxyl radical (M 1 s 1 )<br />
kR second order rate constant for the ozonation of<br />
a given compound (M 1 s 1 )<br />
kLa volumetric mass transfer coefficient (s 1 )<br />
Kow octanol–water partition coefficient for neutral<br />
species (dimensionless)<br />
TOD dose of ozone transferred to the liquid (M)<br />
five minutes on stream, with slightly better results in the<br />
absence of hydrogen peroxide. This result is consistent with<br />
the fact that many PPCP directly react with ozone with large<br />
second order kinetic constants (Huber et al., 2003).<br />
The objective of this research was to verify the occurrence<br />
and fate of 84 pollutants of different classes, mainly PPCP<br />
traced during wastewater treatment in a conventional urban<br />
STP. These include: pharmaceuticals (analgesics, antidepressants,<br />
antiinflammatories, antibiotics, antiepileptics, betablockers<br />
and lipid regulators among others), personal care<br />
products (sunscreen agents, synthetic musks), stimulants<br />
(caffeine, nicotine) and some metabolites (clofibric acid, cotinine,<br />
several metabolites of dipyrone). The effectiveness of<br />
the STP process for the removal of these compounds has been<br />
assessed in a monitoring program undertaken in Alcalá de<br />
Henares (Madrid) during a one-year period with samples<br />
taken before and after a biological activated sludge with<br />
nutrient removal process. The research was also trying to<br />
identify the impact of ozone exposure on individual pollutants<br />
encountered in the secondary effluent. The doses of ozone<br />
required for a given degree of removal of the most representative<br />
individual pollutants has been assessed. For certain<br />
pollutants, the determination of second order kinetic<br />
constants for the ozonation reaction in a real wastewater<br />
matrix was also performed.<br />
2. Material and methods<br />
2.1. Materials and plant description<br />
Wastewater samples were taken every month over a one-year<br />
period from the input and output of the secondary clarifier of<br />
a STP located in Alcalá de Henares (Madrid). This plant treats<br />
a mixture of domestic and industrial wastewater from some<br />
facilities located near the city with a nominal capacity of<br />
3000 m 3 /h of raw wastewater. The Council Directive 91/271/<br />
EEC concerning urban wastewater treatment established<br />
strict limits for the wastewater discharged to sensitive areas<br />
particularly by plants serving an equivalent population of<br />
10 000 or more. In the period prior to the sampling campaign,<br />
the need to adapt STP to the new conditions forced, the<br />
implementation of a biological nutrient removal process<br />
aimed to the elimination of phosphorous and nitrogen. The
580<br />
biological treatment worked under a traditional A2O multistage<br />
configuration with nitrification–denitrification and<br />
enhanced phosphorus removal by phosphorus-accumulating<br />
microorganisms. The treatment takes place in three zones:<br />
anaerobic, anoxic and oxic. The nitrate produced in the oxic<br />
zone is recycled, with mixed liquor to the anoxic zone where<br />
denitrification takes place. The return sludge from the settler<br />
is recycled to the anaerobic zone where the influent and<br />
sludge are mixed under anaerobic conditions. All samples<br />
were immediately processed or stored in a refrigerator (
2.5. Ozonation<br />
The ozonation runs were carried out in a 5-L glass jacketed<br />
reactor operating in semi-batch mode. A temperature of 25 C,<br />
chosen to be close to average ambient temperature, was kept<br />
using a Huber Polystat cc2 and monitored throughout the runs<br />
bymeansofaPt100ResistanceTemperatureDetector(RTD).<br />
Ozone was produced by a corona discharge ozonator (Ozomatic,<br />
119 SWO100) fed by an AirSep AS-12 PSA oxygen<br />
generation unit. The gas containing about 9.7 g/Nm 3 ozone<br />
was bubbled by means of a porous glass disk with a gas flow of<br />
0.36 Nm 3 /h. The reaction vessel was agitated with a Teflon<br />
four-blade impeller at 1000 rpm. The mass transfer coefficient<br />
was determined in transient runs with pure water with<br />
a value of kLa ¼ 0.010 0.005 s 1 . Additional details on the<br />
experimental set-up and procedure are given elsewhere<br />
(Rosal et al., 2008a,b). Throughout the runs, certain samples<br />
were withdrawn for analysis at prescribed intervals. Residual<br />
ozone was removed by bubbling nitrogen in order to prevent<br />
oxidation reactions to continue. For the desorption conditions<br />
used, residual ozone fell down below 10% in less than 20 s.<br />
During the ozonation reactions a moderate increase of pH<br />
took place that can be attributed to the stripping of CO2 from<br />
solution. In conditions that favour the generation of hydroxyl<br />
radicals, this effect is not observed and pH tend to decrease<br />
during ozonation due to the accumulation of carboxylic acids.<br />
The choice of a pH higher than that of the raw wastewater not<br />
only favours hydroxyl radical mediated reactions, but allowed<br />
to keep an almost constant pH value of 8.5 0.1 by using the<br />
feed-back control procedure described above.<br />
3. Results and discussion<br />
3.1. Occurrence<br />
The pollutants analyzed were mainly PPCP and their metabolites<br />
together with some agrochemicals. A list of the<br />
72 anthropogenic emerging pollutants detected in at least one<br />
wastewater sample from the influent to the biological treatment<br />
is detailed in the Appendix together with the analytical<br />
method applied, the molecular formula and the octanol–water<br />
partition coefficient, when available. The limit of quantification<br />
for the biological effluent (LOQ), for most compounds in<br />
the tens of ng/L, is also shown in the Appendix that also lists<br />
the compounds checked but not detected with their limits of<br />
detection (LOD). Several pesticides like chlorfenvinphos and<br />
the herbicide isoproturon were never detected, as expected<br />
considering the urban origin of wastewater. Atrazine,<br />
however, was found in all samples with an average concentration<br />
at the inlet of the biological treatment of 109 ng/L.<br />
Diuron was encountered in two samples and simazine in<br />
three with maximum concentrations of 196 and 32 ng/L,<br />
respectively. Some drugs like paroxetine, the antibiotic cefotaxime<br />
and the antiinflammatory fenoprofen were not<br />
detected in any of the analyzed samples. A similar negative<br />
result was reported for paroxetine by Terzić et al. (2008),<br />
whereas the occurrence of fenoprofen was reported in<br />
concentrations as high as 0.759 mg/L in Canadian wastewater<br />
facilities (Metcalfe et al., 2004).<br />
water research 44 (2010) 578–588 581<br />
The detailed data for the concentrations of the most significant<br />
individual pollutants are shown in Table 2. Itincludes<br />
maximum and minimum values of those compounds encountered<br />
over their quantification limit in at least 4 samples in the<br />
influent of the biological treatment. The compounds excluded<br />
for not complying with this criterion comprise, among other,<br />
carbamazepine-10,11-epoxide that was detected in 2 samples<br />
with a maximum concentration of 63 ng/L, sotalol, detected<br />
three times in the influent with concentrations in the 25–30 ng/L<br />
range or celestolide with a maximum influent concentration of<br />
30 ng/L. In addition, fenofibrate was encountered in one sample<br />
with 101 ng/L, whereas its human metabolite, fenofibric acid<br />
has been detected in all samples at concentrations as high as<br />
117 ng/L. Table 2 also excludes some compounds not systematically<br />
analyzed during the sampling campaign due to<br />
improvements in the analytical procedure. Certain compounds<br />
such as terbutaline or simazine have been encountered in<br />
several samples, but these data have not been considered<br />
statistically significant because they were found at concentrations<br />
so close to the LOQ that removal efficiency in STP could not<br />
be properly assessed and, therefore, they were not included in<br />
Table 2. Among them, diazepam was detected in half of the<br />
samples with a maximum concentration of 8 ng/L in the<br />
influent and 5 ng/L in the effluent, the average in both cases<br />
being near 3 ng/L (that equals LOQ). Also, mepivicaine was<br />
found in 7 samples with an average concentration of 8 ng/L in<br />
the influent and 7 ng/L in the effluent and maximum concentration<br />
in the influent of 14 ng/L.<br />
Paraxanthine, caffeine and acetaminophen were the individual<br />
pollutants usually found in higher concentration, with<br />
averages near 20 ppb in the influent. N-formyl-4-amino-antipiryne<br />
(4-FAA) and galaxolide also exhibited averages in the<br />
mg/L level. The fact that caffeine is the dominant micropollutant<br />
in STP is not new. Caffeine has been detected in<br />
many surface streams and STP effluents in concentration as<br />
high as 230 mg/L (Ternes et al., 2001; Heberer et al., 2002).<br />
Recently, Wilcox et al. (2009) also report caffeine, paraxanthine<br />
and acetaminophen as the most frequently detected<br />
compounds, in the influent of conventional septic systems.<br />
Muñoz et al. (2009) found several substances at mg/L level, with<br />
the highest concentrations corresponding to caffeine, acetaminophen,<br />
atenolol, and paraxanthine, that exceeded 40 mg/L<br />
each. In this work, atenolol was part of the group of<br />
compounds found with averages over the ppb limit that<br />
included ciprofloxacin, hydrochlorthiazide, ibuprofen,<br />
N-acetyl-4-amino-antipiryne (4-AAA), naproxen, nicotine,<br />
and ofloxacin. The occurrence of dipyrone (metamizol) residues<br />
in STP effluents has been less frequently assessed, but<br />
Feldmann et al. (2008) have recently reported concentrations<br />
up to 7 mgL 1 in influents and effluents of STP in Berlin. These<br />
dipyrone residues have been attributed to effluents originated<br />
in hospitals more than to private households. Nicotine was<br />
found in concentrations as high as 12 mg/L in the influent<br />
and 148 ng/L in the effluent. Cotinine, its major urinary<br />
metabolite, was detected in the biological effluent with<br />
a concentration of 100 ng/L. These values agree with those<br />
published by Buerge et al. (2008) who found concentrations of<br />
cotinine of approximately 1–10 mg/L in untreated wastewater,<br />
and 0.01–0.6 mg/L in the treated effluent, corresponding to<br />
elimination efficiencies of over 90%.
582<br />
Table 2 – Concentrations of pollutants in the influent and effluent of the studied STP calculated for compounds detected<br />
over LOQ in at least four influent samples along the monitoring programme. Averages and removal efficiencies have been<br />
calculated excluding concentrations below LOQ.<br />
Compound pKa a<br />
Influent (ng/L) Effluent (ng/L) Removal<br />
Maximum Minimum Average Maximum Minimum Average<br />
efficiency (%)<br />
4-aminoantipyrine (4-AA) 4.3 3325 262 1517 2253 127 676 55.4<br />
4-methylaminoantipyrine (4-MAA) 4.3 1894 314 880 1098 34 291 66.9<br />
Acetaminophen 9.4 37 458 1571 23 202
elow LOQ. It is well known that during wastewater treatment<br />
in STP, PPCP as well as their metabolites partition into the<br />
solid phase or remain dissolved depending on their hydrophobicity.<br />
The hydrophobicity of a neutral compound can be<br />
expressed as its octanol–water partition coefficient, Kow, but in<br />
the case of compounds that can exist in ionized form, the<br />
acid–base equilibrium must also be taken into account. The<br />
pH-dependent or apparent octanol–water distribution coefficient,<br />
D ow, that considers both the dissociation constant of<br />
acidic of basic solutes, pK a, and the current pH of wastewater<br />
can be derived from the Herderson–Hasselbalch equations<br />
(Scheytt et al., 2005). For acidic compounds, like indomethacin<br />
or naproxen, that are dissociated at the pH of wastewater, the<br />
equation yields:<br />
Kow<br />
Dow ¼<br />
1 þ 10pH pKa<br />
In the case of basic drugs, the apparent partition coefficient<br />
can be expressed using the pKa for the corresponding conjugate<br />
acid:<br />
Kow<br />
Dow ¼<br />
1 þ 10pKa pH (2)<br />
Codeine or fluoxethine are among compounds that are<br />
substantially dissociated at ambient pH, whose average was<br />
7.61 0.39, with boundaries representing the 95% confidence<br />
interval. For neutral substances, D ow ¼ K ow.<br />
There is experimental evidence that the removal of organic<br />
pollutants in STP is largely controlled by sorption process with<br />
solid–water distribution coefficients being a function of the<br />
octanol–water distribution coefficient Dow (Carballa et al.,<br />
Removal efficiency (%)<br />
100<br />
90<br />
80<br />
70<br />
60<br />
50<br />
40<br />
30<br />
water research 44 (2010) 578–588 583<br />
6<br />
(1)<br />
1 2<br />
2008). The removal efficiency obtained in this work for the<br />
compounds indicated in Table 2 has been related to Dow and<br />
the results represented in Fig. 1. For a significant group of<br />
compounds, ranging from ketorolac (44%) to galaxolide (88%),<br />
a clear relationship is observed between removal efficiency<br />
and Dow. A group of five compounds were almost completely<br />
removed in the STP even they present relatively low Dow<br />
values. They are the metabolite of caffeine paraxanthine,<br />
caffeine itself, acetaminophen (paracetamol), ibuprofen and<br />
nicotine, compounds otherwise usually found in high<br />
concentrations in raw urban wastewater (Table 2). A similar<br />
result was reported by Muñoz et al. (2008) who reported<br />
concentrations of caffeine, acetaminophen, atenolol, and<br />
paraxanthine that exceeded 40 mg/L each but a substantial<br />
removal in an activated sludge STP (90%, >99%, 43%, and 67%,<br />
respectively). Joss et al. (2005) reported removal efficiencies for<br />
ibuprofen beyond its quantification limit (>90%) and also<br />
coincident are data corresponding to naproxen (50–80%).<br />
This work also identified 14 compounds with removal<br />
efficiencies fell below 20%, but exhibiting intermediate D ow<br />
values ( 2 < D ow < 2.5). They have been marked by the lower<br />
square in Fig. 1 and are the beta-blockers atenolol, metoprolol<br />
and propanolol; the lipid regulator bezafibrate and the<br />
metabolite of fenofibrate fenofibric acid; the antibiotics<br />
erythromycin, sulfamethoxazole and trimethoprim; the antiinflammatories<br />
diclofenac, indomethacin, ketoprofen and<br />
mefenamic acid; the antiepileptic carbamazepine and the<br />
antiacid omeprazole. Joss et al. (2005) reported no significant<br />
removal of sulfamethoxazole and carbamazepine and partial<br />
elimination of diclofenac, results generally coincident with<br />
this work. For musk fragrances, galaxolide and tonalide, Joss<br />
3 4<br />
5<br />
22<br />
17<br />
11<br />
18<br />
19<br />
8<br />
7<br />
10<br />
12<br />
9 14<br />
21<br />
16 15<br />
20<br />
28<br />
10<br />
0<br />
-8.00 -6.00 -4.00<br />
27 32<br />
35<br />
31<br />
39<br />
30 34<br />
29<br />
40<br />
36 37<br />
-2.00 0.00 2.00 4.00 6.00 8.00<br />
log D ow<br />
Fig. 1 – Removal efficiency during conventional activated sludge treatment: (1) paraxanthine, (2) caffeine,<br />
(3) acetaminophen,(4) nicotine, (5) ibuprofen,(6) ketorolac,(7) clofibric acid,(8) furosemide, (9) ciprofloxacin,(10) fluoxethine,<br />
(11) ofloxacin,(12) naproxen,(13) hydrochlorothiazide, (14) 4-aminoantipyrine, (15) metronidazole, (16) N-acetyl-4-aminoantipiryne,<br />
(17) codeine, (18) N-formyl-4-amino-antipiryne, (19) 4-methylaminoantipyrine, (20) ranitidine, (21) antipyrine,<br />
(22) gemfibrozil, (23) benzophenone-3,(24) triclosan,(25) tonalide, (26) galaxolide, (27) atenolol, (28) sulfamethoxazole,<br />
(29) fenofibric acid, (30) metoprolol, (31) bezafibrate, (32) ketoprofen, (33) trimethoprim, (34) diclofenac, (35) indomethacine,<br />
(36) propanolol, (37) mefenamic acid, (38) omeprazole, (39) carbamazepine, (40) erythromycin.<br />
23<br />
24<br />
26<br />
25
584<br />
et al. (2005) reported, however, removal efficiencies somewhat<br />
lower (>50%) than those found in this work (w85%). Carballa<br />
et al. (2004) also reported overall removal efficiencies of the<br />
ranging between 70% and 90% for fragrances in coincidence<br />
with this work and generally higher for antiinflammatories<br />
(40–60%) and sulfamethoxazole. In a further work, Carballa<br />
et al. (2007) indicated different removal efficiencies for several<br />
PPCP during the anaerobic digestion of sewage sludge with<br />
high efficiencies for naproxen or sulfamethoxazole. In<br />
general, a high variability can be expected as a consequence of<br />
different pollutant concentrations, changes in the processes<br />
of the STP as well as operational conditions (Liu et al., 2008).<br />
3.3. Removal by ozonation<br />
The ozonation of a dissolved organic compound may take<br />
place by direct reaction with ozone molecule and also through<br />
the action of secondary oxidants produced from ozone in<br />
aqueous medium. Among them, the strongest oxidant is<br />
hydroxyl radical, generally associated with the oxidation of<br />
the more refractory substances. A mass balance to a given<br />
compound in solution yields the following expression:<br />
dCi<br />
dt ¼ kO 3 CiCO 3 þ kHO$CiCHO$<br />
Elovitz and von Gunten (1999) introduced a parameter R ct<br />
defined as the relationship between the concentrations of<br />
ozone and hydroxyl radical at any moment. Based on the<br />
observation that R ct can be considered constant at least<br />
though certain periods of the ozonation runs, the concentration<br />
of the two main oxidants involved in ozonation reactions<br />
can be related so that Eq. (3) can be solved without experimental<br />
determination of CHO$:<br />
ln Ci<br />
Z t<br />
Z t<br />
¼ kO CO dt þ kHO$<br />
3<br />
3<br />
Cio<br />
0<br />
0<br />
Z t<br />
¼ kR<br />
0<br />
CO dt 3<br />
CHO$dt ¼ kO 3 þ RctkHO$<br />
Z t<br />
CO dt 3<br />
0<br />
The kinetic parameter kR would behave as second order<br />
kinetic constant provided Rct is constant throughout the<br />
sampling period. As shown below, this work shows that Rct is<br />
constant during ozonation at least for the samples taken after<br />
ozone appeared in solution. The preceding findings also lie on<br />
the assumption of slow kinetic regime. The kinetics of<br />
a heterogeneous gas–liquid semicontinuous process is determined<br />
by the relative rates of absorption and chemical reaction<br />
and is characterized by Hatta number. It represents the<br />
maximum rate of chemical reaction relative to the maximum<br />
rate of mass transfer, yielding, for a second order reaction, the<br />
following expression:<br />
Ha ¼<br />
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi<br />
kO Ci;oDO 3 3<br />
kL<br />
In the case of wastewater treatment, ozone reacts with<br />
many compounds in a complex parallel reaction system so<br />
that kO Ci;o can be substituted by 3 P<br />
kO Ci;o. As raw wastewater<br />
3<br />
i<br />
contains a number of compounds whose second order direct<br />
reaction constants with ozone are very large (Huber et al.,<br />
water research 44 (2010) 578–588<br />
(3)<br />
(4)<br />
(5)<br />
2003) mass transfer is likely to limit the ozonation rate during<br />
the first reaction minutes. Once ozone appears in solution, the<br />
following inequality holds:<br />
kLa C O3<br />
CO 3<br />
> X<br />
i<br />
kO 3 Ci;oCO 3<br />
Therefore an upper limit can be obtained for Ha as follows:<br />
rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi<br />
Ha <<br />
kLa C O 3<br />
C O3<br />
kL<br />
1 DO 3<br />
where DO 3 is the diffusivity of ozone in water<br />
(1.77 10 9 m 2 s 1 ). The equilibrium concentration of ozone<br />
C O3 was calculated from Henry’s law using the correlation of<br />
Rischbieter et al. (2000) from the concentration in the gas phase<br />
that was 9.4 g/Nm 3 . The value of the mass transfer coefficient,<br />
kL ¼ 5.5 10 5 ms 1 , was calculated by using the correlation of<br />
Calderbank and Moo-Young (1961). At the beginning of the run,<br />
there was no ozone in solution and Ha is supposed to reach<br />
high values. After about three minutes, the concentration of<br />
ozone that increased during runs approaching equilibrium was<br />
over 2 10 3 mM. This, considering that C O3 ¼ 0:05 mM,<br />
ensures Ha < 0.3 and, therefore, the kinetic regime was slow and<br />
for the sample taken at 4 min and those taken thereafter.<br />
Second order kinetic constants could be obtained as indicated<br />
before for bezafibrate, cotinine, diuron, ketoprofen and<br />
metronidazole. The logarithmic concentration decay is represented<br />
against the integral ozone dose in Fig. 2 in which, the<br />
experimental points represent data from samples taken at 4, 6,<br />
10 and 15 min, the first being considered the initial concentration,<br />
Cio. The comparison with literature data shows that the<br />
transformation of ozone-resistant micropollutants takes place<br />
primarily via indirect radical oxidation inhibited by the wastewater<br />
matrix. The values reported in the literature for<br />
the second order direct and indirect rate constants are:<br />
bezafibrate, kO3 ¼ 590 M 1 s 1 and kHO$ ¼ 7:40 10 9 M 1 s 1<br />
in Huber et al. (2005); ketoprofen, kO3 ¼ 0:4 M 1 s 1 and<br />
kHO$ ¼ 8:40 10 9 M 1 s 1 in Real et al. (2009), diuron,<br />
ln(C io/C i)<br />
4<br />
3<br />
2<br />
1<br />
0<br />
0 0.2 0.4 0.6 0.8 1<br />
Fig. 2 – Logarithmic decay of the concentration of diuron<br />
(B), metronidazole (C), ketoprofen (D) bezafibrate (,) and<br />
cotinine (-) as a function of the integral ozone exposure.<br />
(Two y-axes have been used for clarity.)<br />
6<br />
5<br />
4<br />
3<br />
2<br />
1<br />
0<br />
(6)<br />
(7)<br />
ln(C io/C i)
kO3 ¼ 16:5 M 1 s 1 and kHO$ ¼ 4:6 109 M 1 s 1 in Benitez et al.<br />
(2007) and kO3 ¼ 14:7 M 1 s 1 and kHO$ ¼ 6:6 109 M 1 s 1 in<br />
De Laat et al. (1996) ; metronidazole, kO3 ¼ < 350 M 1 s 1<br />
in Sánchez-Polo et al. (2008) and kHO$ ¼ 1:98 109 M 1 s 1 in<br />
Johnson and Mehrvar (2008). No data have been published<br />
for cotinine. The ozonation of bezafibrate was also studied<br />
by Dantas et al. (2007) who reported a direct kinetic constant<br />
kO3 ¼ 4:24 103 0:66 103 M 1 s 1 for pH 7 and kR ¼ 1.0<br />
10 4<br />
1.07 10 3 M 1 s 1 for pH 8. The reported direct rate<br />
constant for ketoprofen is particularly low and allows to calculate<br />
Rct by dividing the value of kR obtained in this work by kHO$<br />
reportedby Real et al. (2009). The value found, Rct ¼ 3.62 10 7 ,is<br />
very close to those that could be derived using Eq. (4) for bezafibrate<br />
and diuron. The partial contribution of the direct ozone<br />
and radical pathways can be calculated for a given organic<br />
compound as follows:<br />
kHO$CHO$Ci<br />
fOH ¼<br />
kHO$CHO$Ci þ kO Ci 3<br />
¼<br />
kHO$Rct<br />
kHO$Rct þ kO 3<br />
Using literature constants, the fraction degraded by hydroxyl<br />
radicals were 0.819 for bezafibrate, 0.996 for diuron, 1.000 for<br />
ketoprofen and >0.887 for metronidazole. The contribution of<br />
direct ozone reaction is obviously low because the five<br />
compounds studied are relatively refractory to ozonation. By<br />
means of Eq. (8) and taking into account that kR ¼ kO þ kHO$$Rct,<br />
3<br />
the second order rate constants, kR, can be derived from literature<br />
data with the following results: metronidazole<br />
1.07 10 3 M 1 s 1 , bezafibrate 3.27 10 3 M 1 s 1 and diuron<br />
2.41 10 3 M 1 s 1 (Benitez et al., 2007). The two last are in good<br />
agreement with experimental values. In the case of metronidazole,<br />
the low value obtained for Rct suggests that kHO could<br />
have been underestimated.<br />
The efficiency of ozonation for the removal of the main<br />
micropollutants whose concentration in biologically treated<br />
wastewater was >10 ng/L is indicated in Table 3. The table<br />
shows the evolution of the concentration for samples taken<br />
during ozonation up to a reaction time of 15 min and the<br />
amount of ozone required for a given degree of removal. Only<br />
concentrations above LOQ are shown. These LOQ apply for<br />
ozonated samples and are lower than those reported in the<br />
Appendix for untreated samples. They could be reached,<br />
because of the higher preconcentration factor applied to the<br />
ozonated samples (see experimental section) and the reduction<br />
in the matrix effects observed in these cleaner extracts.<br />
The amount of ozone transferred to the liquid at a certain<br />
reaction time, TOD, was determined, from the integration of<br />
the ozone absorption rate equation:<br />
0<br />
1<br />
Z t<br />
TODðtÞ ¼ kLa@C<br />
t CO dtA<br />
(9)<br />
O3 3<br />
0<br />
For the experimental conditions used in this work CO 3 was<br />
always less then 10% of C O3 and, therefore, TOD(t) was<br />
essentially linear with time. The last right column of Table 3<br />
shows the dose of ozone required for the complete removal<br />
(no detection) of a given compound or to achieve certain<br />
removal efficiency in cases where ozonation was unable to get<br />
complete oxidation in less than 15 min corresponding to<br />
a dose of ozone of 0.34 mmol/L of wastewater. It is to be noted<br />
that a change in the pH of wastewater modifies the ozone<br />
water research 44 (2010) 578–588 585<br />
(8)<br />
doses required for a given effect. A pH increase causes<br />
a higher hydroxyl radical exposure increasing Rct and leading,<br />
for a similar integral radical exposure, to lower ozone doses.<br />
Also, it is important to point out that a change of pH may<br />
affect the direct ozonation rates of dissociating compounds. In<br />
our case, raising pH from that of raw wastewater to 8.5, at<br />
which the ozonation took place, could affect the ozonation<br />
rate of some proton-accepting compounds whose pKa w 8,<br />
particularly benzophenone-3, hydrochlorthiazide, nicotine,<br />
ofloxacin and triclosan.<br />
A group of 15 compounds rapidly disappear during the first<br />
120 s on stream, with ozone doses
586<br />
Table 3 – Removal of pollutants contained in wastewater during ozonation. The ozone doses are those required to reach<br />
concentrations below the limit of quantification (LOQ a ) in treated samples.<br />
Ozonation time (min) LOQ a<br />
0 2 4 6 10 15 Ozone doses<br />
for remotion<br />
3-(4-methylbenzylidene) camphor 39 55 50 65 39 72 54 Not removed<br />
4-aminoantipyrine 19 58 – – – – –
4. Conclusions<br />
This research showed the regular presence of over seventy<br />
anthropogenic individual pollutants, some of which are<br />
encountered in relatively high amounts. In raw sewage,<br />
25 compounds were detected in the mg/L range, 15 of which<br />
exceeded this level in yearly averages. Acetaminophen (paracetamol)<br />
and caffeine were persistently detected over 1 ppb<br />
in untreated wastewater. Galaxolide was also encountered in<br />
high concentration, but its occurrence showed a significant<br />
variability. Two metabolites, paraxanthine from caffeine and<br />
4-FAA from metamizol were also found in high amounts in<br />
almost all samples. Other metabolites detected were 4-AA and<br />
4-MAA from dipyrone, fenofibric acid from fenofibrate and<br />
4-AAA also from metamizol. This findings stress the need for<br />
exploring not only the pattern PPCP, but also their metabolic<br />
or photodegradation intermediates.<br />
The efficiency of removal of PPCP in STP was roughly<br />
dependent on its hydrophobicity expressed as apparent<br />
octanol–water distribution coefficient, D ow, a parameter that<br />
takes into account the octanol–water partition coefficient,<br />
Kow, as well as the dissociation constant of acidic or basic<br />
compounds. For most compounds, the removal efficiency<br />
during biological treatment increased with hydrophobicity as<br />
expected considering the higher sorption of non-polar<br />
compounds on sludge. Certain compounds showed important<br />
deviations with a group of relatively polar substances formed<br />
by paraxanthine, caffeine, acetaminophen, ibuprofen and<br />
nicotine that were almost completely removed during biological<br />
treatment. Another group formed by 14 compounds<br />
exhibited removal efficiencies below 20%, even their octanol–<br />
water distribution coefficient was not particularly low. They<br />
are all important pharmaceuticals prescribed and delivered to<br />
sewage in high amounts and include the beta-blockers atenolol,<br />
metoprolol and propanolol; the lipid regulator bezafibrate,<br />
fenofibric acid; the antibiotics erythromycin,<br />
sulfamethoxazole and trimethoprim; the antiinflammatories<br />
diclofenac, indomethacin, ketoprofen and mefenamic acid;<br />
the antiepileptic carbamazepine and the antiacid omeprazole.<br />
An ozonation treatment yielded high removal efficiencies<br />
of most individual pollutants detected in treated wastewater.<br />
The kinetic analysis of the part of the run that takes place in<br />
the slow kinetic regime, allowed the determination of second<br />
order kinetic constants for the ozonation of bezafibrate, cotinine,<br />
diuron, ketoprofen and metronidazole in its wastewater<br />
matrix. This work shows that Rct, a concept developed for<br />
drinking water, can be applied for the ozonation of wastewater<br />
as it was constant at least for the samples taken after<br />
ozone appeared in solution. The ratio of the concentrations of<br />
ozone and hydroxyl radical, Rct, during this period was<br />
3.62 10 7 and the fraction degraded by hydroxyl radicals was<br />
in the 0.819–1.000 range for the aforementioned compounds.<br />
Even when most compounds disappear for doses lower than<br />
340 mM, and most for less than 90 mM, some pollutants,<br />
essentially personal care products were not significantly<br />
removed during ozonation. These were the UV filters<br />
3-(4-methylbenzylidene) camphor and Ethylkexyl methoxycinnamate,<br />
the sunscreen agent Benzophenone-3 and the<br />
aromatic nitro musk xylene. On the other hand, certain polar<br />
water research 44 (2010) 578–588 587<br />
compounds like diclofenac, indomethacin or the betablockers<br />
atenolol, metoprolol and propanolol, which are<br />
poorly removed in the activated sludge conventional treatment,<br />
exhibit large ozonation rates and can be removed from<br />
treated wastewater using moderate ozone doses.<br />
Acknowledgements<br />
The authors wish to express their gratitude to the Ministry of<br />
Education of Spain (Contract CONSOLIDER-INGENIO 2010<br />
CSD2006-00044), the Dirección General de Universidades e<br />
Investigación de la Comunidad de Madrid, Research network<br />
0505/AMB-0395.<br />
Appendix.<br />
Supplementary information<br />
Supplementary data associated with this article can be found<br />
in the doi: 10.1016/j.watres.2009.07.004.<br />
references<br />
Beltrán, F.J., 2004. Ozone Reaction Kinetics for Water and<br />
Wastewater Systems. CRC, Boca Raton, pp. 113–149.<br />
Benitez, F.J., Real, F.J., Acero, J.L., Garcia, C., 2007. Kinetics of the<br />
transformation of phenyl-urea herbicides during ozonation of<br />
natural waters: rate constants. Water Res. 41, 4073–4084.<br />
Bueno, M.J.M., Agüera, A., Gómez, M.J., Hernando, M.D.,<br />
García, J.F., Fernández-Alba, A.R., 2007. Application of liquid<br />
chromatography/quadrupole-linear ion trap mass<br />
spectrometry and time-of-flight mass spectrometry to the<br />
determination of pharmaceuticals and related contaminants<br />
in wastewater. Anal. Chem. 79, 9372–9384.<br />
Buerge, I.J., Kahle, M., Buser, H.R., Müller, M.D., Poiger, T., 2008.<br />
Nicotine derivatives in wastewater and surface waters:<br />
application as chemical markers for domestic wastewater.<br />
Environ. Sci. Technol. 42, 6354–6360.<br />
Calderbank, P.H., Moo-Young, M.B., 1961. The continuous phase<br />
heat and mass transfer properties of dispersions. Chem. Eng.<br />
Sci. 16, 39–54.<br />
Carballa, M., Fink, G., Omil, F., Lema, J.M., Ternes, T., 2008.<br />
Determination of the solid–water distribution coefficient (K d)<br />
for pharmaceuticals, estrogens and musk fragrances in<br />
digested sludge. Water Res. 42, 287–295.<br />
Carballa, M., Omil, F., Lema, J.M., Llompart, M., García-Jares, C.,<br />
Rodríguez, I., Gómez, M., Ternes, T., 2004. Behavior of<br />
pharmaceuticals, cosmetics and hormones in a sewage<br />
treatment plant. Water Res. 38, 2918–2926.<br />
Carballa, M., Omil, F., Ternes, T., Lema, J.M., 2007. Fate of<br />
pharmaceutical and personal care products (PPCPs) during<br />
anaerobic digestion of sewage sludge. Water Res. 41,<br />
2139–2150.<br />
Castiglioni, S., Bagnati, R., Fanelli, R., Pomati, F., Calamari, D.,<br />
Zuccato, E., 2006. Removal of pharmaceuticals in sewage<br />
treatment plants in Italy. Environ. Sci. Technol. 40, 357–363.<br />
Comninellis, C., Kapalka, A., Malato, S., Parsons, S.A., Poulios, I.,<br />
Mantzavinos, D., 2008. Advanced oxidation processes for<br />
water treatment: advances and trends for R&D. J. Chem.<br />
Technol. Biotechnol. 83, 769–776.
588<br />
Dantas, R.F., Centerino, M., Marotta, R., Sans, C., Esplugas, S.,<br />
Andreozzi, A., 2007. Bezafibrate removal by means of<br />
ozonation: primary intermediates, kinetics, and toxicity<br />
assessment. Water Res. 41, 2525–2532.<br />
Daughton, C.G., Ternes, T.A., 1999. Pharmaceuticals and personal<br />
care products in the environment: agents of subtle change?<br />
Environ. Health Perspect. 107, 907–938.<br />
De Laat, J., Maouala-Makata, P., Dore, M., 1996. Rate constants for<br />
reactions of ozone and hydroxyl radicals with several phenylureas<br />
and acetamides. Environ. Technol. 17, 707–716.<br />
Elovitz, M.S., von Gunten, U., 1999. Hydroxyl radical/ozone ratios<br />
during ozonation processes I. The R ct concept. Ozone Sci. Eng.<br />
21, 239–260.<br />
Ernst, M., Jekel, M., 1999. Advanced treatment combination for<br />
groundwater recharge of municipal wastewater by<br />
nanofiltration and ozonation. Water Sci. Technol. 40, 277–284.<br />
Feldmann, D.F., Zuehlke, S., Heberer, T., 2008. Occurrence, fate and<br />
assessment of polar metamizole (dipyrone) residues in hospital<br />
and municipal wastewater. Chemosphere 71, 1754–1764.<br />
Gogate, P.R., Pandit, A.B., 2004a. A review of imperative technologies<br />
for wastewater treatment I: oxidation technologies at ambient<br />
conditions. Adv. Environ. Res. 8, 501–551.<br />
Gogate, P.R., Pandit, A.B., 2004b. A review of imperative<br />
technologies for wastewater treatment II: hybrid methods.<br />
Adv. Environ. Res. 8, 553–597.<br />
Gómez, M.J., Gómez-Ramos, M.M., Agüera, A., Mezcua, M.,<br />
Herrera, S., Fernández-Alba, A.R., 2009. A new gas<br />
chromatography/mass spectrometry method for the<br />
simultaneous analysis of target and non-target organic<br />
contaminants in waters. J. Chromatogr. A, 1216, 4071–4082.<br />
Halling-Sørensen, B., Nielsen, B.N., Lanzky, P.F., Ingerslev, F.,<br />
Lützhoft, H.C.H., Jorgensen, S.E., 1998. Occurrence, fate and<br />
effects of pharmaceutical substances in the environment,<br />
a review. Chemosphere 36, 357–394.<br />
Heberer, T., Reddersen, K., Mechlinski, A., 2002. From municipal<br />
sewage to drinking water: fate and removal of pharmaceutical<br />
residues in the aquatic environment in urban areas. Water Sci.<br />
Technol. 46, 81–88.<br />
Huber, M.M., Canonica, S., Park, G.Y., Gunten, U., 2003. Oxidation<br />
of pharmaceuticals during ozonation and advanced oxidation<br />
processes. Environ. Sci. Technol. 37, 1016–1024.<br />
Huber, M.M., Göbel, A., Joss, A., Hermann, N., Löffler, D.,<br />
Mcardell, C.S., Ried, A., Siegrist, H., Ternes, T.A., von<br />
Gunten, U., 2005. Oxidation of pharmaceuticals during<br />
ozonation of municipal wastewater effluents: a pilot study.<br />
Environ. Sci. Technol. 39, 4290–4299.<br />
Ikehata, K., Naghashkar, N.J., El-Din, M.G., 2006. Degradation of<br />
aqueous pharmaceuticals by ozonation and advanced<br />
ozonation processes: a review. Ozone Sci. Eng. 28, 353–414.<br />
Johnson, M.B., Mehrvar, M., 2008. Aqueous metronidazole<br />
degradation by UV/H 2O 2 process in single- and multi-lamp<br />
tubular photoreactors: kinetics and reactor design. Ind. Eng.<br />
Chem. Res. 47, 6525–6537.<br />
Joss, A., Keller, E., Alder, A.C., Göbel, A., McArdell, C.S., Ternes, T.,<br />
Siegrist, H., 2005. Removal of pharmaceuticals and fragrances<br />
in biological wastewater treatment. Water Res. 39, 3139–3152.<br />
Kupper, T., Plagellat, C., Brändli, R.C., de Alencastro, L.F.,<br />
Grandjean, D., Tarradellas, J., 2006. Fate and removal of<br />
polycyclic musks, UV filters and biocides during wastewater<br />
treatment. Water Res. 40, 2603–2612.<br />
Liu, Z., Kanjo, Y., Mizutani, S., 2008. Removal mechanisms for<br />
endocrine disrupting compounds (EDCs) in wastewater<br />
treatment d physical means, biodegradation, and chemical<br />
advanced oxidation: a review. Sci. Total Environ.. doi:10.1016/<br />
j.scitotenv.2008.08.039.<br />
water research 44 (2010) 578–588<br />
Metcalfe, C., Miao, X.S., Hua, W., Letcher, R., Servos, M., 2004.<br />
Pharmaceuticals in the Canadian environment. In:<br />
Kümmerer, K. (Ed.), Pharmaceuticals in the Environment,<br />
Sources, Fate Effects and Risks. Springer, pp. 67–90.<br />
Muñoz, I., Gómez, M.J., Molina-Díaz, A., Huijbregts, M.A.J.,<br />
Fernández-Alba, A.R., García-Calvo, E., 2008. Ranking potential<br />
impacts of priority and emerging pollutants in urban<br />
wastewater through life cycle impact assessment.<br />
Chemosphere 74, 37–44.<br />
Muñoz, I., Rodríguez, A., Rosal, R., Fernández-Alba, A.R., 2009. Life<br />
cycle assessment of urban wastewater reuse with ozonation<br />
as tertiary treatment. A focus on toxicity-related impacts. Sci.<br />
Total Environ. 407, 1245–1256.<br />
Radjenovic, J., Petrovic, M., Barceló, D., 2009. Fate and distribution<br />
of pharmaceuticals in wastewater and sewage sludge of the<br />
conventional activated sludge (CAS) and advanced membrane<br />
bioreactor (MBR) treatment. Water Res. 43, 831–841.<br />
Raknes, K.L., 2005. Ozone in Drinking Water Treatment: Process<br />
Design, Operation, and Optimization. American Water Works<br />
Association, Denver.<br />
Real, F.J., Benitez, F.J., Acero, J.L., Sagasti, J.J.P., Casas, F., 2009.<br />
Kinetics of the chemical oxidation of the pharmaceuticals<br />
primidone, ketoprofen, and diatrizoate in ultrapure and<br />
natural waters. Ind. Eng. Chem. Res. 48, 3380–3388.<br />
Rischbieter, E., Stein, H., Shumpe, A., 2000. Ozone solubilities in<br />
water and aqueous salt solutions. J. Chem. Eng. Data 45,<br />
338–340.<br />
Rosal, R., Rodríguez, A., Perdigón-Melón, J.A., Mezcua, M.,<br />
Hernando, M.D., Letón, P., García-Calvo, E., Agüera, A.,<br />
Fernández-Alba, A.R., 2008a. Removal of pharmaceuticals and<br />
kinetics of mineralization by O 3/H 2O 2 in a biotreated<br />
municipal wastewater. Water Res. 42, 3719–3728.<br />
Rosal, R., Rodríguez, A., Perdigón-Melón, J.A., Petre, A., García-<br />
Calvo, E., 2008b. Oxidation of dissolved organic matter in the<br />
effluent of a sewage treatment plant by ozone combined with<br />
hydrogen peroxide (O 3/H 2O 2). Chem. Eng. J. 149, 311–318.<br />
Sánchez-Polo, M., Rivera-Utrilla, J., Prados-Joya, G., Ferro-<br />
Garcıa, M.A., Bautista-Toledo, I., 2008. Removal of<br />
pharmaceutical compounds, nitroimidazoles, from waters by<br />
using the ozone/carbon system. Water Res. 42, 4163–4171.<br />
Scheytt, T., Mersmann, P., Lindstädt, R., Heberer, T., 2005.<br />
1-Octanol/water partition coefficients of 5 pharmaceuticals<br />
from human medical care: carbamezepine, clofibric acid,<br />
diclofenac, ibuprofen and propyphenazone. Water Air Soil<br />
Pollut. 165, 3–11.<br />
Tauxe-Wuersch, A., De Alencastro, L.F., Grandjean, D.,<br />
Tarradellas, J., 2005. Occurrence of several acidic drugs in<br />
sewage treatment plants in Switzerland and risk assessment.<br />
Water Res. 39, 1761–1772.<br />
Ternes, T.A., 1998. Occurrence of drugs in German sewage<br />
treatment plants and rivers. Water Res. 32, 3245–3260.<br />
Ternes, T., Bonerz, M., Schmidt, T., 2001. Determination of<br />
neutral pharmaceuticals in wastewater and rivers by liquid<br />
chromatography–electrospray tandem mass spectrometry.<br />
J. Chromatogr. A 938, 175–185.<br />
Terzić, S., Senta, I., Ahel, M., Gros, M., Petrović, M., Barcelo, D.,<br />
Müller, J., Knepper, T., Martí, I., Ventura, F., Jovančić, P.,<br />
Jabučar, D., 2008. Occurrence and fate of emerging wastewater<br />
contaminants in Western Balkan Region. Sci. Total Environ.<br />
399, 66–77.<br />
Wilcox, J.D., Bahr, J.M., Hedman, C.J., Hemming, J.D.C.,<br />
Barman, M.A.E., Bradbury, K.R., 2009. Removal of organic<br />
wastewater contaminants in septic systems using<br />
advanced treatment technologies. J. Environ. Qual. 38,<br />
149–156.
Degradation of the emerging contaminant ibuprofen<br />
in water by photo-Fenton<br />
Fabiola Méndez-Arriaga*, Santiago Esplugas, Jaime Giménez<br />
Departamento de Ingeniería Química, Facultad de Química, Universidad de Barcelona, C/ Martí i Franquès 1, 08028 Barcelona, Spain<br />
article info<br />
Article history:<br />
Received 18 April 2009<br />
Received in revised form<br />
3 July 2009<br />
Accepted 8 July 2009<br />
Available online 15 July 2009<br />
Keywords:<br />
AOP<br />
Photo-Fenton<br />
Non-steroidal anti-inflammatory<br />
drug ibuprofen<br />
1. Introduction<br />
abstract<br />
The recent public interest regarding the presence of pharmaceutical<br />
pollutants in water has raised important concerns due<br />
to the unknown environmental impact and possible damages to<br />
the botany and fauna present in aquatic systems (Halling-Sorensen<br />
et al., 1998; Ternes et al., 2004). One of the most consumed<br />
medications corresponds to the classification of the Non-<br />
Steroidal Anti-Inflammatory Drugs (NSAIDs) with more than 70<br />
million annual prescriptions in the world (Takagi et al., 2006).<br />
Moreover, several recent reports have confirmed the presence of<br />
water research 44 (2010) 589–595<br />
Available at www.sciencedirect.com<br />
journal homepage: www.elsevier.com/locate/watres<br />
* Corresponding author. Tel.: þ34 93 403 9789; fax: þ34 93 402 1291.<br />
E-mail address: fmendeza@ub.edu (F. Méndez-Arriaga).<br />
0043-1354/$ – see front matter ª 2009 Elsevier Ltd. All rights reserved.<br />
doi:10.1016/j.watres.2009.07.009<br />
In this study the degradation of the worldwide Non-Steroidal Anti-Inflammatory Drug<br />
(NSAID) ibuprofen (IBP) by photo-Fenton reaction by use of solar artificial irradiation was<br />
carried out. Non-photocatalytic experiments (complex formation, photolysis and UV/Vis-<br />
H2O2 oxidation) were executed to evaluate the isolated effects and additional differentiated<br />
degradation pathways of IBP. The solar photolysis cleavage of H2O2 generates hydroxylated-IBP<br />
byproducts without mineralization. Fenton reaction, however promotes hydroxylation<br />
with a 10% contamination in form of a mineralization. In contrast photo-Fenton in<br />
addition promotes the decarboxylation of IBP and its total depletion is observed. In absence<br />
of H2O2 a decrease of IBP was observed in the Fe(II)/UV–Vis process due to the complex<br />
formation between iron and the IBP-carboxylic moiety. The degradation pathway can be<br />
described as an interconnected and successive principal decarboxylation and hydroxylation<br />
steps. TOC depletion of 40% was observed in photo-Fenton degradation. The iron-IBP<br />
binding was the key-point of the decarboxylation pathway. Both decarboxylation and<br />
hydroxylation mechanisms, as individual or parallel process are responsible for IBP<br />
removal in Fenton and photo-Fenton systems. An increase in the biodegradability of the<br />
final effluent after photo-Fenton treatment was observed. Final BOD5 of 25 mg L 1 was<br />
reached in contrast to the initial BOD5 shown by the untreated IBP solution (BOD5 < 1mgL 1 ).<br />
The increase in the biodegradability of the photo-Fenton degradation byproducts<br />
opens the possibility for a complete remediation with a final post-biological treatment.<br />
ª 2009 Elsevier Ltd. All rights reserved.<br />
the NSAID ibuprofen (IBP), in effluents of wastewater treatment<br />
plants (WTPs) (Alonso et al., 2006; Gómez et al., 2008). Concentrations<br />
of IBP in the environment are reported between<br />
10 ng L 1 and 169 mgL 1 (Santos et al., 2007). A significant<br />
removal rate of this drug is reached by biological oxidation, in<br />
some cases more than 70% (Carballa et al., 2006). In spite of, its<br />
apparently high biodegradation rate, the ecological risk remains<br />
still high, due to the main byproducts generated during the<br />
biological oxidation. These byproducts, hydroxyl-IBP and carboxy-IBP,<br />
have shown quite similar toxicological consequences<br />
in aquatic environment (Roberts and Thomas, 2006).
590<br />
On the other hand, recent research has shown promising<br />
results in the removal of pharmaceutical pollutants through<br />
the application of Advanced Oxidation Processes (AOPs). The<br />
AOPs are oxidative processes applied for treatment of<br />
contaminants in water, soils and air, based on the presence<br />
and reactivity of the hydroxyl radicals ( OH) which are<br />
generated in atmospheric or subcritical conditions of<br />
temperature and pressure with or without catalyst and/or<br />
reactive energy (electrochemical, UV–Vis or ultrasounds)<br />
(Méndez-Arriaga et al., 2009). The final common byproducts<br />
after AOPs have been reported to be easily degraded by biological<br />
oxidation (Torres et al., 2003). Although AOPs use<br />
different reacting systems (homogeneous or heterogeneous<br />
phases, irradiated or dark conditions, etc.), they are mainly<br />
characterized by the production of hydroxyl radicals with<br />
consecutive unselective attack on the organic material.<br />
Several AOPs have been employed to remove NSAIDs (as<br />
diclofenac, naproxen, IBP, ketoprofen, etc.) but special focus<br />
was placed on the heterogeneous photocatalyst with TiO2,<br />
photo-Fenton solar process, ozonation process, electrochemistry<br />
(Sirés Sadornil Ignacio, 2006) and more recently on<br />
ultrasound energy (Hartmann et al., 2008; Méndez-Arriaga<br />
et al., 2008b). Photo-Fenton reaction is one of the most widely<br />
used AOPs due to its environment friendly application of solar<br />
energy. The extremely complex chemistry of the oxidation<br />
system based on the reaction between iron and hydrogen<br />
peroxide, the called Fenton reaction, has become of great<br />
interest for environmental applications.<br />
Well documented information, in a critical compilation<br />
about the basis of Fenton chemistry as well as its applications<br />
to wastewater treatment, can be founded in Venkatadri and<br />
Peters (1993), Bigda (1995), Chen and Pignatello (1999), Tarr<br />
(2003), Burkitt (2003), Neyens and Baeyens (2003), Safarzadeh-<br />
Amiri et al. (1996) and more recently in a widely recommended<br />
critical review, written by Pignatello et al. (2006). Photo-Fenton,<br />
an attractive alternative for degradation of emerging<br />
contaminants, is based on the redox reaction between Fe(II)<br />
and H2O2 with the concomitant generation of OH and<br />
consecutive catalytic role of iron (Fe(II)/Fe(III)) and it is<br />
improved under UV–Vis irradiation.<br />
Even the Fenton and photo-Fenton reactions have been<br />
applied for wide types of organic contaminants, their use for<br />
the degradation of pharmaceutical compounds has been<br />
applied to a smaller extent than other AOPs like photocatalysis<br />
or O 3.<br />
The Fenton and photo-Fenton processes have been<br />
employed in the treatment of wastewater containing pharmaceutical<br />
pollutants such as azathriopine, asparginase, DCF,<br />
penicillin, a-methyl-phenylglycine, metronidazole, lincomicyn<br />
metabolites, sulfamethazine, sulfamethoxazole, amoxicillin,<br />
bezafibrate, paracetamol, biguanides, guanidine,<br />
triamines, pharmaceutical industry and hospital wastewaters<br />
among others (Klaviaroti et al., 2008; González et al., 2007;<br />
Pérez-Moya et al., 2007; Pérez-Estrada et al., 2005).<br />
Degradation of IBP by photo-Fenton reaction has not been<br />
reported yet. Therefore, the goals of this work were to evaluate<br />
the photo-Fenton ability to degrade IBP in an aqueous<br />
solution with a simulated solar irradiation source as well as to<br />
assess the biodegradability of the identified byproducts in the<br />
effluent after the treatment.<br />
water research 44 (2010) 589–595<br />
2. Reagents, equipments and<br />
analytical procedures<br />
IBP (Sigma–Aldrich), FeSO4$7H2O (Panreac) and H2O2 (30% w/v)<br />
(Panreac) were used without pre-treatment. The experimental<br />
device has been already described in earlier communications<br />
(Méndez-Arriaga et al., 2008a). Initial IBP aqueous solution<br />
(0.87 mM, pH 6.25 0.25) was pumped from the feeding stirred<br />
tank to the Duran-photoreactor (0.08 L) placed inside the<br />
Solarbox (Co.Fo.Me.Gra, Italy). The Xe lamp (Phillips OP 1 kW,<br />
6.9 mEinsteins s 1 in 290–400 nm wavelength range), placed<br />
top-inside of the Solarbox, was switched on, and samples at<br />
different irradiation times were withdrawn. In all experiments,<br />
fresh H2O2 and Fe(II)-acid dilutions were daily<br />
prepared. For photo-Fenton experiments a peristaltic pump<br />
was used to dose solutions of H2O2 and H2SO4 in the entrance<br />
of the photoreactor. Several concentrations of H2O2 were<br />
chosen, from 0 to 0.32 mM (0.04, 0.08, 0.16 and 0.32 mM), with<br />
a dose rate of 0.032 mM min 1 . pH adjust was carried out with<br />
H 2SO 4 diluted solution reaching a final pH 3. IBP concentration<br />
was monitored by HPLC by using an RP-C 18 Trace Extrasil<br />
OD52-5 Micromet 25 0.46 Teknockroma column in isocratic<br />
mode with acetonitrile (99.8% Panreac) and acetic acid (0.25 M)<br />
in 75–25 proportion. Aliquots were also used to measure the<br />
residual concentration of H2O2 by spectrophotometric method<br />
(Nogueira et al., 2005) with a Perkin Elmer UV/VIS Lambda 20.<br />
Total Organic Carbon (TOC) measurements were obtained by<br />
a Shimadzu TOC-V CNS auto sampler instrument. BOD5<br />
determinations were obtained according to Standard Methods<br />
(5120) by respirometric single measuring system OxiTop<br />
procedure. A Liquid Chromatography–Electrospray Ionization<br />
Jasco AS-2050 plus IS mass spectrometer (LC/ESI-MS, TOF<br />
Mariner) was employed to identify several byproducts.<br />
3. Results and discussion<br />
3.1. Photolysis, H 2O 2 and UV–Vis/H 2O 2<br />
photolysis effects<br />
Non-photocatalytic assays were undertaken to evaluate their<br />
isolated influence on the degradation of IBP. For the photolysis<br />
experiment, the IBP solution (initial concentration of 0.87 mM)<br />
was irradiated throughout the photoreactor during 2 h. On the<br />
other hand, several ratios of IBP-H2O2 (ranging from 1:0.001 to<br />
1:10) were mixed in 30 mL vials at free pH conditions (pH<br />
6.25 0.25), constant controlled temperature (30 C) and<br />
continuous dark stirring during 24 h. Furthermore, H2O2<br />
photolysis process was carried out irradiating a solution of<br />
0.87 mM of IBP with 0.32 mM of H 2O 2 during 2 h.<br />
Photolysis let IBP and TOC concentrations unchanged. The<br />
negligible degradation attained by photolysis was not unexpected,<br />
due to the low IBP molar absorption coefficient above<br />
300 nm. Similar results were observed for all ratios IBP-H2O2<br />
after 24 h darkness. Any difference in the original concentration<br />
of IBP was observed and TOC conversion showed an<br />
insignificant change, thus addition of peroxide in the dark<br />
gave a quite stable IBP solution. In contrast, UV–Vis/H2O2<br />
process showed an evident decrease in the concentration of
IBP. When solution was irradiated in presence of H 2O 2, almost<br />
40% of degradation took place after 2 h. The degradation of IBP<br />
can be attributed to the photochemical cleavage of hydrogen<br />
peroxide to yield hydroxyl radicals by light absorption (Tuhkanen,<br />
2004):<br />
H 2O 2 þ hv / 2 OH (1)<br />
In the configuration of our system (Xe lamp and Duranglass<br />
photoreactor) there is an overlap between the emitted<br />
light and the irradiation yet available to be absorbed for H 2O 2<br />
approximately from 290 to 350 nm.<br />
Thus, the elimination of IBP has the unique contribution of<br />
the hydroxyl radical attack produced by H2O2 photolysis but<br />
not due to their separate elements (irradiation or H2O2). An<br />
important remark is the fact that TOC did not decrease. Thus,<br />
the composition of the remained organic solution after 2 h of<br />
UV–Vis/H 2O 2 process could be addressed either to the<br />
hydroxylated byproducts of IBP with weak acidic properties<br />
according to the pH evolution observed from 6.5 to 5.5. The<br />
hydroxylation process of IBP is a consequence of its scavenger<br />
ability in favor of OH. Several indications of the OH scavenger<br />
property in NSAID have been widely reported (Hamburger and<br />
McCay, 1990; Bilodeau et al., 1995; Aruoma and Halliwell, 1988;<br />
Halliwell et al., 1995; Patricò et al., 1999).<br />
3.2. Fenton dark and photo-Fenton reaction<br />
For Fenton dark experiments, several Fe(II) concentrations<br />
(0.15 mM, 0.29 mM; 0.60 mM and 1.2 mM) were selected. Iron<br />
acid solutions were added to 1 L IBP solution (0.87 mM, pH<br />
6.25 0.25) under stirring and controlled temperature (30 C).<br />
H 2O 2 (0.32 mM) was immediately added in the tank and<br />
samples were periodically withdrawn. On the other hand,<br />
photo-Fenton experiments were carried out with similar<br />
conditions in 1.5 L IBP solution as described in Section 2. Iron<br />
acid solution (1.2 mM) was added in the continuously flowed<br />
solution. Several concentrations of H2O2 were chosen, from<br />
0 to 0.32 mM, with a dose rate of 0.032 mM min 1 . pH adjust<br />
was carried out with H2SO4 diluted solution reaching a final<br />
pH 3.<br />
Fig. 1 depicts the IBP conversion after 2 h of Fenton dark<br />
reaction. As higher Fe(II) concentration, higher IBP degradation<br />
is reached. IBP depletion by Fenton involves the chemical<br />
reaction with H 2O 2 as free OH generator (Haber and Weiss,<br />
1934):<br />
Fe(II) þ H2O2 / Fe(III) þ OH þ OH (2)<br />
Fe(III) þ H2O2 / Fe(II) þ O2H þ H þ<br />
Iron acts as catalyst by the simultaneous species Fe(II)/<br />
Fe(III) present during the reaction meanwhile H2O2 is continually<br />
consumed forming OH. The increase of Fe(II) promotes<br />
IBP decrease, in the maximum case ca. of 60% corresponding<br />
to 10% of TOC reduction. Early report (Kennedy et al., 1990)<br />
suggests that for ratios equal or higher to 0.13 of iron/IBP all<br />
water research 44 (2010) 589–595 591<br />
(3)<br />
IBP/ IBPo<br />
1.00<br />
0.75<br />
0.50<br />
0.25<br />
0.00<br />
0 30 60 90 120<br />
Time (min)<br />
Fig. 1 – Effect of Fe(II) concentration on removal of IBP<br />
(0.87 mM) by Fenton dark reaction.(B) 0.15 mM<br />
Fe(II),(:)0.29 mM Fe(II); (,)0.60 mM Fe(II) and (A)1.2 mM<br />
Fe(II). In all cases 0.32 mM of H2O2.<br />
the coordination sites of the metal are occupied by IBP and the<br />
OH production is hampered. However, in this work under the<br />
ratios tested – between 0.12 and 0.33 – the catalyst has shown<br />
to be reactive with H2O2 promoting the degradation of IBP.<br />
On the other hand, in photo-Fenton experiments more<br />
than 80% removal of IBP and almost 40% decrease of the initial<br />
TOC were achieved by using 0.32 mM H2O2 and 1.2 mM Fe(II).<br />
These values are four folds higher than the observed under<br />
dark conditions. Fig. 2 shows the IBP/IBP o for photo-Fenton<br />
process together with photolysis, H 2O 2/UV–Vis, Fenton and<br />
Fe(II)/UV–Vis process, and Fig. 3 shows the concomitant TOC<br />
removal for all the above described processes.<br />
Fig. 2 shows an evident change in the kinetic degradation<br />
of IBP by means of photo-Fenton reaction in comparison with<br />
the dark Fenton series above described. The increase in the<br />
reactivity of the photo-Fenton reaction is attributed to the<br />
simultaneous OH generation from the H2O2/UV–Vis and<br />
photo-Fenton systems. The photo-assisted reaction (photo-<br />
Fenton) gives typically faster rates of degradation and<br />
mineralization than the thermal (‘‘dark’’) reaction. In photo-<br />
Fenton reaction the production of OH is significantly<br />
increased under illuminated conditions due to the stimulating<br />
photoreduction of Fe(III) to Fe(II):<br />
IBP/ IBPo<br />
1.00<br />
0.75<br />
Photolysis<br />
H 2 O 2 /UV-Vis<br />
0.50<br />
Fenton<br />
Fe(II)/UV-Vis<br />
0.25<br />
Photo-Fenton (1.2mM Fe(II))<br />
0.00<br />
0 30 60 90<br />
0.04mM H2O2 0.08mM H2O2 0.16mM H2O2 0.32mM H<br />
120<br />
2O2 Irradiation Time (min)<br />
Fig. 2 – Removal of IBP (0.87 mM) by (-) Photolysis; (,)<br />
H 2O 2/UV–Vis (0.32 mM H 2O 2); (A) Fenton (1.2 mM Fe(II),<br />
0.32 mM H 2O 2); (>)Fe(II)/UV–Vis (1.2 mM Fe(II)); Photo-<br />
Fenton in all cases 1.2 mM Fe(II) and (:) 0.04 mM H 2O 2;(6)<br />
0.08 mM H2O2; (C) 0.16 mM H2O2; (B) 0.32 mM H2O2.
592<br />
% TOC removal<br />
50<br />
40<br />
30<br />
20<br />
10<br />
0<br />
Photo-Fenton 1.2mM Fe(II)<br />
Fig. 3 – TOC removal for H 2O 2/UV–Vis (0.32 mM H 2O 2);<br />
Fenton (1.2 mM Fe(II), 0.32 mM H 2O 2); Fe(II)/UV–Vis (1.2 mM<br />
Fe(II)); Photo-Fenton with 1.2 mM Fe(II) and 0.04 mM H 2O 2,<br />
0.08 mM H 2O 2, 0.16 mM H 2O 2, 0.32 mM H 2O 2. In all cases<br />
0.87 mM initial IBP concentration.<br />
Fe(III) þ H2O þ hv / Fe(II) þ H þ þ OH (4)<br />
Ferrous ion is continuously recycled by irradiation and<br />
therefore it is not depleted during the course of the oxidation,<br />
as stated by Zepp et al. (1992). Additionally Pignatello et al.<br />
(1999) have shown the formation of iron complexes, as<br />
(FeOH) 2þ – predominant in acidic conditions – which have<br />
adsorption bands in the UV–Vis range (between 180 and<br />
410 nm) (Sirés Sadornil Ignacio, 2006), and act as additional<br />
hydroxyl radical source in photo-Fenton reaction.<br />
Fe(II) þ H2O2 / Fe(OH) 2þ þ OH (5)<br />
Degradation of IBP was directly proportional to the amount<br />
of H2O2 employed. However, H2O2 was not in all cases full<br />
consumed. For example, almost 60% of 0.32 mM of H2O2<br />
remained unconsumed. The same proportion is observed by<br />
using 0.16 mM however a clear decrease tendency was<br />
observed between 90 and 120 min. In contrast 0.08 and<br />
0.04 mM of H 2O 2 were almost full consumed after 120 min.<br />
Fenton reaction shows a rapid degradation phase resulting<br />
from a burst of $OH by Eq. (2). However when H2O2 is in large<br />
excess the extent of this burst phase will depend on the Fe/<br />
contaminate molar ratio because the ratio determines the<br />
OH/contaminant ratio in the burst phase (Pignatello et al.,<br />
2006).<br />
Moreover due to the pH was adjusted into the first 30 min,<br />
an early parallel formation of Fe(III) could promote in strong<br />
manner the Fenton-like reaction (Eq. (3)) rather than Fenton<br />
reaction. Since reaction in Eq. (3) is so much slower than<br />
reaction in Eq. (2), the parallel formation of Fe(III) also results<br />
in a slower initial rate of IBP and similar slow consume of<br />
H2O2.<br />
In the dark series more than 90% of the H2O2 remained<br />
after 2 h of reaction. TOC depletion was enhanced under<br />
irradiated conditions in the photo-Fenton reaction. As noted<br />
in Fig. 3 the maximum TOC removal is observed around 40%.<br />
The elimination of organic carbon is also due to the formation<br />
water research 44 (2010) 589–595<br />
of volatile compounds (see also Section 3.5). Thus IBP oxidation<br />
and TOC depletion depend on experimental conditions<br />
and ratios between reactants, in a rather complex way due to<br />
the large number of reactions involved.<br />
3.3. Iron effect: IBP as organic ligand of iron<br />
and its photolysis<br />
An additional important query in Fenton process is the evaluation<br />
of possible complex formation between the organic<br />
compound and the metallic catalyst. Normally in Fenton<br />
applications the iron-complex formation with organic ligands<br />
is employed in order to widen the pH range, to avoid precipitation<br />
of Fe(III)-oxyhydroxide and/or to increase the ironphotolysis<br />
process (Andreozzi et al., 2006). However, undesirable<br />
effects can be also observed if the active coordination<br />
sites in the metal are not able to generate hydroxyl radicals in<br />
the Fenton reaction.<br />
The experiment carried out with 1.2 mM Fe(II) without<br />
H 2O 2, under irradiated condition (> doted in Fig. 2), gave an<br />
unexpected 60% of IBP removal as shown in Fig. 2. The<br />
removal of IBP using Fe(II)/UV–Vis configuration (without<br />
H 2O 2) was in deep studied by additional experiments varying<br />
the iron concentration under irradiated and free pH 5 conditions.<br />
Fig. 4 shows the IBP evolution, for 0.012, 1.2 and 5 mM<br />
FeSO4.7H2O concentrations, during 2 h of irradiation. Elimination<br />
of IBP is directly proportional to the amount of the<br />
irradiated-iron employed. IBP binds iron and under UV–Vis<br />
irradiated conditions this iron–IBP complex is photoactive. At<br />
pH 5, a large fraction of iron is in the form of Fe(III). Coordination<br />
of IBP with Fe(II) is probably negligible and would not be<br />
photolabile in the wavelength range used. Therefore, the<br />
reaction was due to the Fe(III)-complex photolysis.<br />
Iron forms complexes with organic compounds, especially<br />
those acting as polydentate ligands like Fe(III)–carboxylate or<br />
Fe(III)–polycarboxylate complexes (Pignatello et al., 2006). In<br />
the case of NSAID – which are characterized for their<br />
carboxylic acid moiety – some authors have shown their<br />
capacity to bind iron to form coordination complex. For<br />
instance, Kennedy et al. (1990) suggest that the complex<br />
formation between IBP and iron can be reached through the<br />
carboxylic moiety. Under our experimental conditions the<br />
possible production of decarboxylated byproducts and acidic<br />
IBP (mM)<br />
1.00<br />
0.75<br />
0.50<br />
0.25<br />
0.00<br />
0 30 60<br />
Time (min)<br />
90<br />
Fig. 4 – Effect of Fe(III)/IBP coordinated complex photolysis<br />
on IBP degradation during 120 min of UV–Vis irradiation.<br />
Initial Fe(II) acid ion added (C) 0.012 mM; (:) 1.2 mM; (-)<br />
5 mM and 0.87 mM IBP initial concentration.<br />
120
Table 1 – Isolated and combined effects of Fenton and<br />
photo-Fenton reagents on the IBP conversion and TOC<br />
removal observed at 2 h.<br />
Process A þ B / C XIBP TOC<br />
removal<br />
Fe(III) IBP þ hv Decarboxylated byproducts 0.70 20%<br />
H2O2 þ IBP Stable solution without reaction 0.00 0%<br />
H2O2 þ IBP þ hv Hydoxylated bycompounds 0.30 2%<br />
Fe(II) þ H2O2 þ IBP Fe(III)-IBP coordinated complex<br />
þ hydroxylated byproducts<br />
0.50 10%<br />
Fe(II) þ H2O2 þ hv<br />
þ IBP<br />
Decarboxylated þ hydroxylated<br />
þ biodegradable compounds<br />
1.00 40%<br />
aliphatic compounds are related with the change of the pH<br />
during the reaction from initial 5 to final 3.5. Higher iron–IBP<br />
complex amount, higher depletion of IBP through the<br />
carboxylic moiety, in agreement with reports of oxidation in<br />
iron-complex with organic functional group RCOO (Pignatello<br />
et al., 2006). The ligand-to-metal-charge-transfer reduction of<br />
the metal center promotes decarboxylation of IBP, at least<br />
partially, and further degradation of decarboxylated bycompounds<br />
is also driven under whole photo-Fenton system.<br />
3.4. Comparison between photo-Fenton processes and<br />
its isolated or combined reagents<br />
The photo-Fenton degradation of IBP is strongly marked for<br />
the hydroxyl attack and decarboxylation process. In dark<br />
water research 44 (2010) 589–595 593<br />
H 3 C<br />
CH 3<br />
CH3 CH3 CH3 OH<br />
CH3 CH3 CH3 H3C 4<br />
O<br />
HO<br />
5<br />
CH3 H3C 6 OH<br />
H3C 7<br />
O<br />
conditions, the catalytic reaction is less favorable and low TOC<br />
removal is observed. In Table 1 all defined stages of photo-<br />
Fenton process are described together with their simultaneous<br />
contributions of isolated and combined reagents. The<br />
degradation of IBP by iron irradiated process seems to be the<br />
principal way as initial mechanism of degradation even if<br />
H2O2 is not present (0.7 and 20% of IBP conversion and TOC<br />
removal respectively). In photo-Fenton process, the OH yield<br />
improves and simultaneous scavenges the isobutyl moiety of<br />
IBP. Thus the carboxylic moiety of IBP binds the iron while the<br />
isobutyl moiety scavenges the OH radical generated.<br />
3.5. Byproducts from photo-Fenton process<br />
IBP byproducts generated by photo-Fenton treatment (1.2 mM<br />
Fe(II), 0.32 mM H 2O 2, 2 h of irradiation) were identified by an<br />
LC/ESI-TOF-MS, into the m/z range of 65–1000 in negative<br />
ionization mode.<br />
Fig. 5 depicts the proposed reaction pathway with the<br />
corresponding intermediates identified. The remaining TOC<br />
after treatment of IBP (m/z ¼ 206, TR 28) consisted mostly of<br />
two principal types of byproducts: decarboxylated and<br />
hydroxylated metabolites of IBP. Hydroxy-IBP byproducts<br />
identified were the three structures 1 (m/z ¼ 222, TR 16), 2 (m/<br />
z ¼ 222, TR 23) and 3 (m/z ¼ 222, TR 18). The dihydroxylated<br />
compound 10 (m/z ¼ 238, TR 19) was founded at low intensity.<br />
Compound 1 suffers either the cleavage of C 1–C 2 bond of the<br />
isobutyl moiety, forming 4 (m/z ¼ 164, TR 13) and isopropanol<br />
or ketone acetone; or the decarboxylation forming 5<br />
CH 3<br />
CH3 CH3 CH3 OH<br />
CH3 OH<br />
CH3 HO<br />
1 CH3 O<br />
H3C 2 OH<br />
O<br />
H3C 3<br />
CH 3<br />
IBP<br />
O<br />
OH<br />
H3C H3C 8 OH<br />
9<br />
O<br />
CH3 CH3 CH 3<br />
HO<br />
H3C 10<br />
CH 3<br />
CH 3<br />
O<br />
OH<br />
CH3 HO<br />
O<br />
O<br />
OH<br />
Fig. 5 – Byproducts identified and possible degradation pathway by photo-Fenton reaction (1.2 mM Fe(II), 0.32 mM H2O2, 2h<br />
of irradiation, 0.87 mM IBP initial concentration). ID Name, Molecular Mass [Fragments m/z] IBP Ibuprofen, 206 [205 L ; 161 L ].<br />
1 2-HydroxyIBP, 222 [221 L ; 177 L ]. 2 1-HydroxyIBP-, 222 [221 L ; 177 L ]. 3 2-hydroxy-2-[4-(2methylpropyl)phenyl]propanoic<br />
acid, 222 [221 L ; 191 L ]. 4 (2RS)-2-(4-Methylphenyl)propanoic acid, 164 [163 L ; 133 L ]. 5 1-ethyl-4-(2-hydroxy)isobutylbenzene,<br />
178 [177 L , 149 L ]. 6 1-ethyl-4-(1-Hydroxy)isobutylbenzene, 176 [175 L ; 133 L ]. 7 1-[4-(2-Methylpropyl)phenyl]ethanone, 176<br />
[175 L ; 133 L ]. 8 4-(1-hydroxy-2-methylpropyl)acetophenone, 192 [191 L ; 177 L ]. 9 2-Methyl-1-phenylpropane, 134[133 L ;<br />
119 L ]. 10 2-hydroxy-2-[4-(2methylpropyl)phenyl]peroxic acid, 238[237 L ; 221 L ].
594<br />
(m/z ¼ 178, TR 15.5) and formic acid. In the case of 2, the<br />
evidence of its decarboxylation could be represented by the<br />
byproduct 6 (m/z ¼ 178, TR 15.5). However undifferentiated<br />
retention times by MS analysis with its analogous compound<br />
5 were obtained.<br />
In the case of 3, the degradation process follows two<br />
possible pathways by either decarboxylation and deprotonation<br />
to reach 7 (m/z ¼ 176, TR 21) and formic acid, and by<br />
dihydroxylation to obtain the byproduct 10. However<br />
byproduct 10 can be also formed by a non-free radical<br />
pathway involving a nucleophilic exchange of OH of the<br />
carboxylic acid with H 2O 2. 7 can be again hydroxylated in<br />
position 1 of the isobutyl to reach 8 (m/z ¼ 191, TR 25) and also<br />
can be decarboxylated to form 9 (m/z ¼ 133, TR 15) and<br />
acetaldehyde.<br />
An approximation of the main byproducts remained after<br />
photo-Fenton treatment could be assigned to compounds<br />
7 and 9 with 23 and 36% of the total intensities. Thus the<br />
degradation of IBP by photo-Fenton goes through the simultaneous<br />
cleavage of the carboxyl acid moiety to produce formic<br />
acid or acetaldehyde and scavenger of hydroxyl radicals. As<br />
described before the formation of high soluble bycompounds<br />
isopropanol or formic acid and acetone promotes stable<br />
dissolution. The presence of those soluble compounds after 2 h<br />
of reaction improves the biodegradability in the effluent<br />
treated. Initial BOD5 for 0.87 mM of untreated-IBP was
and effects of pharmaceutical substances in the environment -<br />
a review. Chemosphere 36, 357–393.<br />
Hartmann, J., Bartels, P., Mau, U., Witter, M., Tumpling, W.V.,<br />
Hofmann, J., Nietzschmann, E., 2008. Degradation of the drug<br />
diclofenac in water by sonolysis in presence of catalysts.<br />
Chemosphere 70, 453–461.<br />
Kennedy, T.P., Rao, N.V., Noah, W., Michael, J.R., Jafri, M.H.,<br />
Gurtner, G.H., Hoidal, J.R., 1990. Ibuprofen prevents oxidant<br />
lung injury and in vitro lipid-peroxidation by chelating iron.<br />
J. Clin. Invest. 86, 1565–1573.<br />
Klavarioti, M., Mantzavinos, D., Kassinos, D., 2008. Removal of<br />
residual pharmaceuticals from aqueous systems by advanced<br />
oxydation processes. Review article. Environment<br />
International 35 (2), pp. 402–417.<br />
Méndez-Arriaga, F., Esplugas, S., Gimenez, J., 2008a.<br />
Photocatalytic degradation of non-steroidal antiinflammatory<br />
drugs with TiO 2 and simulated solar irradiation.<br />
Water Res. 42, 585–594.<br />
Méndez-Arriaga, F., Torres Palma, R., Pétrier, C., Esplugas, S.,<br />
Gimenez, J., Pulgarin, C., 2008b. Ultrasonic treatment of water<br />
contaminated with ibuprofen. Water Res. 42, 4243–4248.<br />
Méndez-Arriaga F., March 2009. Advanced Oxidation Processes<br />
(Photocatalysis, Photo-Fenton and Sonolysis) for Removal of<br />
Pharmaceutical Pollutants in Water. Doctoral thesis,<br />
Barcelona University.<br />
Neyens, E., Baeyens, J., 2003. A review of the classic Fenton’s<br />
peroxidation as an advanced oxidation technique. J. Hazard.<br />
Mater. 98, 33–50.<br />
Nogueira, R.F.P., Oliveira, M.C., Paterlini, W.C., 2005. Simple and<br />
fast spectrophotometric determination of H2O2 in photo-<br />
Fenton reactions using metavanadate. Talanta 66, 86–91.<br />
Patricò, D., Pasin, M., Barry, O.P., Ghiselli, A., Sabatino, G.A.,<br />
Iuliano, L., FiztGerald, G., Violi, F., 1999. Iron-dependent<br />
human platelet activation and hydroxyl radical formation:<br />
involvement of protein kinase C. Circulation 99, 3118–3124.<br />
Pérez-Estrada, L., Malato, S., Gernjak, W., Agüera, A.,<br />
Thurman, M., Ferrer, I., Fernandez-Alba, A., 2005. Photo<br />
Fenton degradation of diclofenac: identification of main<br />
intermediates and degradation pathway. Environ. Sci.<br />
Technol. 39 (21), 8300–8306.<br />
Pérez-Moya, M., Graells, M., del Valle, L., Centelles, E., Mansilla, H.<br />
, 2007. Fenton and photo-Fenton degradation of<br />
2-chlorophenol: Multivariate analysis and toxicity monitoring.<br />
Catalysis Today, 124, 3-4, 30 2007, 163–171.<br />
Pignatello, J.J., Liu, D., Huston, P., 1999. Evidence for an additional<br />
oxidant in the photoassisted Fenton reaction. Environ. Sci.<br />
Technol. 33, 1832–1839.<br />
water research 44 (2010) 589–595 595<br />
Pignatello, J.J., Oliveros, E., MacKay, A., 2006. Advanced oxidation<br />
processes for organic contaminant destruction based on the<br />
Fenton reaction and related chemistry. Crit. Rev. Environ. Sci.<br />
Technol. 36, 1–84.<br />
Roberts, P.H., Thomas, K.V., 2006. The occurrence of selected<br />
pharmaceuticals in wastewater effluent and surface<br />
waters of the lower Tyne catchment. Sci. Total Environ. 356,<br />
143–153.<br />
Safarzadeh-Amiri, A., Bolton, J.R., Cater, S.R., 1996. The use of iron<br />
in advanced oxidation technologies. J. Adv. Oxid. Technol. 1,<br />
18–26.<br />
Santos, J.L., Aparicio, I., Alonso, E., 2007. Occurrence and risk<br />
assessment of pharmaceutically active compounds in<br />
wastewater treatment plants. A case study: Seville city<br />
(Spain). Environ. Int. 33, 596–601.<br />
Sirés Sadornil Ignacio, 2006. Electrochemical Advanced Oxidation<br />
Processes for the Removal of the Drugs Paracetamol, Clofibric<br />
Acid and Chlorophene from Waters. Doctoral thesis,<br />
Universidad de Barcelona.<br />
Takagi, T., Ramachandran, C., Bermejo, M., Yamashita, S., Yu, L.X.,<br />
Amidon, G.L., 2006. A provisional biopharmaceutical<br />
classification of the top 200 oral drug products in the United<br />
States, Great Britain, Spain, and Japan. Mol. Pharmacol. 3,<br />
631–643.<br />
Tarr, M.A., 2003. Chemical degradation methods for wastes and<br />
pollutants. Chemical Degradation Methods for Wastes and<br />
Pollutants. In: Environmental Science and Pollution Control<br />
Series, vol. 26 165–200.<br />
Ternes, T.A., Herrmann, N., Bonerz, M., Knacker, T., Siegrist, H.,<br />
Joss, A., 2004. A rapid method to measure the solid-water<br />
distribution coefficient (K–d) for pharmaceuticals and musk<br />
fragrances in sewage sludge. Water Res. 38, 4075–4084.<br />
Torres, R., Sarria, V., Torres, W., Peringer, P., Pulgarin, C., 2003.<br />
Electrochemical treatment of industrial wastewater<br />
containing 5-amino-6-methyl-2-benzimidazolone: toward and<br />
electrochemical–biological coupling. Water Res. 37, 7–13.<br />
Tuhkanen, T., 2004. UV/H 2O 2 processes. In: Parsons, S. (Ed.),<br />
Advanced Oxidation Processes for Water and Wastewater<br />
Treatment. IWA, London UK, pp. 86–109.<br />
Venkatadri, R., Peters, R., 1993. Chemical oxidation technologies:<br />
ultraviolet light/hydrogen peroxide, Fenton’s reagent and<br />
titanium dioxide-assisted photocatalysis, hazard. Waste<br />
Hazard. Mater. 10, 107–149.<br />
Zepp, R.G., Faust, B.C., Hoigne, J., 1992. Hydroxyl radical<br />
formation in aqueous reactions (pH 3–8) of iron(II) with<br />
hydrogen-peroxide - the photo-Fenton reaction. Environ. Sci.<br />
Technol. 26, 313–319.
Polar pollutants in municipal wastewater and the water cycle:<br />
Occurrence and removal of benzotriazoles<br />
Thorsten Reemtsma a, *, Ulf Miehe b , Uwe Duennbier c , Martin Jekel b<br />
a<br />
Federal Institute for Risk Assessment, Chemicals Safety, Thielallee 88–92, 14195 Berlin, Germany<br />
b<br />
Technical University of Berlin, Department of Water Quality Control, Sekr KF 4, Strasse des 17 Juni 135, 10623 Berlin, Germany<br />
c<br />
Berliner Wasserbetriebe, Laboratory, Motardstrasse 35, 13629 Berlin, Germany<br />
article info<br />
Article history:<br />
Received 3 April 2009<br />
Received in revised form<br />
9 July 2009<br />
Accepted 11 July 2009<br />
Available online 18 July 2009<br />
Keywords:<br />
Municipal wastewater<br />
Wastewater<br />
Treatment<br />
Benzotriazole<br />
Bank filtration<br />
Ozonation<br />
Activated carbon<br />
Groundwater<br />
Corrosion inhibitor<br />
Rivers<br />
Rhine<br />
Elbe<br />
1. Introduction<br />
abstract<br />
Benzotriazoles, in which a 1,2,3-triazole ring is condensed to<br />
a benzene ring, are a class of high production volume<br />
chemicals (HPVC) used in a wide variety of applications. The<br />
basic benzotriazole components are 1H-benzotriazole (BTri,<br />
also abbreviated as BTZ, BTA or BTAH) and the 1H-methylbenzotriazoles<br />
(tolyltriazoles; TTri, available as technical<br />
water research 44 (2010) 596–604<br />
Available at www.sciencedirect.com<br />
journal homepage: www.elsevier.com/locate/watres<br />
1H-benzo-1,2,3-triazole (BTri) and its methylated analogues (tolyltriazole, TTri) are corrosion<br />
inhibitors used in many industrial applications, but also in households in dishwashing<br />
agents and in deicing fluids at airports and elsewhere. BTri and one of the TTri-isomers<br />
(4-TTri) are typical examples of polar and poorly degradable trace pollutants. Benzotriazole<br />
elimination in four wastewater treatment plants (WWTP) in Berlin ranged from 20 to 70%<br />
for 5-TTRi over 30 to 55% for BTri to insignificant for 4-TTri. WWTP effluent concentrations<br />
were in the range of 7–18 mg/L of BTri, 1–5 mg/L of 4-TTri and 0.8–1.2 mg/L of 5-TTri. BTri and<br />
4-TTri proved to be omnipresent in surface waters of the rivers Rhine and Elbe with<br />
concentrations increasing from
enzotriazoles may dissociate and occur as anions at slightly<br />
basic pH or in contact with metal cations. At neutral pH in<br />
aqueous solution benzotriazoles are expected to be non-ionic<br />
rather than cationic.<br />
Owing to their NH acidity, benzotriazoles show metal<br />
complexing properties. BTri and TTri prevent corrosion of<br />
metals and steels, especially of copper and brass (Hart et al.,<br />
2004), by forming a thin complexing film on the metal surface<br />
(Chadwick and Hashemi, 1978). For this reason BTri or TTri are<br />
used in the metal finishing industry, in semiconductor<br />
industry (Hollingsworth et al., 2005), in milk processing (Verheyen<br />
et al., 2009), and added to various fluids that come in<br />
contact with metals, such as aircraft deicing and antiicing<br />
fluids (Gruden et al., 2001), cooling liquids, brake fluids, metalworking<br />
fluids, etc. BTri is also used for silver protection in<br />
dishwashing machines (Ort et al., 2005). The annual production<br />
of benzotriazoles was reported to be in the range of<br />
9000 tons/year worldwide (Reemtsma et al., 2006). Benzotriazole<br />
can also be employed in the developing process in the<br />
photographic industry (Bergthaller, 2002) and is a biocide<br />
included in Annex I of Council Directive 98/8/EC (European<br />
Commission, 2007).<br />
Structurally more complex hydroxiphenyl-benzotriazoles<br />
are found in personal care products, where they act as stabilizers<br />
for UVA filter components (De Orsi et al., 2006). In the<br />
European Union some benzotriazole derivatives are included<br />
in Annex VII, part I of Council Directive 76/768/EEC as registered<br />
cosmetic ingredients (European Commission, 2000).<br />
Structurally related and partly chlorinated benzotriazole<br />
derivatives are used as stabilizers in plastic material such as<br />
polyethylene terephthalate (PET) (Bentayeb et al., 2007).<br />
First reports on the occurrence of benzotriazoles in the<br />
environment were related to their use in open systems as<br />
aircraft deicing fluids at airports (Cancilla et al., 1998, 2003).<br />
A few years later it was recognized that BTri and TTri are<br />
much more widely distributed in surface waters (Reemtsma<br />
et al., 2006; Voutsa et al., 2006; Giger et al., 2006) in 0.1–1 mg/L<br />
concentrations and are regularly discharged with municipal<br />
wastewater even after state-of-the art biological treatment<br />
in mg/L concentrations (Weiss and Reemtsma, 2005; Voutsa<br />
et al., 2006; Weiss et al., 2006). Only one of these studies<br />
differentiated between the two TTri-isomers (Weiss et al.,<br />
2006) by liquid-chromatography tandem mass spectrometry<br />
(LC-MS/MS) (Weiss and Reemtsma, 2005). With this method<br />
it could be shown that the two isomers behave quite differently<br />
in the environment. While 5-TTri was mineralized<br />
within 15 days in an aerobic biodegradation test with activated<br />
sludge, the 4-TTri isomer remained stable (Weiss et al.,<br />
2006). The much higher stability of 4-TTri as compared to<br />
5-TTri has also been observed in wastewater treatment<br />
and in surface waters (Weiss and Reemtsma, 2005; Weiss<br />
et al., 2006).<br />
Unlike the N-substituted 1,2,4-triazoles, which are used as<br />
herbicides, the benzo-1,2,3-triazoles have only limited biological<br />
activity. EC50 values for the acute toxicity to Vibrio<br />
fisheri were in the range of 7–21 mg/L and NOAEC levels for<br />
invertebrates and vertebrates in the range of 10–100 mg/L<br />
(Pillard et al., 2001). BTri has been shown to exhibit antiestrogenic<br />
activity in vitro in a yeast assay but not in vivo to<br />
fish (Harris et al., 2007).<br />
water research 44 (2010) 596–604 597<br />
Despite these favourable ecotoxicity data the widespread<br />
occurrence of benzotriazoles in the environment should be<br />
a reason to improve our knowledge on the environmental<br />
behaviour of these compounds. Inter alia, the fate of benzotriazoles<br />
in surface waters is not well studied, especially for<br />
the TTri-isomers. It could also be worthwhile to become aware<br />
of those processes that are suitable for the removal of benzotriazoles<br />
from wastewater as well as from raw waters used<br />
for drinking water production.<br />
2. Materials and methods<br />
2.1. Wastewater treatment plants<br />
Samples are taken from the four largest wastewater treatment<br />
plants (WWTPs) in Berlin with treated water flows between<br />
40,000 and 200,000 m 3 /day. All plants are equipped with<br />
primary sedimentation, conventional activated sludge (CAS)<br />
treatment with nutrient removal (N/P removal) and secondary<br />
clarification. The plants differ in the P removal process:<br />
WWTP-R and W use biological P-elimination (Bio-P), WWTP-S<br />
with Bio-P and chemical precipitation and WWTP-M with<br />
chemical precipitation only. Hydraulic retention time (HRT) in<br />
the plants ranged from 11 to 24 h for the biological stage and<br />
the sludge retention time (SRT) from 8 to 25 days. The influent<br />
to WWTP-R had an organic load of 300–900 mg/L chemical<br />
oxygen demand (COD), while it ranged from 700 to 1100 mg/L<br />
COD for the other three plants.<br />
Samples of the influent (raw sewage) and the effluent of the<br />
secondary clarifier were taken as 24 h mixed samples between<br />
June and December 2006. Sampling was performed only at dry<br />
weather flow, and the sampling of effluent was corrected for<br />
hydraulic retention time in each of the plants.<br />
2.1.1. Flocculation-rapid filtration (F-RF)<br />
A polishing of the effluent of WWTP-R was investigated by<br />
rapid filtration on two experimental dual-media filters (0.8 m<br />
sand/1 m anthracite layer) that treated 100–180 L/h. Particle<br />
sizes were 0.7–1.25 mm for sand and 1.4–2.5 mm for anthracite<br />
in Filter 1 and 1.1–1.6 mm for sand and 2.5–4.0 mm for<br />
anthracite in Filter 2. Filter velocities ranged from 6 to 10 m/h.<br />
Filter backwash was performed every 1–2 days.<br />
Before filtration the effluent was treated by flocculation<br />
with either Al 3þ (1 mg/L) or Fe 3þ (2 mg/L). During some phases<br />
10–20 mg/L of powdered activated carbon (PAC; Norit Super<br />
SAE) were added during flocculation (flocculation, PAC, rapid<br />
filtration: F-PAC-RF).<br />
2.2. Field sites and field samples<br />
2.2.1. Rivers Rhine and Elbe<br />
Single grab samples were taken with a bucket in January 2007<br />
at sites used for sampling for the environmental specimen<br />
bank of the Federal Environment Agency (UBA).<br />
2.2.2. River Havel<br />
Samples of the river Havel were taken at the surface water<br />
treatment plants Spandau (upstream, n ¼ 5) and Beelitzhof<br />
(downstream; n ¼ 8) from July to November 2006.
598<br />
2.2.3. Bank filtration<br />
Bank filtration was studied at two sites in Berlin (Germany), at<br />
Lake Tegel and Lake Wannsee/Havel. Details of the bank<br />
filtration, the hydrology of the sites and the transects installed<br />
have been published before (Gruenheid et al., 2005; Massmann<br />
et al., 2008). Samples were withdrawn from several observation<br />
wells and from the production well at a distance of about<br />
100 m from the lakeshore. Samples, taken in April 2008 from<br />
the Tegel site and in May 2008, from both sites, were analyzed<br />
for benzotriazoles.<br />
2.2.4. Surface water treatment plants<br />
Two surface water treatment plants (Tegel and Beelitzhof)<br />
were investigated. These plants are designed to reduce the<br />
total phosphorus load of the surface water (median of 265 mg/L<br />
for Tegel and 129 mg/L for Beelitzhof) to values between 5 and<br />
30 mg/L by coagulation/flocculation, sedimentation and dualmedia<br />
filtration. The influent of the Tegel plant consists of<br />
25–80% effluent of WWTP-S, diluted with surface water. The<br />
Tegel plant treats 130,000–300,000 m 3 /day, the final filters<br />
have a height of 1.8 m and an average filter velocity of 6.6 m/h.<br />
The Beelitzhof plant treats surface water of Lake Wannsee<br />
(part of the River Havel system) downstream of the City of<br />
Berlin. A volume of 12,000 m 3 /day is treated in this plant and<br />
filtration occurs on filters of 1.6 m height and at a filter velocity<br />
of 5 m/h.<br />
2.3. Lab experiment<br />
2.3.1. PAC sorption studies<br />
Batch experiments for isotherm calculation were performed<br />
in either ultrapure water or effluent of WWTP-R with start<br />
concentrations of 100 mg/L of BTri or TTri (technical mix consisting<br />
of 40% 4-TTri and 60% 5-TTri (Weiss and Reemtsma,<br />
2005). Between 1 and 25 mg/L of PAC (Norit Super SAE) were<br />
added from suspensions (0.1–1 g/L) stirred at 600 rpm. The<br />
bottles were continuously mixed on a horizontal shaker<br />
(120 rpm) for 24 h. All experiments were performed in duplicate.<br />
Stability of the compounds was recorded in a control<br />
experiment without PAC addition. Dissolved concentrations<br />
of BTri and TTri (sum of both isomers) were determined after<br />
membrane filtration (0.45 mm cellulose-nitrate membrane<br />
filters) by LC-MS/MS analysis. A Freundlich model was fitted to<br />
the data.<br />
water research 44 (2010) 596–604<br />
2.4. Analysis of benzotriazoles<br />
All field samples were stored in glass bottles, transported to<br />
the lab in ice chests, filtered over 0.45 mm membrane filters<br />
and stored in a refrigerator at 4 C pending analysis, usually<br />
within the next 24 h.<br />
Benzotriazoles were analyzed by LC-MS/MS with an isocratic<br />
elution by methanolic eluents on a diphenyl column<br />
(Pursuit Diphenyl, 3 mm, 150 2 mm) that allowed separation<br />
of the two TTri-isomers. Details have been published before<br />
(Weiss and Reemtsma, 2005). Samples of surface water and<br />
WTTP influents and effluents were analyzed by direct injection<br />
(20 mL for surface water and WWTP effluent; 10 mL for<br />
WWTP influent) into the LC-MS/MS system, while groundwater<br />
samples were analyzed after enrichment of a volume of<br />
50 mL by solid phase extraction (SPE) on a Oasis HLB 200 mg<br />
cartridge. With this procedure an LOQ of 0.15 mg/L (surface<br />
water) and 0.25 mg/L (wastewater) was obtained for direct<br />
injection and 0.05 mg/L after SPE (groundwater) for each of the<br />
three analytes.<br />
3. Results and discussion<br />
3.1. Wastewater treatment plants<br />
Four WWTP in Berlin, treating municipal wastewater as a mix<br />
of household and industrial wastewater were investigated for<br />
the occurrence and removal of benzotriazoles over a 6 month<br />
period (Fig. 1, Table S1). Mean influent concentration for the<br />
four plants ranged from 17 to 44 mg/L for BTri and from 1.1 to<br />
4.9 mg/L for each of the TTri-isomers. BTri and TTri concentrations<br />
were both higher in WWTP-M as compared to the<br />
other three plants. This was not due to a generally more<br />
concentrated wastewater in this plant, because the BTri/DOC<br />
ratio of 0.6& was also twice as high as the value for the other<br />
three plants (0.2–0.3&). The source of this elevated input of<br />
benzotriazoles to this WWTP is unknown.<br />
These influent concentrations are comparable to those<br />
reported in previous studies: mean concentrations of 6.6 mg/L<br />
BTri and 0.7 mg/L for TTri were found in a monitoring of eight<br />
WWTP in Western Europe (Reemtsma et al., 2006). In a study<br />
carried out in Switzerland on 10 WWTP even higher influent<br />
concentrations of 15–73 mg/L for BTri and 1.1–5.6 mg/L for<br />
Fig. 1 – Concentrations of (a) BTri, (b) 5-TTri and (c) 4-TTri in the influent and effluent of the four largest WWTP in Berlin (R<br />
(n [ 5), S (n [ 8), W (n [ 9): 05/2006–04/2007; M (n [ 8): 10/2005–04/2007).
TTri-isomers were recorded (Voutsa et al., 2006). In a previous<br />
study on WWTP-R in Berlin 12.0 mg/L for BTri and 1.3 mg/L for<br />
5-TTri and 2.1 mg/L for 4-TTri were reported as the 2-year<br />
average (Weiss et al., 2006), which is at the lower end of<br />
concentrations determined in this study.<br />
Benzotriazoles in municipal wastewater may originate<br />
from household due to their use as corrosion inhibitors in<br />
dishwashing agents. This could be proven for a residential<br />
area by modeling the loads in its sewer system (Ort et al.,<br />
2005). When wastewater of a larger urban area is collected in<br />
one sewer system, industrial effluents, e.g. from metal finishing<br />
or semiconductor industry, may also be of relevance.<br />
Milligram per litre concentrations of TTri have only recently<br />
been reported in wastewater from milk processing (Verheyen<br />
et al., 2009). In mixed sewer systems runoff from streets<br />
(traffic) may also contribute. Airports, which are a well studied<br />
point source of benzotriazole emission (Cancilla et al., 1998,<br />
2003), may also contribute to the load in municipal wastewater<br />
if their surface runoff is collected and directed to the<br />
municipal sewer system (Weiss et al., 2006).<br />
Of the three benzotriazoles, none was eliminated<br />
completely during wastewater treatment. Generally relative<br />
removal was best for 5-TTri, but highly variable between the<br />
plants; it ranged from 19 to 69%. BTri showed moderate<br />
removal (29–58%), whereas for 4-TTri only one plant showed<br />
a significant but still limited removal (34%) (Fig. 1, Table S1).<br />
Only moderate removal of BTri and TTri (sum of both isomers)<br />
has been reported before in a monitoring of eight WWTP in<br />
Western Europe (Reemtsma et al., 2006) and for a variety of<br />
WWTP in Switzerland (Voutsa et al., 2006).<br />
This different removal of the two TTri-isomers observed<br />
in this study is in agreement with lab experiments, which<br />
showed that 5-TTri can be mineralized by activated sludge<br />
while 4-TTri is persistent (Weiss et al., 2006). These differences<br />
are attributed to the position of the methyl substituent<br />
at the ring system. Previous studies reported that the aerobic<br />
biodegradability of positional isomers of bicyclic aromatic<br />
compounds depends upon the substitution pattern and that<br />
isomers with the ortho-substituent (as in 4-TTri) were not<br />
biodegradable, while the isomer with the meta-substituent (as<br />
in 5-TTri) were degraded (Weiss and Reemtsma, 2005).<br />
The different removal behaviour of the three benzotriazoles<br />
can be highlighted by calculating the concentration<br />
ratios of either BTri or 5-TTri to the most persistent 4-TTri.<br />
Both these ratios tend to decrease during biological treatment<br />
from 6–7 to 4–5 for BTri/4-TTri and from 0.7–0.9 to 0.3–0.8 for<br />
5-TTri/4-TTri (Table S1). These ratios are especially useful to<br />
differentiate between dilution (decrease in concentrations but<br />
no change in the concentration ratios) and biodegradation<br />
(decrease in concentrations and in concentration ratios). This<br />
underlines that an appropriate analytical method that separates<br />
the isomers is essential to adequately study the environmental<br />
fate of TTri (Weiss and Reemtsma, 2005).<br />
Owing to the incomplete removal BTri median values of<br />
7–18 mg/L were found for the effluents of the four WWTP.<br />
Single values up to 38 mg/L were recorded in the effluent of<br />
WWTP-M. In a Swiss study BTri effluent concentrations<br />
ranging from 10 to 100 mg/L have been found (Voutsa et al.,<br />
2006). These are remarkably high concentrations for single<br />
compounds in WWTP effluents. In a study of eight European<br />
water research 44 (2010) 596–604 599<br />
WWTP, in which 28 trace pollutants have been monitored,<br />
BTri ranked second in effluent concentration; it was outcompeted<br />
only by ethylenediamine tetraacetic acid (EDTA)<br />
with 100 mg/L (Reemtsma et al., 2006).<br />
As shown previously the elimination of 5-TTri and BTri in<br />
biological wastewater treatment can be improved to about<br />
60% by using membrane bioreactors. But even with this<br />
upgraded biological treatment system the 4-TTri concentration<br />
was not reduced (Weiss et al., 2006).<br />
As a measure to prioritize trace pollutants in WWTP effluents<br />
with regard to their putative concentration in further<br />
compartments of the water cycle a so-called water cycle<br />
spreading index (WCSI) has been proposed (Reemtsma et al.,<br />
2006). The WCSI is the ratio of the effluent concentration and<br />
the degree of removal in a WWTP. In previous studies the order<br />
of the three benzotriazoles in terms of increasing WCSI was<br />
5-TTri (16 mg/L), BTri (22 mg/L) and 4-TTri (46 mg/L) (Reemtsma<br />
et al., 2006; Weiss et al., 2006). From the influent and effluent<br />
data of these four WWTP the order would be 5-TTri (1–13 mg/L),<br />
4-TTri (4–27 mg/L, 74 mg/L) and BTri (10–54 mg/L). Accordingly,<br />
BTri would be expected to be the benzotriazole occurring in<br />
highest concentration in surface waters and also in later<br />
compartment of a water cycle, closely followed by 4-TTri. From<br />
its WCSI (lower effluent concentration and better removal)<br />
5-TTri would be the least critical components.<br />
3.2. Polishing of WWTP effluents<br />
If the biological processes in WWTP are not suited to remove<br />
polar pollutants from the wastewater, physico-chemical posttreatment<br />
may be used to avoid discharge of these pollutants<br />
into surface waters and their spreading along partially closed<br />
water cycles.<br />
For benzotriazoles, ozonation has been shown to be<br />
a suitable post-treatment process. An ozone dose of 1 mg/mg<br />
DOC in wastewater was sufficient for complete removal of<br />
BTri and TTri (Weiss et al., 2006). No difference in the reactivity<br />
of the three benzotriazoles towards ozone could be<br />
determined.<br />
3.2.1. Flocculation, sand filtration and powdered<br />
activated carbon<br />
In the present study we investigated whether flocculation<br />
followed by rapid sand filtration, with and without addition of<br />
powdered activated carbon (PAC) can also be used for such<br />
a polishing step.<br />
A flocculation and rapid filtration (F-RF) unit (100–180 L/h)<br />
was operated for more than 2 years on site of the WWTP-R<br />
with different flocculants (Al 3þ 1 mg/L, Fe 3þ 2 mg/L) and<br />
different filter velocities (6–7.5 m/h). Based on experiences in<br />
the deironing step of water treatment plants it was hypothesized<br />
that the biomass establishing in the subsequent (bio-)<br />
filters may remove some of the trace pollutants. The flocculation<br />
with either Al 3þ or Fe 3þ had no effect on the concentration<br />
of any of the benzotriazoles. However, 5-TTri showed<br />
an average removal of 50% in the subsequent rapid filtration<br />
(RF) step (Table S1). During some periods the removal was<br />
almost complete (Fig. 2). Thus, biodegradation of 5-TTri on the<br />
filter column must have been comparatively fast, as the residence<br />
time in the filter was about 20 min, only. These data
600<br />
Fig. 2 – Mean concentration of BTri, 4-TTri and 5-TTri in the<br />
feed water and the effluent of a rapid filtration column<br />
(filled with 0.8 m sand and 1 m anthracite) after flocculation<br />
with 1–2 mg/L Fe 3D (F-RF) with and without addition of PAC<br />
(10–20 mg/L) The error bars indicate standard deviation of<br />
three samplings at independent days.<br />
confirm the much better biodegradability of 5-TTri as<br />
compared to BTri and 4-TTri.<br />
The whole process was much more effective when PAC<br />
was added during flocculation (F-PAC-RF) (Fig. 2, Table S1).<br />
With the addition of 20 mg/L of PAC BTri could be reduced by<br />
81% to filter effluent concentrations around 3 mg/L and 5-TTri<br />
as well as 4-TTri were removed to levels below the LOQ of<br />
0.25 mg/L. A PAC concentration of 10 mg/L proved insufficient<br />
in effluent polishing (Fig. 2). In batch experiments with WWTP<br />
effluent, however, 10 mg/L of PAC were able to reduce the BTri<br />
and TTri concentration by 40 and 80%, respectively, in 24 h.<br />
This indicates that the contact time in the rapid filtration<br />
process was too short to make full use of the sorption capacity<br />
of the PAC.<br />
3.3. Occurrence in surface water<br />
3.3.1. Rhine and Elbe<br />
The two major rivers of Germany and also important European<br />
rivers, Rhine (catchment area around 200,000 km 2 ) and Elbe<br />
(catchment area around 150,000 km 2 ), were analyzed for their<br />
benzotriazole concentrations. BTri was quantified (LOQ ¼<br />
0.05 mg/L) in all but one sample with concentrations ranging<br />
from 0.09 to 0.68 mg/L and 4-TTri in six of the nine samples (0.05–<br />
0.46 mg/L; Table 1). Clear trends are seen along both rivers, such<br />
as the increasing concentration of BTri (from 0.13 to 0.35 mg/L<br />
over the 700 km in the Rhine and from
discharged directly into trenches and creeks. However, the<br />
concentration of benzotriazoles in such surface runoffs has<br />
not been analyzed, yet.<br />
3.3.2. River Havel<br />
Also, water of the river Havel that passes the City of Berlin<br />
from the north to the south was analyzed for a period of<br />
5 months, because it is an important component of the<br />
partially closed water cycle of the city (Gruenheid et al., 2005).<br />
While the upstream samples did not show measurable<br />
concentrations of the triazoles, the downstream samples<br />
contained an average 1.6 mg/L of BTri, 0.34 mg/L of 5-TTri and<br />
2.1 mg/L of 4-TTri (Table 1). Here 4-TTri is the most prominent<br />
benzotriazole, which is consistent with our view on the relative<br />
stability of these three compounds.<br />
These surface water concentrations in river Havel downstream<br />
of the city of Berlin are about a factor of 5–10 higher<br />
than those found in Rhine and Elbe (Table 1). This is due to the<br />
high portion of WWTP effluent in the rivers of Berlin with their<br />
low water flow. Increased surface water concentration of BTri<br />
with increasing wastewater portion has been found also for<br />
different Swiss rivers (Giger et al., 2006).<br />
The widespread occurrence of BTri and 4-TTri in surface<br />
waters may also have consequences for the quality of raw<br />
waters used for drinking water production. Along all three<br />
rivers, Rhine, Elbe and Havel, water is abstracted via bank<br />
filtration for drinking water production.<br />
3.4. Treatment of surface water<br />
3.4.1. Flocculation and rapid filtration<br />
At several locations in Berlin (Fig. S1) surface water is treated<br />
by flocculation with Fe 3þ , sedimentation and rapid filtration to<br />
remove dissolved phosphorous and to avoid eutrophication in<br />
critical areas. Because removal of some trace pollutants, such<br />
as para-toluenesulfonamide, has been recorded in these<br />
surface water treatment plants (Richter et al., 2008) we<br />
investigated whether benzotriazoles could also be removed.<br />
However, no significant removal was obtained for any of the<br />
three compounds in each plant. For BTri and 4-TTri, these<br />
results agreed to those obtained for effluent polishing by F-RF<br />
(see above; Table S1). For 5-TTri, a 50% removal had been<br />
obtained in the RF step of the effluent, presumably by<br />
biodegradation. The lack of removal in the corresponding step<br />
with surface water, at the same contact time of 20 min, may<br />
indicate, that either microorganisms from the activated<br />
sludge may be required for this degradation or that the<br />
concentration of 5-TTri in the surface water (0.3–0.5 mg/L) was<br />
too low to induce this biodegradation process.<br />
3.4.2. (River) bank filtration<br />
River bank filtration is an important source of raw water for<br />
drinking water production and the passage of the littoral and<br />
the underground is an important process to improve and<br />
ensure water quality. This is true on a global scale, for<br />
Germany (16% of bank filtration; Hiscock and Grischek, 2002)<br />
but especially for the City of Berlin that gains more than 50% of<br />
the drinking water for its 3.5 Mio inhabitants from bank<br />
filtrate (Gruenheid et al., 2005).<br />
water research 44 (2010) 596–604 601<br />
The concentrations of the three benzotriazoles were<br />
determined along two bank filtration transects towards the<br />
production wells at two different sites (Lake Tegel and Lake<br />
Wannsee) in Berlin (Fig. S1) At both sites the distance of the<br />
production well from the lakeshore is about 100 m, corresponding<br />
to travel times of the water in the range of<br />
4–5 months (Gruenheid et al., 2005; Massmann et al., 2008).<br />
Surface water in Lake Tegel consists of 15–30% effluent of<br />
WWTP-S (Gruenheid et al., 2005). The water abstracted in the<br />
production well consists of approx. 75% lake water and 25%<br />
land-sided groundwater (inland aquifer), corresponding to<br />
a volumetric contribution of previous wastewater of 11–22%<br />
(Gruenheid et al., 2005). Redox conditions are not stable but<br />
fluctuate between anoxic and aerobic depending on water<br />
temperature and the mode of operation of production wells<br />
(Gruenheid et al., 2005). At the second site the hydrological<br />
situation is largely similar.<br />
For two reasons the concentration data (Fig. 3) are a bit<br />
scattered. (i) Not all observation wells sampled at the same<br />
depth and, thus, redox conditions are not identical. (ii) All<br />
samples were taken within 2 days while the water needs<br />
months to travel along the transects from the lakeshores to<br />
the production wells. Thus, not the same body of water was<br />
sampled along one transect. Long-term investigations at these<br />
two sites have shown that seasonal fluctuations can be<br />
recorded in the groundwater body (Massmann et al., 2008).<br />
Despite these difficulties both transects show an ongoing<br />
continuous decrease in BTri concentration from the lakeshore<br />
(distance 0) to the respective production well (Figs. 3a,d). At<br />
Lake Tegel the lakewater concentration of around 2 mg/L is<br />
diminished to 0.2 mg/L in the production well, which corresponds<br />
to a 90% concentration decrease due to dilution and<br />
removal (Fig. 3a). If one considers the dilution by land-sided<br />
groundwater with no BTri contamination, the removal of BTri<br />
during this underground passage would be 86%. At the bank<br />
filtration site at Lake Wannsee the BTri concentration was<br />
reduced by 75% to 0.3 mg/L in the production well (Fig. 3d). The<br />
concentration gradient of BTri at the Lake Tegel transect is<br />
quite parallel to the DOC gradient (Fig. 3a), which may suggest<br />
a cometabolic degradation of BTri.<br />
Removal of the least stable 5-TTri from surface water was<br />
obviously much faster and, likely, complete (Figs. 3c,f). This is<br />
well visible for the Wannsee site, where the first observation<br />
well was placed right in the colmation zone at the lakeshore.<br />
Here the 5-TTri concentration was already diminished to<br />
0.1 mg/L (Fig. 3f), while the concentrations of 4-TTri was only<br />
slightly reduced as compared to the lake water (Fig. 3e).<br />
However, also 4-TTri was removed significantly along both<br />
transects, but the concentration gradient was less steep<br />
(Figs. 3b,e). At Lake Wannsee, where the 4-TTri concentration<br />
in surface water tends to be higher than in Lake Tegel the<br />
concentration in the production well in 100 m distance<br />
remained at 0.13 mg/L (Fig. 3e).<br />
It is unclear by which mechanism 4-TTri was removed.<br />
For sandy aquifer material a retardation factor of 1.1 has<br />
been experimentally determined (Jia et al., 2007). As sandy<br />
material is predominant also at these bank filtration sites,<br />
a removal by sorption appears unlikely. Rather a very slow<br />
microbial degradation may occur. Slow biodegradation of<br />
trace concentrations of organic contaminants taking place in
602<br />
Fig. 3 – Concentration of (a, d) BTri, (b, e) 4-TTri and (c, f) 5-TTri in bank filtration transects at two sites in Berlin. (left) at Lake<br />
Tegel (04/2008 and 05/2008); DOC concentration also shown in (a); (right) at Lake Wannsee (04/2008). PW: production well.<br />
Shaded area below LOQ.<br />
time periods of months are, however, difficult to verify<br />
experimentally.<br />
Whatever process may be responsible, bank filtration was<br />
the most effective natural process for removal of benzotriazoles<br />
from waters. This may also be true for other poorly<br />
degradable polar substances. However, one has to consider<br />
the duration of this treatment process of several months. The<br />
concentration profiles of the transects also show that these<br />
long residence times in the subsurface are essential to remove<br />
poorly degradable polar pollutants such as BTri and 4-TTri,<br />
while for 5-TTri much shorter residence times would have<br />
been sufficient.<br />
These findings consistently show that neither BTri nor<br />
4-TTri are completely eliminated during bank filtration, even<br />
after a travel time of 4 months (120 days).<br />
3.5. Drinking water treatment<br />
Benzotriazoles are widespread in surface waters and<br />
‘‘natural’’ processes such as biological wastewater treatment,<br />
water research 44 (2010) 596–604<br />
biologically active filters and bank filtration are not sufficient<br />
to remove them completely. Thus, BTri and 4-TTri may occur<br />
in raw waters used for drinking water production and technical<br />
treatment processes should be available to remove such<br />
polar pollutants during drinking water treatment. While the<br />
suitability of ozonation for the removal of BTri and TTri with<br />
moderate dosages of ozone has been shown before (Weiss<br />
et al., 2006; Legube et al., 1987), it was investigated here<br />
whether activated carbon treatment could also be used to<br />
remove trace concentrations of benzotriazoles.<br />
Batch experiments with deionized water were performed<br />
to record isotherms for BTri and TTri (as the sum of both<br />
isomers; Fig. 4). The Freundlich parameters K F ¼ 84 (mg/g)/<br />
(mg/L) n for BTri and 187 (mg/g)/(mg/L) n for TTri and n ¼ 0.44<br />
for BTri and 0.52 for TTri indicate a moderate tendency to<br />
adsorb onto activated carbon. It is stronger for the methylated<br />
benzotriazoles (TTri) than for the non-substituted analogue<br />
(BTri).<br />
Based on these isotherm data one can assume that trace<br />
concentrations of TTri that may occur in raw waters are well
Fig. 4 – Isotherms for the adsorption of BTri and TTri (sum<br />
of both isomers) onto PAC (Norit Super SAE) from deionized<br />
water with a start concentration C 0 [ 100 mg/L (mean of<br />
two experiments).<br />
removable by GAC filtrations, as it has been found for the<br />
pharmaceutical diclofenac with a comparable K F values of 140<br />
(mg/g)/(mg/L) n in a previous study (Ternes et al., 2002). For BTri<br />
with its lower KF value, however, the suitability of GAC filtration<br />
is questionable. Rather breakthrough in GAC filtration<br />
may occur, as found for clofibric acid with a comparable KF<br />
value of 71 (mg/g)/(mg/L) n (Ternes et al., 2002). A final evaluation<br />
of the suitability of GAC for the removal of the triazoles<br />
would have to be performed site-specific, as the sorption<br />
tendency and the sorption kinetics in a certain water may be<br />
influenced by its matrix constituents and its pH.<br />
4. Conclusion<br />
Benzotriazoles are typical examples of polar and poorly<br />
degradable pollutants and they behave characteristically in<br />
partially closed water cycles. Biological treatment is insufficient<br />
for their complete removal from water, as shown for<br />
wastewater treatment, surface waters, bank filtration and<br />
biologically active filters. Owing to its wide use and biological<br />
resistance BTri appears to be one of the most prominent<br />
single trace pollutants in WWTP effluents (mg/L concentrations),<br />
and it is omnipresent also in surface waters. Concentrations<br />
of 4-TTri tend to increase along the large European<br />
river systems Rhine and Elbe, indicating the continuous<br />
discharge of this compound into the rivers and its high<br />
stability also in surface waters. Bank filtration with residence<br />
times of several months in the underground proved the most<br />
effective natural process for removing these polar and poorly<br />
degradable trace pollutants.<br />
But neither BTri nor 4-TTri were removed completely and<br />
both compounds may occur at reduced concentration in raw<br />
waters used for drinking water production. Fortunately, these<br />
aromatic compounds of high electron density can be removed<br />
by ozonation and, at least, TTri also by activated carbon<br />
filtration, either in wastewater effluent polishing or in<br />
drinking water treatment. Thus, an intrusion of BTri and TTri<br />
from wastewater into drinking water in partially closed water<br />
water research 44 (2010) 596–604 603<br />
cycles with indirect potable reuse can be avoided by these<br />
technical treatment steps.<br />
Data presented here clearly show an increasing resistance<br />
to biodegradation from 5-TTri over BTri to 4-TTri. Thus, the<br />
widespread contamination of the water cycle with benzotriazoles<br />
could be avoided to a large extent if solely 5-TTri<br />
would be used instead of BTri and of the technical mixture of<br />
5-TTri and 4-TTri.<br />
Acknowledgements<br />
Funding of this research by Berliner Wasserbetriebe (project<br />
‘‘Barrieren’’) is gratefully acknowledged. We are grateful to<br />
Jutta Jakobs for assistance in the laboratory and to Ina Engel<br />
for performing the activated carbon study. The river water<br />
samples were received from the Department of Hydrogeology<br />
of the Free University of Berlin.<br />
references<br />
Bentayeb, K., Batlle, R., Romero, J., Nerin, C., 2007. UPLC-MS as<br />
a powerful technique for screening the nonvolatile<br />
contaminants in recycled PET. Anal. Bioanal. Chem. 388,<br />
1031–1038.<br />
Bergthaller, P., 2002. Couplers in color photography – chemistry<br />
and function – Part 3. Imaging Sci. J. 50, 233–276.<br />
Cancilla, D.A., Martinez, J., Van Aggelen, G.C., 1998. Detection of<br />
aircraft deicing/antiicing fluids in perched water monitoring well<br />
at an international airport. Environ. Sci. Technol. 32, 3834–3835.<br />
Cancilla, D.A., Baird, J.C., Rosa, R., 2003. Detection of aircraft<br />
deicing additives in groundwater and soil samples from<br />
Fairchild Air Force base, a small to moderate user of deicing<br />
fluids. Bull. Environ. Contam. Toxicol. 70, 868–875.<br />
Chadwick, D., Hashemi, T., 1978. Adsorbed corrosion inhibitors<br />
studied by electron spectroscopy: benzotriazoles on copper<br />
and copper alloys. Corros. Sci. 18, 39–51.<br />
Corsi, S.R., Zitomer, D.H., Field, J.A., Cancilla, D.A., 2003.<br />
Nonylphenol ethoxylates and other additives in aircraft<br />
deicers, antiicers and waters receiving airport runoff. Environ.<br />
Sci. Technol. 37, 4031–4037.<br />
De Orsi, D., Gianni, G., Gogliardi, L., Porra, R., Berri, S., Bolasco, A.,<br />
Carpani, I., Tonelli, D., 2006. Simple extraction and HPLC<br />
determination of UV-A and UV-B filters in sunscreen products.<br />
Chromatographia 64, 509–515.<br />
European Commission, 2000. Twenty-Fourth Commission<br />
Directive 2000/6/EC Adapting to Technical Progress Annexes<br />
II, III, VI and VII to Council Directive 76/768/EEC on the<br />
Approximation of the Laws of the Member States Relating to<br />
Cosmetic Products.<br />
European Commission, 2007. Commission Regulation (EC) No.<br />
1451/2007 of Directive 98/8/EC Concerning the Placing of<br />
Biocidal Products on the Market.<br />
Giger, W., Schaffner, C., Kohler, H.-P.E., 2006. Benzotriazole and<br />
tolyltriazole as aquatic contaminants. 1. Input and occurrence<br />
in rivers and lakes. Environ. Sci. Technol. 40, 7186–7192.<br />
Gruden, C.L., Dow, S.M., Hernandez, M.T., 2001. Fate and toxicity<br />
of aircraft deicing fluids additives through anaerobic<br />
digestion. Water Environ. Res. 73, 72–79.<br />
Gruenheid, S., Amy, G., Jekel, M., 2005. Removal of bulk dissolved<br />
organic carbon (DOC) and trace organic compounds by bank<br />
filtration and artificial recharge. Water Res. 39, 3219–3228.
604<br />
Harris, C.A., Routledge, E.J., Schaffner, C., Brian, J.V., Giger, W.,<br />
Sumpter, J.P., 2007. Benzotriazole is antiestrogenic in vitro but<br />
not in vivo. Environ. Toxicol. Chem. 26, 2367–2372.<br />
Hart, D.S., Davis, L.C., Erickson, L.E., Callender, T.M., 2004.<br />
Sorption and partitioning of benzotriazole compounds.<br />
Microchem. J. 77, 9–17.<br />
Hiscock, K.M., Grischek, T., 2002. Attenuation of groundwater<br />
pollution by bank filtration. J. Hydrol. 266, 139–144.<br />
Hollingsworth, J., Sierra-Alvarez, R., Zhou, M., Ogden, K.L., Field, J.A.,<br />
2005. Anaerobic biodegradability and methanogenic toxicity of<br />
key constituents in copper chemical mechanical planarization<br />
effluent of the semiconductor industry. Chemosphere 59,<br />
1219–1228.<br />
Jia, Y., Breedveld, G.D., Aagaard, P., 2007. Column studies on<br />
transport of deicing additive benzotriazole in a sandy aquifer<br />
and a zerovalent iron barrier. Chemosphere 69, 1409–1418.<br />
Legube, B., Guyon, S., Dore, M., 1987. Ozonation of aqueous<br />
solutions of nitrogen heterocyclic compounds: benzotriazoles,<br />
atrazine and amitrole. Ozone Sci. Eng. 9, 233–246.<br />
Loos, R., Gawlik, B.M., Locoro, G., Rimaviciute, E., Contini, S.,<br />
Bidoglio, G., 2009. EU-wide survey of polar organic persistent<br />
pollutants in European river waters. Environ. Pollut. 157,<br />
561–568.<br />
Massmann, G., Sueltenfuss, J., Duennbier, U., Knappe, A.,<br />
Taute, T., Pekdeger, A., 2008. Investigation of groundwater<br />
residence times during bank filtration in Berlin: a multi-tracer<br />
approach. Hydrol. Process 22, 788–801.<br />
Ort, C., Schaffner, C., Giger, W., Gujer, W., 2005. Modeling<br />
stochastic load variations in a sewer system. Water Sci.<br />
Technol. 52, 113–122.<br />
water research 44 (2010) 596–604<br />
Pillard, D.A., Cornell, J.S., Dufresne, D.L., Hernandez, M.T., 2001.<br />
Toxicity of benzotriazole and benzotriazole derivatives to<br />
three aquatic species. Water Res. 35, 557–560.<br />
Reemtsma, T., Weiss, S., Mueller, J., Petrovic, M., Gonzalez, S.,<br />
Barcelo, D., Ventura, F., Knepper, T.P., 2006. Polar pollutant<br />
entry into the water cycle by municipal wastewater:<br />
a European perspective. Environ. Sci. Technol. 40, 5451–5458.<br />
Richter, D., Massmann, G., Duennbier, U., 2008. Identification and<br />
significance of sulphonamides (p-TSA, o-TSA, BSA) in an<br />
urban water cycle (Berlin, Germany). Water Res. 42, 1368–1378.<br />
Ternes, T.A., Meisenheimer, M., McDowell, D., Sacher, F.,<br />
Brauch, H.-J., Haist-Gulde, B., Preuss, G., Wilme, U., Zulei-<br />
Seibert, N., 2002. Removal of pharmaceuticals during drinking<br />
water treatment. Environ. Sci. Technol. 36, 3855–3863.<br />
Verheyen, V., Cruickshank, A., Wild, K., Heaven, M.W., McGee, R.,<br />
Watkins, M., Nash, D., 2009. Soluble, semivolatile phenol and<br />
nitrogen compounds in milk-processing wastewaters. J. Dairy<br />
Sci. 92, 3484–3493.<br />
Voutsa, D., Hartmann, P., Schaffner, C., Giger, W., 2006.<br />
Benzotriazoles, alkylphenols and bisphenol A in municipal<br />
wastewaters and in the Glatt River, Switzerland. Environ. Sci.<br />
Pollut. Res. 13, 333–341.<br />
Weiss, S., Reemtsma, T., 2005. Determination of benzotriazoles<br />
corrosion inhibitors from aqueous environmental samples by<br />
liquid chromatography-electrospray ionization-tandem mass<br />
spectrometry. Anal. Chem. 77, 7415–7420.<br />
Weiss, S., Jakobs, J., Reemtsma, T., 2006. Discharge of three<br />
benzotriazole corrosion inhibitors with municipal wastewater<br />
and improvements by membrane bioreactor treatment and<br />
ozonation. Environ. Sci. Technol. 40, 7193–7199.
Determining the fraction of pharmaceutical residues<br />
in wastewater originating from a hospital<br />
Christoph Ort a, *, Michael G. Lawrence a , Julien Reungoat a , Geoff Eaglesham b ,<br />
Steve Carter b , Jurg Keller a<br />
a<br />
The University of Queensland, Advanced Water Management Centre (AWMC), Brisbane, QLD 4072, Australia<br />
b<br />
Queensland Health Forensic and Scientific Services, Organics Laboratory, QLD 4108, Australia<br />
article info<br />
Article history:<br />
Received 14 May 2009<br />
Received in revised form<br />
21 July 2009<br />
Accepted 1 August 2009<br />
Available online 5 August 2009<br />
Keywords:<br />
Experimental design<br />
Audit data<br />
Quantification<br />
Prediction<br />
Loads<br />
Consumption<br />
1. Introduction<br />
1.1. Brief overview<br />
abstract<br />
Hospital wastewater (HWW) is normally discharged directly,<br />
without pre-treatment, to sewers. Despite mostly being only<br />
water research 44 (2010) 605–615<br />
Available at www.sciencedirect.com<br />
journal homepage: www.elsevier.com/locate/watres<br />
* Corresponding author. Tel.: þ61 7 3365 4730; fax: þ61 7 3365 4726.<br />
E-mail address: c.ort@awmc.uq.edu.au (C. Ort).<br />
0043-1354/$ – see front matter ª 2009 Elsevier Ltd. All rights reserved.<br />
doi:10.1016/j.watres.2009.08.002<br />
Pharmaceutical residues in water are frequently analysed and discussed in connection with<br />
sewage treatment, ecotoxicity and, natural and drinking water quality. Among different<br />
localities hospitals are suspected, or implied, to be a major and highly variable source of<br />
pharmaceuticals that substantially contribute to the total wastewater load. In this study, the<br />
contribution of pharmaceuticals from a hospital to a sewage treatment plant (STP) serving<br />
around 45,000 inhabitants was evaluated. Approximately 200 hospital beds result in a hospital<br />
bed density of 4.4 beds per 1000 inhabitants, which is a typical value for developed world<br />
countries. Prior to sampling, a sound systems analysis was performed, and a sophisticated<br />
continuous flow-proportional sampling regime was applied. Hence, overall experimental<br />
uncertainty was reduced to a minimum, and measurements provide clear evidence that, for 28<br />
of 59 investigated substances, over 85% of the pharmaceutical residue loads do not originate<br />
from the hospital when applying a conservative error estimation. Only for 2 substances,<br />
trimethoprim (18%) and roxithromycin (56%), was the maximum observed contribution of the<br />
hospital >15%. On average, the contribution of the hospital for the compounds detected in<br />
both, hospital effluent and sewage treatment plant influent was small and fairly constant. Five<br />
compounds were only detected in hospital wastewater, and 24 neither in the hospital wastewater<br />
nor in the total wastewater at the influent of the STP. For these compounds no experimental<br />
contribution could be calculated. For the compounds where audit data for both the<br />
national consumption and the specific hospital under investigation were available, a prediction<br />
of the fraction of pharmaceuticals originating from the hospital was performed. Three<br />
quarters of the compounds, classified with the existing audit data, were in the same ‘‘hospital<br />
contribution category’’ as determined by measurements. For most of the other compounds,<br />
plausible reasons could be identified to explain the observed deviations.<br />
ª 2009 Elsevier Ltd. All rights reserved.<br />
a small fraction of the total wastewater volume in the influent<br />
of a sewage treatment plant (STP), HWW has gained increasing<br />
scientific and public attention in the last decade. This is, in part<br />
due to the observation and expectation that HWW is a source<br />
for undesirable constituents, such as (multi-)antibiotic-resistant<br />
bacteria (Baquero et al., 2008; Kummerer, 2004). In other
606<br />
publications, the emission from hospitals was estimated for<br />
antibiotics, anaesthetics, disinfectants, heavy metals, AOX<br />
(Adsorbable Organic Halogens), iodised X-ray contrast media<br />
and cytostatic agents (e.g. Kummerer, 2001). The latter were<br />
also investigated in detail by Lenz et al. (2007). Furthermore,<br />
a number of toxicity assays were performed (Boillot et al., 2008;<br />
Ferk et al., 2009; Hartmann et al., 1998). As a result, it has been<br />
suggested in some studies that pre-treatment of HWW prior to<br />
discharge into the sewers provides a reasonable solution<br />
(Gautam et al., 2007; Lenz et al., 2007; Pauwels and Verstraete,<br />
2006). However, this view is not unanimously supported. The<br />
separate treatment of HWW to reduce the development of<br />
resistant bacteria was questioned (Kummerer, 2009): the<br />
substantial amount of antibiotics used outside of hospitals (in<br />
Germany more than 75%) seems to be a plausible reason that<br />
resistant bacteria are also abundant in wastewater not<br />
receiving any HWW. Additionally, Boillot et al. (2008) found<br />
quantitatively far fewer microorganisms in the effluents of<br />
hospitals than in urban wastewaters which is consistent with<br />
other studies. With regard to pharmaceuticals, Lenz et al.<br />
(2007) report that 1) for some pharmaceuticals merely a small<br />
fraction of the amounts administered in the hospital were<br />
actually found in its effluent (i.e. 0.1–0.2% for doxorubicin, 0.5–<br />
4.5% for 5-fluorouracil and 27–34% for total platinum) and 2)<br />
a complete onsite wastewater treatment process is needed to<br />
significantly remove targeted pharmaceuticals. This includes<br />
full physical and biological treatment steps, not only advanced<br />
processes. Capturing all sources within a hospital (wards,<br />
laboratories) may be further complicated by the fact that<br />
different facilities discharge through different pipes to the<br />
common sewer. This particularly holds true for large existing<br />
hospital complexes.<br />
Therefore, local circumstances need to be considered and<br />
the contribution of an individual hospital needs to be assessed<br />
in relation to the total load in a STP catchment. To our<br />
knowledge, only a few publications explicitly quantify pharmaceutical<br />
residues (subsequently referred to as ‘pharmaceuticals’)<br />
excreted within hospitals compared to the total<br />
pharmaceutical load in the corresponding STP influents<br />
(Feldmann et al., 2008; Heberer and Feldmann, 2005; Thomas<br />
et al., 2007). However, these studies are limited to a small<br />
number of pharmaceuticals, or make an assumption on the<br />
water flow instead of measuring the wastewater flow onsite to<br />
determine actual loads.<br />
In view of the local situation in South East Queensland<br />
where it is proposed to recycle wastewater for indirect potable<br />
reuse, it is sensible to consider whether pre-treatment of<br />
HWW will provide a significant benefit. From two previous<br />
research papers relevant for the region of interest also dealing<br />
with pharmaceuticals the contribution of hospitals cannot be<br />
derived (Khan and Ongerth, 2004; Watkinson et al., 2009).<br />
Therefore, the goal of our study is to determine accurately<br />
the contribution of a hospital to the total pharmaceutical load<br />
found at the inlet of the corresponding STP by means of<br />
measurements. Additionally, this experimentally data<br />
obtained from a limited time period is then compared with<br />
readily available audit data. It shall be assessed whether the<br />
contribution of a hospital can be predicted reliably without<br />
any additional administrative effort, i.e. without extra surveys<br />
on the hospital wards for day-specific consumptions. If<br />
water research 44 (2010) 605–615<br />
measurements matched with the prediction, the same kind<br />
(comprehensiveness and quality) of information can be used<br />
at other locations to make a prediction, a priori without<br />
laborious measurements.<br />
The focus of this research is on dissolved pollutants which<br />
cannot be eliminated in conventional wastewater treatment.<br />
Pollutants showing poor to moderate biological removal need<br />
to be transformed by chemical reactions (e.g. oxidation) or<br />
separated by physical processes (e.g. adsorption onto activated<br />
carbon).<br />
1.2. Systems analysis<br />
The prediction and experimental quantification of pharmaceutical<br />
mass fluxes in the wastewater of a specific STP catchment<br />
are laborious. A sound understanding of the whole system is<br />
required prior to setting up a predictive model, and performing<br />
a confirmative sampling campaign. This particularly holds true<br />
when attempting to attribute different fractions to a multitude<br />
of individual sources, for example if there are several hospitals<br />
and multiple smaller health care facilities in a catchment. Due to<br />
the lack of generally accessible consumption data at sufficiently<br />
high spatial and temporal resolution, models often provide only<br />
a prediction of an average load. Additionally, the latter is prone<br />
to uncertainty due to varying transformations of pharmaceuticals<br />
during human metabolism.<br />
While it would be ideal to have a list of all health care<br />
facilities with size, services provided and precise pharmaceutical<br />
consumption, just obtaining generally available<br />
consumption data is a tedious task in itself. The ‘‘institutional<br />
resolution’’ is often not sufficient without additional administrative<br />
effort, i.e. temporary surveys of the wards in the<br />
hospital(s) under investigation (Feldmann et al., 2008; Kummerer,<br />
2001). Furthermore, the (average) household pharmaceutical<br />
consumption needs to be estimated from national or<br />
state-wide sales and/or prescription data if regional data is not<br />
available.<br />
Moreover, collecting representative samples requires<br />
a thorough knowledge of the sewer layout and awareness of<br />
potentially highly variable concentrations and loads in the<br />
course of a day. Clearly, accurate chemical analysis of a nonrepresentative<br />
sample is not adequate to characterise a real<br />
full-scale system.<br />
1.3. Sampling issues<br />
Accurately quantifying pharmaceutical loads in hospital effluents<br />
or sewers close to any source (sub-catchments, households<br />
or industry) is a demanding undertaking. It requires a substantial<br />
experimental effort and is still prone to uncertainties. The<br />
latter are extremely hard to quantify if sampling is carried out<br />
with conventional (unsophisticated) devices, i.e. auto-samplers<br />
operated in a discrete sampling mode with (too) long time<br />
intervals, or grab samples. Rarely are fluctuations of concentrations<br />
and loads assessed in separate experiments at high<br />
temporal resolution prior to the ‘‘real’’ measuring campaigns.<br />
These pre-experiments are very expensive and may not provide<br />
the data to answer the actual research question. However, if the<br />
applied sampling protocol does not result in the collection of<br />
a representative sample, then the care taken in the following
processes of transport, storage, preparation and chemical<br />
analyses with a sophisticated method cannot make up for this<br />
deficiency (de Gruijter et al., 2006). Subsequent (even sophisticated)<br />
statistical analyses of non-representative samples are<br />
unreliable and the resulting conclusions will therefore be of<br />
limited value. In some cases, the large variation observed in<br />
previous studies may not be ‘‘true natural variation’’ but<br />
instead, may simply be an artefact caused by inadequate<br />
sampling (Ort et al., in preparation).<br />
Therefore, strong emphasis has been put on obtaining<br />
representative samples for this study. In Ort and Gujer (2006)<br />
a method was presented to estimate the required sampling<br />
frequency in order to not exceed a certain sampling error. In<br />
gravity sewers this results in fairly short time intervals if the<br />
substance of interest is contained in a small number of<br />
‘‘wastewater pulses’’ per day (e.g. toilet flushes containing<br />
a specific excreted pharmaceutically active compound).<br />
Sampling frequencies that are too low result in large sampling<br />
uncertainties, especially in the case of only a few patients per<br />
day (Weissbrodt et al., 2009). The often claimed problem of<br />
‘‘limited storage capacity in an auto-sampler’’ can be easily<br />
solved by replacing the glass bottles more than once per day.<br />
This may be more laborious, but it is a much better solution than<br />
using a time-proportional sampling mode, which does not take<br />
samples weightedaccording to the flow inthe sewer. In contrast,<br />
physical boundary conditions such as deep sewers resulting in<br />
long dead times for purging the sampling hose or limited access<br />
to pressurised sewers are more difficult to overcome.<br />
2. Material and methods<br />
2.1. Sewage treatment plant and catchment<br />
characteristics<br />
A total of approximately 45,000 inhabitants in two geographically<br />
separated sub-catchments, Morayfield and Caboolture,<br />
are connected to the South Caboolture STP (subsequently only<br />
referred to as STP) which is operated with two sequencing<br />
batch reactors (SBRs). It treats a daily dry weather flow of<br />
approximately 10,000 m 3 . During long dry periods with high<br />
level water restrictions, this value can drop to 7500 m 3 day 1 .<br />
Morayfield is drained by gravity sewers and contributes<br />
two thirds of the total wastewater. It is only pumped once, at<br />
the STP itself. Caboolture makes up for one third of the total<br />
influent and is a largely pressurised sewer system with<br />
numerous pumping stations. At specific times of the day the<br />
flow is diverted at the influent of the STP and stored in two<br />
large buffer tanks (800 m 3 each) before being pumped to the<br />
SBRs. This combination of sewers and the complex influent<br />
layout of the STP results in very high hydraulic fluctuations<br />
(see Fig. 1). Hours with almost zero flow contrast with hours<br />
around 250–300 L s 1 and in between, the flow varies rapidly<br />
and significantly. During wet weather the relative flow variations<br />
are less significant due to higher base flow.<br />
2.2. Hospital characteristics<br />
Caboolture Public Hospital has 190 beds and offers all services<br />
of a modern regional hospital (listed in Table SI 1, see<br />
water research 44 (2010) 605–615 607<br />
supporting information). A small private hospital providing<br />
mainly day surgery (only around 10 beds) and a small dental<br />
surgery also drain into the same sewer. The wastewater from<br />
the private hospital cannot be accessed separately. Other<br />
small health care facilities within this sewer catchment make<br />
consultations to out-patients, and therefore, the wastewater<br />
from these facilities are not expected to significantly add to<br />
the pharmaceutical load of the STP. The hospital bed density<br />
for the whole STP catchment is 4.4 beds per 1000 inhabitants.<br />
All HWW is collected in a sewage pumping station (SPS CT-51,<br />
subsequently referred to as SPS) before being pumped to the<br />
primary rising main. There is no residential wastewater<br />
contributing to this SPS and the hydraulic residence time in<br />
the main sewer to the STP is approximately 30 min to 1 h<br />
(hydraulic calculations provided by the Regional Council for<br />
the decisive time in the morning when samples at the SPS and<br />
the STP needed to be coordinated). The average daily volume<br />
during dry periods pumped at the SPS is approximately 75 m 3<br />
which is 1% of the total wastewater volume discharged to the<br />
STP. The occupancy of hospital beds in Caboolture during the<br />
sampling period was close to 100% which is representative for<br />
the year to date average.<br />
Unfortunately no comprehensive database exists with<br />
regard to other health care facilities in the catchment of the<br />
STP. Hence, an internet search was performed. Four aged care<br />
facilities with a total capacity of 443 beds were found (297 high<br />
care and 146 low care) with an unknown occupancy rate.<br />
Furthermore, a total of 14 addresses for doctors plus 12<br />
dentists were found. If mass fluxes at the influent of the STP<br />
were significantly higher than expected from average national<br />
consumption and hospital usage, further investigations of<br />
these facilities would be warranted.<br />
2.3. Sampling<br />
Continuous flow-proportional sampling modes were applied<br />
in this study to minimise sampling error. Continuously<br />
diverting a small flow-proportional side stream is conceptually<br />
the best solution to obtain representative samples for<br />
dissolved compounds. However, low velocities in the side<br />
stream prevent proper sampling of solids and long-term<br />
operation may lead to biofilm growth. Due to the limited time<br />
of sampling biofilm growth is not considered problematic in<br />
this instance.<br />
Sampling over consecutive days was preferred to the<br />
alternative option of collecting samples on single days<br />
distributed over a longer period. This drastically reduces the<br />
effect of unknown system behaviour: missing a ‘‘decisive’’<br />
HWW packet at the STP is then limited to the first hour of the<br />
first day and the last hour of the last day. All other water<br />
packets are captured, although they might be attributed to the<br />
STP sample a day later. However, this would merely lead to<br />
higher variability of the hospital’s contribution and not to<br />
a non-quantifiable effect.<br />
2.3.1. Sampling protocol for Caboolture Hospital (SPS CT-51)<br />
The HWW is not easily accessible before it enters the SPS.<br />
Furthermore it would have been very difficult to set up an<br />
accurate flow meter to measure flow in a small open channel<br />
with intermittent, partially very low flows and to use this data
608<br />
Flow [L s -1 ]<br />
500<br />
400<br />
300<br />
200<br />
100<br />
0<br />
Dry weather flow (cv = 0.72)<br />
Wet weather flow (cv = 0.44)<br />
8am 10am 12pm 2pm 4pm 6pm 8pm 10pm 12am 2am 4am 6am 8am<br />
to control the speed of the sampling pump. Instead, plumbers<br />
from the Regional Council fitted a tap in the rising main of the<br />
SPS (see Fig. 2). The tap is upstream of the non-return valve<br />
before the HWW enters the primary rising main leading to the<br />
STP. Electricians from the Regional Council installed an<br />
actuator after the tap which only opens when the pump of the<br />
SPS empties the wet well. Water runs without a sampling<br />
pump due to the pressure in the rising main. Under normal<br />
operating conditions there are about 24 pumping cycles per<br />
day, triggered automatically based on to the water level in the<br />
SPS. While it was found that the flow during one cycle is fairly<br />
constant, it can vary significantly among cycles due to variable<br />
hydraulic conditions in the primary rising main. Therefore,<br />
a manual operating mode was adopted, disabling the auto<br />
level control. This allowed for using the full storage capacity of<br />
the wet well. Starting at 7 AM it was emptied again at 12 PM, 6<br />
PM and 7 AM the following day which required personnel to be<br />
present three times per day (confined space). The pump<br />
Time [h]<br />
Fig. 1 – Two examples for typical flow patterns at the influent of the sewage treatment plant; cv [ coefficient of variation<br />
(standard deviation/mean).<br />
wet well<br />
SPS CT-51<br />
water research 44 (2010) 605–615<br />
pump<br />
control<br />
unit<br />
operates at about 2500 L min 1 and the sampling side stream<br />
was adjusted with the tap to approximately 1 L min 1 ,<br />
resulting in a sampling volume of about 10 L per pump cycle.<br />
In comparison, the dead volume of the tap installation<br />
including hose was 0.5 L (ca. 5% of the sampling volume).<br />
The three samples were collected in separate glass bottles,<br />
and analysed separately. The concentrations of the individual<br />
samples were multiplied with the flow for the corresponding<br />
pump cycle, and summed to obtain a 24-h load. Rough diurnal<br />
variations could also be determined with this sampling<br />
procedure, but they are not relevant for the system and time<br />
scales under investigation, and hence they are not further<br />
discussed in this paper.<br />
2.3.2. Sampling protocol at the sewage treatment plant<br />
To sample for the same ‘‘water packets’’ as at the SPS, sampling<br />
started at 7:45 AM in the influent of the STP. The storage tanks<br />
start filling at 8 AM and are emptied completely during night<br />
dry pit<br />
hose<br />
actuator<br />
tap<br />
non-return<br />
valve<br />
effluent Caboolture Hospital<br />
sample<br />
bottle<br />
to STP<br />
Fig. 2 – Schematic drawing of the sampling point at the sewage pumping station (SPS) CT-51 (not to scale): All hospital<br />
wastewater is discharged to the wet well of the SPS and intermittently pumped to the primary rising main leading to the<br />
sewage treatment plant (STP). Upstream of the non-return valve a stand pipe with a tap and an actuator was fitted. This<br />
allows for taking flow-proportional samples during individual pump cycles.<br />
primary rising main
time, and in the early morning hours. This guarantees that<br />
wastewater is not stored and dragged on over different 24-h<br />
sampling periods. Flow rates in the influent are routinely<br />
measured at high temporal resolution. A wire connected to an<br />
analogue digital converter provides a 4–20 mA signal from the<br />
PLC (programmable logic controller) linear to the flow in the<br />
sewer to control the speed of the sampling pump. The peristaltic<br />
pump (Watson Marlow 520UN, programmable interface, water<br />
proof casing, equipped with a 520R2 pump head and 3.2 mm<br />
tube bore) was tested in the lab to ensure its linear behaviour<br />
over the full speed range under similar physical boundary<br />
conditions (suction height approximately 2 m, pressure height<br />
negligible). The pump speed was set to 0 rpm (revolutions per<br />
minute) for 0 L s 1 in the sewer (pumping 0 mL min 1 )andto<br />
34 rpm for 1000 L s 1 (pumping 69.4 mL min 1 ). The finest<br />
increment of the pump is 0.1 rpm equivalent to 2.9 L s 1<br />
wastewater flow in the influent of the STP. With this setup<br />
approximately 15 L of wastewater were collected in a 20 L glass<br />
bottle which was located in a refrigerated container. Two field<br />
blanks were collected: to this end 0.5 L of MilliQ water was used<br />
to rinse the sampling tube and subsequently 0.5 L MilliQ water<br />
was pumped through the tube to be analysed in the laboratory.<br />
No substances were detected above the limit of quantification.<br />
2.4. Chemical analyses<br />
After collection, the continuously refrigerated samples were<br />
transported to the laboratory where they were filtered the same<br />
day and preserved before analysis. All samples were analysed<br />
for 59 substances by Queensland Health Forensic and Scientific<br />
Services (QHFSS). A detailed description of the method consisting<br />
of solid phase extraction followed by concentration prior<br />
to quantification by LC–MS/MS (liquid chromatography coupled<br />
with tandem mass spectrometry) is given in the supplementary<br />
information SI 2, accompanied with an alphabetical list of all<br />
compounds (Tables SI 2.2 and SI 2.3).<br />
As the method does not allow for correction of absolute<br />
analytical extraction recoveries in raw wastewater samples,<br />
we report relative loads. In order to compare hospital effluent<br />
samples with samples from the influent of the STP, it is<br />
necessary to assume that matrix effects between these<br />
sample types are similar. Any systematic error in recovery is<br />
therefore cancelled out when calculating ratios of loads, i.e.<br />
contribution of HWW to the total influent of the STP.<br />
2.5. Uncertainty assessment<br />
Flows in completely filled pressurised pipes can be measured<br />
more accurately than flows in open water channels (gravity<br />
flow). For this study a maximum error of 10% was assumed,<br />
which equals to 6% (¼10/3 0.5 ) as single standard deviation of<br />
a normal distribution. For chemical analysis a random uncertainty<br />
(reproducibility) of 20% for all compounds was chosen<br />
(see Tables SI 2.1–2.3). The two errors are independent, and<br />
Gaussian error propagation results in an overall uncertainty<br />
estimate for calculated loads of 21% (¼[6 2 þ 20 2 ] 0.5 ).<br />
The flow-proportional continuous sampling procedure<br />
covers all fluctuations in the wastewater over time. Since it is<br />
a reasonable assumption that dissolved compounds are<br />
completely mixed over the whole pipe cross section in the<br />
water research 44 (2010) 605–615 609<br />
influent works, no additional errors need to be taken into<br />
account due to sampling.<br />
2.6. Pharmaceutical audit data<br />
2.6.1. National consumption<br />
An extract from the DUSC database (Drug Utilisation Sub-<br />
Committee) for the year 2008 is listed for the compounds<br />
investigated in this study (see supporting information, Table SI<br />
3). It comprises the sum of subsidised drugs (subsidised under<br />
the Pharmaceutical Benefits Scheme (PBS) and the Repatriation<br />
Pharmaceutical Benefits Scheme (RPBS) and processed by<br />
Medicare Australia) and non-subsidised drugs (under PBS copayment<br />
and private prescriptions). The amounts of non-subsidised<br />
drugs were estimated from continuous data on all<br />
prescriptions dispensed from a validated sample of 370<br />
community based pharmacies. The available data do not<br />
include drugs dispensed to public hospital in-patients, pharmacy<br />
over-the-counter drugs (i.e. non-prescription) and drugs<br />
supplied by supermarkets.<br />
2.6.2. Amounts administered to in-patients in Caboolture<br />
Public Hospital<br />
No specific survey was carried out during the sampling period on<br />
the wards. Routinely stored audit data for a current 12-month<br />
period (2007–2008) was made available by the pharmacy of the<br />
Caboolture Public Hospital. For each pharmaceutical, a specific<br />
database query was performed to derive the amounts exclusively<br />
used for hospitalised in-patients; pharmaceuticals given<br />
to out-patients (in consultations and pharmacy) were not<br />
considered, since they will be taken and excreted at home. The<br />
total annual hospital load was determined after summing the<br />
contributions of all medications containing the pharmaceutically<br />
active compound of interest.<br />
It has to be noted that the amounts derived from this database<br />
are amounts supplied by the pharmacy to the individual<br />
wards and not the amounts effectively administered. However,<br />
it is generally not the hospital’s policy to discard drugs to the<br />
(solid or liquid) waste system, both from a financial and environmental<br />
point of view. Nevertheless, some unused drugs for<br />
in-patients may be collected on the wards and returned to the<br />
pharmacy for reuse or proper disposal. Hence, these drugs do<br />
not contribute to the load in the HWW. However, in discussion<br />
with relevant hospital staff these amounts are considered to be<br />
very limited and are not assessed within this study.<br />
3. Results and discussion<br />
3.1. Evaluation of wastewater volumes<br />
The four consecutive weekdays, mid-February 2009 when<br />
sampling took place, were during a wet period, with flows 1.5–<br />
2 times higher than normal dry weather flow (i.e. surface<br />
runoff in catchments and infiltration to sewage pumping<br />
stations). In Table 1 the flows at the two sampling locations<br />
over the corresponding 24-h periods are summarised. During<br />
the sampling period, the hospital contributed less than 1% of<br />
the total wastewater flow to the STP.
610<br />
Table 1 – Wastewater volumes over 24 h at the SPS CT-51 (hospital wastewater) and the influent to the STP.<br />
Influent STP [m 3 ] Hospital wastewater<br />
(flow at SPS)<br />
7:45 AM – 7:45 AM of the following day 7 AM–7 AM of the following<br />
day [m 3 ]<br />
3.2. Evaluation of relative pharmaceutical loads<br />
To obtain relative pharmaceutical loads, measured concentrations<br />
were multiplied with the corresponding 24-h flow at<br />
each sampling location and normalised by the highest STP<br />
influent load. Four examples representing four different<br />
groups of pharmaceuticals are charted in Fig. 3. Absolute<br />
concentration values are not reported because they are difficult<br />
to compare among different studies; they highly depend<br />
on the sewer system (separate or combined) and on the<br />
hydraulic conditions (dry or wet weather flow). The key figures<br />
chosen for statistical evaluation are presented in Table 2, and<br />
discussed subsequently in detail for one example (atenolol,<br />
a beta-blocker, see also Fig. 3A).<br />
The numbers in black circles ( ) refer to the corresponding<br />
column in Table 2:<br />
Concentration values for atenolol in the influent of the<br />
STP were, on average, 10 times higher than the limit of<br />
quantification (LOQ).<br />
Fraction of influent STP [%]<br />
Day 1 16/2/09 14,064 109 0.8<br />
Day 2 17/2/09 16,921 129 0.8<br />
Day 3 18/2/09 19,059 138 0.7<br />
Day 4 19/2/09 14,347 127 0.9<br />
A<br />
Loads (normalised to<br />
max. STP influent load)<br />
Atenolol<br />
1-3%<br />
1-2%<br />
2-4%<br />
1-2%<br />
fraction of STP influent<br />
C Paracetamol<br />
D<br />
Loads (normalised to<br />
max. STP influent load)<br />
1.2<br />
1.0<br />
0.8<br />
0.6<br />
0.4<br />
0.2<br />
0.0<br />
1.2<br />
1.0<br />
0.8<br />
0.6<br />
0.4<br />
0.2<br />
0.0<br />
day 1<br />
day 1<br />
day 2<br />
day 3<br />
STP influent<br />
day 2<br />
day 3<br />
STP influent<br />
day 4<br />
day 4<br />
day 1<br />
day 1<br />
water research 44 (2010) 605–615<br />
Hospital effluent<br />
day 2<br />
day 3<br />
day 4<br />
Hospital effluent<br />
day 2<br />
day 3<br />
day 4<br />
3-7%<br />
3-7%<br />
3-8%<br />
4-10%<br />
fraction of STP influent<br />
B<br />
The concentrations in the hospital effluent were on<br />
average 2 times higher than in the STP influent.<br />
The STP influent loads show only little day-to-day variation<br />
(cv ¼ 0.06, cv ¼ coefficient of variation ¼ standard<br />
deviation/mean). Day-to-day variation is smaller than<br />
the estimated overall uncertainty.<br />
The loads in the hospital effluent varied more (cv ¼ 0.27).<br />
On average the hospital contributed only 1.8% to the<br />
total atenolol load in the influent of the STP.<br />
For a conservative error estimation, a maximum<br />
contribution of the hospital was calculated by dividing<br />
the upper uncertainty value of the hospital effluent by<br />
the lower uncertainty value of the STP influent for each<br />
day (see Fig. 3). Over all four days, the highest maximum<br />
contribution for atenolol was 3.5%.<br />
Over all four days, the smallest minimum contribution<br />
for atenolol was 0.9% (analogue procedure as in ).<br />
The prediction for an average contribution of the<br />
hospital based on audit data is 0.6% (see more details in<br />
Section 3.4).<br />
0.8<br />
0.6<br />
0.4<br />
0.2<br />
0.0<br />
Gabapentin<br />
1.20 1.2<br />
1.20<br />
Hospital effluent<br />
0.03<br />
1.0<br />
0.04<br />
0.02<br />
0.01<br />
0.00<br />
Trimethoprim<br />
1-2%<br />
1-2%<br />
1-3%<br />
3-7%<br />
fraction of STP influent<br />
1.20 1.2<br />
1.20<br />
Hospital effluent<br />
0.06<br />
0.15<br />
1.0<br />
0.04<br />
0.02<br />
0.00<br />
Loads (normalised to<br />
max. STP influent load)<br />
Loads (normalised to<br />
max. STP influent load)<br />
0.8<br />
0.6<br />
0.4<br />
0.2<br />
0.0<br />
day 1<br />
day 1<br />
day 2<br />
day 2<br />
day 3<br />
STP influent<br />
day 3<br />
STP influent<br />
day 4<br />
day 4<br />
day 1<br />
day 1<br />
day 2<br />
day 2<br />
day 3<br />
day 3<br />
day 4<br />
day 4<br />
8-18%<br />
6-15%<br />
6-13%<br />
7-16%<br />
fraction of STP influent<br />
Fig. 3 – Measured, relative pharmaceutical loads over 24-h periods in the influent of the STP and effluent of the hospital for<br />
four consecutive weekdays. Error bars include uncertainty of flow measurements (±6%) and chemical analysis (±20%),<br />
resulting in an overall uncertainty of ±21% (single standard deviation). Note the different scales for the y-axis of STP influent<br />
and hospital effluent.<br />
0.03<br />
0.02<br />
0.01<br />
0.00<br />
0.10<br />
0.05<br />
0.00
Table 2 – Classification of substances according to the contribution of the hospital to the total load in the influent of the STP<br />
(see Section 3.3 for more explanations of key figures marked with black circles ). LOQs for all compounds are between 0.1<br />
and 2 mgL L1 .<br />
Classification Substance Therapeutic<br />
group f<br />
CSTP<br />
LOQ<br />
C Hospital<br />
CSTP<br />
Coefficient of variation<br />
for loads<br />
Influent<br />
STP<br />
Contribution of hospital<br />
wastewater [% of total STP influent]<br />
Measured average<br />
predicted<br />
with audit<br />
data d<br />
Hospital Min Mean Max<br />
Max 5% Atenolol BB 10.7 2.0 0.06 0.27 0.9 1.8 3.5 0.6<br />
Atorvastatin HL
612<br />
Table 3 – Comparison with other hospital wastewater studies.<br />
Number of hospitals Number of beds<br />
per 1000 inhabitants<br />
Classification of all substances according to maximum<br />
observed contribution from the hospital ( ).<br />
The consistent results for atenolol are reflected across<br />
most of the 30 detected substances. Representatives of other<br />
pharmaceutical groups show also fairly constant loads over<br />
the four-day period: gabapentin (an anticonvulsant), paracetamol<br />
(an analgesic) and trimethoprim (an antibiotic, see<br />
Fig. 3B–D).<br />
From the 59 substances, 5 were detected only in the HWW<br />
but not in the influent of the STP and 24 substances were not<br />
detected above the LOQ in any of the samples. The 30<br />
substances detected at both locations were classified for the<br />
hospital’s contribution to the total influent of the STP. To this<br />
end, the maximum observed contribution including uncertainty<br />
as a conservative estimate was used (see description<br />
before in ). The hospital’s contribution for 17 substances was<br />
at all times ‘‘smaller than 5%’’, 11 additional substances fall in<br />
the category ‘‘smaller than 15%’’ and only 2 substances were<br />
‘‘above 15%’’ (trimethoprim and roxithromycin with a worst<br />
case estimate of 18% and 56% respectively). For most<br />
substances quantified in both STP influent and hospital<br />
effluent, the variations of the loads in the HWW were on<br />
average 2.4 times higher than in the influent to the STP. The<br />
small number of hospital patients compared to the potentially<br />
large number of individuals taking these pharmaceuticals at<br />
home is a valid explanation for this observed difference in<br />
variation.<br />
water research 44 (2010) 605–615<br />
Investigated substances<br />
(% in influent of the corresponding<br />
STP originating from hospitals)<br />
2 4.4 diclofenac (1.4) a<br />
ibuprofen (0.7) a<br />
metoprolol (1.5) a<br />
paracetamol (12) a<br />
tetracycline (0.5) a<br />
trimethoprim (14) a<br />
1 3.6 5 X-ray contrast<br />
media (50) b<br />
cytostatics (max 7.5) b<br />
More than 5 12.1 carbamazepine (15) c<br />
diclofenac (10) c<br />
More than 5 12.1 metamizol (50) c<br />
Four out of the 5 substances only detected in the HWW<br />
were just above the LOQ. With the 100 fold dilution in the<br />
influent of the STP the LOQ would have to be at least three<br />
orders of magnitude lower to reliably quantify the hospital’s<br />
(high) contribution. When assuming that the concentrations<br />
in the influent of the STP were equivalent to the corresponding<br />
LOQ, only a one-sided estimation with regard to the hospital’s<br />
contributions from >5% up to >50% can be made.<br />
However, in some cases this deviates from the prediction<br />
based on audit data (see chapter 3.3).<br />
3.3. Comparison with audit data<br />
Location of study<br />
Oslo, Norway (Thomas et al., 2007)<br />
Winterthur, Switzerland (Weissbrodt et al., 2009)<br />
Berlin, Germany (Heberer and Feldmann, 2005)<br />
Berlin, Germany (Feldmann et al., 2008)<br />
a Concentrations measured over 12 weeks, loads estimated with water consumption, sum of the two major hospitals (an unknown number of<br />
other smaller hospitals/health care facilities are located in the catchment).<br />
b Influent STP was not measured in this study, percentage refers to loads of pharmaceuticals quantified in the hospital’s effluent compared to<br />
day-specific administered amounts.<br />
c Only measured in the effluent of one hospital and then extrapolated for the whole catchment based on audit data of the other hospitals.<br />
If the consumption of pharmaceuticals in a STP catchment<br />
can be estimated from existing national sales or prescription<br />
data, and audit data for the hospital are available, the<br />
contribution of the hospital can be calculated with the<br />
following equation<br />
ConsCab:Hosp:$excretion ratio<br />
contributionðhospitalÞ ¼<br />
ConsCab:Pop:$excretion ratio þ ConsCab:Hosp:$excretion ratio<br />
measured loadðhospitalÞ$recovery$accuracy<br />
y<br />
measured loadðSTP catchmentÞ$recovery$accuracy with ConsCab:Pop: ¼ ConsAUS<br />
$45; 000 ð1Þ<br />
20; 000; 000<br />
where Cons is the consumption, Cab stands for Caboolture,<br />
Pop. for population, AUS for Australia and Hosp. for hospital. It<br />
becomes evident that the transformation due to human<br />
metabolism (excretion ratio) cancels out of the equation when<br />
assumed to be similar for patients in the hospital and for<br />
people at home. The consumption of pharmaceuticals in the<br />
STP catchment is estimated by calculating an average per<br />
capita consumption from the national consumption data<br />
multiplied with the number of inhabitants in the catchment.<br />
The consumption of in-patients in the hospital is added to the
domestic consumption to obtain an estimate for the total STP<br />
influent load (see also Table SI 3).<br />
The prediction for 27 compounds where both national and<br />
hospital audit data were available, in some cases deviated<br />
significantly from the experimentally determined values.<br />
However, only 8 substances would have been classified<br />
differently based on audit data when applying strict boundaries<br />
for the classification which does not change the overall<br />
picture substantially.<br />
Possible reasons for three examples are briefly discussed: 1)<br />
The overestimation in the case of ibuprofen may be reasonably<br />
explained by the fact that the national consumption is likely to<br />
be substantially underestimated because ibuprofen can also be<br />
obtained over the counter and in supermarkets without<br />
prescription. 2) A patient who regularly takes histamine<br />
blockers (at home) is likely to take them with him if he is being<br />
hospitalised (for any treatment not related or interfering with<br />
histamine blockers). This is one of the cases where patients<br />
may bring their own medication to the hospital and is also<br />
assumed to be valid for beta-blockers and diuretics. 3) In some<br />
countries trimethoprim is often applied together with sulphamethoxazole<br />
(combination item) and hence would be<br />
expected in a similar ratio. In Australia, the consumption<br />
pattern is different: 70% of trimethoprim is sold as single item<br />
(general public) and in the hospital under investigation even<br />
90% are administered as individual compound.<br />
In other cases the explanation may be sought in a higher or<br />
lower than average number of patients being treated during the<br />
sampling period in the hospital. However, if the number of<br />
treated patients shall be estimated from measurements, the<br />
excretion ratio and absolute recoveries for chemical analyses<br />
need to be taken in to account (see Eq. (1)). This makes it difficult<br />
to compare measured influent loads from a STP with audit data<br />
from an individual health care facility to reliably calculate the<br />
health care facility’s contribution to the total influent of the STP.<br />
3.4. Comparison with other studies<br />
The Caboolture catchment, with 4.4 beds per 1000 inhabitants,<br />
is comparable with two other studies (3.6 and 4.4 beds per 1000<br />
inhabitants, see Table 3). Without audit data for the hospitals<br />
and general public, the load estimations based on measured<br />
concentrations and an estimate for wastewater based on<br />
average water consumption in the study by Thomas et al. (2007)<br />
make a direct comparison difficult. However, higher contributions<br />
were also found for paracetamol and trimethoprim. In<br />
the study by Weissbrodt et al. (2009) the loads at the influent of<br />
the STP were not measured. The percentage determined in this<br />
study is the amounts measured in the sewer divided by the<br />
amounts administered on the corresponding days. The<br />
compounds investigated in the Swiss study are iodinated X-ray<br />
contrast media and cytostatics, both compounds almost<br />
exclusively administered in hospitals. Only 50% of the X-ray<br />
contrast media and a maximum of 7.5% of the cytostatics were<br />
quantified in the hospital’s effluent, implying that the<br />
remaining part is most likely ‘‘carried home’’ by patients and<br />
excreted in household toilets. In the studies by Heberer and<br />
Feldmann (2005) and Feldmann et al. (2008) the hospital bed<br />
density is significantly higher (12.1 beds per 1000 inhabitants)<br />
with a sub-catchment bed density of 24. Pharmaceutical loads<br />
water research 44 (2010) 605–615 613<br />
were measured in the influent of the STP and in selected<br />
hospital effluents. With day-specific hospital consumption<br />
data the contribution of the other hospitals was estimated,<br />
resulting in a total hospital contribution of 15% (carbamazepine),<br />
10% (diclofenac) and 50% (metamizole, not measured in<br />
our study). Although the results seem to be in good agreement<br />
with our study, the limited number of compounds, the various<br />
approaches used, and the different catchment characteristics<br />
preclude a comprehensive comparison.<br />
3.5. Hospital wastewater treatment and catchments in<br />
South East Queensland<br />
Over 800 pharmaceuticals, disinfectants and other substances<br />
are recorded in the DUSC and the hospital database. Whilst<br />
the 59 substances analysed for in this study presents one of<br />
the more comprehensive studies of the relative contribution<br />
of a hospital to total load in wastewater, we do not claim that<br />
these results can be extrapolated for each of these 800<br />
substances, at all hospitals, or medical research activities in<br />
general. As is often the case, the selection of these 59<br />
substances was based upon the availability of a validated<br />
analytical method. Despite this ‘‘limitation’’, even if there<br />
were substances that originate almost exclusively from<br />
hospital wastewater, or if measures were taken to prevent<br />
pharmaceutical residues entering hospital wastewater<br />
(source control, separate collection of urine and faeces<br />
(Heinzmann et al., 2008)) or if hospital wastewater was treated<br />
on site, over 85% of the total load for the majority of the<br />
pharmaceuticals investigated in this study would still reach to<br />
the STP because they are excreted by the public at home in<br />
their households. Even for very specific compounds, almost<br />
exclusively administered in hospitals, the trends in many<br />
health care systems are moving towards shorter hospitalisations<br />
or even treatment of out-patients (particularly diagnostics).<br />
Two examples are the iodinated X-ray media and<br />
cytostatics: although administered in high amounts in<br />
hospitals, they cannot be recovered to 100% and hence solely<br />
attributed to hospital effluent (Weissbrodt et al., 2009).<br />
Relevance to other catchments in South East Queensland<br />
(SEQ): The three catchments of main interest within SEQ, the<br />
ones with advanced water treatment plants for providing<br />
purified recycled water to the region (for planned indirect<br />
potable reuse scheme), have approximately 8 hospital beds<br />
per 1000 inhabitants (Luggage Point, eleven hospitals), 0.4<br />
(Gibson Island, one hospital) and 1.7 (Bundamba, five hospitals).<br />
While the hospital in Caboolture (this study) contributes<br />
4.2 beds per 1000 inhabitants (total in the catchment 4.4), the<br />
biggest individual hospital in the catchment of Luggage Point<br />
accounts for only 1.5. A desktop exercise analysing audit data<br />
from the sum of all hospitals in these catchments is proposed<br />
to evaluate if further steps are required. This includes the<br />
planning of future sampling campaigns and the potential<br />
benefit of treating some hospitals’ wastewater at the source.<br />
4. Conclusions<br />
Measurements: For several, widely applied pharmaceuticals,<br />
an individual hospital seems to be a small additional point
614<br />
source in the catchment of a sewage treatment plant. In<br />
this study a hospital with 4.4 hospital beds per 1000<br />
inhabitants contributed less than 15% to the total load in<br />
the influent of the sewage treatment plant for 28<br />
substances, detected in both hospital effluent and STP<br />
influent, which is in good agreement with estimates from<br />
other studies. Considering a conservative worst case<br />
uncertainty estimation, the hospital contribution only<br />
exceeded 15% for two substances, roxithromycin (max.<br />
56%) and trimethoprim (max. 18%).<br />
Audit data: The contribution of the hospital calculated with<br />
audit data and the chosen classification reveals good<br />
agreement with actual measurements for three quarters of<br />
the substances. National audit data to calculate the<br />
consumption by the general public in a catchment and<br />
hospital data for in-patients appear to be good predictors.<br />
This approach can be used with some confidence for<br />
substances where no analytical method exists to experimentally<br />
determine concentrations and loads or where the<br />
LOQ is not low enough. This needs to be tested for other<br />
countries (dependant upon the comprehensiveness and<br />
quality of national and hospital audit data).<br />
Sampling in general: Sampling campaigns in hospital wastewater<br />
are prone to high uncertainty due to a highly dynamic<br />
system (flow and concentrations). All effort should be<br />
undertaken to understand the system (behaviour) prior to<br />
setting up a sound sampling protocol to ensure that representative<br />
samples can be obtained.<br />
Other catchments in South East Queensland: The preliminary<br />
analysis based on hospital bed densities suggests focusing<br />
on the catchment of the STP at Luggage Point (approximately<br />
8 hospital beds per 1000 inhabitants). However it has<br />
to be noted that this hospital bed density consists of 3<br />
major public hospitals and a series of private hospitals.<br />
Since measurements will be very expensive to assess all<br />
hospitals’ contributions. A detailed desktop analysis of all<br />
audit data is planned to identify if there are major sources<br />
and if measurements at selected locations may be<br />
appropriate.<br />
Hospital wastewater treatment: If, for whatever motivation,<br />
hospital wastewater shall be treated separately onsite, it<br />
must be noted, that for many substances no major overall<br />
reduction can be achieved since many pharmaceuticals are<br />
taken on a regular basis at home. With the current trend to<br />
shorter hospitalisations and treatments (diagnostics) of outpatients,<br />
this also holds true for compounds mainly<br />
administered in hospitals.<br />
Acknowledgments<br />
We would like to acknowledge the following institutions and<br />
individuals who contributed to this study: the South East<br />
Queensland Urban Water Security Research Alliance (financial<br />
support); David Fillmore, Rick Jones and Wayne Batchler<br />
from Moreton Bay Water (assistance in setting up proper<br />
sampling points and support during the campaign);<br />
Benjamin Tan and Mary Hodge from Queensland Health<br />
water research 44 (2010) 605–615<br />
Forensic and Scientific Services (logistics and processing<br />
samples); John Doonan, Greg Jackson and Daniel Field from<br />
Queensland Health (making contacts and providing hospital<br />
data); Maxine Robinson and Chris Raymond from The Drug<br />
Utilisation Sub-Committee of the Pharmaceutical Benefits<br />
Advisory Committee, Commonwealth Department of Health<br />
and Aged Care (providing Australian drug-use statistics);<br />
Christa McArdell from Eawag (valuable discussion and<br />
review of manuscript); and the Swiss National Science<br />
Foundation (Grant PBEZP2-122958 awarded to the first<br />
author).<br />
Appendix.<br />
Supplementary information<br />
Information about the supplementary material can be found<br />
at doi:10.1016/j.watres.2009.08.002.<br />
references<br />
Baquero, F., Martínez, J.-L., Cantón, R., 2008. Antibiotics and<br />
antibiotic resistance in water environments. Current Opinion<br />
in Biotechnology 19, 260–265.<br />
Boillot, C., Bazin, C., Tissot-Guerraz, F., Droguet, J., Perraud, M.,<br />
Cetre, J.C., Trepo, D., Perrodin, Y., 2008. Daily physicochemical,<br />
microbiological and ecotoxicological fluctuations of a hospital<br />
effluent according to technical and care activities. Science of<br />
the Total Environment 403 (1–3), 113–129.<br />
de Gruijter, J., Brus, D., Bierkens, M., Knotters, M., 2006. Sampling<br />
for Natural Resource Monitoring. Springer.<br />
Feldmann, D.F., Zuehlke, S., Heberer, T., 2008. Occurrence, fate<br />
and assessment of polar metamizole (dipyrone) residues in<br />
hospital and municipal wastewater. Chemosphere 71 (9),<br />
1754–1764.<br />
Ferk, F., Misik, M., Grummt, T., Majer, B., Fuerhacker, M.,<br />
Buchmann, C., Vital, M., Uhl, M., Lenz, K., Grillitsch, B.,<br />
Parzefall, W., Nersesyan, A., Knasmuller, S., 2009. Genotoxic<br />
effects of wastewater from an oncological ward. Mutation<br />
Research – Genetic Toxicology and Environmental<br />
Mutagenesis 672 (2), 69–75.<br />
Gautam, A.K., Kumar, S., Sabumon, P.C., 2007. Preliminary study<br />
of physico-chemical treatment options for hospital<br />
wastewater. Journal of Environmental Management 83 (3),<br />
298–306.<br />
Hartmann, A., Alder, A.C., Koller, T., Widmer, R.M., 1998.<br />
Identification of fluoroquinolone antibiotics as the main<br />
source of umuC genotoxicity in native hospital wastewater.<br />
Environmental Toxicology and Chemistry 17 (3), 377–382.<br />
Heberer, T., Feldmann, D., 2005. Contribution of effluents from<br />
hospitals and private households to the total loads of<br />
diclofenac and carbamazepine in municipal sewage effluents –<br />
modeling versus measurements. Journal of Hazardous<br />
Materials 122 (3), 211–218.<br />
Heinzmann, B., Schwarz, R.-J., Schuster, P., Pineau, C., 2008.<br />
Decentralized collection of iodinated X-ray contrast media in<br />
hospitals – results of the feasibility study and the practice test<br />
phase. Water Science and Technology 57 (2).<br />
Khan, S.J., Ongerth, J.E., 2004. Modelling of pharmaceutical<br />
residues in Australian sewage by quantities of use and<br />
fugacity calculations. Chemosphere 54 (3), 355–367.
Kummerer, K., 2001. Drugs in the environment: emission of<br />
drugs, diagnostic aids and disinfectants into wastewater by<br />
hospitals in relation to other sources – a review. Chemosphere<br />
45 (6–7), 957–969.<br />
Kummerer, K., 2004. Resistance in the environment. Journal of<br />
Antimicrobial Chemotherapy 54 (2), 311–320.<br />
Kummerer, K., 2009. Antibiotics in the aquatic environment –<br />
a review – part II. Chemosphere 75, 435–441.<br />
Lenz, K., Mahnik, S.N., Weissenbacher, N., Mader, R.M., Krenn, P.,<br />
Hann, S., Koellensperger, G., Uhl, M., Knasmuller, S., Ferk, F.,<br />
Bursch, W., Fuerhacker, M., 2007. Monitoring, removal and<br />
risk assessment of cytostatic drugs in hospital wastewater.<br />
Water Science and Technology 56 (12), 141–149.<br />
Ort, C., Gujer, W., 2006. Sampling for representative<br />
micropollutant loads in sewer systems. Water Science and<br />
Technology 54 (6–7), 169–176.<br />
Ort, C., Lawrence, M.G., Joss, A. Sampling for pharmaceuticals<br />
and personal care products to assess representative<br />
water research 44 (2010) 605–615 615<br />
environmental loads: sewers and wastewater treatment<br />
plants, in preparation.<br />
Pauwels, B., Verstraete, W., 2006. The treatment of hospital<br />
wastewater: an appraisal. Journal of Water and Health 4 (4),<br />
405–416.<br />
Thomas, K.V., Dye, C., Schlabach, M., Langford, K.H., 2007.<br />
Source to sink tracking of selected human pharmaceuticals<br />
from two Oslo city hospitals and a wastewater treatment<br />
works. Journal of Environmental Monitoring 9 (12), 1410–<br />
1418.<br />
Watkinson, A.J., Murby, E.J., Kolpin, D.W., Costanzo, S.D., 2009.<br />
The occurrence of antibiotics in an urban watershed: from<br />
wastewater to drinking water. Science of the Total<br />
Environment 407, 2711–2723.<br />
Weissbrodt, D., Kovalova, L., Ort, C., Pazhepurackel, V., Moser, R.,<br />
Hollender, J., Siegrist, H., McArdell, C., 2009. Mass flows of Xray<br />
contrast media and cytostatics in hospital wastewater.<br />
Environmental Science & Technology 43 (13), 4810–4817.
A three-compartment model for micropollutants sorption<br />
in sludge: Methodological approach and insights<br />
Maialen Barret, Dominique Patureau, Eric Latrille, Hélène Carrère*<br />
INRA, UR050, Laboratoire de Biotechnologie de l’Environnement, Avenue des Etangs, 11100 Narbonne, France<br />
article info<br />
Article history:<br />
Received 11 June 2009<br />
Received in revised form<br />
17 August 2009<br />
Accepted 20 August 2009<br />
Available online 27 August 2009<br />
Keywords:<br />
Activated sludge<br />
Bioavailability<br />
Isotherm<br />
Sorption model<br />
Xenobiotic<br />
1. Introduction<br />
abstract<br />
Organic micropollutants such as Polycyclic Aromatic Hydrocarbons<br />
(PAHs) present very low water solubility and are<br />
highly hydrophobic. Due to these properties, they tend to sorb<br />
to organic matter. In the context of wastewater treatment,<br />
sorption phenomena firstly impact on physical fate of these<br />
compounds, determining the fluxes distribution throughout<br />
the physical separation steps (eg. settling, centrifugation). But<br />
these physical mechanisms only determine micropollutants<br />
dispersion, whereas biological mechanisms are mainly<br />
responsible for their removal. Biological mechanisms may<br />
also be influenced by sorption, throughout bioavailability<br />
limitation towards biodegradation. Bioavailability might be<br />
the main limiting factor of PAHs biodegradation in sludge.<br />
water research 44 (2010) 616–624<br />
Available at www.sciencedirect.com<br />
journal homepage: www.elsevier.com/locate/watres<br />
* Corresponding author. Tel.: þ33 468 425 168; fax: þ33 468 425 160.<br />
E-mail address: carrere@supagro.inra.fr (H. Carrère).<br />
0043-1354/$ – see front matter ª 2009 Elsevier Ltd. All rights reserved.<br />
doi:10.1016/j.watres.2009.08.029<br />
In sludge resulting from wastewater treatment, organic micropollutants sorb to particles<br />
and to dissolved/colloidal matter (DCM). Both interactions may influence their physical and<br />
biological fate throughout the wastewater treatment processes. To our knowledge, sludge<br />
has never been considered as a three-compartment matrix, in which micropollutants<br />
coexist in three states: freely dissolved, sorbed-to-particles and sorbed-to-DCM. A methodology<br />
is proposed to concomitantly determine equilibrium constants of sorption to<br />
particles (Kpart) and to DCM (KDCM). Polycyclic Aromatic Hydrocarbons (PAHs) were chosen<br />
as model compounds for the experiments. The logarithm of estimated equilibrium<br />
constants ranged from 3.1 to 4.3 and their usual correlation to PAH hydrophobicity was<br />
verified. Moreover, PAH affinities for particles and for DCM could be compared. Affinity for<br />
particles was found to be stronger, probably due to their physical and chemical characteristics.<br />
This work provided a useful tool to assess the freely dissolved, sorbed-to-particles<br />
and sorbed-to-DCM concentrations of contaminants, which are necessary to accurately<br />
predict their fate. Besides, guidelines to investigate the link between sorption and the<br />
fundamental concept of bioavailability were proposed.<br />
ª 2009 Elsevier Ltd. All rights reserved.<br />
Micropollutants bioavailability has been widely investigated<br />
in soils and many tools were subsequently designed to quantify<br />
it (Reid et al., 2000; Tang et al., 2002; van Straalen et al., 2005;<br />
Oleszczuk, 2007; Hickman et al., 2008; Oleszczuk, 2008).<br />
However, these studies referred to different definitions of<br />
bioavailability. Moreover, bioavailability and bioaccessibility<br />
were sometimes confused. Semple et al. (2004, 2007) gathered<br />
most of the encountered concepts into two definitions:<br />
a bioavailable compound is ‘‘freely available to cross an<br />
organism’s cellular membrane from the medium the organism<br />
inhabits at a given time’’; a bioaccessible compound is ‘‘available<br />
to cross an organism’s cellular membrane from the environment,<br />
if the organism has access to the chemical’’. These<br />
definitions imply that a bioaccessible compound can become<br />
bioavailable after some time.
Nomenclature<br />
DCM Dissolved and Colloidal Matter<br />
DM Dry Matter of the sludge, which includes particles<br />
and DCM<br />
PAH Polycyclic Aromatic Hydrocarbon<br />
Cfree Concentration of freely dissolved PAHs (mg/mL)<br />
CDCM Concentration of sorbed-to-DCM PAHs (mg/gDCM) Cpart Concentration of sorbed-to-particles PAHs<br />
(mg/gPART) On the contrary, micropollutants bioavailability in sludge<br />
has been little studied. Related concepts and mechanisms are<br />
less precisely described than in soil. Some of them may be<br />
similar between soil and sludge, although matrix characteristics<br />
and micropollutants/matrix contact time completely differ.<br />
From a general point of view, micropollutants are usually<br />
assumed to be bioavailable when they are located in the<br />
aqueous phase and bioaccessible when they are initially sorbed-to-particles<br />
and can be transferred to the aqueous phase<br />
during the process (Byrns, 2001; Artola-Garicano et al., 2003;<br />
Langford et al., 2005; Urase and Kikuta, 2005; Dionisi et al., 2006).<br />
Otherwise, the assumption that a fraction of sorbed-to-particles<br />
compounds would be directly bioavailable can also be<br />
encountered (Fountoulakis et al., 2006). Anyway, even if the<br />
nature of the link between sorption phenomena and bioavailability<br />
was never demonstrated and is still subject to controversy<br />
when studying sludge, the existence of a link is generally<br />
admitted. Thus, to study sorption phenomena appears to be the<br />
first step towards bioavailability investigation.<br />
In published studies dealing with micropollutants sorption<br />
in sludge, sludge have been considered as a two-compartment<br />
system, corresponding to two states for micropollutants: sorbed-to-particles<br />
and aqueous (Byrns, 2001; Artola-Garicano<br />
et al., 2003; Ternes et al., 2004; Dionisi et al., 2006; Carballa<br />
et al., 2008). However, sorption phenomena also occur inside<br />
aqueous phase. Indeed, micropollutants have been shown to<br />
interact with dissolved and colloidal matter (DCM) of various<br />
origins (Chiou et al., 1986; Perminova et al., 2001) including<br />
sludge from wastewater treatment plant (Holbrook et al., 2004).<br />
The sorption to DCM is likely to influence micropollutants<br />
bioavailability in sludge as well as sorption to particles<br />
(Mackay and Fraser, 2000; Vinken et al., 2004). Thus, the freely<br />
dissolved and sorbed-to-DCM micropollutants states should be<br />
differentiated. Sludge should then be considered as a threecompartment<br />
matrix, with three states of micropollutants:<br />
freely dissolved, sorbed-to-DCM and sorbed-to-particles. To<br />
our knowledge, such a model has been used to study sorption<br />
phenomena in aquatic environmental systems such as sediments<br />
(Schrap and Opperhuizen, 1992; Lee and Kuo, 1999;<br />
Amiri et al., 2005) but it has never been applied to sludge.<br />
Micropollutants sorption equilibria in environmental<br />
samples including sludge are often modelled by Freundlich<br />
isotherms with a Freundlich coefficient close to 1, which is<br />
equivalent to a linear isotherm (Hung et al., 2004; Ivashechkin<br />
et al., 2004; Arias-Estevez et al., 2007). As a consequence, the<br />
equilibrium of sorption to particles is usually assumed to fit<br />
linear equations (Ternes et al., 2004; Carballa et al., 2006;<br />
Dionisi et al., 2006; Katsoyiannis et al., 2006) as well as the<br />
water research 44 (2010) 616–624 617<br />
C aqu<br />
K DCM<br />
Kpart<br />
Kglobal<br />
Concentration of apparently dissolved<br />
micropollutant (mg/mL)<br />
Equilibrium constant of PAHs sorption to DCM<br />
(mL/gDCM) Equilibrium constant of PAHs sorption to particles<br />
(mL/gPART)<br />
Coefficient of PAHs partition between aqueous<br />
phase and particles (mL/gPART)<br />
equilibrium of sorption to DCM (Yamamoto et al., 2003; Zhou<br />
et al., 2007).<br />
The objective of this study was to propose and to validate<br />
a methodological approach to study micropollutants sorption<br />
equilibria in sludge, considering sludge as a three-compartment<br />
matrix. Even though this investigation focused on PAHs,<br />
the methodology and its implications may be transposed to<br />
a wider range of micropollutants.<br />
2. Materials and methods<br />
2.1. Sludge source<br />
All experiments were performed using activated sludge from<br />
a urban wastewater treatment plant. The influent was entirely<br />
domestic. The plant consisted in preliminary treatments,<br />
primary settling, aerated tank, secondary settling and thermophilic<br />
anaerobic digestion of sewage sludge. Hydraulic<br />
retention time in the aerated tank was 0.36 day. After collection,<br />
sludge was divided into several samples and stored at<br />
20 C in order to guaranty that all following experiments<br />
were performed with exactly the same matrix. DCM fraction<br />
represented the dissolved and colloidal matters which were<br />
not differentiated, defined as the centrifugation supernatant<br />
(12 000 g, 20 min, 35 C) filtered at 1.2 mm (Whatman GF/C<br />
glass microfiber filter). As some particles were unwillingly<br />
suspended back after centrifugation because of little pellet<br />
cohesiveness, the filtration step was performed to remove<br />
these particles. The DCM mass retained by the filter was<br />
negligible, as the DCM fraction beyond the size of 1.2 mm was<br />
insignificant (data not shown). On total sludge and on DCM<br />
fraction, the dry matter (DM) was measured by weighing the<br />
samples after heating at 105 C during 24 h and the organic<br />
matter after heating at 550 C during 2 h. The defrosted activated<br />
sludge contained 65.4 1.2 g DM/L, constituted of 7 1%<br />
of DCM and 93 2% of particles. Organic matter respectively<br />
accounted for 72 and 80% of particles and DCM dry matter.<br />
2.2. Chemicals<br />
All solvents were purchased from J.T.Baker. Mixtures are<br />
indicated in volume percentage. Fluorene, phenanthrene,<br />
anthracene, fluoranthene, pyrene, benzo(a)anthracene,<br />
chrysene, benzo(b)fluoranthene, benzo(k)fluoranthene,<br />
benzo(a)pyrene, dibenzo(a,h)anthracene, benzo(g,h,i)perylene<br />
and indeno(1,2,3,c,d)pyrene powders were obtained from<br />
Dr Ehrenstorfer GmbH. Each PAH was dissolved in
618<br />
dichloromethane at 1 g/L. The spiking mix was prepared from<br />
these individual concentrated solutions, adding 5 mL of each,<br />
evaporating solvent under gentle nitrogen flow and dissolving<br />
in 50 mL of acetonitrile. Final concentrations were 100 mg/L<br />
for each PAH. This mixture was used in equilibrium experiments,<br />
as competitive phenomena are often assumed not to<br />
occur at low concentration (Krauss and Wilcke, 2005; Dionisi<br />
et al., 2006; Zhou et al., 2007).<br />
The 10 mg/L standard solution of PAHs in acetonitrile was<br />
provided by Dr Ehrenstorfer GmbH. Dilution factors from 10 to<br />
1000 were applied to obtain 6 calibration levels. Standards<br />
were stored at –20 C.<br />
2.3. PAHs quantification in sludge<br />
400 mL of sludge were centrifuged during 20 min at<br />
12 000 g, at 35 C. The supernatant was filtered at 1.2 mm<br />
with glass microfiber filters (GF/C Whatman) to obtain the<br />
aqueous phase containing the free and sorbed-to-DCM<br />
micropollutants, to which 1% of formaldehyde at 37% was<br />
added for conservation. The particulate phase (pellet) was<br />
frozen at 20 C and freeze-dried.<br />
Extraction from aqueous phase was performed using SPE<br />
cartridges with 1 g C18 sorbent from Macherey-Nagel.<br />
Cartridges were washed and conditioned with 2 5mL of<br />
methanol:toluene:acetonitrile (33:33:33), 2 5 mL of methanol<br />
and 2 5 mL of ultra pure water. 100 mL of sample were<br />
percolated through the extraction columns at a flow rate of<br />
2 mL/min. After the sample percolated, the sorbent was dried<br />
in a stream of atmospheric air during 30 min. Adsorbates were<br />
eluted with 2 5 mL of methanol:toluene (50:50). Extracts<br />
were dried under nitrogen flow and dissolved into 1 mL of<br />
acetonitrile. Each sample was extracted in duplicate and corrected<br />
by the recovery performance determined on the same<br />
aqueous sample (same DCM concentration) spiked at<br />
approximately the same PAHs concentration.<br />
Extraction from solid phase was carried out as described by<br />
Trably et al. (2004).<br />
PAHs quantification in extracts from aqueous and solid<br />
phases was performed according to the method optimised by<br />
the same authors (Trably et al., 2004). The sum of PAHs<br />
concentration in aqueous and in solid phase was compared to<br />
the spiking concentration to ensure that no significant losses<br />
occurred during the experiment.<br />
2.4. The three-compartment model<br />
A micropollutant present in sludge can be located in one<br />
among three physical compartments (Fig. 1): the freely dissolved<br />
one (concentration Cfree, mg/mL), the sorbed-to-DCM<br />
one (concentration CDCM, mg/gDCM) and the sorbed-to-particles<br />
one (concentration Cpart, mg/gPART).<br />
At equilibrium, the three-compartment system can be<br />
described by the two following equations:<br />
Kpart ¼ Cpart<br />
and<br />
Cfree<br />
water research 44 (2010) 616–624<br />
(1)<br />
Micropollutant<br />
KDCM ¼ CDCM<br />
Cfree<br />
where K part is the equilibrium constant of micropollutant<br />
sorption to particles (mL/g PART) andK DCM is the equilibrium<br />
constant of sorption to DCM (mL/g DCM). C free is very difficult to<br />
obtain experimentally whereas Caqu (concentration of apparently<br />
dissolved micropollutant, sum of the freely dissolved and<br />
sorbed-to-DCM states, mg/mL) and Cpart are easily measurable.<br />
From Caqu and Cpart measurement, Kglobal (mL/gPART) can be<br />
estimated:<br />
Kglobal ¼ Cpart<br />
Caqu<br />
KpartCfree<br />
Kpart<br />
¼<br />
¼<br />
KDCMCfree½DCMŠþCfree KDCM½DCMŠþ1<br />
where [DCM] represents the DCM concentration (g/mL). K global<br />
is not a thermodynamical equilibrium constant but a partition<br />
coefficient, which is system-dependant. Equation (3) was<br />
previously used in several studies dealing with the effect of<br />
DCM on micropollutants sorption in environmental aquatic<br />
systems (Schrap and Opperhuizen, 1992; Amiri et al., 2005).<br />
2.5. Equilibration time<br />
The kinetics of PAHs sorption was monitored in order to<br />
determine the equilibration time. For this, eight sludge<br />
samples were firstly mixed at 100 rpm at 35 C in centrifugation<br />
flasks in order to stabilize the matrix. Then, each sample<br />
was spiked at 5 mg PAH/g DM to start with the sorption kinetics.<br />
The aqueous PAHs concentration was measured after 1, 3, 4, 6,<br />
8, 10, 14 and 24 h of incubation (extractions were carried out in<br />
duplicate).<br />
2.6. Linearity verification<br />
Particle<br />
K part KDCM<br />
Dissolved and<br />
colloidal matter<br />
(DCM)<br />
Fig. 1 – Representation of the three-compartment model of<br />
a micropollutant in sludge.<br />
At constant DCM concentration, K global is constant. As<br />
a consequence, the linearity of C part ¼ f(C aqu) is a linearity<br />
indicator of equilibria (1) and (2). Five samples at 0.7 g DCM/L<br />
and 4.9 g PART/L were therefore directly put in centrifugation<br />
flasks and equilibrated at 35 C on a mixing table (100 rpm).<br />
After 30 min of equilibration, they were spiked with the<br />
spiking mix at different concentrations ranging from 1 to 5<br />
mgHAP/gDM and incubated again at 35 C for 4 h. The samples<br />
were then centrifuged at 35 C and both Cpart and Caqu were<br />
measured according to quantification section.<br />
(2)<br />
(3)
2.7. Determination of sorption equilibrium constants<br />
Kpart and KDCM can be extracted from equation (3) by assessing<br />
Kglobal with various DCM concentrations for each PAH. Variations<br />
of DCM concentrations from 0.6 to 4.5 gDCM/L were<br />
obtained by diluting sludge with its own supernatant previously<br />
separated and with water in different proportions. To<br />
ensure that this dilution method did not modify pH and that<br />
no particulate matter was released into DCM, the pH and the<br />
chemical oxygen demand (COD) were measured in 10 samples<br />
diluted by 1.2 (3.7 gDCM/L) to 200 (0.02 gDCM/L) fold factor. The<br />
COD was chosen here because it is a fast and little volume<br />
consuming method.<br />
The same procedure previously detailed was applied to<br />
equilibrate, spike, incubate and quantify Cpart and Caqu in the<br />
samples from 0.6 to 4.5 gDCM/L. Hence, Kglobal dependence to<br />
DCM concentration was measured. A non linear regression<br />
algorithm of Levenberg-Marquardt type (Marquardt, 1963) was<br />
used to minimize the sum of square errors and to estimate the<br />
two parameters of the K global model: K part and K DCM. In some<br />
cases, the algorithm did not converge, due to the measurement<br />
noise and to high K DCM[DCM] values compared to 1.<br />
Indeed, when the ‘‘þ1’’ term becomes negligible in the<br />
denominator of equation (3), this leads to Equation (4):<br />
Kglobalz<br />
Kpart<br />
KDCM½DCMŠ<br />
¼ a<br />
½DCMŠ<br />
An infinity of couples (Kpart;KDCM) can be solution of this<br />
equation, defined by the ratio a¼Kpart/KDCM. a could be<br />
precisely determined by the algorithm. Afterwards, introduction<br />
of a into equation (3) allowed us to estimate Kpart for<br />
each Kglobal measurement and its corresponding [DCM]:<br />
Kglobal<br />
Kpart ¼<br />
1 Kglobal ½DCMŠ<br />
a<br />
The average of calculated Kpart values and the corresponding<br />
KDCM were used with the Levenberg-Marquardt algorithm<br />
in order to evaluate standard deviations and correlation<br />
coefficient. This procedure was applied to dibenzo(a,h)anthracene,<br />
benzo(g,h,i)perylene and indeno(1,2,3,c,d)pyrene.<br />
3. Results and discussion<br />
3.1. Hypotheses verification<br />
From the initial DCM concentration in sludge (12 700 mgO2/L),<br />
the DCM concentration in diluted samples (from 1.2 to 200 fold<br />
factor) could be predicted, assuming that no matter was<br />
transferred from particles into aqueous phase. These predicted<br />
concentrations were compared to measured values. The ratio<br />
between them was very close to 1 (Fig. 2). Only for dilution<br />
factors higher than 20, a little fraction of particulate matter<br />
(15%) was found to be released, since the ratio reached 1.15.<br />
The highest dilution factor applied in sorption experiments<br />
was 7.5, which ensures that no particulate matter release<br />
occurred during these experiments. As a consequence, the<br />
DCM concentrations estimated thanks to the dilution factor<br />
were reliable. Moreover the pH was not modify by the dilution<br />
water research 44 (2010) 616–624 619<br />
(4)<br />
(5)<br />
Measured COD (mgO2/L) /<br />
theoretical COD (mgO2/L)<br />
1,3<br />
1,2<br />
1,1<br />
1<br />
0,9<br />
0,8<br />
1 10 100<br />
5,4<br />
1000<br />
Dilution factor<br />
Fig. 2 – Ratio between measured and theoretical aqueous<br />
COD (mg/L) (black circles) and pH (grey triangle) as<br />
a function of sludge dilution factor. The dotted frame<br />
indicates the range of dilution factors used for sorption<br />
experiments.<br />
(Fig. 2), implying that the physicochemical properties of DCM<br />
can be assumed to be unchanged.<br />
Besides, the absence of particulate matter release in<br />
diluted samples indicates that sludge DCM does not behave<br />
like sorbable/desorbable matter. Indeed, matter is likely to be<br />
either dissolved/colloidal or particulate according to intrinsic<br />
parameters. This implies that PAHs sorption to particles via<br />
the sorption of DCM on which PAHs are sorbed does not occur,<br />
which validates the ‘‘three-compartment model / two equilibria’’<br />
description (Fig. 1).<br />
The kinetics of fluorene and benzo(a)pyrene sorption in<br />
sludge are presented in Fig. 3. The sorption equilibrium state<br />
for this two PAHs was achieved within 1 h shaking, and it was<br />
the same for all other PAHs (data not shown). This value of 1 h<br />
for equilibrium achievement is consistent with literature<br />
(Dionisi et al., 2006) and suggests that during wastewater<br />
treatment processes, bioavailability is not limited by dynamic<br />
phenomena, as contact between xenobiotics and sludge lasts<br />
for several hours or days. Furthermore, a 4 h incubation time<br />
was selected for equilibrium experiments.<br />
The linear regressions performed to extract K global from<br />
C part ¼ f(C aqu) produced regression coefficients R 2 ranging<br />
from 0.85 to 0.95 (Table 1). This is consistent with the<br />
hypothesis of linearity assumed in the model. Moreover,<br />
Kglobal was plotted as a function of DCM concentration (Fig. 4)<br />
for each PAH. The obtained models fitted very well the<br />
experimental data with R 2 ranging from 0.68 to 0.95, which<br />
again reinforced the model.<br />
3.2. K global dependence to DCM concentration<br />
According to the considered PAH, the measured partition<br />
coefficients varied of a ten fold factor in the set of tested<br />
conditions (Fig. 4). Indeed, a little shift of DCM concentration<br />
was shown to imply an important variation of Kglobal, above all<br />
at low concentrations. Moreover, when sludge is not considered<br />
as a three-compartment system and measured Kglobal is<br />
equated to an equilibrium constant, this so-called constant<br />
can be largely underestimated. For example, measured Kglobal<br />
6,4<br />
6,2<br />
6<br />
5,8<br />
5,6<br />
pH
620<br />
Aqueous fraction (%)<br />
100<br />
80<br />
60<br />
40<br />
20<br />
Fluorene<br />
for benzo(b)fluoranthene at 0.6 g DCM/L was about 2000 mL/<br />
g PART whereas the real equilibrium constant K part was estimated<br />
at 6920 mL/gPART (Fig. 4). In this case, assimilating<br />
Kglobal to the equilibrium constant would lead to an underestimation<br />
of 71%. Overall, this underestimation varied from 57<br />
to 94% within the PAHs family and the more hydrophobic the<br />
PAH, the higher the error. Such an underestimation prevents<br />
from accurately predicting micropollutants sorption in sludge.<br />
All these results highlighted the necessity to consider the<br />
third phase to precisely assess the micropollutants fate.<br />
The obtained partition coefficients could not be strictly<br />
compared to literature data. Indeed, the studies published<br />
until now about PAHs sorption in sludge are very few, and<br />
authors neither take into account the three compartments nor<br />
measure DCM concentration. Moreover, our results underlined<br />
the strong Kglobal dependence to this concentration,<br />
which obliges to compare Kglobal values at equivalent DCM<br />
concentrations.<br />
3.3. Comparison between K part and K DCM<br />
0<br />
0 10 20 30<br />
0<br />
0 10 20 30<br />
t (h)<br />
t (h)<br />
The logarithm of thermodynamical equilibrium constants<br />
Kpart and KDCM extracted from Kglobal dependence to DCM<br />
concentration are presented in Table 1. Standard deviation<br />
water research 44 (2010) 616–624<br />
Aqueous fraction (%)<br />
100<br />
80<br />
60<br />
40<br />
20<br />
Benzo(a)pyrene<br />
Fig. 3 – Percentage of initially spiked PAHs that remain in aqueous phase as a function of equilibration time. Since PAHs are<br />
spiked in liquid form, at time 0, the aqueous fraction is assumed to be 100%.<br />
was higher for highly hydrophobic compounds than for<br />
weakly hydrophobic ones, due to mathematical difficulties<br />
mentioned in the paragraph 2.7. The first way to reduce<br />
standard deviations would be to multiply measurements in<br />
order to decrease experimental uncertainty due to (i) sludge<br />
heterogeneity and (ii) quantification procedure. The second<br />
way would be to optimize measurements distribution on the<br />
DCM concentration scale. However, this option is limited for<br />
practical reasons. Indeed, the upper DCM concentration limit<br />
comes from the DCM concentration in raw sludge that cannot<br />
be passed over. Besides, to minimize DCM concentration, no<br />
supernatant volume should be added and a small sludge<br />
volume should be diluted with water. Nevertheless, particles<br />
concentration would concomitantly decrease and particles<br />
quantity needed for PAH quantification would constitute the<br />
lower DCM concentration limit. Nonetheless, standard deviation<br />
values for most PAHs were acceptable in comparison<br />
with the literature data on adsorption topic, since standard<br />
deviations, although very seldom estimated, are in this range<br />
(Kordel et al., 1997; Dionisi et al., 2006; Carballa et al., 2008).<br />
Again, K part comparison with previously published data<br />
was hardly feasible. However, at low particulate matter<br />
concentration (1 g/L), which indicates a probable very low<br />
DCM concentration, Dionisi et al. (2006) measured a K global of<br />
Table 1 – log Kow, partition coefficient Kglobal and coefficient R 2 obtained from linear regression of Cpart [ f(Caqu) at 0.7 gDCM/L,<br />
and equilibrium constants extracted from Kglobal dependence to DCM concentration in the range 0.6–4.5 gDCM/L.<br />
Compound log Kow Partition coefficient Equilibrium constants<br />
K global (mL/ g part) R 2<br />
log K part<br />
log K DCM<br />
Fluorene 4.18 638 0.89 3.21 0.23 3.24 0.37<br />
Phenanthrene 4.46 2142 0.92 3.56 0.24 3.35 0.36<br />
Anthracene 4.5 1493 0.92 3.45 0.17 3.11 0.29<br />
Fluoranthene 4.9 2357 0.85 4.01 0.44 3.46 0.61<br />
Pyrene 4.88 3196 0.95 3.71 0.37 3.20 0.62<br />
Benzo(a)anthracene 5.63 2712 0.92 3.78 0.21 3.33 0.32<br />
Chrysene 5.63 1833 0.92 3.86 0.40 3.47 0.56<br />
Benzo(b)fluoranthene 6.04 2381 0.89 3.84 0.38 3.49 0.52<br />
Benzo(k)fluoranthene 6.21 1857 0.90 4.27 0.76 4.02 0.85<br />
Benzo(a)pyrene 6.06 566 0.89 4.01 0.55 3.83 0.65<br />
Dibenzo(a,h)anthracene 6.86 1701 0.90 4.34 1.57 4.12 1.71<br />
Benzo(g,h,i)perylene 6.78 1542 0.89 4.11 1.10 4.00 1.20<br />
Indeno(1,2,3,c,d)pyrene 6.58 1035 0.87 4.03 0.83 3.81 0.97
)<br />
( mL/<br />
gP<br />
ART<br />
mL/<br />
gP<br />
AR<br />
) Kglobal ( T<br />
mL/<br />
gP<br />
AR<br />
) Kglobal ( T<br />
P ) Kglobal Kglobal ( mL/<br />
g ART<br />
)<br />
Kglobal ( mL/<br />
gP<br />
ART<br />
1200<br />
800<br />
400<br />
Fluorene<br />
K part = 1 606<br />
K DCM = 1 717<br />
R 2 = 0.857<br />
0<br />
0 1 2 3 4 5<br />
8000<br />
Fluoranthene<br />
Kpart = 10 225<br />
6000<br />
4000<br />
KDCM = 2 894<br />
R2 = 0.823<br />
2000<br />
0<br />
0 1 2 3 4 5<br />
Chrysene<br />
5000<br />
4000<br />
3000<br />
Kpart = 7 251<br />
KDCM = 2 940<br />
R2 = 0.854<br />
2000<br />
1000<br />
3000<br />
2000<br />
1000<br />
9700 5000 mL/g PART for pyrene in an activated sludge. This<br />
value might be compared to our K part, asK part corresponds to<br />
K global limit when DCM concentration tends to zero. We<br />
calculated a K part in the range 2200–12 000 mL/g PART for pyrene<br />
(Table 1), which is in good agreement with Dionisi et al.’s<br />
(2006) values.<br />
KDCM is easier to compare with literature data than Kpart,<br />
since several studies were performed in absence of particles. As<br />
an illustration, Holbrook et al. (2004) studied pyrene adsorption<br />
onto sludge DCM. Depending on sample origin and size<br />
Phenanthrene<br />
K part = 3 659<br />
K DCM = 2 244<br />
R 2 = 0.902<br />
0<br />
0 1 2 3 4 5<br />
4000<br />
Pyrene<br />
Kpart = 5 089<br />
3000<br />
2000<br />
KDCM = 1 594<br />
R2 = 0.680<br />
1000<br />
0<br />
0 1 2 3 4 5<br />
Benzo(b)fluoranthene<br />
5000<br />
4000<br />
3000<br />
Kpart = 6 920<br />
KDCM = 3 073<br />
R2 = 0.877<br />
2000<br />
1000<br />
3000<br />
2000<br />
1000<br />
Anthracene<br />
K part = 3 636<br />
K DCM= 2 117<br />
R 2 = 0.833<br />
0<br />
0 1 2 3 4 5<br />
Benzo(a)anthracene<br />
5000<br />
4000<br />
3000<br />
Kpart = 7 941<br />
KDCM = 3 337<br />
R2 = 0.876<br />
2000<br />
1000<br />
0<br />
0 1 2 3 4 5<br />
8000<br />
Benzo(k)fluoranthene<br />
Kpart = 18 604<br />
6000<br />
4000<br />
KDCM = 10 429<br />
R2 = 0.950<br />
0<br />
0 1 2 3 4 5<br />
0<br />
0 1 2 3 4 5<br />
0<br />
0 1 2 3 4 5<br />
5000<br />
4000<br />
3000<br />
Kpart = 10 283<br />
KDCM = 6 745<br />
R<br />
2000<br />
1000<br />
2 8000<br />
= 0.940<br />
6000<br />
4000<br />
Kpart = 22 076<br />
KDCM = 13 202<br />
R<br />
2000<br />
2 5000<br />
= 0.863<br />
4000<br />
3000<br />
Kpart = 12 770<br />
KDCM = 9 992<br />
R<br />
2000<br />
1000<br />
2 Benzo(a)pyrene Dibenzo(a,h)anthracene Benzo(g,h,i)perylene<br />
α = 1.672<br />
α = 1.278<br />
= 0.895<br />
5000<br />
4000<br />
3000<br />
2000<br />
1000<br />
0<br />
0 1 2 3 4 5<br />
Indeno(1,2,3,c,d)pyrene<br />
Kpart = 10 599<br />
KDCM = 6 465<br />
R2 α = 1.639<br />
= 0.823<br />
0<br />
0 1 2 3 4 5<br />
DCM (g/L)<br />
water research 44 (2010) 616–624 621<br />
0<br />
0 1 2 3 4 5<br />
DCM (g/L)<br />
2000<br />
0<br />
0 1 2 3 4 5<br />
DCM (g/L)<br />
Fig. 4 – Measured (mark) and modelled (line) K global (mL/g PART) as a function of DCM concentration (g/L). The a coefficient (for<br />
dibenzo(a,h)anthracene, benzo(g,h,i)perylene and indeno(1,2,3,c,d)pyrene), Kpart,KDCM and regression coefficients estimated<br />
thanks to the algorithm are presented on each graph. The corresponding logarithmic values and standard deviation for Kpart<br />
and KDCM are presented in Table 1.<br />
fractionation, the authors measured some K DCM values ranging<br />
from below 1000–80 000 mL/g OC (g of organic carbon). Since<br />
organic carbon usually represents approximately 25% of the<br />
total matter in sludge (Dignac et al., 2000), these values are<br />
widespread from below 250 to 20 000 mL/gDCM. This range is in<br />
accordance with our estimation of 380–6607 mL/gDCM (Table 1).<br />
The common and general correlation between the logarithm<br />
of equilibrium constants and hydrophobicity of PAHs<br />
(Karickhoff et al., 1979; Poerschmann and Kopinke, 2001) was<br />
observed for both Kpart and KDCM (Fig. 5). This demonstrates
622<br />
log K part , log K DCM<br />
7<br />
6<br />
5<br />
4<br />
3<br />
2<br />
y = 0.308x + 2.13<br />
y = 0.348x + 1.61<br />
1<br />
4 4.5 5 5.5 6 6.5 7<br />
log K ow<br />
Fig. 5 – log Kpart (black rounds) and log KDCM (grey rounds)<br />
as a function of PAHs log K ow.<br />
that PAHs sorption to particles and to DCM may be driven by<br />
hydrophobic interactions. Moreover, the linear regressions<br />
revealed very similar slopes (0.31 and 0.35) but different<br />
origins (2.13 and 1.61), resulting in different values (Fig. 5).<br />
Indeed, the PAHs affinity for particles was found to be stronger<br />
than for DCM. This finding can be related to the compartments<br />
characteristics since equilibrium constants reflect the<br />
affinity of PAHs for the corresponding compartment. In this<br />
work, the compartments were differentiated according to<br />
their physical criteria, as defined in the 2.1. Section: the<br />
density criteria imposed by centrifugation, followed by the<br />
size criteria of filtration. Other physical properties such as<br />
hydrophobicity, surface charge and specific surface area<br />
differentiate particles and DCM. In addition of physical properties,<br />
Bougrier et al. (2008) demonstrated that biochemical<br />
composition (chemical oxygen demand, carbohydrates and<br />
proteins content) is different in particles and DCM. These<br />
properties are influenced by sludge history in terms of<br />
substrate (wastewater), involved microbiological population,<br />
and process operating parameters. For example, high sludge<br />
retention time in activated sludge process reduces particles<br />
specific surface area (Liss et al., 2002) and leads to more<br />
hydrophobic particles (Liao et al., 2001) with higher lipid<br />
content (Liss et al., 2002). Quantifying the influence of physical<br />
and biochemical sludge characteristics on the sorption equilibrium<br />
constants of PAHs would be very interesting.<br />
3.4. Assessment of PAHs compartment distribution<br />
Kpart and KDCM determination and dependence to hydrophobicity<br />
provided the parameters needed to simulate PAHs<br />
distribution between the three sludge compartments.<br />
Aqueous and freely dissolved percentages were simulated in<br />
two theoretical and extreme cases: for a weakly hydrophobic<br />
PAH and for a strongly hydrophobic one (Fig. 6).<br />
The simulation demonstrated that freely dissolved fraction<br />
does not exceed 0.4% of the total concentration for a very<br />
hydrophobic compound and 2.5% for a weakly hydrophobic<br />
one. This fraction is slightly influenced by DCM concentration<br />
in both cases.<br />
On the contrary, aqueous fraction of both PAHs appeared<br />
to strongly depend on DCM concentration. This result is of<br />
water research 44 (2010) 616–624<br />
%<br />
35<br />
30<br />
25<br />
20<br />
15<br />
10<br />
5<br />
0<br />
0 2 4<br />
6 8<br />
DCM (g/L)<br />
Fig. 6 – Percentage of free (plots) and aqueous (linked plots)<br />
concentrations modelled for a theoretical weakly<br />
hydrophobic PAH (log K ow [ 4.2, square) and for<br />
a theoretical very hydrophobic PAH (log Kow [ 6.8, triangle)<br />
as a function of dissolved and colloidal matter<br />
concentration, particulate concentration being fixed at<br />
10 g/L. For this simulation, Kpart and KDCM of both PAHs<br />
were estimated thanks to the relationship presented in<br />
Fig. 5. The reversion point is located at 6.01 gDCM/L, which<br />
corresponds to 25.5% of aqueous PAHs. Over this value,<br />
aqueous fraction is slightly higher for the very<br />
hydrophobic PAH than for the weakly hydrophobic one.<br />
great concern since aqueous fraction is usually equated to<br />
bioavailable fraction, as previously mentioned. It means that<br />
low DCM concentrations favour higher percentage in aqueous<br />
phase for weakly hydrophobic PAHs than for very hydrophobic<br />
ones, whereas reasonable high concentrations<br />
discriminate less PAHs. The extrapolation of the model to<br />
extremely concentrated conditions ([DCM] > 6 g/L) shows that<br />
the aqueous concentration of a very hydrophobic PAH could<br />
equal and even exceed a weakly hydrophobic one (Fig. 6). In<br />
the usual bioavailability hypothesis ‘‘aqueous is bioavailable’’,<br />
this would mean that a very hydrophobic PAH could be more<br />
bioavailable than a weakly hydrophobic one. This case is<br />
really atypical, since micropollutants are usually assumed to<br />
be less bioavailable when they are more hydrophobic (Mackay<br />
and Fraser, 2000). It is an important insight derived from the<br />
three-compartment approach. The reversion point would be<br />
located at [DCM] ¼ 6.0 g/L for the studied activated sludge, as<br />
particulate matter concentration was fixed at 10 g/L for the<br />
simulation. However, it has to be reminded that the model<br />
was established for DCM concentrations up to 4.5 g/L. Moreover,<br />
DCM at the reversion point would represent 37.5% of the<br />
total matter, which is much higher than common values<br />
(Bougrier et al., 2008). As a conclusion, the latter case is little<br />
probable in real systems, but it could be interesting to create it<br />
in laboratory investigations.<br />
Apart from the descriptive issue previously discussed, such<br />
simulations constitute a new tool for micropollutants physical<br />
fate assessment. They will allow to predict the fate of PAHs<br />
throughout the separation steps encountered in wastewater
treatment plants. In the example of a settler, the flux of sorbedto-particles<br />
PAHs in solid effluent (for agricultural disposal)<br />
can be quantified as well as the flux of sorbed-to-DCM and<br />
dissolved PAHs carried by clarified water (discharged into<br />
surface water). Indeed, Katsoyiannis and Samara (2007)<br />
observed an influence of DCM concentration on micropollutants<br />
fate, without quantifying this effect. The threecompartment<br />
approach could provide the missing data to<br />
quantitatively take this interaction between micropollutants<br />
and DCM into account. As a result, the methodology presented<br />
here constitutes a real improvement in the assessment of<br />
micropollutants fate throughout treatment plants and of their<br />
environmental discharges.<br />
3.5. Towards bioavailability<br />
Quantifying PAHs distribution in sludge samples has an<br />
important outcome concerning its consequences on<br />
bioavailability topic. Indeed, it may be useful to verify the<br />
usual hypothesis of identity between aqueous and bioavailable<br />
fractions (Artola-Garicano et al., 2003; Dionisi et al., 2006).<br />
This assumption may be verified by measuring biodegradation<br />
at different aqueous concentrations. Initial biodegradation<br />
rate in batch experiments could be an appropriate<br />
indicator for biodegradation. The assumption would be<br />
confirmed if biodegradation rate correlates with aqueous<br />
concentration. Guidelines for such an experiment are<br />
provided by the present work: it suggests that different<br />
aqueous concentrations of one PAH might be achieved thanks<br />
to different DCM concentrations.<br />
On the other hand, investigation on the link between freely<br />
dissolved and bioavailable fractions of one PAH appears more<br />
difficult, as freely dissolved concentration varied little in the<br />
presented conditions. It means that this link might be investigated<br />
through either the comparison between different<br />
compounds in one matrix or the comparison of one<br />
compound in various matrixes for which this compound has<br />
different affinities.<br />
These guidelines are of great importance for the modeling<br />
of micropollutants biological fate in sludge, since until now,<br />
models have been built on hypotheses which have never been<br />
clearly demonstrated.<br />
4. Conclusions<br />
The three-compartment methodology proposed to investigate<br />
micropollutants sorption in sludge was validated. The results<br />
underlined the fact that a reliable tool for micropollutants<br />
behaviour prediction in sludge should integrate a threecompartment<br />
model since DCM concentration influences<br />
micropollutants distribution in a significant way. Indeed, the<br />
usual biphasic approach was shown to lead to an important<br />
underestimation of equilibrium constants. Despite the lack of<br />
published data about PAH sorption in sludge, the results could<br />
be confronted to a few studies. The confrontation showed the<br />
consistency of equilibrium constant values and of dependency<br />
to PAH hydrophobicity.<br />
Besides, PAH affinities for particles and for DCM could be<br />
compared. Affinity for particles was found to be stronger. This<br />
water research 44 (2010) 616–624 623<br />
may be due to the differences between particles and DCM in<br />
term of physical and chemical characteristics. Further investigation<br />
is necessary to include the influence of both sludge<br />
and micropollutant characteristics in the three-compartment<br />
model. This research, being presently undergone by the<br />
authors, will enable to predict PAHs behaviour in any sludge,<br />
as a function of its characteristics. The future predictive<br />
model would allow to estimate the PAH distribution in any<br />
sludge in order (i) to determine fluxes throughout physical<br />
separation steps and (ii) to confront PAHs distribution<br />
between compartments to their biodegradation, which might<br />
help to elucidate bioavailability issue.<br />
Acknowledgements<br />
This study was funded by the Institut National de Recherche<br />
Agronomique and the Ministère de l’Enseignement Supérieur<br />
et de la Recherche du Gouvernement Français, France, and the<br />
authors gratefully acknowledge them.<br />
references<br />
Amiri, F., Bornick, H., Worch, E., 2005. Sorption of phenols onto<br />
sandy aquifer material: the effect of dissolved organic matter<br />
(DOM). Water Research 39, 933–941.<br />
Arias-Estevez, M., Fernandez-Gandara, D., Garcia-Falcon, M.S.,<br />
Garcia-Rio, L., Mejuto, J.C., Simal-Gandara, J., 2007. Sorption of<br />
PAHs to colloid dispersions of humic substances in water.<br />
Bulletin of Environmental Contamination and Toxicology 79,<br />
251–254.<br />
Artola-Garicano, E., Borkent, I., Damen, K., Jager, T., Vaes, W.H.J.,<br />
2003. Sorption kinetics and microbial biodegradation activity<br />
of hydrophobic chemicals in sewage sludge: model and<br />
measurements based on free concentrations. Environmental<br />
Science and Technology 37, 116–122.<br />
Bougrier, C., Delgenès, J.P., Carrère, H., 2008. Effects of thermal<br />
treatments on five different waste activated sludge samples<br />
solubilisation, physical properties and anaerobic digestion.<br />
Chemical Engineering Journal 139, 236–244.<br />
Byrns, G., 2001. The fate of xenobiotic organic compounds in<br />
wastewater treatment plants. Water Research 35, 2523–2533.<br />
Carballa, M., Fink, G., Omil, F., Lema, J.M., Ternes, T., 2008.<br />
Determination of the solid-water distribution coefficient (K-d)<br />
for pharmaceuticals, estrogens and musk fragrances in<br />
digested sludge. Water Research 42, 287–295.<br />
Carballa, M., Omil, F., Alder, A.C., Lema, J.M., 2006. Comparison<br />
between the conventional anaerobic digestion of sewage<br />
sludge and its combination with a chemical or thermal pretreatment<br />
concerning the removal of pharmaceuticals and<br />
personal care products. Water Science and Technology 53,<br />
109–117.<br />
Chiou, C.T., Malcolm, R.L., Brinton, T.I., Kile, D.E., 1986. Water<br />
solubility enhancement of some organic pollutants and<br />
pesticides by dissolved humic and fulvic acids. Environmental<br />
Science and Technology 20, 502–508.<br />
Dignac, M.F., Ginestet, P., Rybacki, D., Bruchet, A., Urbain, V.,<br />
Scribe, P., 2000. Fate of wastewater organic pollution during<br />
activated sludge treatment: nature of residual organic matter.<br />
Water Research 34, 4185–4194.<br />
Dionisi, D., Bertin, L., Bornoroni, L., Capodicasa, S., Papini, M.P.,<br />
Fava, F., 2006. Removal of organic xenobiotics in activated<br />
sludges under aerobic conditions and anaerobic digestion of
624<br />
the adsorbed species. Journal of Chemical Technology and<br />
Biotechnology 81, 1496–1505.<br />
Fountoulakis, M.S., Stamatelatou, K., Batstone, D.J., Lyberatos, G.,<br />
2006. Simulation of DEHP biodegradation and sorption during<br />
the anaerobic digestion of secondary sludge. Water Science<br />
and Technology 54, 119–128.<br />
Hickman, Z.A., Swindell, A.L., Allan, I.J., Rhodes, A.H., Hare, R.,<br />
Semple, K.T., Reid, B.J., 2008. Assessing biodegradation<br />
potential of PAHs in complex multi-contaminant matrices.<br />
Environmental Pollution 156, 1041–1045.<br />
Holbrook, R.D., Love, N.G., Novak, J.T., 2004. Investigation of<br />
sorption behavior between pyrene and colloidal organic<br />
carbon from activated sludge processes. Environmental<br />
Science and Technology 38, 4987–4994.<br />
Hung, N.V., Tateda, M., Ike, M., Fujita, M., Tsunoi, S., Tanaka, M.,<br />
2004. Sorption of biodegradation end products of nonylphenol<br />
polyethoxylates onto activated sludge. Journal of<br />
Environmental Sciences-China 16, 564–569.<br />
Ivashechkin, P., Corvini, P.F.X., Dohmann, M., 2004. Behaviour of<br />
endocrine disrupting chemicals during the treatment of<br />
municipal sewage sludge. Water Science and Technology 50,<br />
133–140.<br />
Karickhoff, S.W., Brown, D.S., Scott, T.A., 1979. Sorption of<br />
hydrophobic pollutants on natural sediments. Water Research<br />
13, 241–248.<br />
Katsoyiannis, A., Samara, C., 2007. The fate of dissolved<br />
organic carbon (DOC) in the wastewater treatment process<br />
and its importance in the removal of wastewater<br />
contaminants. Environmental Science and Pollution<br />
Research 14, 284–292.<br />
Katsoyiannis, A., Zouboulis, A., Samara, C., 2006. Persistent<br />
organic pollutants (POPs) in the conventional activated sludge<br />
treatment process: model predictions against experimental<br />
values. Chemosphere 65, 1634–1641.<br />
Kordel, W., Hennecke, D., Franke, C., 1997. Determination of the<br />
adsorption-coefficients of organic substances on sewage<br />
sludges. Chemosphere 35, 107–119.<br />
Krauss, M., Wilcke, W., 2005. Persistent organic pollutants in soil<br />
density fractions: distribution and sorption strength.<br />
Chemosphere 59, 1507–1515.<br />
Langford, K.H., Scrimshaw, M.D., Birkett, J.W., Lester, J.N., 2005.<br />
The partitioning of alkylphenolic surfactants and<br />
polybrominated diphenyl ether flame retardants in activated<br />
sludge batch tests. Chemosphere 61, 1221–1230.<br />
Lee, C., Kuo, L., 1999. Quantification of the dissolved organic<br />
matter effect on the sorption of hydrophobic organic<br />
pollutant: application of an overall mechanistic sorption<br />
model. Chemosphere 38, 807–821.<br />
Liao, B.Q., Allen, D.G., Droppo, I.G., Leppard, G.G., Liss, S.N., 2001.<br />
Surface properties of sludge and their role in bioflocculation<br />
and settleability. Water Research 35, 339–350.<br />
Liss, S.N., Liao, B.Q., Droppo, I.G., Allen, D.G., Leppard, G.G., 2002.<br />
Effect of solids retention time on floc structure. Water Science<br />
and Technology 46, 431–438.<br />
Mackay, D., Fraser, A., 2000. Bioaccumulation of persistent<br />
organic chemicals: mechanisms and models. Environmental<br />
Pollution 110, 375–391.<br />
Marquardt, D., 1963. An algorithm for least-squares estimation of<br />
nonlinear parameters. SIAM Journal of Applied Mathematics<br />
11, 431–441.<br />
Oleszczuk, P., 2007. Investigation of potentially bioavailable<br />
and sequestrated forms of polycyclic aromatic<br />
water research 44 (2010) 616–624<br />
hydrocarbons during sewagesludgecomposting.<br />
Chemosphere 70, 288–297.<br />
Oleszczuk, P., 2008. Tenax-TA extraction as predictor for free<br />
available content of polycyclic aromatic hydrocarbons (PAHs)<br />
in composted sewage sludges. Journal of Environmental<br />
Monitoring 10, 883–888.<br />
Perminova, I.V., Grechishcheva, N.Y., Kovalevskii, D.V.,<br />
Kudryavtsev, A.V., Petrosyan, V.S., Matorin, D.N., 2001.<br />
Quantification and prediction of the detoxifying properties of<br />
humic substances related to their chemical binding to<br />
polycyclic aromatic hydrocarbons. Environmental Science and<br />
Technology 35, 486–490.<br />
Poerschmann, J., Kopinke, F.D., 2001. Sorption of very<br />
hydrophobic organic compounds (VHOCs) on dissolved humic<br />
organic matter (DOM). 2. Measurement of sorption and<br />
application of a Flory–Huggins concept to interpret the data.<br />
Environmental Science and Technology 35, 1142–1148.<br />
Reid, B.J., Stokes, J.D., Jones, K.C., Semple, K.T., 2000.<br />
Nonexhaustive cyclodextrin-based extraction technique for<br />
the evaluation of PAH bioavailability. Environmental Science<br />
and Technology 34, 3174–3179.<br />
Schrap, S.M., Opperhuizen, A., 1992. On the contradictions<br />
between experimental sorption data and the sorption<br />
partitioning model. Chemosphere 24, 1259–1282.<br />
Semple, K.T., Doick, K.J., Jones, K.C., Burauel, P., Craven, A.,<br />
Harms, H., 2004. Defining bioavailability and bioaccessibility<br />
of contaminated soil and sediment is complicated.<br />
Environmental Science and Technology 38, 228A–231A.<br />
Semple, K.T., Doick, K.J., Wick, L.Y., Harms, H., 2007. Microbial<br />
interactions with organic contaminants in soil: definitions,<br />
processes and measurement. Environmental Pollution 150,<br />
166–176.<br />
Tang, J.X., Liste, H.H., Alexander, M., 2002. Chemical assays of<br />
availability to earthworms of polycyclic aromatic<br />
hydrocarbons in soil. Chemosphere 48, 35–42.<br />
Ternes, T.A., Herrmann, N., Bonerz, M., Knacker, T., Siegrist, H.,<br />
Joss, A., 2004. A rapid method to measure the solid-water<br />
distribution coefficient (K-d) for pharmaceuticals and musk<br />
fragrances in sewage sludge. Water Research 38, 4075–4084.<br />
Trably, E., Delgenes, N., Patureau, D., Delgenes, J.P., 2004.<br />
Statistical tools for the optimization of a highly reproducible<br />
method for the analysis of polycyclic aromatic hydrocarbons<br />
in sludge samples. International Journal of Environmental<br />
Analytical Chemistry 84, 995–1008.<br />
Urase, T., Kikuta, T., 2005. Separate estimation of adsorption and<br />
degradation of pharmaceutical substances and estrogens in<br />
the activated sludge process. Water Research 39, 1289–1300.<br />
van Straalen, N.M., Donker, M.H., Vijver, M.G., van Gestel, C.A.M.,<br />
2005. Bioavailability of contaminants estimated from uptake rates<br />
into soil invertebrates. Environmental Pollution 136, 409–417.<br />
Vinken, R., Hollrigl-Rosta, A., Schmidt, B., Schaffer, A., Corvini, P.F.X.,<br />
2004. Bioavailability of a nonylphenol in dependence on the<br />
association to dissolved humic substances. Water Science and<br />
Technology 50, 277–283.<br />
Yamamoto,H.,Liljestrand,H.M.,Shimizu,Y.,Morita,M.,2003.Effects<br />
of physical–chemical characteristics on the sorption of selected<br />
endocrine disruptors by dissolved organic matter surrogates.<br />
Environmental Science and Technology 37, 2646–2657.<br />
Zhou,J.L.,Liu,R.,Wilding,A.,Hibberd,A.,2007.Sorptionof<br />
selected endocrine disrupting chemicals to different aquatic<br />
colloids. Environmental Science and Technology 41,<br />
206–213.
Removal of micropollutants and reduction of biological<br />
activity in a full scale reclamation plant using ozonation<br />
and activated carbon filtration<br />
J. Reungoat a, *, M. Macova b , B.I. Escher b , S. Carswell b,c , J.F. Mueller b , J. Keller a<br />
a<br />
The University of Queensland, Advanced Water Management Centre (AWMC), Qld 4072, Australia<br />
b<br />
The University of Queensland, National Research Centre for Environmental Toxicology (EnTox), QLD 4108, Australia<br />
c<br />
Queensland Health Forensic and Scientific Services, Organics Laboratory, QLD 4108, Australia<br />
article info<br />
Article history:<br />
Received 12 May 2009<br />
Received in revised form<br />
14 September 2009<br />
Accepted 21 September 2009<br />
Available online 1 October 2009<br />
Keywords:<br />
Water reuse<br />
Oxidation<br />
Adsorption<br />
Pesticides<br />
Pharmaceuticals<br />
Bioassays<br />
water research 44 (2010) 625–637<br />
Available at www.sciencedirect.com<br />
journal homepage: www.elsevier.com/locate/watres<br />
abstract<br />
* Corresponding author. Tel.: þ61 734466251; fax: þ61 733654726.<br />
E-mail address: j.reungoat@awmc.uq.edu.au (J. Reungoat).<br />
0043-1354/$ – see front matter ª 2009 Elsevier Ltd. All rights reserved.<br />
doi:10.1016/j.watres.2009.09.048<br />
Pharmaceutical compounds are found in secondary treated effluents up to mgL -1 levels and<br />
therefore discharged into surface waters. Since the long term effects of these compounds<br />
on the environment and human health are, to date, largely unknown, implementation of<br />
advanced treatment of wastewaters is envisaged to reduce their discharge. This is of<br />
particular relevance where surface waters are used as drinking water sources and when<br />
considering indirect potable reuse. This study aimed at assessing the removal of organic<br />
micropollutants and the concurrent reduction of their biological activity in a full scale<br />
reclamation plant treating secondary effluent. The treatment consists of 6 stages: denitrification,<br />
pre-ozonation, coagulation/flocculation/dissolved air flotation and filtration<br />
(DAFF), main ozonation, activated carbon filtration and final ozonation for disinfection. For<br />
that purpose, representative 24-hour composite samples were collected after each stage.<br />
The occurrence of 85 compounds was monitored by LC/MS-MS. A battery of 6 bioassays<br />
was also used as a complementary tool to evaluate non-specific toxicity and 5 specific toxic<br />
modes of action. Results show that, among the 54 micropollutants quantified in the<br />
influent water, 50 were removed to below their limit of quantification representing more<br />
than 90% of concentration reduction. Biological activity was reduced, depending on the<br />
specific response that was assessed, from a minimum of 62% (AhR response) to more than<br />
99% (estrogenicity). The key processes responsible for the plant’s performances were the<br />
coagulation/flocculation/DAFF, main ozonation and activated carbon filtration. The effect<br />
of these 3 processes varied from one compound or bioassay to another but their combination<br />
was almost totally responsible for the overall observed reduction. Bioassays yielded<br />
complementary information, e.g. estrogenic compounds were not detected in the<br />
secondary effluent by chemical analysis, but the samples had an estrogenic effect. The<br />
main ozonation formed oxidation by-products of the organic micropollutants but<br />
decreased the level of non-specific toxicity and other specific toxic modes of action,<br />
demonstrating that the mixture of oxidation by-products was less potent than the mixture<br />
of the parent compounds for the considered effects.<br />
ª 2009 Elsevier Ltd. All rights reserved.
626<br />
1. Introduction<br />
In the last decade, numerous studies around the world have<br />
demonstrated the presence of pharmaceutical compounds in<br />
domestic wastewaters. These compounds are removed to<br />
different degrees by the commonly used biological activated<br />
sludge processes. While some compounds (e.g. ibuprofen,<br />
paracetamol) are effectively removed, others (e.g. carbamazepine,<br />
diclofenac) are barely affected by the biological treatment<br />
(Onesios et al., 2009). As a result, some pharmaceutical<br />
compounds are released into the environment, contaminating<br />
surface waters. This situation is of concern as these<br />
compounds have been designed to affect biological systems<br />
and long term exposure effects on the biological elements in<br />
the environment are still largely unknown. Human health is<br />
also at risk when treated wastewater is discharged to water<br />
bodies that are used as drinking water sources or considered<br />
for indirect potable reuse. Therefore, advanced treatment of<br />
wastewaters has to be envisaged to reduce the load of these<br />
compounds in discharged effluents.<br />
Several technologies have proven to be effective in<br />
removing pharmaceutical compounds from water: activated<br />
carbon adsorption (Nowotny et al., 2007; Snyder et al., 2007;<br />
Ternes et al., 2002; Westerhoff et al., 2005; Yu et al., 2008),<br />
ozonation and advanced oxidation processes (Esplugas et al.,<br />
2007; Huber et al., 2003; Huber et al., 2005; Kim et al., 2008;<br />
Nakada et al., 2007; Ternes et al., 2003; Zwiener and Frimmel,<br />
2000) and membrane filtration (Kimura et al., 2004; Snyder<br />
et al., 2007; Yoon et al., 2007). Activated carbon adsorption and<br />
ozonation are considered to be economically feasible for<br />
tertiary treatment of wastewaters by Joss et al. (2008). Nevertheless,<br />
none of these processes can remove all the<br />
compounds of concern and to date, no extensive study has<br />
been carried out on the efficiency of the sequential combination<br />
of these two processes for the removal of micropollutants<br />
from secondary treated wastewater.<br />
Chemical analysis of micropollutants in water has gone<br />
through major developments in recent years and today’s<br />
analytical techniques can measure some compounds at<br />
concentrations as low as a few ng L 1 . Chemical analysis is<br />
powerful to measure the concentration of targeted<br />
compounds but does not cover the whole range of chemicals<br />
that might be present in the effluent of a wastewater treatment<br />
plant. Analytical techniques usually focus on pharmaceutically<br />
active parent compounds whereas a variable<br />
fraction is excreted from the human body as metabolites.<br />
Moreover, biological treatment of wastewater may form yet<br />
unidentified compounds through biodegradation reactions;<br />
e.g. 17b-estradiol is degraded to estrone (Ternes et al., 1999).<br />
Subsequent chemical treatments such as ozonation can also<br />
form by-products. Chemical analysis of a limited suite of<br />
compounds does not allow assessment of the potential biological<br />
adverse effects of the wastewater as the cumulative<br />
effect of the mixture of chemicals that may be present cannot<br />
be easily integrated. Bioassays have been utilised as complementary<br />
monitoring tools for the assessment of possible biological<br />
effects of chemicals that are present in a particular<br />
water sample (Escher et al., 2008b; Muller et al., 2007). These<br />
bioanalytical tools are designed to quantify non-specific<br />
water research 44 (2010) 625–637<br />
toxicity or particular toxic modes of action (e.g. estrogenicity,<br />
genotoxicity, phytotoxicity) induced by a sample on a biological<br />
organism or a biological process. To date, bioassays have<br />
not been widely used in the evaluation of water treatment<br />
processes and have rarely been accompanied with chemical<br />
analysis (Macova et al., 2010).<br />
The present study assessed the removal of selected<br />
micropollutants (pharmaceuticals and pesticides) and the<br />
decrease of biological activity along a full scale water reclamation<br />
plant using ozonation and activated carbon adsorption<br />
to treat a secondary effluent. This aimed at identifying the<br />
key steps in the treatment process as well as the key operating<br />
parameters. Results of chemical analysis and bioassays were<br />
compared to determine whether the use of both techniques<br />
was redundant or allowed a deeper understanding of the<br />
performances of individual treatment processes. The performance<br />
and relevance of the treatment train is further discussed<br />
in the context of indirect potable reuse.<br />
2. Materials and methods<br />
2.1. South Caboolture Water Reclamation Plant<br />
The South Caboolture Water Reclamation Plant was designed<br />
to reduce riverine pollution from the 40,000 population<br />
equivalent wastewater treatment plant and to provide recycled<br />
water to industry and community consumers. Whilst the<br />
plant provides water for non-potable applications, it has been<br />
designed to meet drinking water standards. The treatment<br />
process detailed in Fig. 1 incorporates biological denitrification,<br />
pre-ozonation, coagulation/flocculation/dissolved air<br />
flotation-sand filtration (DAFF), main ozonation, biological<br />
activated carbon filtration and final ozonation for disinfection.<br />
The activated carbon was renewed in March 2008 after 9 years<br />
of service, its adsorption capacity was assumed not to be<br />
exhausted at the time sampling took place, 4 months later.<br />
van Leeuwen et al. (2003) published more details on the<br />
process and its performances.<br />
2.2. Sample collection<br />
Four sets of samples were collected over winter 2008 under<br />
dry weather conditions including three during week days and<br />
one during the weekend (11-07-08, 22-07-08, 27-07-08 and 06-<br />
08-08). Water temperature across the plant was 22 2 C and<br />
pH was 7.0 0.5. Samples were collected at 7 sampling points<br />
along the treatment train, labelled S1 to S7 on Fig. 1, in order to<br />
evaluate the performance of individual treatment steps. As<br />
the flow rate in the reclamation plant is constant (8 megalitres<br />
per day), representative samples were collected as time<br />
proportional 24-h composites. For S1 and S7, refrigerated<br />
auto-samplers collected 200 mL every 15 minutes; for the<br />
remaining sampling locations, a continuous flow pump<br />
collected samples at a flow rate of 7 mL min -1 . At each point,<br />
samples were collected into a glass bottle pre-washed with<br />
MilliQ water and HPLC grade acetone. The samples were<br />
protected from light and refrigerated during collection. For<br />
micropollutant analysis, 1 L was transferred into solvent
washed amber glass bottles and preserved with sodium thiosulfate.<br />
For the bioassays, 2 L were transferred into solvent<br />
washed amber glass bottles and hydrochloric acid (36%) was<br />
added to a final concentration of 5 mM for preservation. For<br />
dissolved organic carbon (DOC) measurements, 100 mL were<br />
filtered through a 0.45 mm PTFE membrane and collected in<br />
plastic (HDPE) bottles rinsed with the sample beforehand. All<br />
samples were transported on ice and stored frozen or at 4 C.<br />
2.3. Chemical analyses<br />
2.3.1. Dissolved organic carbon<br />
The dissolved organic carbon (DOC) was measured as nonpurgable<br />
organic carbon (NPOC) with an Analytik Jena multi<br />
N/C 3100 instrument at the Advanced Water Management<br />
Centre (AWMC). For each sample, 2d3 replicates were<br />
measured, giving a relative standard deviation of less than 3%.<br />
2.3.2. Micropollutants<br />
Micropollutant analysis was carried out by Queensland Health<br />
Forensic and Scientific Services (QHFSS, accredited by the<br />
National Association of Testing Authorities, Australia, and ISO<br />
9001 certified). The method consisted of solid phase extraction<br />
(SPE), concentration and quantification by liquid chromatography<br />
coupled with tandem mass spectrometry (LC/MSdMS).<br />
This method, described in details in the supporting information<br />
SI 1, allowed the quantification of 85 compounds (listed in<br />
Table SI 3 in the supporting information SI 2) selected on the<br />
water research 44 (2010) 625–637 627<br />
Fig. 1 – South Caboolture Water Reclamation Plant process (adapted from van Leeuwen et al. (2003)). S1 to S7: sampling<br />
locations. HRT: hydraulic retention time. SRT: sludge retention time.<br />
basis of quantity of usage of the particular compounds, their<br />
potential toxicity and their resistance to degradation. Their<br />
limit of quantification (LOQ) was 0.01 mgL -1 in most cases.<br />
Concentrations were calculated using an internal calibration<br />
method. Results were not corrected for recovery efficiency of<br />
the extraction method. Recoveries of single compounds are<br />
assumed to be consistent along the treatment as the matrix is<br />
similar therefore; calculation of the removal efficiency is not<br />
affected. Indeed, Gros et al. (2009) determined the recoveries<br />
of 73 pharmaceutical compounds by SPE in raw wastewater<br />
and treated effluent and their results show that the recovery<br />
for a given compound generally vary by less than 20% from<br />
one matrix to the other despite the fact that these matrices are<br />
very different. The standard deviation of the method is 20%<br />
(Table SI 1 in the supporting information SI 1). When the<br />
reported outlet concentration was below the LOQ of the<br />
compound, removal efficiency was calculated as a minimum<br />
value using the LOQ as outlet concentration.<br />
2.3.3. Bioassays<br />
Six bioassays, discussed in details in Macova et al. (2010), were<br />
applied to the samples collected in the study. Baseline toxicity<br />
was determined with the non-specific Vibrio fischeri bioluminescence<br />
inhibition test. Five additional specific toxic modes<br />
of action were also evaluated: estrogenic activity (E-SCREEN<br />
assay), arylhydrocarbon receptor response (CAFLUX assay),<br />
neurotoxicity (acetylcholinesterase inhibition assay), phytotoxicity<br />
(PSII inhibition I-PAM assay) and genotoxicity (umuC
628<br />
assay). Water samples were extracted by SPE using Oasis HLB<br />
cartridges. Full dose response curves were determined for<br />
a serial dilution of the extract for each bioassay. Results were<br />
expressed as toxic equivalent concentrations (TEQ) except for<br />
the umuC assay. The TEQ represents the concentration of<br />
a given reference compound that would be required to<br />
produce the same effect as the mixture of the various<br />
compounds in the sample. When the outlet TEQ was below<br />
the LOQ of the bioassay, removal efficiency was calculated as<br />
a minimum value using the LOQ as outlet TEQ. In the umuC<br />
assay, the response is determined as an induction ration (IR),<br />
an IR 1.5 is considered genotoxic. For genotoxic samples,<br />
ECIR1.5 corresponds to how many times the sample must be<br />
concentrated or diluted to elicit an IR of 1.5. Results are<br />
expressed as 1/ECIR1.5 therefore a higher number represents<br />
a higher genotoxic effect.<br />
3. Results and discussion<br />
3.1. Micropollutants<br />
In the reclamation plant’s influent, 54 of the 85 targeted<br />
compounds had a median concentration above their LOQ<br />
confirming that conventional activated sludge treatment does<br />
not completely remove these micropollutants from wastewater<br />
(Fig. 2). The concentrations ranged from 0.01 to<br />
2.10 mgL -1 with the exception of gabapentin, which was<br />
consistently found at higher concentrations ranging from<br />
5.60–6.50 mgL -1 (data provided in the supporting information<br />
in Table SI 3). The factor between the minimum and the<br />
maximum concentrations measured for each individual<br />
compound was generally close to or lower than 2, with<br />
a maximum of 3.6 observed for iopromide. No clear pattern<br />
Fig. 2 – Number of compounds quantified and dissolved<br />
organic carbon (DOC) after indicated stage along the<br />
treatment train. Bars represent the number of compounds<br />
(left y-axis) with a median concentration above the LOQ<br />
(4 samples). Dots represent DOC on two different sampling<br />
days (right y-axis). Denit: denitrification; Pre-O3: preozonation;<br />
DAFF: dissolved air flotation-sand filtration;<br />
Main O3: main ozonation; AC: activated carbon; Fin O3:<br />
final ozonation.<br />
water research 44 (2010) 625–637<br />
could be distinguished between the different sampling days.<br />
The increase or decrease of single compound concentrations<br />
from one day to another appeared to be random, even when<br />
comparing the sample collected during the weekend to<br />
samples collected during week days. Twenty-five compounds<br />
had an influent median concentration above 0.10 mgL –1 , their<br />
fate is further detailed hereafter. The removal efficiency of<br />
these compounds was determined in each treatment step<br />
except when the concentration before treatment was lower<br />
than ten times the LOQ and below LOQ after treatment. This<br />
criterion was used to allow the determination of removals up<br />
to 90% in any case. The DOC was measured for two sets of<br />
samples (22-07-08 and 06-08-08), and varied from 14.2 to<br />
19.7 mg L -1 in the influent water. The following paragraphs<br />
discuss the performance of each individual stage of the<br />
treatment by comparing concentrations of the selected<br />
compounds in the inlet and outlet water for the considered<br />
process.<br />
3.1.1. Denitrification<br />
The number of compounds above LOQ was unchanged after<br />
denitrification and the concentrations of the 25 selected<br />
compounds decreased by less than 20% (with the exception of<br />
atenolol, 38%). To our knowledge, the removal of micropollutants<br />
by denitrifying bacteria has not been specifically<br />
studied to date. In the present case, methanol is added to the<br />
water prior to the denitrification reactor and the bacteria use it<br />
as the primary electron donor which may prevent or reduce<br />
the degradation of micropollutants. The observed increase in<br />
DOC after the denitrification was probably due to residual<br />
methanol.<br />
3.1.2. Pre-ozonation<br />
The number of compounds quantified was stable through the<br />
pre-ozonation process (Fig. 2) and the concentration of the 25<br />
selected compounds decreased by less than 30%. The ozone<br />
dose applied in the pre-ozonation step is approximately<br />
2mgL -1 and the DOC was approximately 20 mg L -1 corre-<br />
-1<br />
sponding to an ozone dose of 0.1 mgO3 mgDOC.<br />
In such conditions,<br />
ozone is rapidly consumed by the organic matter (Buffle<br />
et al., 2006a,b) therefore micropollutants oxidation is limited.<br />
Ozonation has been proved to be very effective for oxidising<br />
various micropollutants in secondary treated wastewaters<br />
(Hollender et al., 2009; Huber et al., 2005; Snyder et al., 2006;<br />
Ternes et al., 2003; Wert et al., 2009) but with higher ozone<br />
-1<br />
doses of at least 0.25 to 0.50 mgO3 mgDOC. Snyder et al. (2006)<br />
also investigated a full scale wastewater treatment plant<br />
using pre-ozonation with a low ozone dose (not specified)<br />
prior to biological activated carbon filtration, and similarly,<br />
observed removal of micropollutants inferior to 30%.<br />
3.1.3. Coagulation/flocculation/DAFF<br />
In the coagulation/flocculation/DAFF process, the negatively<br />
charged colloids are neutralised by the addition of a chemical<br />
agent (aluminium sulphate) which causes them to aggregate<br />
in floccs (flocculation) which are subsequently removed by<br />
dissolved air flotation and sand filtration. The main purpose of<br />
this stage is to decrease the DOC, which was achieved effectively<br />
as demonstrated by the 40–50% DOC reduction (Fig. 2).<br />
This reduction of the DOC is important to achieve maximal
performance of the main ozonation process as discussed<br />
above. After the DAFF, 53 compounds were quantified and the<br />
concentrations of the 25 selected compounds decreased by<br />
less than 20% except for atenolol (42%), caffeine (29%), gabapentin<br />
(44%), gemfibrozil (32%) and roxithromycin (37%). It is<br />
also interesting to note that paracetamol concentration<br />
increased by 44% in the process, which could not be explained.<br />
No literature reporting investigations of the removal of<br />
micropollutants from treated wastewaters by coagulationflocculation<br />
could be identified. However, several studies on<br />
lab-scale and full scale drinking water treatment systems<br />
have reported removals lower than 50% for several pharmaceuticals<br />
and pesticides by such solids removal processes<br />
(Adams et al., 2002; Ternes et al., 2002; Thuy et al., 2008; Vieno<br />
et al., 2006; Westerhoff et al., 2005). Suarez et al. (2009) also<br />
showed similar limited removal in hospital wastewaters.<br />
Nevertheless, Suarez et al. (2009) and Westerhoff et al. (2005),<br />
showed that removals of up to 80% can be achieved for highly<br />
hydrophobic compounds (i.e logKow > 6) suggesting that the<br />
micropollutants removal during coagulation-flocculation<br />
occurs via hydrophobic interactions with neutral particles.<br />
Most of the pharmaceuticals and pesticides measured in this<br />
study are hydrophilic or of moderate hydrophobicity with<br />
logKow < 4(Table SI 3 in supporting information SI 2); therefore<br />
little removal can be expected in the coagulation/flocculation/<br />
DAFF stage. Among the exceptions observed, the removal of<br />
gemfibrozil can be explained by its more hydrophobic nature<br />
(logKow ¼ 4.77, EPI SUITE 4.0) but other compounds are very<br />
hydrophilic and no explanation for their removal could be<br />
advanced.<br />
3.1.4. Main ozonation<br />
-1<br />
The main ozonation stage (approximately 0.5 mgO3 mgDOC) decreased the concentration of 26 compounds below LOQ<br />
(Fig. 2). Among the 25 selected compounds, 9 showed<br />
water research 44 (2010) 625–637 629<br />
a reduction of more than 90% and 13 others were reduced by<br />
more than 70% while iopromide and gabapentin were reduced<br />
by 55% (Fig. 3). Removal of naproxen was not determined<br />
because its concentration was lower than ten times its LOQ<br />
before treatment and below LOQ after. The higher removal<br />
efficiency obtained by the main ozonation stage compared<br />
with the pre-ozonation is due to the higher ozone dosage<br />
relative to the DOC content. These results are in agreement<br />
with previous findings (Buffle et al., 2006b; Hollender et al.,<br />
2009; Huber et al., 2005; Snyder et al., 2006; Ternes et al., 2003;<br />
Wert et al., 2009). The reaction of organic compounds with<br />
molecular ozone is selective and only certain groups of<br />
compounds react rapidly, e.g. aliphatic molecules with double<br />
bonds, deprotonated amines and aromatics with an activating<br />
group. Most of the pharmaceuticals and pesticides possess<br />
one of these moieties and would be expected to react with<br />
molecular ozone with high second order rate constants (kO3).<br />
The results show that the removal of micropollutants<br />
increased with increasing value of kO3 but even compounds<br />
weakly reactive with ozone (i.e. iopromide and MCPA; Table 1)<br />
were oxidised. Oxidation in ozonation processes can also<br />
occur via the attack of hydroxyl radicals formed by ozone<br />
decomposition in water. Hydroxyl radicals are highly reactive<br />
with almost any organic molecule; the second order rate<br />
constants of oxidation by hydroxyl radicals are typically in the<br />
10 9 M -1 s -1 range (Table 1). Therefore, the observed removals of<br />
iopromide and MCPA suggest a substantial exposure to<br />
hydroxyl radicals. Indeed, Buffle et al. (2006a,b) showed that<br />
the ozonation of wastewaters leads to high exposure to<br />
hydroxyl radicals; in the first instants of the reaction, the<br />
hydroxyl radical production is equivalent to, or greater than<br />
what can be obtained with advanced oxidation processes. To<br />
our knowledge, gabapentin oxidation by ozone has not been<br />
reported in the literature and no reaction rate data could be<br />
found. Gabapentin possesses both carboxylic and amine<br />
Fig. 3 – Median removal of selected compounds (median of influent concentration > 0.10 mgL -1 ) by the main ozonation stage<br />
and the combination of the main ozonation and the activated carbon (AC) filtration stages. Error bars represent minimum<br />
and maximum removal, no error bar means that the compound was below LOQ after treatment; therefore removal was<br />
calculated as a minimum using the LOQ. C S4: concentration before main ozonation; C S5: concentration after main ozonation;<br />
CS6: concentration after activated carbon filtration.
630<br />
Table 1 – Removal of selected pharmaceuticals in the main ozonation stage and their first order reaction rate constants with<br />
ozone (kO3) and hydroxyl radicals (kHO ). The temperature (T ) and pH given are the experimental conditions used for the<br />
determination of the reaction rates.<br />
Compound Removal kO3 (M 1 s 1 ) pH T ( C) kHO (109.M 1 s 1 ) pH T ( C) Ref.<br />
Iopromide 55% 91% 3.0 10 4 –1.0 10 6<br />
2 20<br />
Trimethoprim >93% 2.7 10 5<br />
7 20 6.5 0.2 7 25<br />
Sulphamethoxazole >93% ~2.5 10 6<br />
7 20 5.5 0.7<br />
5.5 10 5<br />
7 20 7 25<br />
Diclofenac >94% ~10 6<br />
7 20 7.5 1.5<br />
(1.84 0.15) 10 4<br />
6 25 7 25<br />
Paracetamol >96% 4.11<br />
6 k<br />
10 7 25 2.2 0.017 5.5 25<br />
Carbamazepine >98% ~3 10 5<br />
7 20 8.8 1.2 7 25<br />
(7.81 1.31) 10 4<br />
25<br />
2.05 0.14 5 room<br />
a Huber et al. (2003).<br />
b Benitez et al. (2004).<br />
c Benner et al. (2008).<br />
d Song et al. (2008).<br />
e Dodd et al. (2006).<br />
f Rivas et al. (2009).<br />
g Vogna et al. (2004b).<br />
h Andreozzi et al. (2003).<br />
i Andreozzi et al. (2002).<br />
j Vogna et al. (2004a).<br />
k Calculated from Andreozzi et al (2003).<br />
groups, their pKa are 3.89 and 9.56 respectively (Zour et al.,<br />
1992). Therefore gabapentin is present as a zwitterion under<br />
usual wastewater pH conditions. Protonated amines and<br />
deprotonated carboxylic acids are typically not highly reactive<br />
with ozone (Hoigné and Bader, 1983) and gabapentin oxidation<br />
can be expected to be controlled by hydroxyl radical reactions.<br />
As a primary amine, gabapentin has an expected second order<br />
rate constant for the reaction with ozone of about 50 M -1 s -1<br />
which is significantly higher than for iopromide. However, the<br />
similar behaviour of the two substances shows that their<br />
decrease is controlled by hydroxyl radical reactions. The DOC<br />
decreased by less than 10% in the process, showing that<br />
ozonation does not lead to substantial mineralization of dissolved<br />
organic matter.<br />
3.1.5. Activated carbon filtration<br />
Activated carbon filtration further removed the micropollutants<br />
and only 2 compounds were quantified above their<br />
LOQ downstream: roxithromycin (0.01 mgL -1 ) and gabapentin<br />
(0.70 mgL -1 ). The removal efficiency of the compounds having<br />
a median concentration of at least ten times their LOQ prior to<br />
activated carbon filtration was calculated: oxazepam, tramadol<br />
and venlafaxine were removed by more than 90% and<br />
gabapentin was removed by 53%. Adsorption onto activated<br />
carbon also removed 20–30% of the DOC showing a less<br />
effective adsorption of organic matter than of micropollutants.<br />
Previous studies (Ormad et al., 2008; Snyder et al.,<br />
2007; Ternes et al., 2002; Westerhoff et al., 2005) demonstrated<br />
water research 44 (2010) 625–637<br />
that powdered and granular activated carbon can efficiently<br />
remove micropollutants from natural water sources used for<br />
drinking water. To date, the removal of micropollutants from<br />
wastewaters by activated carbon has not been extensively<br />
studied; Snyder et al. (2007) found that a water reuse facility<br />
using granular activated carbon without regular replacement/<br />
regeneration provided little removal. In the present work, the<br />
activated carbon had been replaced about 4 months before the<br />
collection of the samples so it is assumed that its adsorption<br />
capacity was not yet exhausted. Adsorption onto activated<br />
carbon is driven mainly by hydrophobic interactions and<br />
Westerhoff et al. (2005) observed that the trend in the removal<br />
efficiency of micropollutants by activated carbon can be predicted<br />
from logKow values. Gabapentin is zwitterionic at<br />
neutral pH as shown above and much more hydrophilic<br />
(logKow ¼ -1.37) compared to oxazepam, tramadol and venlafaxine<br />
(logKow ¼ 2.32, 3.01 and 3.28 respectively). This<br />
explains the difference observed in the removal efficiencies of<br />
these compounds. It can be expected that, with time, the<br />
adsorption capacity of the activated carbon filter will decrease<br />
and eventually be exhausted while biological activity will<br />
develop in the filter and contribute to biodegradation of the<br />
by-products formed by ozonation (Simpson, 2008).<br />
3.1.6. Final ozonation<br />
In the final effluent, after final ozonation, 4 compounds had<br />
a median concentration above the LOQ: gabapentin (0.45 mgL -1 ),<br />
roxithromycin (0.01 mgL -1 ), DEET (0.03 mgL -1 ) and caffeine<br />
a<br />
b<br />
c<br />
d<br />
a<br />
e<br />
c<br />
d<br />
f<br />
e<br />
a<br />
e<br />
a<br />
g<br />
h<br />
a<br />
i<br />
j
(0.02 mgL -1 ). The median concentrations of caffeine and DEET<br />
are very close to the LOQ and are below the LOQ prior to the<br />
final ozonation stage; therefore one can assume that their<br />
quantification is due to analytical variations rather than an<br />
actual concentration variation. It is difficult to evaluate the<br />
efficiency of the final ozonation in the removal of micropollutants<br />
because they all had concentrations equal or below<br />
the LOQ before the treatment except gabapentin. The median<br />
removal of gabapentin in the final ozonation was only 20%,<br />
which is low compared to the main ozonation, and is due to<br />
-1<br />
the lower ozone dose employed (approximately 0.3 mgO3 mgDOC). The DOC was not removed in this treatment step.<br />
3.1.7. Overall treatment<br />
The full treatment decreased the concentration of 50 of the 54<br />
compounds quantified in the influent water to levels below<br />
LOQ (Fig. 2). Concomitantly, DOC was also reduced by 55–60%<br />
in the effluent compared to the influent water. Overall, among<br />
the 25 selected compounds, 11 were removed by more than<br />
95% and 11 by more than 89%. The median removal of gabapentin<br />
was 86% and the removals of naproxen and iopromide<br />
were not calculated because their concentration was lower<br />
than 10 times their LOQ in the influent and below their LOQ in<br />
the effluent. The 4 remaining compounds were gabapentin<br />
(0.45 mgL -1 ), roxithromycin (0.01 mgL -1 ), DEET (0.03 mgL -1 ) and<br />
caffeine (0.02 mgL -1 ).<br />
The first 3 stages of the treatment train (i.e. denitrification,<br />
pre-ozonation and coagulation/flocculation/DAFF) did not<br />
decrease the number of micropollutants quantified in the<br />
water (Fig. 2). After these 3 stages, the concentrations of the 25<br />
compounds that had an influent median concentration of at<br />
least 0.10 mgL -1 were generally still greater than 50% compared<br />
to the influent concentration (Fig. 4). The main ozonation and<br />
water research 44 (2010) 625–637 631<br />
the activated carbon adsorption played a key role in the<br />
treatment bringing the concentration of 26 and 25 compounds<br />
below LOQ respectively (Fig. 2). The main ozonation generally<br />
decreased the micropollutants to less than 20% of their<br />
influent concentration and activated carbon filtration further<br />
removed the compounds to levels below LOQ except for<br />
gabapentin and roxythromycin (Fig. 4). The combined effects<br />
of the main ozonation and the activated carbon filtration<br />
processes decreased the concentration of 10 of the 25 selected<br />
micropollutants by more than 95% and by more than 89% for<br />
12 of the 15 remaining compounds compared to their<br />
concentration prior to the main ozonation (Fig. 3). Gabapentin<br />
concentration was reduced by 79%. These results show that<br />
ozonation followed by activated carbon filtration is a very<br />
effective combination of processes to remove micropollutants<br />
from secondary treated wastewater. The key steps in the<br />
removal of the DOC were the DAFF and the activated carbon<br />
adsorption. Although the DAFF reduced the concentration of<br />
micropollutants by less than 30%, it also played a key role<br />
indirectly by reducing the DOC which enhanced the performances<br />
of the main ozonation.<br />
3.2. Bioanalytical results<br />
The influent biological activity was higher than the blank<br />
(MilliQ water) in all the bioassays (Table 2). The effect of the<br />
treatment train on each bioassay is discussed individually<br />
hereafter.<br />
3.2.1. Baseline toxicity<br />
The Vibrio fischeri bioluminescence inhibition test is a nonspecific<br />
bacterial toxicity test widely recognised in the field of<br />
ecotoxicology as the standard assay for acute cytotoxicity. The<br />
Fig. 4 – Median relative concentrations of selected compounds (median of influent concentration > 0.10 g L -1 ) after the<br />
dissolved air flotation-sand filtration (DAFF, S4), the main ozonation (S5) and the activated carbon (S6) stages (error bars<br />
represent maximum and minimum values). C is the concentration after the specified treatment step and the reference<br />
concentration, C0, is the concentration in the influent water of the reclamation plant.
632<br />
Table 2 – Maximum, median and minimum biological activity in the influent and effluent of the reclamation plant and<br />
overall maximum, median and minimum decrease observed.<br />
Bioassay Result expression Blank Influent Effluent Decrease (%)<br />
assay reflects the general ‘‘energy status’’ of the bacteria and<br />
can indicate the toxic potency of a broad spectrum of<br />
compounds with different modes of action. Denitrification and<br />
pre-ozonation did have a slight stimulatory effect on the<br />
baseline toxicity whereas the following treatment steps were<br />
able to substantially decrease baseline toxicity (Fig. 5). The TEQ<br />
was reduced to 62, 37 and 21% of the influent water level after<br />
the coagulation/flocculation/DAFF, the main ozonation and<br />
the activated carbon filtration respectively. As is discussed in<br />
more details in Macova et al. (2010), an almost linear correlation<br />
exists between DOC level and TEQ. The slight increase in<br />
TEQ after denitrification cannot be related to the residual<br />
methanol added before this process, which is still partially<br />
present after denitrification as is discussed above, but is likely<br />
to be related to some non-volatile organic chemicals. The 52%<br />
decrease of TEQ in the DAFF stage is accompanied by a 40–50%<br />
reduction in DOC, whereas micropollutants concentrations<br />
were generally reduced by less than 30% in the meantime.<br />
Although the SPE that is performed prior to toxicity testing<br />
should be able to remove a substantial fraction of the DOC and<br />
will remove all of the residual methanol, some DOC, most<br />
likely smaller breakdown products that have similar<br />
Max Med Min Max Med Min Max Med Min<br />
Baseline toxicity Baseline toxicity EqC a (TEQ, mg L 1 ) 0.21 2.9 2.1 2.0 0.72 0.52 0.31 84 78 67<br />
Estrogenicity Estradiol EqC (EEQ, ng L 1 )
Activated carbon filtration reduced the baseline toxicity by<br />
50% and the DOC by 30 to 35%. Activated carbon effectively<br />
adsorbs the more hydrophobic compounds, which is again<br />
consistent with the general trend discussed above that the<br />
more hydrophobic compounds have a higher toxic activity<br />
than the more hydrophilic ones. Based on this fact, identification<br />
of the compounds exhibiting a high toxic activity could<br />
start with the identification of the more hydrophobic<br />
compounds.<br />
The final ozonation did not further reduce the baseline<br />
toxicity compared to after activated carbon filtration. The<br />
effluent TEQ was approximately 80% lower than the<br />
influent TEQ (Fig. 5) and only 2.5 times higher than<br />
the blank (Table 2). This , indicates that the residual toxicity<br />
is of no concern, unless the residual organic chemicals and<br />
organic matter inducing this effect were of very specific<br />
potency. This latter question was tested with a series of<br />
specific endpoints that respond to environmentally relevant<br />
modes of toxic action.<br />
3.2.2. Estrogenic activity<br />
The E-SCREEN assay specifically responds to natural<br />
hormones and other compounds that can mimic the activity<br />
of the female sex hormone estradiol. The estrogenic activity of<br />
the samples is expressed as an estradiol equivalent concentration<br />
(EEQ). The median influent EEQ was 5.8 ng L -1 ; higher<br />
than levels previously reported in South East Queensland.<br />
Most of the effluents from 12 activated sludge wastewater<br />
treatment plants tested by Leusch et al. (2006) had EEQs below<br />
4ngL -1 and sometimes below 1 ng L -1 . Hormones and endocrine<br />
disrupting compounds (EDCs) were not investigated in<br />
this study but an earlier sampling campaign of the secondary<br />
treated wastewater showed that the concentrations of<br />
measured estrogenic compounds (17b-estradiol, 17a-ethynylestradiol,<br />
estrone, estriol, bisphenol A, nonylphenol) were all<br />
below the LOQ of 1 ng L -1 (Table SI 4 in supporting information<br />
SI 3). This demonstrates the relevance of using bioassays as<br />
complementary tools to chemical analysis for the assessment<br />
of water quality and process performances.<br />
Denitrification did not affect the estrogenicity (Fig. 5). Preozonation<br />
with an ozone dose of approximately 0.10 mgO3<br />
-1<br />
mgDOC reduced the EEQ by 34% compared to the influent. This<br />
is higher than the removal previously observed by Snyder<br />
et al. (2006) who measured the EEQ reduction induced by<br />
various ozone doses in treated wastewater with a DOC of<br />
6.38 mg L -1 . They found that an ozone dose of 2.1 mg L -1<br />
1<br />
(0.33 mgO3 mgDOC)<br />
only removed 18% of the EEQ but, with<br />
ozone doses of 3.6 mg L -1 1<br />
(0.56 mgO3 mgDOC)<br />
and above, 90%<br />
or more removal could be achieved. In a recent study on a full<br />
scale ozonation in a Swiss sewage treatment plant (STP), the<br />
dose-dependency of removal of micropollutants yielded<br />
similar results (Escher et al., 2009). While most endpoints<br />
showed a clear dose-dependency of reduction of effects, the<br />
reduction of estrogenicity was already large at low ozone<br />
doses and depended more on the EEQ than on the ozone<br />
dose. When estrogenicity was already below a certain level,<br />
which was very close to the detection limit, the quantification<br />
of further reduction became difficult and prone to large<br />
uncertainty. For the remaining samples, ozone doses of 1.6–<br />
5.3 mg L -1 in the presence of 4.2–6.0 mg L -1 DOC lead to more<br />
water research 44 (2010) 625–637 633<br />
than 90% reduction of estrogenicity. This is consistent with<br />
laboratory experiments that demonstrated that almost all<br />
first-generation transformation products of estrogenic<br />
chemicals had severely decreased estrogenic potency (Lee<br />
et al., 2008). Thus ozonation can be considered as a fairly<br />
selective oxidation, where even low doses selectively target<br />
one of the most environmentally relevant modes of toxic<br />
action, namely estrogenicity.<br />
After the coagulation/flocculation/DAFF stage the EEQ<br />
increased drastically by a median factor of 3.3 compared to the<br />
level prior to treatment. At this treatment step, the concentration<br />
of DOC is greatly reduced (by 40–50%), and there is<br />
a likelihood that the estrogenic chemicals that were bound to<br />
DOC were released during this treatment step. We have<br />
previously observed with another estrogenicity assay that<br />
DOC appears to reduce the bioavailability of estrogens<br />
(B. Escher, unpublished results). Estrogenic chemicals are<br />
typically relatively hydrophobic and bind well to DOC (Neale<br />
et al., 2008). In general DOC is not bioavailable in bioassays<br />
(the discussion on the small breakdown products above is an<br />
exception of this general paradigm) and micropollutants sorbed<br />
to DOC would not be bioavailable either. A large fraction of<br />
the matrix and also the DOC is supposed to be removed by SPE<br />
but, given the color of the extracts, it is possible that<br />
a substantial fraction of larger DOC is co-extracted. In addition,<br />
for the E-SCREEN test, it was demonstrated that the<br />
presence of serum proteins modulates the free and bioavailable<br />
concentration of estrogenic chemicals (Heringa et al.,<br />
2004). This effect was also hydrophobicity dependent and was<br />
much more pronounced for the more hydrophobic octylphenol<br />
than for the less hydrophobic estradiol. Protein binding is<br />
generally less important than binding to DOC or lipids,<br />
therefore while the effect on bioavailability was not very large<br />
for estradiol in the study of Heringa et al. (2004); it might well<br />
be relevant under the conditions of the present study. This<br />
hypothesis needs to be evaluated in the future by exploring<br />
the correlation between size distribution of naturally occurring<br />
DOC and effect on bioavailability, estrogenicity and<br />
toxicity.<br />
The main ozonation reduced the EEQ by a median value of<br />
92 and 95% compared to the level of the reclamation plant’s<br />
influent and to the level before treatment respectively<br />
whereas DOC was not affected. It can be concluded that the<br />
mixture of by-products formed by the oxidation of the estrogenic<br />
compounds by ozone and hydroxyl radicals have<br />
a much lower estrogenic activity than the mixture of parent<br />
compounds, which is consistent with expectations as discussed<br />
above and in Lee et al. (2008).<br />
Activated carbon filtration was able to efficiently adsorb<br />
residual estrogenic compounds and further reduced the EEQ<br />
by another 95% to levels below the detection limit of 0.02 ng L -1<br />
and the final effluent concentration was below the quantification<br />
limit of 0.06 ng L -1 . The overall treatment efficiency for<br />
the removal of estrogenic activity was greater than 99%. This<br />
is in good agreement with observations on a full scale ozonation<br />
in a Swiss STP (Escher et al., 2009). As discussed above<br />
the analytically determined concentrations of (xeno)estrogens<br />
were below the quantification limit, therefore for this<br />
endpoint the very sensitive bioassay poses a great advantage<br />
despite the observed limitations due to matrix effects.
634<br />
3.2.3. Ah-receptor response<br />
The CAFLUX assay targets dioxins and dioxin-like compounds<br />
such as polychlorinated biphenyls (PCBs) but can also respond<br />
to other chemicals such as polycyclic aromatic hydrocarbons<br />
(PAHs) (Macova et al., 2010). The results of the test are<br />
expressed as 2,3,7,8-tetrachlorodibenzo-p-dioxin equivalent<br />
concentration (TCDDEQ). The median TCDDEQ of the influent<br />
water was 0.82 ng L -1 and there was no significant variation<br />
along the first three steps of the treatment process; i.e denitrification,<br />
pre-ozonation and coagulation/flocculation/DAFF<br />
(Fig. 5). The main ozonation removed about 50% of the TCDDEQ<br />
but subsequent activated carbon filtration and final ozonation<br />
did not show further important removal and the median<br />
TCDDEQ of the final effluent was approximately 3.9 times<br />
higher than the blank (Table 2). Two sets of samples were<br />
submitted to a sulphuric acid silica gel clean up procedure that<br />
aims at removing organic chemicals except those that are not<br />
oxidised such as polychlorinated dibenzodioxins, -furans and<br />
PCBs. The samples were then tested again with the CAFLUX<br />
assay to evaluate the contribution of these very persistent<br />
chemicals (i.e dioxins, furans and dioxin-like PCBs). Results<br />
showed that after clean up the TCDDEQ was not significantly<br />
different from the blank (values ranged from 0.09 to 0.11 ng L -1 ).<br />
This shows that the effect induced by the samples without<br />
sulphuric acid silica gel clean up is not due to the presence of<br />
dioxins, furans or dioxin-like PCBs but was caused by other<br />
chemicals. Since none of these groups of chemicals were<br />
quantified by chemical analysis in this study, no comparison<br />
between chemical and biological analysis is possible.<br />
3.2.4. Genotoxicity<br />
The umuC assay responds specifically to genotoxic compounds<br />
that cause DNA damages. To detect genotoxic effects caused by<br />
metabolites, the test is also performed in presence of a rat liver<br />
extract that can transform indirect genotoxicants to metabolites<br />
that are DNA damaging compounds. The median influent<br />
1/ECIR1.5 were 0.19 and 0.060 in the absence and presence of the<br />
rat liver extract respectively, showing that the sample was less<br />
genotoxic after metabolisation. This is what one would<br />
commonly expect, an exception would be PAHs that are activated<br />
by metabolism. Denitrification and pre-ozonation did not<br />
have a substantial influence on genotoxicity (Fig. 5). The coagulation/flocculation/DAFF<br />
stage decreased 1/EC IR1.5 by 59%<br />
compared to the influent. The main ozonation drastically<br />
reduced the genotoxicity, 1/EC IR1.5 was reduced by 80 and 93%<br />
compared to the DAFF effluent and to the influent of the plant<br />
respectively. After activated carbon filtration as well as in the<br />
final effluent, 1/ECIR1.5 was below the LOQ of the bioassay (Table 2).<br />
In every case, the genotoxicity of the metabolised sample was<br />
lower than the non-metabolised sample, indicating that the<br />
type of chemicals inducing the genotoxic effect did not change<br />
over the treatment.<br />
3.2.5. Neurotoxicity<br />
Neurotoxicity is measured by the inhibition of the enzyme<br />
acetylcholinesterase (AChE). Organophosphate and carbamate<br />
pesticides specifically bind to this enzyme and the<br />
results are expressed as parathion equivalent concentration<br />
(PTEQ). The median PTEQ in the secondary treated wastewater<br />
was 3.1 mgL -1 ; denitrification and pre-ozonation did not<br />
water research 44 (2010) 625–637<br />
reduce the PTEQ whereas DAFF decreased it by 31% compared<br />
to influent (Fig. 5). Unlike the other bioassays, the effect of the<br />
main ozonation on PTEQ was not significant but activated<br />
carbon filtration reduced it drastically to level below the<br />
quantification limit of the bioassay (0.30 mgL -1 ) which represents<br />
more than an 80% and 90% decrease compared to the<br />
main ozonation effluent and the plant influent water respectively.<br />
This observation is consistent with theoretical expectation,<br />
as it is known that compounds like diazinon and<br />
chlorpyrifos, which often constitute a large fraction of the<br />
acetylcholinesterase inhibitors, are not well oxidised by<br />
ozone. In contrast, these compounds are fairly hydrophobic<br />
(logKow ¼ 3.96 and 4.66 respectively), therefore sorption to<br />
activated carbon can be expected. A similar removal pattern<br />
has been observed for acetylcholinesterase inhibitors in the<br />
above-mentioned Swiss STP: none of the single removal steps<br />
(biological treatment, ozonation, sand filtration) had a high<br />
removal efficiency but all steps taken together produced<br />
a satisfactory overall removal (Escher et al., 2009).<br />
3.2.6. Phytotoxicity<br />
The I-PAM assay is sensitive to herbicides that directly inhibit<br />
photosynthesis; the results are reported as a diuron equivalent<br />
concentration (DEQ). The DEQ of the influent water ranged from<br />
0.05 to 0.22 mgL -1 with a median value of 0.10 mgL -1 (Table 2).<br />
Diuron concentrations were measured by chemical analysis; it<br />
was reported in every sample of the influent water from 0.02 to<br />
0.04 mgL -1 , suggesting that its contribution to the effect observed<br />
was limited. Among the other herbicides quantified, only<br />
simazine is also a photosystem II inhibitor with a relative<br />
potency of 0.15 (Muller et al., 2008). Simazine concentrations in<br />
the influent ranged from 0.05 to 0.19 mgL -1 . These two<br />
compounds considered together accounted for 17–93% of the<br />
measured DEQ demonstrating the interest of this bioassay to<br />
take into account non-measured compounds. The DEQ<br />
increased by factors of 2.2 and 3.5 after denitrification and preozonation<br />
respectively but variation from one day to another<br />
was large therefore it is difficult to draw a conclusion (Fig. 5). This<br />
increase was accompanied with a slight increase in baseline<br />
toxicity and could therefore be caused by baseline toxicants<br />
interfering with the measurement of the photosynthesis yield<br />
(Macova et al., 2010). The coagulation/filtration/DAFF stage<br />
reduced DEQ by 67% and 88% compared to the plant’s influent<br />
water and to the pre-ozonated water respectively. After the<br />
main ozonation the DEQ was not further significantly reduced;<br />
Diuron’s concentrations were equal to or below the LOQ of<br />
0.01 mgL -1 and simazine was removed by approximately 50%,<br />
their contribution accounting for 16–38% of the observed DEQ.<br />
Subsequent activated carbon filtration and final ozonation did<br />
not further affect the DEQ but diuron and simazine were<br />
removed below their LOQ. The overall treatment achieved 75%<br />
median decrease of DEQ, the effluent median DEQ was 0.03 mgL -1<br />
(Table 2).<br />
3.2.7. Overall treatment<br />
The biological activity of the effluent was lower compared to<br />
the influent and close or equal to the blank level showing that<br />
the treatment process could effectively decrease the biological<br />
adverse effects observed with the bioassays; from 62% for the<br />
AhR response to more than 99% for estrogenicity (Table 2). The
key treatment steps responsible for the decrease of biological<br />
activity are the DAFF stage, the main ozonation and the activated<br />
carbon adsorption (Fig. 5).<br />
3.3. Indirect potable reuse considerations<br />
Micropollutants reported concentrations were compared to the<br />
guidelines values for indirect potable reuse given in the<br />
Australian Guidelines for Water Recycling: Augmentation of<br />
Drinking Water Supplies (Table SI 3 in supporting information<br />
SI 2). The reported concentrations of the measured compounds<br />
were found to be below the guideline values in the influent<br />
water of the reclamation plant before any treatment. After<br />
removal by the advanced treatment process, concentrations<br />
were several orders of magnitude below the guideline values.<br />
For information purpose, median equivalent concentrations<br />
obtained with the bioassays were compared to the corresponding<br />
reference compound’s guideline value when<br />
available. Note however, that the effect caused by a mixture<br />
cannot be compared directly to a guideline value of a single<br />
compound. Moreover, the bioassays used here are acute tests<br />
and no conclusions can be drawn about chronic effects.<br />
Nevertheless such a comparison gives an impression on the<br />
expected hazard of the mixture but must be communicated<br />
with caution to a lay audience. For estrogenicity, neurotoxicity<br />
and phytotoxicity the reference compounds were estradiol,<br />
parathion and diuron and the guidelines values were 175 ng L -1 ,<br />
10 mgL -1 and 30 mgL -1 respectively. Similarly to individual<br />
compounds concentrations, the bioassays equivalent<br />
concentrations were already below the guidelines values in<br />
the water entering the reclamation plant. Final effluent<br />
median equivalent concentrations were also several orders of<br />
magnitude below the corresponding guideline values, i.e.<br />
more than 2900, 33 and 428 fold for estrogenicity, neurotoxicity<br />
and phytotoxicity respectively.<br />
For the parameters considered, the water quality complies<br />
with the requirements of the Australian Guidelines for Water<br />
Recycling: Augmentation of Drinking Water Supplies. This<br />
suggests that such a treatment train could be considered as an<br />
alternative to the combination of microfiltration and reverse<br />
osmosis for indirect potable reuse schemes. It has the advantage<br />
of not producing a waste stream and would be certainly<br />
less energy intensive. Nevertheless, before this process can be<br />
recommended for indirect potable reuse, additional consideration<br />
needs to be given to the overall risk management<br />
strategies of the treatment train. Moreover, the removal of<br />
pathogens such as viruses and bacteria has to be assessed as<br />
well as the potential to form disinfection by-products due to<br />
the remaining DOC levels.<br />
4. Conclusions<br />
The assessment of a tertiary treatment train regarding the<br />
removal of micropollutants and decrease of biological activity<br />
leads to the following conclusions:<br />
1) The treatment train reduced the concentration of the 54<br />
micropollutants quantified in the secondary treated<br />
wastewater as well as the biological activity observed in<br />
water research 44 (2010) 625–637 635<br />
bioassays. Overall concentration reductions were typically<br />
higher than 90% and most of the compounds were removed<br />
to levels lower than 0.01 mgL 1 . The observed effects<br />
decreased by 62% to more than 90% depending on the assay<br />
and levels in the final effluent were close to or equivalent to<br />
the blank’s. The effect of individual processes varied from<br />
one compound and bioassay to another but the combination<br />
of the coagulation/flocculation/DAFF stage, the main<br />
ozonation and the activated carbon filtration was responsible<br />
for the major part of the observed reduction.<br />
2) The use of a battery of bioassays as complementary tools to<br />
chemical analysis yielded valuable additional information on<br />
the water quality and the process efficiency. The results<br />
showed the limitations of chemical analysis to assess potential<br />
biological adverse effects and the ability of bioassays to<br />
take into account the presence of non-measured compounds,<br />
formed transformation products and/or mixture effects.<br />
3) Concurrent measurement of DOC, micropollutants and<br />
biological activity showed that the DOC level can influence<br />
the effect observed in the bioassays. Baseline toxicity was<br />
shown to correlate well with the DOC levels. It is supposed<br />
that non-targeted organic matter is co-extracted with the<br />
targeted organic micropollutants in the SPE purification<br />
step. More understanding of the nature and quantity of<br />
these co-extracted compounds is needed.<br />
4) Ozone dose relative to DOC content is a key parameter for<br />
the ozonation performance. A low ozone dose of 0.1 mgO3<br />
-1<br />
mgDOC did not affect the concentration of micropollutants<br />
-1<br />
or biological activity whereas a dose of 0.5 mgO3 mgDOC was<br />
able to remove most of the micropollutants by more than<br />
70% and substantially decrease the biological activity.<br />
5) Degradation products are a major concern in ozonation<br />
processes. Here, the results from the battery of bioassays<br />
show that the main ozonation leads to lower baseline and<br />
specific toxic effects. It can be concluded that the mixture of<br />
degradation products formed have an overall less harmful<br />
potential than the mixture of parent compounds.<br />
Acknowledgements<br />
This work was co-funded by the Urban Water Security<br />
Research Alliance under the Enhanced Treatment Project, the<br />
CRC Water Quality and Treatment Project No. 2.0.2.4.1.1 –<br />
Dissolved Organic Carbon Removal by Biological Treatment<br />
and by EnTox. The National Research Centre for Environmental<br />
Toxicology (EnTox) is a joint venture of the University<br />
of Queensland and Queensland Health Forensic and Scientific<br />
Services. The authors acknowledge the following institution<br />
and individuals who contributed to this study: Moreton Bay<br />
Water for access to the South Caboolture Water Reclamation<br />
Plant; Ray McSweeny and Paul McDonnell (Moretom Bay<br />
Water) for their help during sampling; Chris Pipe-Martin<br />
(Ecowise) for providing information on the South Caboolture<br />
Water Reclamation Plant; Geoff Eaglesham, Steve Carter and<br />
Mary Hodge (Queensland Health Forensic and Scientific<br />
Services) for the analysis of micropollutants and discussions<br />
on the analytical method; Wolfgang Gernjak, Michael G.<br />
Lawrence, Christoph Ort and Maria Jose Farre Olalla
636<br />
(Advanced Water Management Centre) for fruitful discussions<br />
and proof reading of the manuscript.<br />
Appendix.<br />
Supplementary data<br />
Supplementary data associated with this article can be found<br />
in the online version, at doi:10.1016/j.watres.2009.09.048.<br />
references<br />
Adams, C., Wang, Y., Loftin, K., Meyer, M., 2002. Removal of<br />
antibiotics from surface and distilled water in conventional<br />
water treatment processes. Journal of Environmental<br />
Engineering–ASCE 128 (3), 253–260.<br />
Andreozzi, R., Caprio, V., Marotta, R., Vogna, D., 2003.<br />
Paracetamol oxidation from aqueous solutions by means of<br />
ozonation and H2O2/UV system. Water Research 37 (5),<br />
993–1004.<br />
Andreozzi, R., Marotta, R., Pinto, G., Pollio, A., 2002.<br />
Carbamazepine in water: persistence in the environment,<br />
ozonation treatment and preliminary assessment on algal<br />
toxicity. Water Research 36 (11), 2869–2877.<br />
Benitez, F.J., Acero, J.L., Real, F.J., Roman, S., 2004. Oxidation of<br />
MCPA and 2,4-D by UV radiation, ozone, and the combinations<br />
UV/H2O2 and O-3/H2O2. Journal of Environmental Science<br />
and Health Part B-Pesticides Food Contaminants and<br />
Agricultural Wastes 39 (3), 393–409.<br />
Benner, J., Salhi, E., Ternes, T., von Gunten, U., 2008. Ozonation of<br />
reverse osmosis concentrate: Kinetics and efficiency of beta<br />
blocker oxidation. Water Research 42 (12), 3003–3012.<br />
Buffle, M.O., Schumacher, J., Meylan, S., Jekel, M., von Gunten, U.,<br />
2006a. Ozonation and advanced oxidation of wastewater:<br />
effect of O-3 dose, pH, DOM and HO center dot-scavengers on<br />
ozone decomposition and HO center dot generation. Ozone:<br />
Science and Engineering 28 (4), 247–259.<br />
Buffle, M.O., Schumacher, J., Salhi, E., Jekel, M., von Gunten, U.,<br />
2006b. Measurement of the initial phase of ozone decomposition<br />
in water and wastewater by means of a continuous quench-flow<br />
system: application to disinfection and pharmaceutical<br />
oxidation. Water Research 40 (9), 1884–1894.<br />
Dodd, M.C., Buffle, M.O., Von Gunten, U., 2006. Oxidation of<br />
antibacterial molecules by aqueous ozone: Moiety-specific<br />
reaction kinetics and application to ozone-based wastewater<br />
treatment. Environmental Science & Technology 40 (6),<br />
1969–1977.<br />
Escher, B.I., Bramaz, N., Mueller, J.F., Quayle, P., Rutishauser, S.,<br />
Vermeirssen, E.L.M., 2008a. Toxic equivalent concentrations<br />
(TEQs) for baseline toxicity and specific modes of action as<br />
a tool to improve interpretation of ecotoxicity testing of<br />
environmental samples. Journal of Environmental Monitoring<br />
10 (5), 612–621.<br />
Escher, B.I., Bramaz, N., Ort, C., 2009. Monitoring the treatment<br />
efficiency of a full scale ozonation on a sewage treatment<br />
plant with a mode-of-action based test battery. Journal of<br />
Environmental Monitoring 11 (10), 1836–1846.<br />
Escher, B.I., Bramaz, N., Quayle, P., Rutishauser, S.,<br />
Vermeirssen, E.L.M., 2008b. Monitoring of the ecotoxicological<br />
hazard potential by polar organic micropollutants in sewage<br />
treatment plants and surface waters using a mode-of-action<br />
based test battery. Journal of Environmental Monitoring 10 (5),<br />
622–631.<br />
Esplugas, S., Bila, D.M., Krause, L.G.T., Dezotti, M., 2007.<br />
Ozonation and advanced oxidation technologies to remove<br />
water research 44 (2010) 625–637<br />
endocrine disrupting chemicals (EDCs) and pharmaceuticals<br />
and personal care products (PPCPs) in water effluents. Journal<br />
of Hazardous Materials 149 (3), 631–642.<br />
Gros, M., Petrovic, M., Barcelo, D., 2009. Tracing pharmaceutical<br />
residues of different therapeutic classes in environmental<br />
waters by using liquid chromatography/quadrupole-linear ion<br />
trap mass spectrometry and automated library searching.<br />
Analytical Chemistry 81 (3), 898–912.<br />
Heringa, M.B., Schreurs, R., Busser, F., Van Der Saag, P.T., Van Der<br />
Burg, B., Hermens, J.L.M., 2004. Toward more useful in vitro<br />
toxicity data with measured free concentrations.<br />
Environmental Science and Technology 38 (23), 6263–6270.<br />
Hoigné, J., Bader, H., 1983. Rate constants of reactions of ozone with<br />
organic and inorganic compounds in water–II: dissociating<br />
organic compounds. Water Research 17 (2), 185–194.<br />
Hollender, J., Zimmermann, S.G., Koepke, S., Krauss, M.,<br />
McArdell, C.S., Ort, C., Singer, H., von Gunten, U., Siegrist, H.,<br />
2009. Elimination of organic micropollutants in a municipal<br />
wastewater treatment plant upgraded with a full-scale post-<br />
Ozonation followed by sand filtration. Environmental Science<br />
and Technology 43 (20), 7862–7869.<br />
Huber, M.M., Canonica, S., Park, G.Y., von Gunten, U., 2003.<br />
Oxidation of pharmaceuticals during ozonation and advanced<br />
oxidation processes. Environmental Science and Technology<br />
37 (5), 1016–1024.<br />
Huber, M.M., Gobel, A., Joss, A., Hermann, N., Loffler, D.,<br />
McArdell, C.S., Ried, A., Siegrist, H., Ternes, T.A., von<br />
Gunten, U., 2005. Oxidation of pharmaceuticals during<br />
ozonation of municipal wastewater effluents: a pilot study.<br />
Environmental Science and Technology 39 (11), 4290–4299.<br />
Joss,A.,Siegrist,H.,Ternes,T.A.,2008.Areweaboutto<br />
upgrade wastewater treatment for removing organic<br />
micropollutants? Environmental Science and Technology<br />
57 (2), 251–255.<br />
Kim, I.H., Tanaka, H., Iwasaki, T., Takubo, T., Morioka, T., Kato, Y.,<br />
2008. Classification of the degradability of 30 pharmaceuticals<br />
in water with ozone, UV and H2O2. Water Science and<br />
Technology 57 (2), 195–200.<br />
Kimura, K., Toshima, S., Amy, G., Watanabe, Y., 2004. Rejection of<br />
neutral endocrine disrupting compounds (EDCs) and<br />
pharmaceutical active compounds (PhACs) by RO membranes.<br />
Journal of Membrane Science 245 (1–2), 71–78.<br />
Lee, Y., Escher, B.I., Von Gunten, U., 2008. Efficient removal of<br />
estrogenic activity during oxidative treatment of waters<br />
containing steroid estrogens. Environmental Science and<br />
Technology 42 (17), 6333–6339.<br />
Leusch, F.D.L., Chapman, H.F., van den Heuvel, M.R., Tan, B.L.L.,<br />
Gooneratne, S.R., Tremblay, L.A., 2006. Bioassay-derived<br />
androgenic and estrogenic activity in municipal sewage in<br />
Australia and New Zealand. Ecotoxicology and Environmental<br />
Safety 65 (3), 403–411.<br />
Macova, M., Escher, B.I., Reungoat, J., Carswell, S., Lee Chue, K.,<br />
Keller, J., Mueller, J.F., 2010. Monitoring the biological activity<br />
of micropollutants during enhanced wastewater treatment<br />
with ozonation and activated carbon filtration. Water<br />
Research 44 (2), 477–492.<br />
Muller, R., Schreiber, U., Escher, B.I., Quayle, P., Nash, S.M.B.,<br />
Mueller, J.F., 2008. Rapid exposure assessment of PSII<br />
herbicides in surface water using a novel chlorophyll<br />
a fluorescence imaging assay. Science of the Total<br />
Environment 401 (1–3), 51–59.<br />
Muller, R., Tang, J.Y.M., Thierb, R., Mueller, J.F., 2007. Combining<br />
passive sampling and toxicity testing for evaluation of mixtures<br />
of polar organic chemicals in sewage treatment plant effluent.<br />
Journal of Environmental Monitoring 9 (1), 104–109.<br />
Nakada, N., Shinohara, H., Murata, A., Kiri, K., Managaki, S.,<br />
Sato, N., Takada, H., 2007. Removal of selected<br />
pharmaceuticals and personal care products (PPCPs) and
endocrine-disrupting chemicals (EDCs) during sand filtration<br />
and ozonation at a municipal sewage treatment plant. Water<br />
Research 41 (19), 4373–4382.<br />
Neale, P.A., Escher, B.I., Schafer, A.I., 2008. Quantification of<br />
solute–solute interactions using negligible-depletion solidphase<br />
microextraction: measuring the affinity of estradiol to<br />
bulk organic matter. Environmental Science and Technology<br />
42 (8), 2886–2892.<br />
Nowotny, N., Epp, B., vonSonntag, C., Fahlenkamp, H., 2007.<br />
Quantification and modeling of the elimination behavior of<br />
ecologically problematic wastewater micropollutants by<br />
adsorption on powdered and granulated activated carbon.<br />
Environmental Science and Technology 41 (6), 2050–2055.<br />
Onesios, K.M., Yu, J.T., Bouwer, E.J., 2009. Biodegradation and<br />
removal of pharmaceuticals and personal care products in<br />
treatment systems: a review. Biodegradation 20 (4), 441–466.<br />
Ormad, M.P., Miguel, N., Claver, A., Matesanz, J.M., Ovelleiro, J.L.,<br />
2008. Pesticides removal in the process of drinking water<br />
production. Chemosphere 71 (1), 97–106.<br />
Rivas, J., Gimeno, O., Encinas, A., Beltrán, F., 2009. Ozonation of<br />
the pharmaceutical compound ranitidine: Reactivity and<br />
kinetic aspects. Chemosphere 76 (5), 651–656.<br />
Simpson, D.R., 2008. Biofilm processes in biologically active<br />
carbon water purification. Water Research 42 (12), 2839–2848.<br />
Snyder, S.A., Adham, S., Redding, A.M., Cannon, F.S., DeCarolis, J.,<br />
Oppenheimer, J., Wert, E.C., Yoon, Y., 2007. Role of membranes<br />
and activated carbon in the removal of endocrine disruptors<br />
and pharmaceuticals. Desalination 202 (1–3), 156–181.<br />
Snyder, S.A., Wert, E.C., Rexing, D.J., Zegers, R.E., Drury, D.D.,<br />
2006. Ozone Oxidation of Endocrine Disruptors and<br />
Pharmaceuticals in Surface Water and Wastewater. Ozone:<br />
Science and Engineering 28 (6), 445–460.<br />
Song, W.H., Cooper, W.J., Mezyk, S.P., Greaves, J., Peake, B.M.,<br />
2008. Free radical destruction of beta-blockers in aqueous<br />
solution. Environmental Science & Technology 42 (4),<br />
1256–1261.<br />
Suarez, S., Lerna, J.M., Omil, F., 2009. Pre-treatment of hospital<br />
wastewater by coagulation-flocculation and flotation.<br />
Bioresource Technology 100 (7), 2138–2146.<br />
Ternes, T.A., Kreckel, P., Mueller, J., 1999. Behaviour and<br />
occurrence of estrogens in municipal sewage treatment plants –<br />
II. Aerobic batch experiments with activated sludge. Science of<br />
the Total Environment 225 (1–2), 91–99.<br />
Ternes,T.A.,Meisenheimer,M.,McDowell,D.,Sacher,F.,Brauch,H.J.,<br />
Gulde, B.H., Preuss, G., Wilme, U., Seibert, N.Z., 2002. Removal of<br />
water research 44 (2010) 625–637 637<br />
pharmaceuticals during drinking water treatment.<br />
Environmental Science and Technology 36 (17), 3855–3863.<br />
Ternes, T.A., Stuber, J., Herrmann, N., McDowell, D., Ried, A.,<br />
Kampmann, M., Teiser, B., 2003. Ozonation: a tool for removal<br />
of pharmaceuticals, contrast media and musk fragrances from<br />
wastewater? Water Research 37 (8), 1976–1982.<br />
Thuy, P.T., Moons, K., van Dijk, J.C., Anh, N.V., Van der Bruggen, B.,<br />
2008. To what extent are pesticides removed from surface<br />
water during coagulation-flocculation? Water and Environment<br />
Journal 22 (3), 217–223.<br />
van Leeuwen, J., Pipe-Martin, C., Lehmann, R.M., 2003. Water<br />
reclamation at South Caboolture, Queensland, Australia.<br />
Ozone: Science and Engineering 25 (2), 107–120.<br />
Vieno, N., Tuhkanen, T., Kronberg, L., 2006. Removal of<br />
pharmaceuticals in drinking water treatment: effect of chemical<br />
coagulation. Environmental Technology 27 (2), 183–192.<br />
Vogna, D., Marotta, R., Andreozzi, R., Napolitano, A., d’Ischia, M.,<br />
2004a. Kinetic and chemical assessment of the UV/H2O2<br />
treatment of antiepileptic drug carbamazepine. Chemosphere<br />
54 (4), 497–505.<br />
Vogna, D., Marotta, R., Napolitano, A., Andreozzi, R., d’Ischia, M.,<br />
2004b. Advanced oxidation of the pharmaceutical drug<br />
diclofenac with UV/H2O2 and ozone. Water Research 38 (2),<br />
414–422.<br />
Wert, E.C., Rosario-Ortiz, F.L., Snyder, S.A., 2009. Effect of ozone<br />
exposure on the oxidation of trace organic contaminants in<br />
wastewater. Water Research 43 (4), 1005–1014.<br />
Westerhoff, P., Yoon, Y., Snyder, S., Wert, E., 2005. Fate of<br />
endocrine-disruptor, pharmaceutical, and personal care<br />
product chemicals during simulated drinking water treatment<br />
processes. Environmental Science and Technology 39 (17),<br />
6649–6663.<br />
Yoon, Y., Westerhoff, P., Snyder, S.A., Wert, E.C., Yoon, J., 2007.<br />
Removal of endocrine disrupting compounds and<br />
pharmaceuticals by nanofiltration and ultrafiltration<br />
membranes. Desalination 202 (1–3), 16–23.<br />
Yu, Z.R., Peldszus, S., Huck, P.M., 2008. Adsorption characteristics<br />
of selected pharmaceuticals and an endocrine disrupting<br />
compound – naproxen, carbamazepine and nonylphenol – on<br />
activated carbon. Water Research 42 (12), 2873–2882.<br />
Zour, E., Lodhi, S.A., Nesbitt, R.U., Silbering, S.B., Chaturvedi, P.R.,<br />
1992. Stability studies of gabapentin in aqueous-solutions.<br />
Pharmaceutical Research 9 (5), 595–600.<br />
Zwiener, C., Frimmel, F.H., 2000. Oxidative treatment of<br />
pharmaceuticals in water. Water Research 34 (6), 1881–1885.
Removal of natural hormone estrone from secondary effluents<br />
using nanofiltration and reverse osmosis<br />
Xue Jin 1 , Jiangyong Hu*, Say Leong Ong<br />
Division of Environmental Science and Engineering, National University of Singapore, 10 Kent Ridge Crescent, 119260, Singapore<br />
article info<br />
Article history:<br />
Received 19 December 2008<br />
Received in revised form<br />
1 June 2009<br />
Accepted 23 September 2009<br />
Available online 13 October 2009<br />
Keywords:<br />
Estrone<br />
Nanofiltration<br />
Reverse osmosis<br />
Rejection<br />
Treated effluent<br />
Binding<br />
1. Introduction<br />
abstract<br />
Recently, the use of treated municipal wastewater for<br />
groundwater recharge and indirect drinking water reuse has<br />
become promising worldwide. Nanofiltration (NF) and reverse<br />
osmosis (RO) membranes are widely used in this kind of<br />
environmental application (Schäfer et al., 2005). However, the<br />
presence of endocrine disrupting chemicals (EDCs) in the<br />
aquatic environment has raised great consumer concern due<br />
to their potential health risk (Tim, 1997; Adams, 1998; Croley<br />
water research 44 (2010) 638–648<br />
Available at www.sciencedirect.com<br />
journal homepage: www.elsevier.com/locate/watres<br />
The rejection of steroid hormone estrone by nanofiltration (NF) and reverse osmosis (RO)<br />
membranes in treated sewage effluent was investigated. Four NF/RO membranes with<br />
different materials and interfacial characteristics were utilized. To better understand<br />
hormone removal mechanisms in treated effluent, effluent organic matters (EfOM) were<br />
fractionated using column chromatographic method with resins XAD-8, AG MP-50 and IRA-<br />
96. The results indicate that the presence of EfOM in feed solution could enhance estrone<br />
rejection significantly. Hydrophobic acid (HpoA) organic fraction made a crucial contribution<br />
to this ‘‘enhancement effect’’. Hydrophobic base (HpoB) could also improve estrone<br />
rejection while hydrophobic neutral (HpoN) and hydrophilic acid (HpiA) with low aromaticity<br />
had little effects. The increment in estrone rejection was predominantly attributed to<br />
the binding of estrone by EfOM in feed solutions, which led to an increase in molecular<br />
weight and appearance of negative charge (for the HpoA case) and thus an increased level<br />
of estrone rejection. However, the improvement of estrone rejection by HpoA decreased<br />
with increasing calcium ion concentration. The important conclusion of this study is, first,<br />
hydrophobic acid macromolecules are recommended to be added into feed water to<br />
improve the rejection of trace hormone during NF/RO membrane process, and, second,<br />
removal of calcium ions via pretreatment and application of membrane with more negative<br />
charge at its interface can greatly intensify this ‘‘enhancement effect’’.<br />
ª 2009 Elsevier Ltd. All rights reserved.<br />
et al., 2000; Baker, 2001; Iguchi et al., 2001). Amongst many<br />
kinds of EDCs, the impact of steroid hormones such as<br />
estrone, estradiol and ethinylestradiol is prominent because<br />
their potency can be several thousand times higher than that<br />
of others (Desbrow et al., 1998; Ternes et al., 1999; Johnson and<br />
Sumpter, 2001). It is noteworthy to emphasize that concentrations<br />
of some steroid hormones in secondary effluent are<br />
often high enough to harm wildlife although it is within ng/L<br />
range (Baronti et al., 2000; Williams et al., 2003; Soliman et al.,<br />
2004). In the light of the magnitude of this problem, NF/RO<br />
* Corresponding author. Tel.: þ65 65164540; fax: þ65 67744202.<br />
E-mail addresses: jinxuesky@ucla.edu (X. Jin), esehujy@nus.edu.sg (J. Hu), cveongsl@nus.edu.sg (S.L. Ong).<br />
1<br />
Current address: Civil & Environmental Engineering Department, University of California, 5732 Boelter Hall, Los Angeles, CA 90095-<br />
1593, USA<br />
0043-1354/$ – see front matter ª 2009 Elsevier Ltd. All rights reserved.<br />
doi:10.1016/j.watres.2009.09.057
membranes and different post-treatment technologies are<br />
essential to remove EDCs from treated effluent before water<br />
reuse and ensure the safety of drinking water.<br />
Removal of EDCs using NF/RO membranes in single-organic<br />
solution (the experiments were conducted with target EDCs<br />
spiked in synthetic model solutions in the absence of other<br />
organics) has been reported in several recent studies (Kimura<br />
et al., 2003, 2004; Schäfer et al., 2003; Nghiem et al., 2004a, b).<br />
Many factors may influence the interaction between EDCs and<br />
membrane and thus their rejection including physicochemical<br />
properties of EDCs, such as molecular weight (MW), molecular<br />
size, charge (as a function of dissociation constant), hydrophobicity<br />
and polarity; membrane properties, such as material,<br />
surface charge (zeta potential), hydrophobicity, molecular<br />
weight cut-off (MWCO), porosity and desalting degree; and<br />
solution chemistry, such as pH, ionic strength and presence of<br />
divalent cations (Bellona et al., 2004, Schäfer et al., 2005).<br />
Another important factor influencing the rejection efficiency is<br />
initial concentration of target EDCs in feed solution. Kimura<br />
et al. (2003) reported that experiments conducted at a ng/l<br />
concentration range resulted in a lower rejection efficiency as<br />
compared to experiments conducted at a mg/l range. This<br />
observation substantiates that experiments conducted within<br />
the ‘‘realistic’’ concentration range are necessary to provide an<br />
accurate value of rejection efficiency. It should, however, be<br />
pointed out that the results achieved from a single-organic<br />
solution cannot be extrapolated to the mixtures typically<br />
encountered in a real aquatic environment.<br />
In previous works, we have investigated the influence of<br />
other dissolved organic matter (DOM) in feed solution on the<br />
removal of steroid hormone estrone during NF processes (Jin<br />
et al., 2007). The results indicated that estrone rejection depended<br />
on the type of DOM co-present. Neutral and highly hydrophilic<br />
dextran had little effect on estrone rejection. The presence<br />
of humic acid without phenolic groups but great aromaticity<br />
improved estrone rejection slightly. In contrast, estrone rejection<br />
was obviously increased by the presence of hydrophobic<br />
acid fraction (HpoA) derived from secondary effluent. Therefore,<br />
it is expected that the rejection mechanism of EDCs in real water<br />
matrix, such as secondary effluent, is very complicated.<br />
Organic matter in secondary effluent is referred to as<br />
effluent organic matter (EfOM) and can be categorized into<br />
fractions based on the difference in hydrophobicity and<br />
charge properties using a column chromatographic method<br />
(Leenheer, 1981; Thurman and Malcolm, 1981; Imai et al., 2002;<br />
Hu et al., 2003). Adsorption of some fractions in EfOM onto<br />
membranes would be expected to alter the membrane surface<br />
properties, which subsequently results in changes in the<br />
rejection of EDCs (Ng and Elimelech, 2004; Yoon et al., 2004; Xu<br />
et al., 2006; Steinle-Darling et al., 2007). Alternatively, EDCs<br />
can bind to some fractions in EfOM and be retained together,<br />
which can also affect their rejection characteristics (Agbekodo<br />
et al., 1996; Devitt et al., 1998; Majewska-Nowak et al., 2002;<br />
Nghiem et al., 2004a, b; Hu et al., 2007; Jin et al., 2007). Both<br />
interactions mentioned above (EfOM-membrane and EfOM-<br />
EDCs) are markedly influenced by the physicochemical characteristics<br />
of EfOM. Therefore, different fractions of EfOM<br />
would demonstrate diverse impacts on EDCs rejection by<br />
membranes. To our knowledge, however, very little information<br />
is available on the rejection of trace EDCs by NF/RO<br />
water research 44 (2010) 638–648 639<br />
membranes with the presence of different fractions of EfOM.<br />
In order to better understand the removal mechanism of<br />
EDCs in secondary effluent by NF/RO membranes and optimize<br />
membrane performance, it is important to clarify this<br />
issue.<br />
It is known that divalent cations (especially calcium) affect<br />
the characteristics of EfOM, membrane charges and thus the<br />
two above-mentioned interactions (EfOM–membrane and<br />
EfOM–EDCs). For example, Devitt et al. (1998) reported that<br />
atrazine–NOM association decreased in the presence of<br />
calcium. Therefore, calcium ion is predicted to affect the<br />
rejection of EDCs with the presence of other EfOM. However,<br />
to date, the intricate relationship between physicochemical<br />
characteristics of EfOM, calcium ion concentration and<br />
membrane performance on EDC rejection remains poorly<br />
understood. In order to gain insight into the removal mechanism<br />
of EDCs in treated effluent, more research effort is<br />
needed to better understand the above relationship.<br />
In this study, we investigated the removal of steroid hormone<br />
estrone from secondary effluent using a bench-scale cross-flow<br />
NF/RO system. We examined the effect of various organic fractions<br />
isolated from secondary effluent and membrane characteristics<br />
on the removal of estrone by this process. After<br />
determining the key organic fraction which makes a crucial<br />
contribution to the enhanced estrone rejection by EfOM, the<br />
influence of calcium ion concentration on estrone rejection with<br />
the presence of the key fraction was investigated.<br />
2. Materials and methods<br />
2.1. NF/RO membranes and characterization<br />
Four NF/RO membranes, denoted as DL, CK, AK and CG (Sepa<br />
membranes, OSMONICS, USA), were used. DL and AK are polyamide<br />
(PA) composite membranes. CK and CG are made of<br />
cellulose acetate (CA) polymer. New membranes were soaked in<br />
ultrapure water for 24 h with the water being replaced regularly<br />
prior to use. Membrane properties were characterized as<br />
described elsewhere (Jin et al., 2007) and are shown in Table 1.<br />
Briefly, contact angle was determined with ultrapure water at<br />
pH 5.8 0.2 using a goniometer (Ramé-hart contact angle<br />
goniometer, Imaging System, Mountain Lakes, NJ, USA). Zeta<br />
potential of the membrane surface was measured in the electrolyte<br />
background solution (1 mM NaHCO3 and 8 mM NaCl, pH<br />
7) using an Electro Kinetic Analyzer (EKA) from Anton Paar<br />
(Austria). To characterize the membrane pore size, MWCO was<br />
estimated by filtering the electrolyte background solution containing<br />
polyethylene glycol (PEG) with MW of 200, 400, 600 and<br />
1000 Da, purchased from Fluka (Switzerland). Filtration tests<br />
were performed using the same cross-flow membrane test unit<br />
as used in the estrone filtration tests with clean membrane.<br />
A PEG concentration of approximately 6 ppm as dissolved<br />
organic carbon (DOC) was used for each measurement.<br />
2.2. Membrane filtration unit<br />
A laboratory-scale cross-flow membrane filtration unit was<br />
used (Fig. S1). Membrane coupon (13 cm by 13 cm) was housed<br />
in a stainless steel membrane cell. Feed solution was fed from
640<br />
Table 1 – Membrane properties.<br />
Membrane Membrane type Contact angle ( ) Zeta potential a<br />
a reservoir by a pump, capable of providing a maximum<br />
pressure of 4137 kPa and a maximum flow of 20 l/min. The<br />
temperature inside the feed reservoir was maintained<br />
constant (22 0.1 C) by a recirculating chiller. The pressure<br />
(1379 kPa with DL, CK and AK, and 2758 kPa with CG) inside<br />
the membrane cell was set with a back-pressure regulator and<br />
a bypass valve.<br />
Each experiment was conducted in two steps: (1) the<br />
membrane was compacted with ultrapure water until the<br />
water flux remained constant; (2) the reservoir was emptied<br />
and filled with test solution. The solution was then filtered at<br />
constant transmembrane pressure. All experiments were<br />
conducted for 24 h. During filtration, both permeate and<br />
concentrate were recirculated back to the feed tank. At specified<br />
intervals, permeate and feed samples were collected for<br />
analysis. Rejection (R) is defined as R ¼ 100 ð1 Cp=C fÞ,<br />
where C p and C f are the permeate and feed concentrations,<br />
respectively. The effect of EfOM on estrone rejection was<br />
determined based on the rejection measured at the end of<br />
filtration where stable rejection was achieved.<br />
2.3. EfOM fractionation method<br />
Secondary effluent was collected from a local municipal<br />
wastewater treatment plant. The water was first filtered through<br />
amicrofiltration(MF)unit(0.05mm, Desal/Osmonics, USA) to<br />
simulate pretreatment ahead of NF/RO processes in the real<br />
plant, which is commonly used to remove suspended solids and<br />
colloids. The total organic carbon (TOC) and total dissolved solid<br />
(TDS) of the water samples ranged from 9.2 to 10.8 ppm and 455<br />
to 618 ppm, respectively. Other characteristics of the MF treated<br />
secondary effluent are summarized in Table S1.<br />
The filtrates were subsequently fractionated into six<br />
subcomponents: HpoA, hydrophobic bases (HpoB), hydrophobic<br />
neutrals (HpoN), hydrophilic acids (HpiA), hydrophilic<br />
bases (HpiB) and hydrophilic neutrals (HpiN) using a method<br />
developed by Leenheer (1981). SupeliteÔ XAD-8 resin<br />
(SUPELCO, USA.), AG-MP-50 cation exchange resin (BioRad,<br />
USA) and Amberlite IRA-96 anion exchange resin (Rohm and<br />
Haas, France) were the resins used. Details on the instruments<br />
and fractionation methods have been published previously<br />
(Hu et al., 2003). The DOC fractionation results for the<br />
secondary effluent are presented in Table 2. HpoA predominated<br />
in secondary effluent and accounted for 36.6–40.3% of<br />
the total amount of dissolved organic matter. These results do<br />
not agree with an earlier study by Imai et al. (2002) who found<br />
different results on the fractions with HpiA as the most<br />
water research 44 (2010) 638–648<br />
MWCO (Da) b<br />
abundant fraction for the samples taken from Japan. Thus,<br />
DOM composition in secondary effluent varies with the origin<br />
of wastewater and the type of treatment process used. In this<br />
study, the experimental results obtained in MF treated<br />
secondary effluent should be analyzed based on its specific<br />
DOM-fraction distribution as reported here.<br />
2.4. Chemical and solution chemistry<br />
Typical flux/pressure<br />
(m/s per Pa) c<br />
DL Polyamide NF 30.7 2.7 16.3 2.7 490 1:46 10 5 =6:89 10 5<br />
CK Cellulose acetate NF 54.2 1.4 6.0 0.3 560 1:32 10 5 =1:52 10 6<br />
AK Polyamide RO 41.7 1.3 19.9 0.2
In all feed waters, initial estrone concentration was maintained<br />
at 100 ng/L. Solution with different calcium ion<br />
concentrations (0, 0.3 and 0.6 mM) were accomplished by<br />
adding CaCl2 stock solution, while the total ionic strength was<br />
kept constant by adjusting NaCl concentration.<br />
2.5. Analytical methods<br />
2.5.1. Estrone detection<br />
Estrone was extracted from aqueous sample by OASIS HLB<br />
cartridge (500 mg) from Waters, Singapore. Before the analytes<br />
were extracted, 80 ng/l of E1-d4 was added to the sample<br />
water to increase accuracy of the analytical data. The<br />
cartridge was subsequently conditioned by 6 ml of diethylether,<br />
5 ml of methanol and 10 ml of ultrapure water. The<br />
sample water (400 ml) was loaded through the cartridge at<br />
flow rate of 3 ml/min with the aid of a vacuum pump. After<br />
sample loading, the cartridge was eluted with 10 ml of<br />
a methanol–diethylether solution (10:90, v/v). The eluant was<br />
collected in a brown glass vial and was allowed to dry under<br />
a gentle stream of nitrogen. After drying, the residuals were<br />
re-dissolved in 0.4 ml of methanol and then analyzed by liquid<br />
chromatography–tandem mass spectrometry (MDS Sciex API<br />
2000 tandem triple-quadropole mass spectrometer) from<br />
Applied Biosystems, USA. Analytes were chromatographed on<br />
a 2.1 150 mm column filled with 3.5 mm C18 reversed phase<br />
packing (Agilent, Singapore). The multiple reaction monitoring<br />
(MRM) mode was chosen for quantification. After<br />
observing collision-induced dissociation (CID) spectra<br />
obtained by full-scan production experiments, the following<br />
MRM pairs were chosen: E1: 269.2/145.0; 269.2/143.0; E1-d4:<br />
273.0/146.9. Estrone recovery was not matrix-dependent due<br />
to using internal standard. The mean recovery and relative<br />
standard deviation (RSD) was 99 7.4% (n ¼ 16). The limit of<br />
detection was 0.5 ng/l, calculated using a signal-to-noise ratio<br />
of 3. Details with respect to basic information on this method<br />
are provided separately (Jin, 2007).<br />
2.5.2. TOC and UV 254 analysis<br />
In all cases, the concentrations of EfOM were detected by<br />
a TOC analyzer (Shimadzu TOC-Vcsh, Japan). TOC and DOC<br />
are used interchangeably in this study because secondary<br />
effluent was filtered through 0.05 mm membrane before using.<br />
The specific ultraviolet absorption of selected organic fraction<br />
at 254 nm (SUVA254) was calculated by dividing the absorbance<br />
at 254 nm measured using the Shimadzu UV-1700<br />
UV-Visible Spectrophotometer (Japan) by the DOC level of the<br />
solution.<br />
2.5.3. Molecular Weight Measurement<br />
High performance size exclusion chromatography (HPSEC)<br />
was performed to determine both weight- and number-averaged<br />
molecular weights. Instrumentation consisted of a SHI-<br />
MADZU (Japan) HPLC instrument LC-10AT VP with a Polymer<br />
Laboratories aquagel-OH column and a SHIMADZU SPD-M10A<br />
diode array detector operating at 260 cm. The HPSEC system<br />
was calibrated using polystyrene sulfonates standards of<br />
known average MW between 210 and 13000, and acetone.<br />
Mobile phase was composed of ultrapure water buffered with<br />
phosphate to pH of 6.8.<br />
water research 44 (2010) 638–648 641<br />
3. Results and discussion<br />
3.1. Rejection of estrone by NF/RO membranes in<br />
electrolyte background solution<br />
Filtration of estrone in electrolyte background solution (1 mM<br />
NaHCO 3 and 8 mM NaCl, pH 7) provides the baseline from<br />
which the effect of other organic matter on estrone rejection<br />
can be found. The open symbols in Fig. 1 present the observed<br />
rejections of estrone in electrolyte background solution by<br />
four kinds of NF/RO membranes. It appeared that estrone<br />
rejection was initially higher than 90% and decreased<br />
dramatically and then stabilized at later filtration stage.<br />
The excellent removal performances of all membranes at<br />
the initial filtration stage can be attributed to their adsorption<br />
capabilities and steric hindrance. However, the adsorption<br />
effect can only contribute to the short-term removal of<br />
estrone. As the feed solution is continuously filtered through<br />
the membrane, more and more available sites on the<br />
membrane are occupied by adsorbed estrone. When the<br />
partition of estrone between feed solution and membrane<br />
reaches equilibrium, there is no further net adsorption effect<br />
taking place and thus the contribution from adsorption would<br />
be negligible. Under this condition, size exclusion would<br />
become the overriding removal mechanism at the later<br />
filtration stage. Consistent with the MWCO values of the<br />
membranes, the ultimate rejection of estrone followed the<br />
order: AK > CG > DL > CK. AK membrane with MWCO value<br />
below the MW of estrone can remove estrone efficiently.<br />
However, for the other three kinds of membranes with MWCO<br />
values larger than the MW of estrone, ultimate rejection was<br />
extremely low. To avoid overestimation of estrone rejection,<br />
in this paper, all of the determined rejections were based on<br />
23 h of filtration when stable rejection was achieved.<br />
3.2. Rejection of estrone by NF/RO membranes in<br />
secondary effluent matrix<br />
Estrone rejection in MF treated secondary effluent, compared<br />
with that in electrolyte background solution, using four kinds<br />
of NF/RO membranes was monitored and the results are presented<br />
in Fig. 1, whereas the corresponding DOC rejection in<br />
secondary effluent is presented in Fig. 2. Estrone removal<br />
efficiencies in MF treated secondary effluent matrix for all<br />
NF/RO membranes used were consistently higher than the<br />
results obtained in electrolyte background solution. The<br />
improvements in ultimate estrone rejection were 6.5%, 32.5%,<br />
21.6% and 11.6% for AK, DL, CK and CG membranes, respectively.<br />
The results obtained here are consistent with an earlier<br />
study by Comerton et al. (2008), who investigated the influence<br />
of water matrix on EDC rejection by RO and NF membranes.<br />
They reported a significant increase in EDCs rejection by tight<br />
NF membranes in membrane bioreactor effluent and natural<br />
waters when compared to the ultrapure water.<br />
The improvement in estrone rejection could be due to the<br />
following reasons. Firstly, permeate flux decline was observed in<br />
MF treated secondary effluent matrix during separation<br />
processes (in electrolyte background solution, reduction in<br />
permeate flux was not observed for any membrane tested,
642<br />
a b<br />
Estrone Rejection (%)<br />
Estrone Rejection (%)<br />
100<br />
80<br />
60<br />
40<br />
20<br />
0<br />
100<br />
80<br />
60<br />
40<br />
20<br />
0<br />
Electrolyte background<br />
Secondary effluent<br />
0 5 10 15 20 25<br />
Filtration Time (h)<br />
which can be attributed to the very low concentration of estrone<br />
in the feed solution). After 23 h of filtration, the flux decreased by<br />
about 13.0%, 8.9%, 33.8% and 17.5% for DL, CK, AK and CG<br />
membrane, respectively. This suggests that the membranes<br />
were fouled by EfOM. The entrapment of EfOM onto the<br />
membrane active layer restricted the passage of contaminants,<br />
resulting in an increase in the observed rejection of both DOC<br />
(Fig. 2) and estrone at the ng/l level (Fig. 1). Similar increases in<br />
trace organic rejection as a result of organic fouling have been<br />
reported in previous studies (Agenson and Urase, 2007; Bellona<br />
and Drewes, 2007; Nghiem and Hawkes, 2007). However, simple<br />
EfOM fouling may not be the main reason, because the sequence<br />
Estrone Rejection (%)<br />
c d<br />
Electrolyte background<br />
Secondary effluent<br />
0 5 10 15 20 25<br />
Filtration Time (h)<br />
Estrone Rejection (%)<br />
100<br />
80<br />
60<br />
40<br />
20<br />
0<br />
100<br />
80<br />
60<br />
40<br />
20<br />
Electrolyte background<br />
Secondary effluent<br />
0 5 10 15 20 25<br />
Filtration Time (h)<br />
Electrolyte background<br />
Secondary effluent<br />
0<br />
0 5 10 15 20 25<br />
Filtration Time (h)<br />
Fig. 1 – Estrone rejection in electrolyte background solution and MF treated secondary effluent by (a) DL, (b) CK, (c) AK and (d) CG.<br />
DOC Rejection (%)<br />
100<br />
95<br />
90<br />
85<br />
80<br />
75<br />
70<br />
65<br />
60<br />
DL<br />
CK<br />
AK<br />
CG<br />
0 5 10 15 20 25<br />
Filtration Time (h)<br />
Fig. 2 – DOC rejection in MF filtered secondary effluent by<br />
four kinds of NF/RO membranes.<br />
water research 44 (2010) 638–648<br />
of the improvement in estrone rejection (DL > CK > CG > AK)<br />
was mismatched with the sequence based on permeate flux<br />
decline in secondary effluent matrix (AK > CG > DL > CK).<br />
Therefore the enhanced estrone rejection can be mainly attributed<br />
to the other reason, namely that, secondly, estrone molecules<br />
may bind to some fractions of EfOM in bulk solution and be<br />
retained together, resulting in a reduction in the passage of<br />
estrone molecules through membrane. The remainder of this<br />
paper is focused on proving this hypothesis.<br />
3.3. Rejection of estrone in the presence of specific<br />
organic fractions at same TOC level<br />
In order to optimize membrane performance, it is essential to<br />
identify which organic fraction in secondary effluent would<br />
make a crucial contribution to the ‘‘enhancement effect’’ on<br />
estrone removal. Amongst the six fractions derived from<br />
EfOM, HpiB was present in insignificant quantities and thus<br />
was expected to have a negligible influence on estrone<br />
removal. HpiN, as the last fraction derived from the resin<br />
fractionation system, inevitably contains some impurities of<br />
other fractions (Hu et al., 2003). Moreover, as shown in our<br />
previous work (Jin et al., 2007), dextran representing hydrophilic<br />
neutral organics had little effect on estrone rejection.<br />
Therefore, in this study, the research was focused on the<br />
influences of HpoA, HpoB, HpoN and HpiA on estrone rejection<br />
by NF/RO membranes.<br />
As shown earlier, DL and CK membranes demonstrated<br />
a greater improvement in estrone rejection in MF treated<br />
secondary effluent compared to AK and CG membranes.<br />
Investigations were firstly conducted to compare the capability<br />
of each isolated fraction to improve estrone removal
a<br />
Estrone Rejection (%)<br />
b<br />
Estrone Rejection (%)<br />
100<br />
80<br />
60<br />
40<br />
20<br />
0<br />
100<br />
80<br />
60<br />
40<br />
20<br />
0<br />
Electrolyte background<br />
With HpoA<br />
With HpoB<br />
With HpoN<br />
With HpiA<br />
0 5 10 15 20 25<br />
Filtration Time (h)<br />
Electrolyte background<br />
With HpoA<br />
With HpoB<br />
With HpoN<br />
With HpiA<br />
0 5 10 15 20 25<br />
Filtration Time (h)<br />
Fig. 3 – Observed rejection of estrone with the presence of<br />
different organic fractions at same DOC level (1 ppm) by (a)<br />
DL and (b) CK.<br />
using DL and CK membranes. To avoid the influence of DOC<br />
values, the initial concentration of each organic fraction was<br />
adjusted to a DOC value of approximately 1 ppm, which was<br />
close to the original concentrations of HpoB, HpoN and HpiA<br />
present in secondary effluent (Table 2).<br />
Estrone rejection in the presence of each fraction over<br />
a 24 h filtration period is presented in Fig. 3. It is evident that<br />
estrone rejection was significantly improved by the presence<br />
of HpoA, rising from 15.3% to 41.9% with DL membrane and<br />
from 8.2% to 15.3% with CK membrane. The presence of HpoB<br />
also resulted in an improvement in estrone rejection of as<br />
high as 3.6% and 4.1% with DL and CK, respectively. In<br />
contrast, HpoN and HpiA did not exhibit any obvious increase<br />
in estrone rejection for any of the membranes tested,<br />
although HpoN and HpiA were removed more efficiently than<br />
HpoB (Fig. 4). The differences in the ‘‘enhancement effect’’<br />
between HpoA and HpiA indicate that hydrophobic organic<br />
fractions have a greater contribution to the enhanced estrone<br />
rejection in secondary effluent than hydrophilic organic<br />
fractions.<br />
In all experiments, no obvious permeate flux decline (
644<br />
a<br />
DOC Rejection (%)<br />
b<br />
DOC Rejection (%)<br />
100<br />
90<br />
80<br />
70<br />
60<br />
50<br />
40<br />
30<br />
20<br />
10<br />
0<br />
100<br />
90<br />
80<br />
70<br />
60<br />
50<br />
40<br />
30<br />
20<br />
10<br />
0<br />
0 5 10 15 20 25<br />
Filtration Time (h)<br />
HpoA<br />
HpoB<br />
HpoN<br />
HpiA<br />
HpoA<br />
HpoB<br />
HpoN<br />
HpiA<br />
0 5 10 15 20 25<br />
Filtration Time (h)<br />
Fig. 4 – Observed rejection of bulk organics (as DOC) with<br />
the presence of different organic fractions at same DOC<br />
level (1 ppm) by (a) DL and (b) CK.<br />
molecules can bind to bulk DOM in feed matrix and subsequently<br />
be retained together. In this case, the ‘‘enhancement<br />
effect’’ on estrone removal was controlled not only by the<br />
binding between estrone and DOM in feed solution, but also by<br />
the rejection efficiency of the DOM with which estrone<br />
molecules are associated. Therefore, the weak effect of HpoB<br />
on estrone rejection is due to the inefficient removal of HpoB<br />
by CK membrane (Fig. 7b). It is apparent that the HpoA fraction,<br />
for all membranes used, had a significantly greater<br />
‘‘enhancement effect’’ than the other two hydrophobic fractions<br />
on ultimate estrone removal. After 23 h of filtration, the<br />
presence of HpoA increased estrone rejection by 29.8%, 13.3%<br />
and 20.5% for DL, CK and CG membranes, respectively. This<br />
could be attributed to a combination of the following reasons.<br />
Firstly, the structural characteristics of HpoA (phenolic group)<br />
water research 44 (2010) 638–648<br />
Fig. 5 – Schematic of possible hydrogen bonding between<br />
phenolic groups of estrone molecule and HpoA and<br />
between phenolic group in estrone molecule and aromatic<br />
amine group in HpoB (A–E stand for any possible functional<br />
groups within HpoB fraction).<br />
allowed estrone molecules to bind with HpoA in feed solution<br />
through hydrogen bonding, thus estrone could be removed<br />
together with HpoA. Secondly, HpoA with the highest<br />
concentration (Table 2) in feed solution would have the<br />
highest chance to associate with estrone molecules. Thirdly,<br />
HpoA with the largest MW of more than 4000 Da and a negative<br />
charge could be removed effectively (Fig. 7). The formation<br />
of estrone-HpoA complex in bulk solution led to<br />
a significant increase in MW (from 270 Da to more than<br />
4000 Da) and the appearance of negative charge and thus an<br />
increased level of estrone rejection by negative charged<br />
membranes based on the mechanisms of size exclusion and<br />
charge repulsion. Therefore, it could be concluded that the<br />
HpoA fraction was the greatest contributor to the ‘‘enhancement<br />
effect’’ on estrone rejection in treated effluent.<br />
It is worthwhile to note that the ‘‘enhancement effect’’ of<br />
HpoA followed the order of DL > CG > CK. This sequence<br />
agrees with the sequence based on the negative charge of the<br />
membrane surface (Table 1). According to the above discussion,<br />
the binding of estrone with HpoA in feed solution was<br />
critically important for the ‘‘enhancement effect’’. Less negative<br />
charge of membrane surface would result in more HpoA<br />
adsorption onto membrane due to the weaker electrostatic<br />
repulsion between HpoA and membrane. Consequently, less<br />
HpoA molecules would be available in feed solution for binding<br />
with estrone, resulting in a weaker improvement in estrone<br />
rejection. The results suggest that a more negatively-charged<br />
Table 3 – Characteristics of specific organic fractions derived from secondary effluent.<br />
Parameter HpoA HpoB HpoN HpiA<br />
MW (Da) 4192–4442 762–1751 1406–2658 3628–3825<br />
r 1.29–1.43 1.01–1.84 1.01–1.27 1.18–1.32<br />
SUVA254 (m 1 L/mg C) 2.58–4.25 2.06–3.81 1.19–1.57 0.75–0.81
a<br />
Estrone Rejection (%)<br />
b<br />
Estrone Rejection (%)<br />
c<br />
Estrone Rejection (%)<br />
100<br />
80<br />
60<br />
40<br />
20<br />
0<br />
100<br />
80<br />
60<br />
40<br />
20<br />
0<br />
100<br />
80<br />
60<br />
40<br />
20<br />
0<br />
Electrolyte background<br />
With HpoA<br />
With HpoB<br />
With HpoN<br />
0 5 10 15 20 25<br />
Filtration Time (h)<br />
Electrolyte background<br />
With HpoA<br />
With HpoB<br />
With HpoN<br />
0 5 10 15 20 25<br />
Filtration Time (h)<br />
Electrolyte background<br />
With HpoA<br />
With HpoB<br />
With HpoN<br />
0 5 10 15 20 25<br />
Filtration Time (h)<br />
Fig. 6 – Observed rejection of estrone with the presence of<br />
different hydrophobic organic fractions (at their original<br />
concentration in secondary effluent) by (a) DL, (b) CK and (c) CG.<br />
membrane would benefit the interaction between HpoA and<br />
estrone and thus estrone rejection.<br />
3.5. Effect of calcium ion concentration on estrone<br />
rejection in HpoA-containing solution<br />
Based on the discussion above, the estrone-binding capacity<br />
of HpoA is the major contributor to its significant ‘‘enhancement<br />
effect’’ on estrone rejection by NF/RO membranes. It is<br />
known that divalent cations, such as calcium, can influence<br />
the binding of trace contaminants by humic substances<br />
water research 44 (2010) 638–648 645<br />
a<br />
DOC Rejection (%)<br />
b<br />
DOC Rejection (%)<br />
c<br />
DOC Rejection (%)<br />
100<br />
90<br />
80<br />
70<br />
60<br />
50<br />
40<br />
30<br />
20<br />
10<br />
0<br />
100<br />
90<br />
80<br />
70<br />
60<br />
50<br />
40<br />
30<br />
20<br />
10<br />
0<br />
100<br />
90<br />
80<br />
70<br />
60<br />
50<br />
40<br />
30<br />
20<br />
10<br />
0<br />
HpoA<br />
HpoB<br />
HpoN<br />
0 5 10 15 20 25<br />
Filtration Time (h)<br />
HpoA<br />
HpoB<br />
HpoN<br />
0 5 10 15 20 25<br />
Filtration Time (h)<br />
HpoA<br />
HpoB<br />
HpoN<br />
0 5 10 15 20 25<br />
Filtration Time (h)<br />
Fig. 7 – Observed rejection of bulk organics (as DOC) with the<br />
presence of different hydrophobic organic fractions (at their<br />
original concentration in SE) by (a) DL, (b) CK and (c) CG.<br />
(Schlautman and Morgan, 1993). Therefore, calcium ion<br />
concentration in feed solution is predicted to affect the<br />
estrone rejection in HpoA-containing solution.<br />
Fig. 8 presents the influence of HpoA on estrone rejection<br />
by DL membrane at different calcium concentrations. It is<br />
evident that while the presence of HpoA could increase the<br />
rejection of estrone, a higher calcium ion concentration tended<br />
to reverse this effect. In particular, the addition of 0.3 mM<br />
calcium ion in feed solution resulted in the ‘‘enhancement
646<br />
a<br />
Estrone Rejection (%)<br />
b<br />
Estrone Rejection (%)<br />
c<br />
Estrone Rejection (%)<br />
100<br />
80<br />
60<br />
40<br />
20<br />
0<br />
100<br />
80<br />
60<br />
40<br />
20<br />
0<br />
100<br />
80<br />
60<br />
40<br />
w/o HpoA<br />
with HpoA<br />
0 5 10 15 20 25<br />
Filtration Time (h)<br />
w/o HpoA<br />
with HpoA<br />
0 5 10 15 20 25<br />
Filtration Time (h)<br />
w/o HpoA<br />
with HpoA<br />
0 5 10 15 20 25<br />
Filtration Time (h)<br />
Fig. 8 – Influence of HpoA on estrone rejection by DL<br />
membrane at different calcium ion concentration (a) 0 mM<br />
Ca 2D , (b) 0.3 mM Ca 2D and (c) 0.6 mM Ca 2D .<br />
effect’’ of HpoA on ultimate estrone rejection by DL<br />
membrane decreasing to 18.0% as compared to the improvement<br />
of 29.8% with the absence of calcium ion. When calcium<br />
ion concentration was increased to 0.6 mM, HpoA did not<br />
exhibit observable improvement in estrone rejection.<br />
In addition, negligible differences in permeate flux decline<br />
and DOC rejection were observed with increasing calcium<br />
concentration (data are not shown, but are available on<br />
request). This observation is unexpected and also differs<br />
water research 44 (2010) 638–648<br />
from other published investigations (Hong and Elimelech,<br />
1997; Tang et al., 2007), in which a more serious permeate<br />
flux decline with an increased calcium concentration was<br />
reported. This phenomenon was possibly due to the relatively<br />
low calcium ion concentration added in this study.<br />
Such low calcium ion concentration might not be enough to<br />
overcome the electrostatic repulsive energy barrier between<br />
HpoA and membrane and thus cannot result in a severe<br />
permeate flux decline. In this case, the diminished effect of<br />
HpoA on estrone rejection by membrane with increasing<br />
calcium concentration can most probably be attributed to the<br />
lower estrone–HpoA complex formation. Calcium ions can<br />
interact specifically with acidic functional groups (predominantly<br />
carboxylic groups) of humic substances (Hong and<br />
Elimelech, 1997). Such interactions reduce the charge repulsion<br />
between negative functional groups on HpoA molecules<br />
due to charge shielding and neutralization. As a result, the<br />
structure of HpoA tends to change from a stretched and<br />
linear configuration to a coiled and compact configuration<br />
with increased calcium concentration (Clark and Lucas,<br />
1998). This would result in the interactive sites within HpoA<br />
molecules becoming less easily reachable by estrone molecules<br />
by restricting access to interior sorption sites. In this<br />
case, the increase in calcium concentration would decrease<br />
the ability of HpoA to bind estrone molecules in feed water.<br />
This observation is similar to the results of Schlautman and<br />
Morgan (1993) who reported that the presence of calcium at<br />
neutral pH decreased the ability of humic acid to bind perylene.<br />
They attributed the finding to the fact that the<br />
compression of the HA structure decreased the size of voids<br />
in humic acid into which target solute could partition. Since<br />
the presence of calcium ions was not conducive to the<br />
binding of estrone by HpoA, the effect of increasing calcium<br />
concentration, even at relatively low levels, was to push<br />
estrone rejection towards the ‘‘free’’ estrone limit of rejection<br />
obtained in the absence of HpoA.<br />
4. Conclusions<br />
The results reported in this paper demonstrate that NF and<br />
loose RO membranes cannot remove trace steroid hormone<br />
efficiently in single-organic solution. However, the removal<br />
efficiency can be enhanced significantly in the presence of<br />
EfOM in feed solution. The HpoA fraction played a paramount<br />
role in the ‘‘enhancement effect’’. This enhancement is<br />
mainly attributed to the binding of estrone molecules to HpoA<br />
in feed solutions and being removed with the negatively<br />
charged HpoA macromolecules based on the mechanisms of<br />
size exclusion and charge repulsion. However, the presence<br />
of calcium ion tended to diminish this effect. Application of<br />
membrane with more negative-charge at its interface can<br />
greatly intensify this ‘‘enhancement effect’’.<br />
Supplementary data<br />
Supplementary data associated with this article can be found<br />
in online version at doi:10.1016/j.watres.2009.09.057.
eferences<br />
Adams, N.R., 1998. Clover phyto-oestrogens in sheep in Western<br />
Australia. Pure and Applied Chemistry 70 (9), 1855–1862.<br />
Agbekodo, K.M., Legube, B., Dard, S., 1996. Atrazine and simazine<br />
removal mechanisms by nanofiltration: influence of natural<br />
organic matter concentration. Water Research 30 (11), 2535–2542.<br />
Agenson, K.O., Urase, T., 2007. Change in membrane performance<br />
due to organic fouling in nanofiltration (NF)/reverse osmosis<br />
(RO) applications. Separation and Purification Technology 55<br />
(2), 147–156.<br />
Baker, V.A., 2001. Endocrine disrupters – testing strategies to<br />
assess human hazard. Toxicology in Vitro 15 (4-5), 413–419.<br />
Baronti, C., Curini, R., D’Ascenzo, G., Di Corcia, A., Gentili, A.,<br />
Samperi, R., 2000. Monitoring natural and synthetic estrogens<br />
at activated sludge sewage treatment plants and in a receiving<br />
river water. Environmental Science & Technology 34 (24),<br />
5059–5066.<br />
Bellona, C., Drewes, J.E., 2007. Viability of a low-pressure nanofilter<br />
in treating recycled water for water reuse applications: a pilotscale<br />
study. Water Research 41 (17), 3948–3958.<br />
Bellona, C., Drewes, J.E., Xu, P., Amy, G., 2004. Factors affecting<br />
the rejection of organic solutes during NF/RO treatment –<br />
a literature review. Water Research 38 (12), 2795–2809.<br />
Clark, M.M., Lucas, P., 1998. Diffusion and partition of humic acid<br />
in a porous ultrafiltration membrane. Journal of Membrane<br />
Science 143 (1–2), 13–25.<br />
Comerton, M.A., Andrews, C.R., Bagley, M.D., Hao, C., 2008. The<br />
rejection of endocrine disrupting and pharmaceutically active<br />
compounds by NF and RO membranes as a function of<br />
compounds and water matrix properties. Journal of<br />
Membrane Science 313 (1-2), 323–335.<br />
Croley, T.R., Hughes, R.J., Koenig, B.G., Metcalfe, C.D., March, R.E.,<br />
2000. Mass spectrometry applied to the analysis of estrogens<br />
in the environment. Rapid Communications in Mass<br />
Spectrometry 14 (13), 1087–1093.<br />
Desbrow, C., Routledge, E.J., Brighty, G.C., Sumpter, J.P., Waldock, M.,<br />
1998. Identification of estrogenic chemicals in STW effluent. 1.<br />
Chemical fractionation and in vitro biological screening.<br />
Environmental Science & Technology 32 (11), 1549–1558.<br />
Devitt, E.C., Ducellier, F., Cote, P., Wiesner, M.R., 1998. Effects of<br />
natural organic matter and the raw water matrix on the<br />
rejection of atrazine by pressure-driven membranes. Water<br />
Research 32 (9), 2563–2568.<br />
Hong, S., Elimelech, M., 1997. Chemical and physical aspects of<br />
natural organic matter (NOM) fouling of nanofiltration<br />
membranes. Journal of Membrane Science 132 (2), 159–181.<br />
Hu, J.Y., Ong, S.L., Shan, J.H., Kang, J.B., Ng, W.J., 2003. Treatability of<br />
organic fractions derived from secondary effluent by reverse<br />
osmosis membrane. Water Research 37 (19), 4801–4809.<br />
Hu, J.Y., Jin, X., Ong, S.L., 2007. Rejection of estrone by<br />
nanofiltration: influence of solution chemistry. Journal of<br />
Membrane Science 302 (1-2), 188–196.<br />
Iguchi, T., Watanabe, H., Katsu, Y., 2001. Developmental effects of<br />
estrogenic agents on mice, fish, and frogs: a mini-review.<br />
Hormones and Behavior 40 (2), 248–251.<br />
Imai, A., Fukushima, T., Matsushige, K., Kim, Y.H., Choi, K.,<br />
2002. Characterization of dissolved organic matter in<br />
effluents from wastewater treatment plants. Water Research<br />
36 (4), 859–870.<br />
Jin, X., 2007. Rejection of steroid hormone estrone by NF/RO<br />
membranes. PhD thesis. Department of Civil Engineering,<br />
National University of Singapore, Singapore.<br />
Jin, X., Hu, J.Y., Ong, S.L., 2007. Influence of dissolved organic<br />
matter on estrone removal by NF membranes and the role of<br />
their structures. Water Research 41 (14), 3077–3088.<br />
water research 44 (2010) 638–648 647<br />
Johnson, A.C., Sumpter, J.P., 2001. Removal of endocrinedisrupting<br />
chemicals in activated sludge treatment works.<br />
Environmental Science & Technology 35 (24), 4697–4703.<br />
Kimura, K., Amy, G., Drewes, J.E., Heberer, T., Kim, T.U.,<br />
Watanabe, Y., 2003. Rejection of organic micropollutants<br />
(disinfection by-products, endocrine disrupting compounds,<br />
and pharmaceutically active compounds) by NF/RO<br />
membranes. Journal of Membrane Science 227 (1–2), 113–121.<br />
Kimura, K., Toshima, S., Amy, G., Watanabe, Y., 2004. Rejection of<br />
neutral endocrine disrupting compounds (EDCs) and<br />
pharmaceutical active compounds (PhACs) by RO membranes.<br />
Journal of Membrane Science 245 (1–2), 71–78.<br />
Leenheer, J.A., 1981. Comprehensive approach to preparative<br />
isolation and fractionation of dissolved organic carbon from<br />
natural waters and wastewaters. Environmental Science &<br />
Technology 15 (5), 578–587.<br />
Majewska-Nowak, K., Kabsch-Korbutowicz, M., Dod, M.,<br />
Winnicki, T., 2002. The influence of organic carbon<br />
concentration on atrazine removal by UF membranes.<br />
Desalination 147 (1–3), 117–122.<br />
Ng, H.Y., Elimelech, M., 2004. Influence of colloidal fouling on<br />
rejection of trace organic contaminants by reverse osmosis.<br />
Journal of Membrane Science 244 (1–2), 215–226.<br />
Nghiem, L.D., Hawkes, S., 2007. Effects of membrane fouling<br />
on the nanofiltration of pharmaceutically active<br />
compounds (PhACs): Mechanisms and role of membrane<br />
pore size. Separation and Purification Technology 57 (1),<br />
176–184.<br />
Nghiem, L.D., Manis, A., Soldenhoff, K., Schäfer, A.I., 2004a.<br />
Estrogenic hormone removal from wastewater using NF/RO<br />
membranes. Journal of Membrane Science 242 (1–2),<br />
37–45.<br />
Nghiem, L.D., Schäfer, A.I., Elimelech, M., 2004b. Removal of<br />
natural hormones by nanofiltration membranes:<br />
measurement, modeling, and mechanisms. Environmental<br />
Science & Technology 38 (6), 1888–1896.<br />
Schäfer, A.I., Fane, A.G., Waite, T.D., 2005. Nanofiltration:<br />
Principles & Applications. Elsevier Advanced Technology,<br />
Oxford, New York.<br />
Schäfer, A.I., Nghiem, L.D., Waite, T.D., 2003. Removal of the<br />
natural hormone estrone from aqueous solutions using<br />
nanofiltration and reverse osmosis. Environmental Science &<br />
Technology 37 (1), 182–188.<br />
Schlautman, M.A., Morgan, J.J., 1993. Effects of aqueous chemistry<br />
on the binding of polycyclic aromatic hydrocarbons by<br />
dissolved humic materials. Environmental Science &<br />
Technology 27 (5), 961–969.<br />
Soliman, M.A., Pedersen, J.A., Suffet, I.H., 2004. Rapid gas<br />
chromatography–mass spectrometry screening method for<br />
human pharmaceuticals, hormones, antioxidants and<br />
plasticizers in water. Journal of Chromatography A 1029 (1-2),<br />
223–237.<br />
Steinle-Darling, E., Zedda, M., Plumlee, H.M., Ridgway, F.H.,<br />
Reinhard, M., 2007. Evaluating the impacts of membrane<br />
type, coating, fouling, chemical properties and water<br />
chemistry on reverse osmosis rejection of seven<br />
nitrosoalklyamines, including NDMA. Water Research 41 (17),<br />
3959–3967.<br />
Tang, Y.C., Kwon, Y.N., Leckie, O.J., 2007. Fouling of reverse<br />
osmosis and nanofiltration membranes by humic acid –<br />
effects of solution composition and hydrodynamic<br />
conditions. Journal of Membrane Science 290 (1–2),<br />
86–94.<br />
Ternes, T.A., Stumpf, M., Mueller, J., Haberer, K., Wilken, R.D.,<br />
Servos, M., 1999. Behavior and occurrence of estrogens in<br />
municipal sewage treatment plants - I. Investigations in<br />
Germany, Canada and Brazil. The Science of the Total<br />
Environment 225 (1–2), 81–90.
648<br />
Thurman, E.M., Malcolm, R.L., 1981. Preparative isolation of<br />
aquatic humic substances. Environmental Science &<br />
Technology 15 (4), 463–466.<br />
Tim, Z., 1997. In vitro bioassays for assessing estrogenic substances.<br />
Environmental Science & Technology 31 (3), 613–623.<br />
Williams, R.J., Johnson, A.C., Smith, J.J., Kanda, R., 2003. Steroid<br />
estrogens profiles along river stretches arising from sewage<br />
treatment works discharges. Environmental Science &<br />
Technology 37 (9), 1744–1750.<br />
Xu, P., Drewes, J.E., Kim, T.U., Bellona, C., Amy, G., 2006. Effect of<br />
membrane fouling on transport of organic contaminants in<br />
water research 44 (2010) 638–648<br />
NF/RO membrane applications. Journal of Membrane Science<br />
279 (1–2), 165–175.<br />
Yamamoto, H., Liljestrand, H.M., Shimizu, Y., Morita, M., 2003.<br />
Effects of physical-chemical characteristics on the sorption of<br />
selected endocrine disruptors by dissolved organic matter<br />
surrogates. Environmental Science and Technology 37 (12),<br />
2646–2657.<br />
Yoon, Y., Westerhoff, P., Yoon, J., Snyder, S.A., 2004. Removal of<br />
17 beta estradiol and fluoranthene by nanofiltration and<br />
ultrafiltration. Journal of Environmental Engineering 130 (12),<br />
1460–1467.
Screening of antimycotics in Swedish sewage treatment<br />
plants – Waters and sludge<br />
Richard H. Lindberg*, Jerker Fick, Mats Tysklind<br />
Department of Chemistry, Umea˚ University, SE-901 87 Umea˚, Sweden<br />
article info<br />
Article history:<br />
Received 25 March 2009<br />
Received in revised form<br />
19 October 2009<br />
Accepted 23 October 2009<br />
Available online 31 October 2009<br />
Keywords:<br />
Antimycotics<br />
Sewage water<br />
Sludge<br />
Mass flow<br />
1. Introduction<br />
abstract<br />
There are approximately 40 antimycotics, i.e. pharmaceuticals<br />
that can be used to treat fungal infections in humans,<br />
many of which are imidazoles, triazoles or allylamines. Antimycotics<br />
are usually administered topically, although a few,<br />
such as ketoconazole and fluconazole, are also used perorally.<br />
All antimycotics currently used are general cytochrome P450<br />
inhibitors that reportedly affect CYP families 1–3. For instance,<br />
the imidazoles and triazoles strongly inhibit CYP3A4 and 2C9,<br />
while the allylamines are known to affect CYP2D6 (Sweetman,<br />
2007). This has potentially significant implications for their<br />
use as pharmaceuticals since the P450s are responsible for the<br />
initial metabolism of ca. 75% of xenobiotics in humans, hence<br />
water research 44 (2010) 649–657<br />
Available at www.sciencedirect.com<br />
journal homepage: www.elsevier.com/locate/watres<br />
* Corresponding author. Tel.: þ46 90 786 5464; fax: þ46 90 12 7655.<br />
E-mail address: richard.lindberg@chem.umu.se (R.H. Lindberg).<br />
0043-1354/$ – see front matter ª 2009 Elsevier Ltd. All rights reserved.<br />
doi:10.1016/j.watres.2009.10.034<br />
Concentrations of six pharmaceutical antimycotics were determined in the sewage water,<br />
final effluent and sludge of five Swedish sewage treatment plants (STPs) by solid phase<br />
extraction, liquid/solid extraction, and liquid chromatography–electrospray tandem mass<br />
spectrometry. The antimycotics were quantified by internal standard calibration. The<br />
results were used to estimate national flows that were compared to predictions based on<br />
sales figures. Fluconazole was the only one of the six investigated antimycotics that was<br />
detected (at concentrations ranging from 90 to 140 ng L 1 ) in both raw sewage water and<br />
final effluent. Negligible amounts of this substance were removed from the aqueous phase,<br />
and its levels were below the limit of quantification in all of the analyzed sludge samples.<br />
In contrast, clotrimazole, ketoconazole and econazole were present in all of the sludge<br />
samples, at concentrations ranging between 200 and 1000 mgkg 1 , dry weight. There were<br />
close correlations between the national measured and predicted antimycotic mass flows.<br />
Antimycotic fate analysis, based on sales figures, indicated that 53% of the total amount of<br />
fluconazole sold appeared in the final effluents of the STPs, while 1, 155, 35, 209 and 41% of<br />
the terbinafine, clotrimazole, ketoconazole, econazole and miconazole sold appeared in the<br />
digested dewatered sludge.<br />
ª 2009 Elsevier Ltd. All rights reserved.<br />
they could have undesirable effects on patients’ ability to<br />
eliminate other substances (Sweetman, 2007; Guengerich,<br />
2008). More specifically, the imidazole and triazole antimycotics<br />
inhibit the essential activity of CYP51 (also known as<br />
ERG11 in fungi) in ergosterol synthesis in fungal pathogens,<br />
thereby increasing the permeability of their membranes,<br />
inhibiting their growth and/or killing them (Graybill and<br />
Drutz, 1980; Yoshida et al., 2000; Lepesheva and Waterman,<br />
2007). The antimycotic activity of allylamine antimycotics is<br />
also based on interference with ergosterol biosynthesis, but by<br />
inhibiting squalene epoxidase (Ryder, 1992).<br />
Gunnarson et al. have recently shown that six species<br />
commonly used in aquatic ecotoxicology tests possess CYP51<br />
orthologs that have more than 70% homology to the drug
650<br />
targets (Gunnarsson et al., 2008). Further, imidazole and triazole<br />
antimycotics have been shown to inhibit CYP19, which<br />
participates in the conversion of androgens to estrogens<br />
(Trösken et al., 2004; Gyllenhammar et al., 2009), with serious<br />
potential consequences for sex differentiation in exposed<br />
vertebrates, including testicular development in female<br />
salmon (Piferrer et al., 1994). Gyllenhammar et al. (2009) have<br />
also shown that exposure to the imidazole clotrimazole can<br />
have complex effects on aromatase activity in the gonads and<br />
brain of Xenopus tropicalis larvae; decreasing brain aromatase<br />
activity during their gonadal differentiation and increasing<br />
gonadal aromatase activity during metamorphosis.<br />
Concentrations of antimycotics have been analyzed<br />
(generally by solid phase extraction, SPE, followed by liquid (or<br />
gas) chromatography mass spectrometry, LC(GC)–MS/MS) in<br />
various aquatic environments and sewage flows (Kolpin et al.,<br />
2004; Thomas and Hilton, 2004; Roberts and Bersuder, 2006;<br />
Roberts and Thomas, 2006; Peschka et al., 2007; Kahle et al.,<br />
2008; Van De Steene and Lambert, 2008). Clotrimazole levels in<br />
the Main and Rhine rivers in Germany, and both the raw<br />
sewage water and final effluent from a German sewage<br />
treatment plant (STP), have been reported to be in the low<br />
ng L 1 range (Peschka et al., 2007). Comparable concentrations<br />
of clotrimazole have also been reported in two UK studies,<br />
which assessed sewage water from the Howton STP and<br />
samples collected along the Tyne estuary (Thomas and Hilton,<br />
2004; Roberts and Thomas, 2006). Somewhat higher levels of<br />
clotrimazole and fluconazole, between 10 and 110 ng L 1 , have<br />
been found in Swiss STP waters, along with fluconazole at<br />
levels close to its limit of quantification (LOQ), 1 ng L 1 ,in<br />
Swiss lakes (Kahle et al., 2008). In addition, quantifiable levels<br />
of miconazole have been found in both surface and sewage<br />
waters in the UK (Roberts and Bersuder, 2006), although it was<br />
not found at levels exceeding its reporting limit (17.5 ng L 1 )in<br />
an analysis of the contribution of wastewater contaminants to<br />
US streams (Kolpin et al., 2004). However, no published studies<br />
to our knowledge have analyzed the concentrations of antimycotics<br />
in digested sewage sludge.<br />
The six substances examined here were chosen to represent<br />
a range, in terms of applications and physicochemical<br />
properties, of antimycotics commonly used in Sweden.<br />
Information on these six substances and the internal standard<br />
sulconazole (which is not used in this country) is presented in<br />
Table 1. Little is known about the fate and ecotoxic potential of<br />
these substances, thus the aim of the presented study was to<br />
increase information on their occurrence, levels and fate<br />
within STPs. In addition, the collected data were cross-validated<br />
against predictive equivalents. To our knowledge, this is<br />
the first published study to present national screening data on<br />
antimycotics in both sewage water and sludge.<br />
2. Experimental<br />
2.1. Chemicals<br />
All of the antimycotics were bought from Sigma Aldrich<br />
(Stockholm, Sweden) except for terbinafine, which was<br />
obtained from LGC Standards (Bora˚s, Sweden). Formic acid<br />
and methanol (HPLC-grade) were purchased from JT Baker<br />
water research 44 (2010) 649–657<br />
(Deventer, the Netherlands), acetonitrile (HPLC-grade) from<br />
Fischer chemicals (Zurich, Switzerland), and sulfuric acid<br />
from Merck (Darmstadt, Germany). Water was purified (to<br />
18.2 MU cm resistivity) using a MAXIMA HPLC ultra pure<br />
water system (ELGA, High Wycombe Bucks, England). Stock<br />
and working (including calibration standards) solutions of<br />
individual antimycotics were prepared in methanol and<br />
diluted in water, respectively.<br />
2.2. Sample sites, collection and pre-extraction<br />
procedure<br />
Raw sewage water, final effluent and sludge samples were<br />
taken from five Swedish STPs during October and November<br />
2007. The STPs surveyed were Stockholm (Henriksdal), Gothenburg<br />
(Ryaverken), Umea˚ , Alingsa˚s (Nolhaga), and Bollebygd,<br />
which were selected to represent a wide range of STPs in<br />
terms of size and catchment area. Sewage water from the<br />
Umea˚ county hospital (UCH) was also collected during the<br />
same time period. In addition, raw sewage water, raw sewage<br />
water particles, raw sludge and digested dewatered sludge<br />
were collected from the Umea˚ STP in April 2008 to assess the<br />
fate of antimycotics during the treatment process. The locations<br />
of the STPs and data on their waste flows and catchment<br />
areas are provided in Fig. 1. All of the STPs treat sewage water<br />
mechanically, chemically and biologically. Nitrogen is also<br />
actively removed at the Henriksdal, Ryaverken and Nolhaga<br />
STPs, and the final solid product of all the STPs is anaerobically<br />
digested sludge, except at Bollebygd, where only clarifiers<br />
are used.<br />
Flow-proportional samples of raw sewage water and final<br />
effluents were collected from each of the five STPs in each<br />
case over a full single day, while grab samples of the UCH<br />
sewage water and digested dewatered sludge were collected.<br />
After collection, the sewage water (2 L in brown glass bottles)<br />
and sludge samples (500 g in brown glass beakers) were<br />
immediately stored at 18 C prior to extraction. Grab<br />
sampling was also used in the antimycotic fate study at the<br />
Umea˚ STP, in which raw sewage water (1 þ 10 L), final effluent<br />
(2 L), primary sludge (2 L) and 500 g of digested dewatered<br />
sludge samples were collected between 8:30 and 9:00 on each<br />
of two consecutive days in April 2008 (except for the digested<br />
dewatered sludge, which was only collected on the first of the<br />
sampling days due to technical problems). The raw sewage<br />
water (1 L) and final effluent samples for SPE were filtered<br />
(through 0.45 mm filters) immediately after collection, while<br />
the remaining raw sewage water (10 L) and primary sludge<br />
samples were centrifuged at 4800 rpm for 20 min, to concentrate<br />
the particles sufficiently for analysis (Lindberg et al.,<br />
2006), then air-dried for 24 h. All of the samples were then<br />
stored at 18 C awaiting extraction.<br />
2.3. Sample preparation – sewage waters<br />
The aqueous samples were thawed, filtered through 0.45 mm<br />
MFÔ membrane filters (Millipore, Sundbyberg, Sweden),<br />
acidified to pH 3 with sulfuric acid and 500 ng of the internal<br />
standard (IS) sulconazole was added to 500 mL portions of the<br />
resulting samples, which were then subjected to SPE using<br />
Evolute ABN (6 mL, 200 mg) solid sorbent columns (Biotage,
Table 1 – Information on the antimycotics investigated in this study.<br />
Terbinafine<br />
(allylamine)<br />
Clotrimazole<br />
(imidazole)<br />
Ketoconazole<br />
(imidazole)<br />
Fluconazole<br />
(triazole)<br />
Econazole<br />
(imidazole)<br />
Miconazole<br />
(imidazole)<br />
IS sulconazole<br />
(imidazole)<br />
O<br />
N<br />
N<br />
N<br />
Cl<br />
N N O<br />
Cl<br />
Cl<br />
Cl<br />
N<br />
Cl<br />
N<br />
HO<br />
N<br />
N<br />
N<br />
N<br />
N<br />
H<br />
O<br />
N<br />
O<br />
N<br />
Cl<br />
F F<br />
N<br />
O<br />
O<br />
N<br />
S<br />
Cl<br />
water research 44 (2010) 649–657 651<br />
N<br />
Cl<br />
N<br />
Cl<br />
N<br />
Cl<br />
Cl<br />
Cl<br />
Cl<br />
Application Use a<br />
Topical 43<br />
Topical 11<br />
(kg year 1 )<br />
EAS b Log K ow c<br />
g<br />
g<br />
PEC d<br />
(ng L 1 )<br />
UD e BCF f<br />
5.81 6 m 5370<br />
6.26 2 m 13,183<br />
Topical/peroral 399/22 –/0% 4.45 (4.35) 60 r 447<br />
Peroral 121 80% u 0.25 20 r 3<br />
Topical 67<br />
Topical 134<br />
Not used<br />
in Sweden<br />
g<br />
g<br />
5.61 10 r 4169<br />
6.25 20 r 12,882<br />
a Estimated from Apoteket AB’s Swedish sales during 2007.<br />
b Excretion of active substance through urine and/or feces (LIF, 2008).<br />
c Log Kow values estimated by KowWin, EPI suiteÔ (EPI SuiteÔ) (experimental values in parentheses).<br />
d Predicted environmental concentrations in surface waters, calculated according to EMEA guidelines Phase 1, except that Swedish sales values<br />
were used rather than the supposed penetration factor (Fpen), in combination with the daily defined dose (European Medicines Agency, 2006).<br />
e Ultimate degradation (m ¼ months, r ¼ recalcitrant) estimated by BioWin, EPI suiteÔ (EPI SuiteÔ).<br />
f Estimated bioconcentration factor as estimated by BcfWin, EPI suiteÔ (EPI SuiteÔ).<br />
g No excretion due to topical application.<br />
6.46
652<br />
Uppsala, Sweden) that had been conditioned with 6 mL of<br />
methanol and 6 mL water adjusted to pH 3 with sulfuric acid.<br />
Each sample was applied at a flow rate of 5 mL min 1 , cleaned<br />
using 5 mL of aqueous methanol (5%), and eluted by two<br />
washes of 3 mL methanol with 0.5% formic acid. The eluates<br />
were evaporated almost to dryness and finally reconstituted<br />
with 0.7 mL aqueous acetonitrile (10%).<br />
2.4. Sample preparation – sludge<br />
The sludge samples were thawed, then 2.0 g portions from<br />
each sample were placed in separate glass centrifuge tubes.<br />
The IS was added, 500 ng, and the samples were subjected to<br />
liquid/solid extraction step by adding 10.0 mL of methanol to<br />
each tube and ultrasonically agitating them for 10 min. They<br />
were subsequently centrifuged at 4800 rpm for 10 min, and<br />
the supernatants were placed in separate 30 mL glass vials.<br />
The residues were then re-extracted in the same manner and<br />
the two supernatants were combined. The combined extracts<br />
(approximately 19 mL) were made up to 200 mL with purified<br />
water research 44 (2010) 649–657<br />
Fig. 1 – Mass flows of the measured antimycotics (g day L1 ) in the raw sewage water, final effluent and digested dewatered<br />
sludge of five Swedish sewage treatment plants.<br />
water, acidified to pH 3, and finally subjected to SPE using the<br />
protocol described above for the aqueous samples.<br />
The dry weights of the samples were determined by<br />
weighing portions after drying them at 105 C for 24 h.<br />
2.5. Liquid chromatography–mass spectrometry<br />
Sample extracts and calibration solutions (10 mL) were injected<br />
using an AS 3000 autoinjector (Thermo Finnigan, San Jose, CA,<br />
US) into a Hydrosphere C18 guard column (10 4 mm i.d.,<br />
5 mm particle size; YMC Inc., Wilmington, NC, US) followed by<br />
an analytical column of the same type (150 4.6 mm i.d., 5 mm<br />
particle size). The antimycotics were then eluted by a linear<br />
gradient of 90:10 to 70:30 A:B (v/v) over 17 min followed by<br />
a further 4-min linear gradient to 30:70 (v/v), where A and B are<br />
water and acetonitrile, respectively (both acidified with 0.1%<br />
formic acid), at a 0.8 mL min 1 flow rate, generated using<br />
a P4000 HPLC pump (Spectra system, Thermo Finnigan) at<br />
a constant temperature of 25 C.<br />
The eluting antimycotics were detected by an LCQ Duo ion<br />
trap mass spectrometer (Thermo Finnigan) with an
electrospray ion source operating in positive ion mode. The<br />
source voltage was maintained at a constant 6.0 kV and the<br />
heated capillary temperature was set to 250 C. The MS/MS<br />
parameters and the collision energies were optimized semiautomatically<br />
and manually, respectively.<br />
2.6. Extraction yield assessment<br />
Absolute extraction yields (including matrix suppression of<br />
precursor/product ions during instrumental analysis) were<br />
investigated by fortifying 500 ng of each analyte and the IS to<br />
raw sewage water (500 mL, n ¼ 4) and digested sludge samples<br />
(2.0 g, n ¼ 3). The SPE of the water samples began immediately<br />
but the sludge samples were left for an hour prior to the liquid/<br />
solid extraction. The extraction yield was evaluated by<br />
a comparison of the analyte and IS peak areas of the samples<br />
to the corresponding peak areas in a calibration solution. The<br />
same experimental set up was used in order to assess corrected<br />
extraction yields with the exception that the IS was<br />
added post extraction. The corrected extraction yields were<br />
based on peak area ratios (analyte peak area/IS peak area) of<br />
the samples to the corresponding peak areas ratios in a calibration<br />
solution.<br />
2.7. Quantification<br />
The antimycotics were identified by retention time, precursor<br />
and product ion m/z. The most abundant transitions were<br />
monitored and the IS calibration method (peak area ratios of<br />
analytes and IS) was used for all substances except fluconazole<br />
(external calibration without recovery correction). The<br />
ten-point calibration curve, based on 500 ng of sulconazole<br />
and ranging from 1 ng mL 1 to 1000 ng mL 1 of the analytes,<br />
was prepared in aqueous acetonitrile (10%).<br />
2.8. Mass flow calculations<br />
Mass flows (g day 1 ) of each antimycotic substance at each<br />
STP were calculated using the obtained concentrations and<br />
the aqueous and solid flows during the sample collection<br />
periods, and the daily production of digested dewatered<br />
sludge (dw) obtained from monthly production figures<br />
water research 44 (2010) 649–657 653<br />
supplied by staff of the STPs. The predicted mass flows<br />
(g day 1 ) were based on total sales of the active substances to<br />
both institutional and non-institutional care during October<br />
and November 2007 (Svensson, 2008), the number of people<br />
connected to each STP and the total population of Sweden<br />
(Statistics Sweden).<br />
3. Results and discussion<br />
3.1. Quantification<br />
At least two transitions were monitored for each of the antimycotics<br />
(see Table 2), except ketoconazole, for which only<br />
one detectable product ion was formed, high CEs only resulted<br />
in additional low m/z product ions with weak signal intensities.<br />
An approach that could have been used to improve the<br />
level of confirmation is the use of the other Cl-isotope of<br />
ketoconazole, unfortunately it was not considered in this<br />
work. In addition, although two product ions were recorded<br />
for fluconazole, signals from one of them were often severely<br />
suppressed by matrix components, so it was not used for<br />
quantification. In general, matrix suppression of the analyte<br />
and IS precursor/product ion signals was immense and in this<br />
context it should be noted that the Evolute ABN sorbent<br />
column used in the SPE process was designed specifically for<br />
quantifying analytes in dirty extracts, since it reduces interference<br />
by matrix components such as macromolecules, but<br />
the sewage water and sludge extracts were brownish in color,<br />
indicating that it failed to remove substantial proportions of<br />
low molecular weight substances from these samples. It<br />
should also be noted that IS calibration was not used to<br />
quantify fluconazole, since its extraction yields were much<br />
higher than those of the other antimycotics (including the IS),<br />
and it correlated most poorly with the IS in terms of retention<br />
time and Log Kow. In addition, the IS was not natively present<br />
in the samples, thus an underestimation of the antimycotics<br />
was avoided. Terconazole was also considered as an IS,<br />
however, traces of it were found in the sewage waters and<br />
sludge from the southern part of Sweden, most likely due to<br />
its use as a pesticide in the agricultural areas, and it was<br />
therefore not used.<br />
Table 2 – Monitored ions, retention times, method detection limits and extraction yields of the antimycotics in aqueous and<br />
solid matrices.<br />
MS/MS transitions CE a (AU) t r (min) MLOQs b<br />
Extraction<br />
yields (%)<br />
IS corrected<br />
extraction yields (%)<br />
Waters (ng L 1 ) Sludge (mgkg 1 ) Waters Solids Waters Solids<br />
Terbinafine 292.8 > 141.4, 205.4, 170.3 31 17.3 5 3 27 9 40 7 133 48 109 26<br />
Clotrimazole 277.2 > 165.5, 242.5 35 17.2 5 3 23 5 34 8 117 31 79 6<br />
Ketoconazole 532.9 > 490.7 33 14.9 20 10 9 5 48 15 104 21 108 9<br />
Fluconazole 307.3 > 238.3 23 13.3 50 40 104 12 64 6 – –<br />
Econazole 383.0 > 125.3, 193.1 34 18.6 80 40 10 6 53 16 115 21 115 10<br />
Miconazole 417.7 > 161.4, 229.4 38 19.8 100 50 5 3 41 12 69 28 95 17<br />
Sulconazole (IS) 399.3 > 330.8, 183.2 22 19.1 – 13 10 43 10 – –<br />
a Collision energy in LC–MS/MS (arbitrary units).<br />
b Method limits of quantification based on the second calibration curve point in terms of mass per liter (liquid samples) and dry weight (sludge<br />
samples).
654<br />
3.2. Concentrations and mass flows of the antimycotics<br />
in sewage water and sludge<br />
The results from the Umea˚ STP antimycotic fate and the<br />
national screening study are presented in Table 3. The<br />
concentrations of the antimycotics in both the aqueous and<br />
solid matrices were generally similar to, or slightly lower than,<br />
previously reported levels of antimycotics (Kahle et al., 2008)<br />
and other pharmaceuticals, such as antibiotics (Golet et al.,<br />
2003; Carballa et al., 2004; Göbel et al., 2005a,b; Lindberg et al.,<br />
2005, 2006). However, the only substances found at higher<br />
levels than their respective LOQs in the aqueous samples were<br />
fluconazole (in both the Umea˚ STP and hospital samples) and<br />
ketoconazole (in the hospital samples). The levels found in the<br />
final effluent were in close correlation with the raw sewage<br />
water (just below LOQ in raw sewage water of Stockholm but<br />
the chromatographic peak was identified and above three<br />
times the signal to noise ratio) hence elimination from the<br />
aqueous phase is minimal. Fluconazole was not present in<br />
any of the sludge samples thus it is the only one of the<br />
monitored antimycotics that will be directly released into<br />
recipient waters of the STPs. Kahle et al. came to the same<br />
conclusion regarding the fate of fluconazole in their study<br />
made at Swiss STPs (Kahle et al., 2008). Three of the other<br />
antimycotics (clotrimazole, ketoconazole and econazole) were<br />
detected in sludge samples collected from all of the studied<br />
STPs. However, miconazole was not detected in any samples<br />
Table 3 – Sewage water and sludge antimycotic concentrations.<br />
National screening study a,b<br />
STP Um STP St STP Al STP Go STP Bo UCH<br />
RSW, ng L 1<br />
from the Alingsa˚s or Bollebygd STPs, and terbinafine was not<br />
detected in samples from the Umea˚ and Alingsa˚s STPs.<br />
According to EMEA guidelines (see Table 1), predicted environmental<br />
surface water concentrations can only be considered<br />
acceptable for fluconazole, and at best misleading for the<br />
other substances, since sludge is not included in the equation.<br />
It should be noted here that the distribution of the antimycotics<br />
between the aqueous and solid phases seems to be<br />
consistent with their respective Log Kow values. However,<br />
models based on Log K ow are not suitable to predict the environmental<br />
fate of pharmaceuticals (Tolls, 2001), and<br />
substances with similar Log K ow values have been shown to<br />
sorb to sludge to widely differing degrees (Lindberg et al., 2005,<br />
2006).<br />
Antimycotic mass flows (g day 1 ) were established to: (a)<br />
investigate the absolute distribution of these substances<br />
between the aqueous and solid phases, and (b) cross-validate<br />
them with mass flows based on sales statistics. The<br />
recorded mass flows from the STP screening study are<br />
presented together with the predicted values in Fig. 1. The<br />
recorded antimycotic mass flows were similar (within an<br />
order of magnitude) to their respective predicted equivalents<br />
at each STP with two exceptions: a) the measured<br />
mass flows in Bollebygd STP were ca. 100-fold lower than<br />
predicted, possibly because it is the smallest of the investigated<br />
STPs, and that Bollebygd is a small town lacking<br />
both a hospital and large industries; and b) the much lower<br />
RSW, ng L 1<br />
Terbinafine d d d d d d d<br />
Clotrimazole d d d d d d d<br />
Ketoconazole d d d d d 30 d<br />
Fluconazole 90 d d d d 570 120<br />
Econazole d d d d d d d<br />
Miconazole d d d d d d d<br />
FE, ng L 1<br />
water research 44 (2010) 649–657<br />
FE, ng L 1<br />
Terbinafine d d d d d d<br />
Clotrimazole d d d d d d<br />
Ketoconazole d d d d d d<br />
Fluconazole 140 100 d d d 100<br />
Econazole d d d d d d<br />
Miconazole d d d d d d<br />
Fate in Umea˚ STP study a,c<br />
DDS, mgkg 1 (dw) RSP, mgkg 1 (dw) RS, mgkg 1 (dw) DDS, mgkg 1 (dw)<br />
Terbinafine d 4 d 7 30 d 40 d<br />
Clotrimazole 30 120 120 110 50 10 310 60<br />
Ketoconazole 910 910 480 670 280 980 1300 1800<br />
Fluconazole d d d d d d d d<br />
Econazole 590 1000 660 680 210 470 240 550<br />
Miconazole 160 970 d 190 d d d 370<br />
a RSW ¼ raw sewage water, FE ¼ Final effluent, DDS ¼ digested dewatered sludge, RSP ¼ raw sewage particles, and RS ¼ raw sludge.<br />
b STP Um ¼ Umea˚ ; STP St ¼ Stockholm, Henriksdal; STP Al ¼ Alingsa˚ s; STP Go ¼ Gothenburg, Ryaverken; STP Bo ¼ Bollebygd; UCH ¼ Umea˚<br />
county hospital (Norrlands Universitetssjukhus).<br />
c Mean concentrations from two days of sampling.<br />
d Below LOQ.
observed mass flow of terbinafine, which suggests that this<br />
topically used substance (of the ones included) is only to<br />
a small extent present in the sewage waters post application.<br />
In Umea˚, the use of ketoconazole and fluconazole in<br />
the institutional care accounts for ca. 23% and 56% of the<br />
total amounts sold, respectively. However, while the mass<br />
flow of fluconazole in the UCH sewage water represented<br />
31% of the total mass flow in the STP raw sewage water,<br />
a substantially smaller proportion of the ketoconazole of<br />
UCH (1%) was found in the solid flows of the STP (digested<br />
dewatered sludge). In addition, mean mass flow values<br />
normalized to the number of people connected to each STP<br />
indicated that 53% of the purchased fluconazole eventually<br />
appeared in the final effluents, while 1, 35, 41, 155 and<br />
209% of terbinafine, ketoconazole, miconazole, clotrimazole<br />
and econazole, respectively, appeared in digested dewatered<br />
sludge. In addition, the relative distribution between<br />
the antimycotics based on sales statistics (total sales nonand<br />
institutional care) in comparison to the observed<br />
results in dewatered digested sludge, with the exception of<br />
final effluents for fluconazole, can be seen in Fig. 2.<br />
Although the comparison would be more accurate by using<br />
raw sewage water and raw sewage particles and sales<br />
statistics of each catchment area of the STPs, the posttreatment<br />
findings at the STPs show similar distribution of<br />
the antimycotics as to the predicted equivalent.<br />
water research 44 (2010) 649–657 655<br />
Fig. 2 – The relative distribution of antimycotics based on<br />
sales records compared to observed results. Abbreviations:<br />
STP Um [ Umea˚; STP St [ Stockholm, Henriksdal; STP<br />
Al [ Alingsa˚s; STP Go [ Gothenburg, Ryaverken; STP<br />
Bo [ Bollebygd; STP mean [ mean value of the<br />
distribution of the antimycotics.<br />
3.3. Fate analysis within Umea˚ sewage treatment plant<br />
The mass flow results from the fate analysis in Umea˚ STP are<br />
given in Fig. 3. Again, fluconazole was not detected in any of<br />
the solid matrices, and its mass flows in the raw sewage water<br />
Fig. 3 – Mass flows of antimycotics (g day L1 ) within the Umea˚ sewage treatment plant.
656<br />
and the final effluent were very similar (removal efficiency ca.<br />
13%). Thus, these results corroborate the indications from the<br />
screening study that fluconazole is at best marginally affected<br />
by current sewage water treatments. In contrast, clotrimazole,<br />
ketoconazole and econazole were all present in the solid flow<br />
of the raw sewage water, and they were also, together with<br />
terbinafine, detected in the raw sludge. The results for clotrimazole<br />
are consistent with data presented by Kahle et al.<br />
(2008), who found extractable amounts of this substance in<br />
the suspended solids of raw sewage water of Swiss STPs.<br />
Clotrimazole, ketoconazole, econazole and terbinafine were<br />
not strongly affected by sludge digestion either, and all of<br />
them (apart from terbinafine) were found at levels exceeding<br />
their respective LOQs in the resultant sludge. Traces of<br />
miconazole were also observed post-digestion.<br />
Our results show that fluconazole is released at significant<br />
levels into the aquatic environment, and although this<br />
substance is reportedly the least effective CYP19 aromatase<br />
inhibitor of the azole antimycotics (Trösken et al., 2004), it is<br />
a moderate Cyp3A4 and a potent Cyp2C9 inhibitor, and thus<br />
may have significant direct ecotoxicological effects. The<br />
possibility that it may also have adverse indirect environmental<br />
effects cannot be excluded, since (for example) ketoconazole<br />
has been shown to increase the sensitivity of<br />
rainbow trout to ethynylestradiol (Hasselberg et al., 2008).<br />
However, sludge is the major route for antimycotics into the<br />
environment, and thus their exposure levels and potential<br />
negative effects will depend on the fate of the sludge. Sludge is<br />
commonly used as a fertilizing agent (e.g. for golf courses and<br />
forests) or as landfill material, so we recommend that future<br />
environmental risk assessments of this group of pharmaceuticals,<br />
not only focus on the aqueous environment, but also<br />
consider their potential impact on terrestrial organisms.<br />
4. Conclusions<br />
The levels of fluconazole in the raw sewage water were<br />
similar to those of the final effluent and in addition it was<br />
not observed in digested dewatered sludge.<br />
Terbinafine, clotrimazole, ketoconazole, econazole and<br />
miconazole were only present in sludge samples, in turn,<br />
not strongly affected by sludge digestion.<br />
Observed antimycotic mass flows were in general similar to<br />
their respective predicted equivalents at each STP except for<br />
terbinafine and for the smallest STP included in this study.<br />
Future environmental risk assessments of antimycotics<br />
should consider both the aqueous and the terrestrial<br />
environment.<br />
Acknowledgements<br />
We gratefully acknowledge financial support from the<br />
Swedish Research Council for Environment, Agricultural<br />
Sciences and Spatial Planning (FORMAS). We also thank the<br />
personnel at the sewage treatment plants for their assistance.<br />
water research 44 (2010) 649–657<br />
references<br />
Carballa, M., Omil, F., Lema, J.M., Llompart, M., Garcia-Jares, C.,<br />
Rodriguez, I., Gomez, M., Ternes, T., 2004. Behaviour of<br />
pharmaceuticals, cosmetics and hormones in a sewage<br />
treatment plant. Water Res. 38, 2918–2926.<br />
EPI SuiteÔ was Developed by the EPA’s Office of Pollution<br />
Prevention Toxics and Syracuse Research Corporation (SRC),<br />
2000. U.S. Environmental Protection Agency.<br />
European Medicines Agency, 2006. Pre-authorisation Evaluation<br />
of Medicines for Human Use. http://www.emea.europa.int/<br />
pdfs/human/swp/444700en.pdf (accessed 11.12.2008).<br />
Göbel, A., Thomsen, A., McArdell, C.S., Joss, A., Giger, W., 2005a.<br />
Occurrence and sorption behaviour of sulfonamides,<br />
macrolides, and trimethoprim in activated sludge treatment.<br />
Environ. Sci. Technol. 39, 3981–3989.<br />
Göbel, A., McArdell, C.S., Suter, M.J.-F., Giger, W., 2005b. Trace<br />
determination of macrolide and sulfonamide antimicrobials,<br />
a human sulfonamide metabolite, and trimethoprim in<br />
wastewater using liquid chromatography coupled to<br />
electrospray tandem mass spectrometry. Anal. Chem. 76,<br />
4756–4764.<br />
Golet, E.M., Xifra, I., Siegrist, H., Alder, A.C., Giger, W., 2003.<br />
Environmental exposure assessment of fluoroquinolone<br />
antibacterial agents from sewage to soil. Environ. Sci. Technol.<br />
37, 3243–3249.<br />
Graybill, J.R., Drutz, D.J., 1980. Ketoconazole – a major innovation<br />
for treatment of fungal disease. Ann. Int. Med. 93, 921–923.<br />
Guengerich, P.F., 2008. Cytochrome P450 and chemical toxicology.<br />
Chem. Res. Toxicol. 21, 70–83.<br />
Gunnarsson, L., Jauhiainen, A., Kristiansson, E., Nerman, O.,<br />
Larsson, D.G.J., 2008. Evolutionary conservation of human<br />
drug targets in organisms used for environmental risk<br />
assessments. Environ. Sci. Technol. 42, 5807–5813.<br />
Gyllenhammar, I., Eriksson, H., Söderqvist, A., Lindberg, R.H.,<br />
Fick, J., Berg, C., 2009. Clotrimazole exposure modulates<br />
aromatase activity in gonads and brain during gonadal<br />
differentiation in Xenopus tropicalis frogs. Aquat. Toxicol. 91,<br />
102–109.<br />
Hasselberg, L., Westerberg, S., Wassmur, B., Celander, M.C., 2008.<br />
Ketoconazole, an antifungal imidazole, increases the<br />
sensitivity of rainbow trout to 17a-ethynylestradiol exposure.<br />
Aquat. Toxicol. 86, 256–264.<br />
Kahle, M., Buerge, I.J., Hauser, A., Müller, M.D., Poiger, T., 2008.<br />
Azole fungicides: occurrence and fate in wastewaters and<br />
surface waters. Environ. Sci. Technol. 42, 7193–7200.<br />
Kolpin, D.W., Skopec, M., Meyer, M.T., Furlong, E.T., Zaugg, S.D.,<br />
2004. Urban contribution of pharmaceuticals and other<br />
organic wastewater contaminants to streams during differing<br />
flow conditions. Sci. Total Environ. 328, 119–130.<br />
Lepesheva, G.I., Waterman, M.R., 2007. Sterol 14<br />
alpha-demethylase cytochrome P450 (CYP51), a P450 in all<br />
biological kingdoms. Biochim. Biophys. Acta-General Subjects<br />
1770, 467–477.<br />
LIF, 2008. Läkemedelsindustriföreningen (the Swedish<br />
association of the pharmaceutical industry). FASS. Elanders,<br />
Kungsbacka.<br />
Lindberg, R.H., Wennberg, P., Johansson, M.I., Tysklind, M.,<br />
Andersson, B.A.V., 2005. Screening of human antibiotic<br />
substances and determination of weekly mass flows in five<br />
sewage treatment plants in Sweden. Environ. Sci. Technol. 39,<br />
3421–3429.<br />
Lindberg, R.H., Olofsson, U., Rendahl, P., Johansson, M.I.,<br />
Tysklind, M., Andersson, B.A.V., 2006. Behavior of<br />
fluoroquinolones and trimethoprim during mechanical,<br />
chemical and active sludge treatment of sewage water and<br />
digestion of sludge. Environ. Sci. Technol. 40, 1042–1048.
Peschka, M., Roberts, P.H., Knepper, T.P., 2007. Analysis, fate<br />
studies and monitoring of the antifungal agent clotrimazole in<br />
the aquatic environment. Anal. Bioanal. Chem. 389, 959–968.<br />
Piferrer, F., Zanuy, S., Carillo, M., Solar, I.I., Devlin, R.H.,<br />
Donaldson, E.M., 1994. Brief treatment with an aromatase<br />
inhibitor during sex differentiation causes chromosomally<br />
female salmon to develop as normal, functional males. J. Exp.<br />
Zool. 270, 255–262.<br />
Roberts, P.H., Bersuder, P., 2006. Analysis of OSPAR priority<br />
pharmaceuticals using high-performance liquid<br />
chromatography–electrospray ionisation tandem mass<br />
spectrometry. J. Chromatogr. A 1134, 143–150.<br />
Roberts, P.H., Thomas, K.V., 2006. The occurrence of selected<br />
pharmaceuticals in wastewater effluent and surface waters of<br />
the lower Tyne catchment. Sci. Total Environ. 356, 143–153.<br />
Ryder, N.S., 1992. Terbinafine: mode of action and properties of the<br />
squalene epoxidase inhibition. Br. J. Dermatol. 126 (Suppl. 39), 2–7.<br />
Statistics Sweden. http://www.scb.se/templates/<br />
tableOrChart____78315.asp (accessed 02.10.08).<br />
Svensson, Elin, 2008. Apoteket AB, Sweden. Personal<br />
communication.<br />
water research 44 (2010) 649–657 657<br />
Sweetman, S.C. (Ed.), 2007. Martindale: The Complete Drug<br />
Reference. Pharmaceutical Press, London Electronic version,<br />
Edition 070214.<br />
Thomas, K.V., Hilton, M.J., 2004. The occurrence of selected<br />
human pharmaceutical compounds in UK estuaries. Mar.<br />
Pollut. Bull. 49, 436–444.<br />
Tolls, J., 2001. Sorption of veterinary pharmaceuticals in soils:<br />
a review. Environ. Sci. Technol. 35, 3397–3406.<br />
Trösken, E.R., Scholz, K., Lutz, R.W., Volkel, W., Zarn, J.A., Lutz, W.<br />
K., 2004. Comparative assessment of the inhibition of<br />
recombinant human CYP19 (aromatase) by azoles used in<br />
agriculture and as drugs for humans. Endocr. Res. 30, 387–394.<br />
Van De Steene, J.C., Lambert, W.E., 2008. Validation of a solidphase<br />
extraction and liquid chromatography–electrospray<br />
tandem mass spectrometric method for the determination of<br />
nine basic pharmaceuticals in wastewater and surface water<br />
samples. J. Chromatogr. A 1182, 153–160.<br />
Yoshida, Y., Aoyama, Y., Noshiro, M., Gotoh, O., 2000. Sterol 14demethylase<br />
P450 (CYP51) provides a breakthrough for the<br />
discussion on the evolution of cytochrome P450 gene<br />
superfamily. Biochem. Biophys. Res. Commun. 273 (3), 799–804.
Pharmaceuticals and personal care products in archived U.S.<br />
biosolids from the 2001 EPA national sewage sludge survey<br />
Kristin McClellan, Rolf U. Halden*<br />
Center for Environmental Biotechnology, The Biodesign Institute at Arizona State University, 1001 S. McAllister Avenue, Tempe,<br />
AZ 85287-5701, USA<br />
article info<br />
Article history:<br />
Received 1 July 2009<br />
Received in revised form<br />
15 December 2009<br />
Accepted 21 December 2009<br />
Available online 4 January 2010<br />
Keywords:<br />
Municipal sludge<br />
Organic wastewater contaminants<br />
Risk assessment<br />
Land application<br />
1. Introduction<br />
abstract<br />
Pharmaceuticals and personal care products (PPCPs) are<br />
common contaminants of the environment, and have been<br />
detected in surface water (Kolpin et al., 2002; Cahill et al.,<br />
2004; Moldovan, 2006; Roberts and Thomas, 2006; Tamtam<br />
et al., 2008), groundwater (Heberer et al., 2000; Lindsey et al.,<br />
water research 44 (2010) 658–668<br />
Available at www.sciencedirect.com<br />
journal homepage: www.elsevier.com/locate/watres<br />
In response to the U.S. National Academies’ call for a better assessment of chemical<br />
pollutants contained in the approximately 7 million dry tons of digested municipal sludge<br />
produced annually in the United States, the mean concentration of 72 pharmaceuticals and<br />
personal care products (PPCP) were determined in 110 biosolids samples collected by the<br />
U.S. Environmental Protection Agency (EPA) in its 2001 National Sewage Sludge Survey.<br />
Composite samples of archived biosolids, collected at 94 U.S. wastewater treatment plants<br />
from 32 states and the District of Columbia, were analyzed by liquid chromatography<br />
tandem mass spectrometry using EPA Method 1694. Thirty-eight (54%) of the 72 analytes<br />
were detected in at least one composite sample at concentrations ranging from 0.002 to<br />
48 mg kg 1 dry weight. Triclocarban and triclosan were the most abundant analytes with<br />
mean concentrations of 36 8 and 12.6 3.8 mg kg 1 (n ¼ 5), respectively, accounting for<br />
65% of the total PPCP mass found. The loading to U.S. soils from nationwide biosolids<br />
recycling was estimated at 210–250 metric tons per year for the sum of the 72 PPCPs<br />
investigated. The results of this nationwide reconnaissance of PPCPs in archived U.S.<br />
biosolids mirror in contaminant occurrences, frequencies and concentrations, those<br />
reported by the U.S. EPA for samples collected in 2006/2007. This demonstrates that PPCP<br />
releases in U.S. biosolids have been ongoing for many years and the most abundant PPCPs<br />
appear to show limited fluctuations in mass over time when assessed on a nationwide<br />
basis. The here demonstrated use of five mega composite samples holds promise for<br />
conducting cost-effective, routine monitoring on a regional and national basis.<br />
ª 2009 Elsevier Ltd. All rights reserved.<br />
2001; Fick et al., 2009), and drinking water (Stackelberg et al.,<br />
2004; Loraine and Pettigrove, 2006; Loos et al., 2007; Focazio<br />
et al., 2008), as well as in agricultural soils subject to land<br />
application of digested municipal sludge (Kinney et al., 2008;<br />
Kupper et al., 2004; Wu et al., 2009), also known as biosolids.<br />
Wastewater treatment plants were identified as one possible<br />
source for surface water contamination. Over-the-counter<br />
Abbreviations: DL, Detection limit; EPA, Environmental protection agency; NEBRA, North east biosolids and residuals association;<br />
NSSS, National sewage sludge survey; PPCP, Pharmaceuticals and personal care products; RPD, Relative percent difference;<br />
TNSSS, Targeted national sewage sludge survey.<br />
* Corresponding author. Tel.: þ1 480 727 0893; fax: þ1 480 727 0889.<br />
E-mail addresses: Kristin.McClellan@asu.edu (K. McClellan), Halden@asu.edu (R.U. Halden).<br />
0043-1354/$ – see front matter ª 2009 Elsevier Ltd. All rights reserved.<br />
doi:10.1016/j.watres.2009.12.032
and prescription drugs enter the wastewater via excretion of<br />
urine and feces containing parental drugs and their conjugates<br />
as well as other metabolites, or from disposal of<br />
unwanted or expired medications (Halling-Sorensen et al.,<br />
1998; Fent et al., 2006). Similarly, chemical constituents of<br />
personal care products may be directly disposed of into<br />
domestic wastewater. Removal of PPCPs during municipal<br />
wastewater treatment is rarely complete, thereby creating<br />
a pathway for entry of these compounds into aquatic environments<br />
via wastewater reclamation (Halling-Sorensen<br />
et al., 1998; Ternes, 1998; Daughton and Ternes, 1999; Hirsch<br />
et al., 1999) and into terrestrial environments via land<br />
application of biosolids (Ternes et al., 2004a).<br />
Of the more than 7 million tons of sewage sludge produced<br />
in the United States in 2004, about 50% was applied to land as<br />
fertilizer or soil amendment, and 45% was disposed of in<br />
landfills or as landfill cover (NEBRA, 2007). Terrestrial environments<br />
can offer effective biological, physical, and chemical<br />
attenuation mechanisms for manmade pollutants. However,<br />
they also can act as a source term for chemical migration into<br />
surface and groundwater from biosolids runoff and leachate.<br />
Pharmaceutical compounds are designed to be biologically<br />
active and therefore may have effects on non-target organisms<br />
even at trace concentrations extant in terrestrial and<br />
aquatic environments. While acute toxic effects of pharmaceuticals<br />
on non-target organisms have been investigated for<br />
some compounds, chronic toxicity and potential subtle environmental<br />
effects are only scarcely known (Fent et al., 2006).<br />
Also insufficiently investigated is the effect of mixtures of<br />
pharmaceuticals on aquatic organisms, although biochemical<br />
interactions of drugs in humans are well known. Of additional<br />
concern is the possible uptake of contaminants into food<br />
crops grown on agricultural fields that were fertilized with<br />
biosolids (Kumar et al., 2005; Dolliver et al., 2007). Currently,<br />
no regulation exists in the United States for PPCPs contained<br />
in biosolids, and a need for more information on the occurrence<br />
of and risk from these compounds has been noted by<br />
the National Research Council of the National Academies of<br />
the United States (National Research Council, 2002). In the<br />
past, analytical methods were limited, especially for trace<br />
analyses of complex environmental samples. In 2007, the<br />
release of U.S. EPA method 1694 (USEPA, 2007a) for the analysis<br />
of PPCPs in various matrices afforded the opportunity to<br />
analyze biosolids samples using a standardized protocol.<br />
The U.S. EPA has performed national sewage sludge surveys<br />
(NSSS) in 1989, 2001, and 2007. The survey conducted in 2001<br />
served to evaluate the potential need for regulations of trace<br />
levels of dioxins and polychlorinated biphenyls (USEPA, 2007b).<br />
After the 2001 survey was completed, unused samples were<br />
released to a nationwide repository of biosolids samples now<br />
maintained at the Biodesign Institute at Arizona State University.<br />
This investigation evaluated occurrences and concentrations<br />
of PPCPs in biosolids from the year 2001 to enable risk<br />
assessments and to establish a national baseline for evaluating<br />
temporal trends of PPCPs in U.S. biosolids. The analysis<br />
of composite samples by EPA Method 1694 was employed to<br />
determine average concentrations of 72 PPCPs in archived<br />
biosolids collected by the EPA, as a representative sample of<br />
the more than 16,000 treatment plants located in the contiguous<br />
United States.<br />
water research 44 (2010) 658–668 659<br />
2. Materials and methods<br />
2.1. Sampling procedure<br />
Biosolids samples with solid contents between 1% and 30%<br />
were obtained from 94 wastewater treatment plants in 32<br />
states and the District of Columbia for the 2001 National<br />
Sewage Sludge Survey (USEPA, 2007b). They were selected by<br />
the U.S. EPA to obtain a representative estimate of the<br />
occurrence of chemical contaminants in sewage sludge that is<br />
disposed of primarily by land application. Information on the<br />
exact sampling locations is available in Table S1 of the<br />
supplementary material (SM). This survey aimed to estimate<br />
levels of dioxins, dibenzofurans, and coplanar polychlorinated<br />
biphenyls in biosolids. The sampling was conducted<br />
by the U.S. EPA between February and March 2001, and<br />
samples were collected according to sampling procedures<br />
developed by the U.S. EPA (USEPA, 2001). Samples were only<br />
taken from fully processed sewage sludges intended for<br />
disposal. Eighty-nine of the 94 WWTPs had one single system<br />
for sludge treatment, therefore one sample was collected. Five<br />
facilities had two systems for treating their sludges, therefore<br />
two samples were taken from each of these plants. In addition,<br />
duplicate samples were collected from 15% of facilities<br />
(14 samples) for precision analysis. This amounted to 113<br />
samples overall. After completion of the 2001 NSSS, the<br />
samples were acquired by the Halden laboratory for further<br />
studies. For the 7-year period between acquisition and analysis<br />
in the spring of 2008, samples were stored at 20 C.<br />
2.2. Composite sample preparation<br />
From the 113 biosolids samples acquired from the EPA, three<br />
were excluded from analysis because the sample containers<br />
were broken or compromised; the remaining 110 samples were<br />
randomly grouped into five groups. Composite samples were<br />
prepared by weighing out approximately 1 g of dry weight from<br />
each sample and pooling it to obtain 5 composites each containing<br />
solids from between 21 and 24 individual samples.<br />
A duplicate of composite sample #3 was prepared to serve as<br />
a blind duplicate.<br />
2.3. Sample analysis<br />
The samples were analyzed by AXYS Analytical Services (2045<br />
Mills Road West, Sydney, British Columbia, Canada V8L 3S8)<br />
according to EPA method 1694 (USEPA, 2007a). For the purpose<br />
of compound detection, the 72 analytes were divided into four<br />
groups. All analytes were separated by liquid chromatography<br />
and detected by tandem mass spectrometry. For compounds<br />
with a respective labeled analog, the concentration was<br />
determined using the isotope dilution technique. The corresponding<br />
concentrations are deemed to be of high quality. For<br />
compounds where a labeled analog was not available, the<br />
concentration was determined using an external calibration.<br />
Quantitative data for analytes not determined by the isotope<br />
dilution method are judged to be less robust. More detailed<br />
information on the analysis method and accuracy and precision<br />
criteria for acceptance of analytical data is available in
660<br />
supplementary material (SM). Further information on study<br />
limitations can be found in the discussion section.<br />
2.4. Quality assurance<br />
To ensure system and laboratory performance, several tests<br />
were performed before sample analysis. Calibration accuracy<br />
was verified using a calibration standard solution with labeled<br />
and native analytes. Retention times of native and labeled<br />
compounds had to be within 15 seconds of the respective<br />
retention time established during the previous calibration. In<br />
addition, ongoing precision and recovery were ensured. Lab<br />
blanks were analyzed before each sample analysis. A duplicate<br />
sample analysis was performed by the lab for each batch<br />
between 7 and 20 samples. In addition to these standard<br />
procedures, a blind duplicate was included in the sample set to<br />
evaluate analysis precision. Precision is expressed as relative<br />
percent difference (RPD) between each pair of measured<br />
concentrations. It was calculated using the following equation,<br />
RPD½%Š ¼ jCsample Cduplicatej 100<br />
Csample þ Cduplicate<br />
2<br />
where C sample and C duplicate are, respectively, the concentration<br />
detected in the original sample and in its duplicate<br />
sample, and RPD is the relative percent difference.<br />
2.5. Modeling of soil and porewater concentration<br />
For assessment of their potential environmental impact, the<br />
concentrations of PPCPs after mixing with agricultural soil<br />
were calculated. The mixing ratio with soil was assumed<br />
based on the EPA-recommended rate of biosolids application<br />
of up to 4.5 dry kg per m 2 and by further assuming incorporation<br />
into soil to a depth of 10 cm (USEPA, 1994). The soil bulk<br />
density was assumed to equal 1.3 g cm 3 and the biosolids<br />
bulk density 1.6 g cm 3 . Though soil moisture content is<br />
known to vary widely depending on soil properties, for the<br />
purpose of this calculation, a soil moisture content of 22% (v/v)<br />
was used, as reported by others for agricultural soil (De Lannoy<br />
et al., 2006). A typical soil organic carbon fraction was<br />
assumed (Causarano et al., 2008), as was a biosolids organic<br />
carbon fraction of 0.4 (USEPA, 2007b). The concentration in soil<br />
and porewater at equilibrium was calculated based on the<br />
organic carbon fraction of the soil-biosolids mixture and the<br />
compound specific organic-carbon distribution coefficient<br />
(K OC) of each analyte (Chalew and Halden, 2008). Calculations<br />
took into consideration the soil moisture content, as well as<br />
the volume of biosolids added to the soil, and were conducted<br />
for an environmentally relevant pH range of 7–9. The<br />
concentrations of PPCPs in porewater of soil/biosolid mixtures<br />
were calculated using the equations below,<br />
mbiosolids<br />
Cbiosolids<br />
msoilþbiosolids<br />
Cporewater ¼<br />
fporewater rporewater þ KOC fOC<br />
msoilþbiosolids<br />
Csoil ¼ mbiosolids<br />
Cbiosolids<br />
msoilþbiosolids<br />
fporewater rporewater Cporewater<br />
msoilþbiosolids<br />
where m is the dry mass in kg m 3 of the solid matrix, C the<br />
concentration in mgkg 1 , r the density in kg m 3 , and fporewater<br />
water research 44 (2010) 658–668<br />
(1)<br />
(2)<br />
(3)<br />
and f OC the dimensionless fractions of, respectively, porewater<br />
and organic carbon in the soil/biosolids mixture.<br />
2.6. Drug usage and ecotoxicity data<br />
Information of prescription drug consumption was obtained<br />
from Internet sources (www.drugtopics.com). Exotoxicity<br />
data were taken from EPA’s Ecotox database (www.epa.gov/<br />
ecotox).<br />
2.7. Modeling of annual loading to agricultural soil<br />
The annual loading of PPCPs contained in biosolids was based<br />
on a production of 5.6–7 million dry tons of sewage sludge in<br />
the U.S., of which 50–60% is applied to land (National Research<br />
Council, 2002; Jones-Lepp and Stevens, 2007; NEBRA, 2007).<br />
3. Results<br />
3.1. Data quality assurance<br />
Lab blanks showed no detections above the detection limit for<br />
any of the analytes except for ciprofloxacin [61 mg kg 1 dry<br />
weight (dw); detection limit (DL) 32 mg kg 1 dw] and erythromycin-H2O<br />
(2.6 mg kg 1 dw; DL 1.5 mg kg 1 dw)]. However,<br />
concentrations detected in biosolids were 100- and 40-times<br />
greater than the background level detected in lab blanks.<br />
Therefore, measured concentrations for both analytes were<br />
accepted.<br />
Recovery for all analytes typically was good with an average<br />
of 112% but some notable outliers were observed leading to<br />
a range of 12–493% (See Table 1 and Table S2, SM). A total of<br />
10 analytes exceeded the methods’ lower and upper control<br />
limits. For 5 analytes (anhydrotetracycline, azithromycin,<br />
cimetidine, 4-epianhydrochlortetracycline, and 4-epianhydrotetracycline)<br />
recovery rates were below the method’s lower<br />
control limit of 40%. The concentrations reported for these<br />
analytes in biosolids may represent underestimates. The<br />
recovery rates of 5 analytes (clinafloxacin, enrofloxacin,<br />
lomefloxacin, ofloxacin, and sarafloxacin) were above the<br />
method’s upper control limits. The respective concentrations<br />
reported for these analytes may represent overestimations. All<br />
of the above mentioned analytes were quantified using labeled<br />
surrogate standards. In the case of clinafloxacin, enrofloxacin,<br />
lomefloxacin, ofloxacin, and sarafloxacin, 13 C3- 15 N-ciprofloxacin<br />
was used as an isotope-labeled surrogate standard.<br />
The recoveries were determined from spiked quality control<br />
samples, where recovery of 13 C3- 15 N-ciprofloxacin was below<br />
the method’s lower control limit. This led to extremely high<br />
recoveries of the above mentioned analytes. In the biosolids<br />
samples, recovery of 13 C 3- 15 N-ciprofloxacin was always within<br />
the method’s control limits. However, the lack of adequate<br />
recovery of 13 C 3- 15 N-ciprofloxacin in quality control samples<br />
suggests that concentrations reported in biosolids for clinafloxacin,<br />
enrofloxacin, lomefloxacin, ofloxacin, and sarafloxacin<br />
should be interpreted with caution.<br />
The duplicate analysis revealed 18% relative percent<br />
difference (RPD) for all analytes. The blind duplicate analysis<br />
of a subset of 10 analytes revealed 28% RPD. Both RPD values
Table 1 – Analytical results and summary statistics for pharmaceuticals and personal care products detected in U.S. biosolids collected in 2001 and reported on a dry weight basis.<br />
Substance name CAS RN Detection<br />
limit<br />
[mg kg 1 ]<br />
Anhydrotetracycline g<br />
Azithromycin g<br />
Standard<br />
deviation<br />
of detection<br />
limit [mgkg 1 ]<br />
n ¼ 5<br />
Frequency<br />
measured<br />
[%]<br />
Maximum<br />
concentration<br />
[mg kg 1 ]<br />
Mean detected<br />
concentration<br />
[mg kg 1 ]<br />
Standard<br />
deviation<br />
of mean<br />
concentration<br />
[mg kg 1 ] n ¼ 5<br />
Recovery<br />
[%]<br />
Isotope<br />
dilution<br />
quantification<br />
Use Lowest effect<br />
concentration<br />
for aquatic<br />
biota [mg L 1 ]<br />
Projected<br />
annual<br />
land application<br />
[kg yr 1 ] f<br />
4496-85-9 74.1 7.4 100 880 392 249 51 Antibiotic 1300–1600<br />
83905-01-5 5.6 1.6 100 1220 838 224 12 Antibiotic 2800–3500<br />
Caffeine 58-08-2 59.0 16.4 100 643 248 200 78 Stimulant<br />
Carbamazepine 298-46-4 5.6 1.6 100 238 163 56.4 139 Anticonvulsant 100 b (Jos<br />
Chlortetracycline 57-62-5 22.9 6.2 60 43.5 23.4 16.9 159 Antibiotic 36 c (Brain<br />
Cimetidine g<br />
0.05 a (Bantle<br />
et al., 1994)<br />
et al., 2003)<br />
et al., 2004)<br />
51481-61-9 6.4 2.0 100 893 504 208 41 Antacid 1700–2100<br />
Ciprofloxacin 85721-33-1 20.9 6.4 100 10800 6858 2348 98 Antibiotic<br />
5 d (Halling-<br />
Sorensen, 2001)<br />
Clarithromycin 81103-11-9 5.6 1.6 100 94.6 66.2 25.5 114 Antibiotic 220–270<br />
Codeine 76-57-3 11.2 3.2 20 29.7 n.d. 114 Analgesic<br />
Cotinine 486-56-6 9.9 3.3 100 38.6 28.1 8.3 103<br />
Nicotine<br />
metabolite<br />
Diltiazem 42399-41-7 1.3 0.1 100 109 45.2 34.2 110 Antianginal 150–190<br />
Diphenhydramine 58-73-1 2.3 0.6 100 1740 1166 516 101 Antihistamine 3900–4800<br />
Doxycycline 564-25-0 23.9 6.3 100 1780 966 436 85 Antibiotic 3200–4000<br />
Enrofloxacin h<br />
93106-60-6 19.5 6.5 60 28.6 n.d. 190 Antibiotic 49 d (Robinson<br />
et al., 2005)<br />
4-Epianhydrotetracycline<br />
g<br />
4465-65-0 77.1 20.2 100 399 261 71.7 39 Antibiotic 900–1100<br />
4-Epichlortetracycline<br />
14297-93-9 56.3 16.0 40 93.0 n.d. 161 Antibiotic<br />
4-Epitetracycline 23313-80-6 74.4 18.8 100 3040 2376 517 82 Antibiotic 8000–9800<br />
Erythromycin-H 2O 114-07-8 1.6 0.2 100 183 81.5 52.3 97 Antibiotic<br />
Fluoxetine 54910-89-3 8.2 1.0 100 258 171 46.6 89 Antidepressant<br />
Gemfibrozil 25812-30-0 12.2 12.3 100 159 152 13.2 107<br />
Antihyperlipidemic<br />
22 700 c (Williams<br />
et al., 1992)<br />
36 h, c (Flaherty<br />
and Dodson, 2005)<br />
30,040 c (Zurita<br />
et al., 2007)<br />
Ibuprofen 15687-27-1 122 123 80 359 246 121 109 Anti-inflammatory 830–1000<br />
Isochlortetracycline 514-53-4 22.5 6.4 60 36.0 n.d. 70 Antibiotic<br />
Lomefloxacin h<br />
98079-51-7 13.2 2.1 40 16.1 n.d. 388 Antibiotic 106 e (Robinson<br />
et al., 2005)<br />
Metformin 657-24-9 119 32.5 80 456 305 152 116 Antidiabetic 1000–1300<br />
Miconazole 22916-47-8 5.8 1.3 100 1100 777 266 86 Antifungal 2600–3200<br />
Minocycline 10118-90-8 1880 524 80 2630 1884 939 54 Antibiotic 6300–7800<br />
830–1000<br />
550–680<br />
80–100<br />
23,000–28,000<br />
100–120<br />
270–340<br />
580–710<br />
510–630<br />
(continued on next page)
Table 1 (continued)<br />
Substance name CAS RN Detection<br />
limit<br />
[mg kg 1 ]<br />
Standard<br />
deviation<br />
of detection<br />
limit [mgkg 1 ]<br />
n ¼ 5<br />
Frequency<br />
measured<br />
[%]<br />
Maximum<br />
concentration<br />
[mg kg 1 ]<br />
Mean detected<br />
concentration<br />
[mg kg 1 ]<br />
Standard<br />
deviation<br />
of mean<br />
concentration<br />
[mg kg 1 ] n ¼ 5<br />
Naproxen 22204-53-1 24.3 24.6 100 273 119 79 106<br />
Recovery<br />
[%]<br />
Isotope<br />
dilution<br />
quantification<br />
Antiinflammatory<br />
Use Lowest effect<br />
concentration<br />
for aquatic<br />
biota [mg L 1 ]<br />
Projected<br />
annual<br />
land application<br />
[kg yr 1 ] f<br />
Norfloxacin 70458-96-7 63.5 6.3 100 418 289 74.0 136 Antibiotic 18 000 b (Yang<br />
et al., 2008)<br />
970–1200<br />
Ofloxacin h<br />
82419-36-1 7.7 4.2 100 8140 5446 1941 493 Antibiotic 21 d (Robinson<br />
et al., 2005)<br />
18000–23,000<br />
Oxytetracyclin 79-57-2 22.8 5.8 100 114 87.5 22.2 90 Antibiotic 50 d (Hanson<br />
et al., 2006)<br />
300–360<br />
Ranitidine 66357-35-5 6.5 1.7 100 30.1 21.0 8.0 51 Antacid 70–90<br />
Sulfamethoxazole 723-46-6 2.9 0.7 20 3.3 n.d. 98 Antibiotic<br />
9 e (Brain<br />
et al., 2008)<br />
Sulfanilamide 63-74-1 56.2 16.0 40 87.3 n.d. 101 Antibiotic<br />
Tetracycline 60-54-8 57.5 15.5 100 2790 1914 691 124 Antibiotic 47 e (Brain<br />
et al., 2004)<br />
Thiabendazole 148-79-8 5.9 1.2 100 370 110 131 103 Fungicide 310 c (EPA, 2000) 370–460<br />
Triclocarban 101-20-2 183 67.0 100 48100 36060 8049 96 Disinfectant<br />
Triclosan 3380-34-5 487 493 100 19700 12640 3816 105 Disinfectant<br />
Trimethoprim 738-70-5 10.6 3.8 60 60.5 26.0 21.5 97 Antibiotic<br />
0.101 c<br />
(EPA/OTS, 1992)<br />
0.12 b (Wilson<br />
et al., 2003)<br />
16 000 b (Luetzhoft<br />
et al., 1999)<br />
400–500<br />
6400–7900<br />
120,000–150,000<br />
42,000–52,000<br />
a Amphibium.<br />
b Green algae.<br />
c Crustacean.<br />
d Cyanobacteria.<br />
e Macrophyte.<br />
f Projected deposition rates of PPCPs on land based on 5.6–6.9 million dry tons of annual sewage sludge production and a land application rate of 60% (National Research Council, 2002; USEPA, 2003; USGS, 2006;<br />
Jones-Lepp and Stevens, 2007).<br />
g Concentration may be underestimated due to low recovery in quality control samples.<br />
h Concentration may be overestimated due to high recovery in quality control samples.<br />
90–110
were in control with respect to the target RPD of 30% or less.<br />
The RPD value improved to 11% for 9 analytes, when excluding<br />
results for metformin, whose lack of detection in the blind<br />
duplicate unfavorably increased the summary statistic for<br />
measurement precision. A non-blinded duplicate analysis,<br />
performed by the contract laboratory for these 10 analytes,<br />
showed an RPD value of 11%.<br />
3.2. Study representativeness and sample integrity<br />
The prolonged storage of samples between sampling event<br />
and analysis may have allowed for the chemical degradation<br />
of labile analytes to occur. Therefore, any results of this study<br />
are conservative with respect to the detection frequency and<br />
concentration of compounds found. In other words, due to<br />
pooling of a large number of samples, analytes occurring<br />
infrequently and at low concentrations may have been diluted<br />
out to below the detection limit. A comparison of the presented<br />
data with the EPA data (USEPA, 2009), reveals that the<br />
mean concentrations of all analytes show no statistically<br />
significant difference within the 95th percentile confidence<br />
interval. Therefore, the prolonged storage did not impair the<br />
detection of multiple analytes at elevated concentrations in<br />
archived samples.<br />
3.3. Occurrence of PPCPs in biosolids<br />
All composites tested positive for at least 26 analytes. Of the 72<br />
PPCPs targeted, 38 (54%) were detected at concentrations<br />
ranging from the low parts-per-billion (ppb) to the parts-permillion<br />
(ppm) range. These 38 PPCPs include 8 that have not<br />
previously been reported in biosolids in the peer-reviewed<br />
literature but that also were observed in the U.S. EPA’s TNSSS<br />
published online (USEPA, 2009). The remaining 34 PPCPs were<br />
Concentration in biosolids [µg kg -1 dry weight]<br />
100<br />
000<br />
10<br />
000<br />
1000<br />
100<br />
10<br />
1<br />
triclocarban<br />
triclosan<br />
‡<br />
ciprofloxacin<br />
ofloxacin<br />
*<br />
4-epitetracycline<br />
tetracycline<br />
minocycline<br />
water research 44 (2010) 658–668 663<br />
‡<br />
diphenhydramine<br />
doxycycline<br />
azithromycin<br />
miconazole<br />
cimetidine<br />
‡<br />
‡<br />
*<br />
anhydrotetracycline<br />
metformin<br />
norfloxacin<br />
‡<br />
*<br />
not detected in any of the composite biosolids samples. The<br />
mean total concentration of all targeted PPCPs combined in<br />
the five composite samples was 74.4 mg kg 1 dw of sewage<br />
sludge 21.4 mg kg 1 standard deviation (n ¼ 5).<br />
The two most abundant contaminants were the disinfectants<br />
triclocarban (48% of total detected PPCP mass) and triclosan<br />
(17%) (Fig. 1). Their mean concentrations were 36 8<br />
and 12.6 3.8 mg kg 1 dw (n ¼ 5), respectively (Table 1). The<br />
second most abundant class of PPCPs found was antibiotics. In<br />
order of decreasing concentration, ciprofloxacin, ofloxacin, 4epitetracycline,<br />
tetracycline, minocycline, doxycycline and<br />
azithromycin were found at concentrations between 6.8 2.3<br />
and 0.8 0.2 mg kg 1 dw (n ¼ 5) (Table 1). The combined mass<br />
of all antibiotics constituted about 29% of the total mass of<br />
PPCPs per sample.<br />
In addition to the 3 tetracyclines reported above, three<br />
tetracycline antibiotics were found for which peer-reviewed<br />
occurrence data in sewage sludge thus far were lacking.<br />
Anhydrotetracycline, 4-epianhydrotetracycline, and chlortetracycline<br />
(in order of decreasing concentration) together<br />
accounted for 6.9 2.5 mg kg 1 dw (n ¼ 5) (Table 1). Also not<br />
previously reported in biosolids were two prescription drugs,<br />
metformin and ranitidine. They together accounted for 0.4%<br />
of the total mass of PPCPs found in sewage sludge composites.<br />
4. Discussion<br />
4.1. Study limitations<br />
For this study, a relatively large number of individual samples<br />
were combined to form 5 composite pools or mega composites.<br />
This approach served to reduce the number of samples to be<br />
analyzed in order to obtain a defensible estimate of mean<br />
*<br />
4-epianhydrotetracycline<br />
caffeine<br />
ibuprofen<br />
fluoxetine<br />
carbamazepine<br />
gemfibrozil<br />
naproxen<br />
thiabendazole<br />
oxytetracycline<br />
erythromycin-H2O clarithromycin<br />
diltiazem<br />
sulfanilamide<br />
*<br />
*<br />
4-epichlortetracycline<br />
cotinine<br />
trimethoprim<br />
chlortetracycline<br />
ranitidine<br />
*<br />
‡*<br />
ppm<br />
ppb<br />
*<br />
‡*<br />
isochlortetracycline<br />
enrofloxacin<br />
codeine<br />
lomefloxacin<br />
Fig. 1 – Rank order of mean concentrations for 38 PPCPs detected in composites of a total of 110 U S biosolids samples from<br />
94 treatment plants in 32 states and the District of Columbia. Newly detected compounds are shown in darker hue. Error<br />
bars depict ± one standard deviation (n [ 5). Some concentrations represent estimates only (z) and some analytes were<br />
detected inconsistently (*).<br />
sulfamethoxazole *
664<br />
analyte concentrations across the various treatment facilities<br />
represented. While being efficient and economical for the<br />
intended purpose, this approach was not well suited to capture<br />
the full spectrum of concentrations of individual PPCPs as<br />
a function of plant type, treatment processes employed, populations<br />
served, as well as of geographical locations and<br />
climate zones represented. As a comparison with the EPA<br />
TNSSS data reveals (USEPA, 2009), the detection frequency of<br />
less abundant analytes was significantly reduced in composite<br />
samples compared to individual sample analysis. Therefore,<br />
analytes that were not detected will represent a conservative<br />
estimate, and may still occur at detectable concentrations in<br />
individual samples from specific plants. While the mega<br />
composite approach cannot serve to determine variability<br />
between the large numbers of WWTPs studied, it was found to<br />
be suitable for identifying major contaminants of concern as<br />
well as their average concentration in a large sample set.<br />
4.2. Sanitizing agents<br />
Triclocarban, which in previous U.S. studies had been found in<br />
concentrations ranging from 5.97 to 51 mg kg 1 dry weight<br />
(dw) (Heidler et al., 2006; Chu and Metcalfe, 2007; Sapkota et al.,<br />
2007), was detected in every composite sample assayed. Also<br />
found in significant amounts was triclosan, which has been<br />
observed in a number of U.S. studies of sewage sludge with<br />
reported concentrations ranging from 0.53 to 30 mg kg 1 dw<br />
(Chu and Metcalfe, 2007; Heidler and Halden, 2007; Kinney<br />
et al., 2008; McAvoy et al., 2002; USEPA, 2003). The EPA TNSSS<br />
(USEPA, 2009) found triclocarban and triclosan at concentrations<br />
of up to 441 and 133 mg kg 1 dw, respectively, with mean<br />
concentrations at 38.7 59.7 and 12 18 mg kg 1 dw (n ¼ 74),<br />
respectively. The relatively high mean concentrations of triclocarban<br />
and triclosan reported here and by the U.S. EPA<br />
(USEPA, 2009) in U.S. biosolids are in line with the intense<br />
usage of these antimicrobials and their high octanol/water<br />
partitioning coefficient (log KOW) of 4.9 and 4.8, respectively<br />
(both at neutral pH), which indicates significant potential of<br />
both compounds for sorption to biosolids (Halden and Paull,<br />
2005). In addition, triclosan was found to be persistent in<br />
biosolids after aeration (Ying et al., 2007). Both triclosan and<br />
triclocarban concentrations found in the present study fall<br />
within the range of concentrations previously reported.<br />
Information on previously reported concentrations of PPCPs in<br />
biosolids is available in Table S2, SM.<br />
4.3. Antibiotics<br />
The most abundant antibiotic was ciprofloxacin (6.8 2.3 mg<br />
kg 1 dw; n ¼ 5), which is among the 30 most prescribed drugs<br />
in the United States, according to Internet sources. Ciprofloxacin,<br />
a metabolite of enrofloxacin, is polar and therefore<br />
prone to electrostatic interactions with the negatively charged<br />
surfaces of microbes that are found in high concentrations<br />
especially in secondary sludge (Ternes et al., 2004b). Ciprofloxacin<br />
has been detected in sewage sludge by several studies<br />
conducted in Sweden and Switzerland. Detected concentrations<br />
ranged from 6 10 5 –11 mg kg 1 dw (Golet et al., 2002;<br />
Lindberg et al., 2005, 2006, 2007). Therefore, the mean<br />
concentration found in the present study (6.8 2.3 mg kg 1<br />
water research 44 (2010) 658–668<br />
dw; n ¼ 5) falls in the mid range of concentrations reported<br />
from outside of the U.S. The EPA TNSSS also identified ciprofloxacin<br />
as an abundant microcontaminant in sewage sludge,<br />
which was detected in every sample analyzed, yielding<br />
a mean concentration of 8.7 8.5 mg kg 1 dw.<br />
At a mean concentration of 5.4 1.9 mg kg 1 dw, ofloxacin<br />
was the fourth most abundant contaminant found in biosolids.<br />
It is fairly hydrophilic (log KOW of 0.2) and does not appear on<br />
the list of the top 200 prescription drugs. Its detection at<br />
elevated levels in every composite sample came as a surprise.<br />
However, ofloxacin had been found in concentrations of up to<br />
2mgkg 1 dw in biosolids from Sweden (Lindberg et al., 2005). It<br />
can be speculated that the carboxyl group contained in ofloxacin<br />
may form an ion complex with exchangeable cations<br />
associated with negatively charged surfaces.<br />
Among the 10 most abundant PPCPs of the present study<br />
were three tetracycline antibiotics, 4-epitetracycline, tetracycline<br />
and minocycline, which had not been reported in<br />
biosolids before in the peer-reviewed literature. In addition,<br />
doxycycline was found as the ninth most abundant PPCP at<br />
a mean concentration of 1 0.4 mg kg 1 dw. It had not been<br />
detected in biosolids from the U.S. before and only once in<br />
a Swedish study that found concentrations similar to those<br />
reported here (Lindberg et al., 2005). All tetracycline antibiotics<br />
are fairly hydrophilic (log KOW of 1.33 for tetracycline/4-epitetracycline,<br />
0.42 for minocycline, and 1.36 for doxycycline).<br />
Despite their hydrophilic character, the mean concentrations<br />
reported here are around 1–2 mg kg 1 dw. Tetracycline antibiotics<br />
are known to precipitate with ions of magnesium, calcium<br />
and ferric iron, and therefore accumulate in the solid fraction<br />
during wastewater treatment. Tetracycline, doxycycline and<br />
minocycline also rank among the 200 prescription drugs most<br />
widely used in the U.S. Other tetracycline antibiotics analyzed<br />
in this study were only found at mean concentrations below<br />
0.4 mg kg 1 dw, but the concentrations of anhydrotetracycline,<br />
4-epianhydrotetracycline and 4-epianhydrochlortetracycline<br />
were possibly underestimated due to low recoveries. The sum<br />
of tetracycline antibiotics found in sewage sludge constitute<br />
about 8 1.3mgkg 1 dw, which is similar to the findings of the<br />
EPA TNSSS that found about 5 mg kg 1 dw (USEPA, 2009).<br />
Azithromycin was found at 0.8 0.2 mg kg 1 dw. It ranked<br />
as the 6th most frequently prescribed drug in 2007 and is also<br />
fairly hydrophobic. Due to both these properties and the fact<br />
that biodegradation of azithromycin was found to be insignificant<br />
(Ericson, 2007), one would expect high concentrations<br />
in biosolids. Yet, there are 7 drugs (not counting triclocarban<br />
and triclosan) that were found at higher concentrations.<br />
However, a low recovery of azithromycin of only 12% may<br />
indicate that actual azithromycin concentrations in biosolids<br />
are much higher than the detected concentration. Azithromycin<br />
has been detected in previous studies at concentrations<br />
of up to 6.5 mg kg 1 dw in the U.S. (Jones-Lepp and<br />
Stevens, 2007) and at up to 0.16 mg kg 1 dw in sludge from<br />
Germany and Switzerland (Gobel et al., 2005a, 2005b).<br />
4.4. Bioavailability and soil/porewater equilibria<br />
To explore the importance of bioavailability of PPCPs<br />
sequestered in biosolids, the concentrations of individual<br />
PPCPs that are anticipated to occur in the solid and liquid
phases upon mixing of land applied biosolids into soil were<br />
calculated (Fig. 2). In fully equilibrated soil-biosolids mixtures<br />
(assuming an EPA-recommended mixing ratio of approximate<br />
25:1), concentrations of PPCPs on soil particles are expected to<br />
fall into the ppb range for most analytes, except for triclocarban,<br />
which was projected to be present at around 1 ppm(w/w)<br />
dw. Since most of the PPCPs found in biosolids are fairly<br />
hydrophobic, their calculated concentrations in porewater at<br />
equilibrium typically were quite low (Fig. 2 B) (Kinney et al.,<br />
2008). A comparison of these estimated dissolved PPCP levels<br />
in porewater with the lowest effect concentrations for aquatic<br />
organisms (red circles in Fig. 2 B) suggests that the leaching of<br />
dissolved PPCPs into surface waters probably does not present<br />
an important mechanism for exposure of aquatic biota for the<br />
majority of analytes detected. Concentrations calculated for<br />
soil porewater typically were several orders of magnitude<br />
below the lowest effect concentration reported for aquatic<br />
organisms. Notable exceptions were 6 analytes (Fig. 2) that are<br />
expected to yield potentially problematic concentrations in<br />
porewater after land application and partitioning of biosolidsderived<br />
PPCPs. These include the antibiotics ciprofloxacin,<br />
ofloxacin and tetracycline, as well as the stimulant caffeine,<br />
Conc. in dry weight soil at<br />
pH 7-9 [µg kg-1 ]<br />
Conc. in porewater<br />
at pH 7-9 [µg L-1 ]<br />
1000<br />
100<br />
10<br />
1<br />
0.<br />
1<br />
0.<br />
01<br />
100<br />
000<br />
10<br />
000<br />
1000<br />
100<br />
10<br />
1<br />
0.<br />
1<br />
0.<br />
01<br />
0.<br />
001<br />
0.<br />
0001<br />
‡<br />
*<br />
‡<br />
‡<br />
*<br />
‡<br />
*<br />
‡*<br />
and the two sanitizing agents triclosan and triclocarban<br />
(Fig. 2).<br />
4.5. Risk assessment data gaps<br />
These results suggest that aside from the 7 notable exceptions<br />
discussed above, the majority of the PPCPs detected in<br />
biosolids in this study likely exert no acute effects on aquatic<br />
organisms, assuming that biosolids are applied as regulated<br />
by the EPA (USEPA, 1994) and that the migration of solids from<br />
agricultural land to surface water via soil erosion and runoff<br />
can be completely prevented. While the former assumption is<br />
plausible, the latter may not always apply, as soil erosion is<br />
a common phenomenon.<br />
However, chronic toxicity as well as effects from<br />
mixtures of PPCPs on non-target organisms cannot be<br />
assessed due to lack of appropriate toxicity data. It has been<br />
speculated that the presence of sub-therapeutic concentrations<br />
of antibiotics may adversely affect soil microbial<br />
community structures as well as induce spreading of resistance<br />
among bacterial pathogens. In addition to influencing<br />
microbial populations, it has been shown that some<br />
*<br />
*<br />
‡*<br />
*<br />
4-epianhydrotetracycline<br />
4-epichlortetracycline<br />
4-epitetracycline<br />
anhydrotetracycliine<br />
azithromycin<br />
caffeine<br />
carbamazepine<br />
chlortetracycline<br />
cimetidine<br />
ciprofloxacin<br />
clarithromycin<br />
codeine<br />
cotinine<br />
diltiazem<br />
diphenhydramine<br />
doxycycline<br />
enrofloxacin<br />
erythromycin-H2O fluoxetine<br />
gemfibrozil<br />
ibuprofen<br />
isochlortetracycline<br />
lomefloxacin<br />
metformin<br />
miconazole<br />
minocycline<br />
naproxen<br />
norfloxacin<br />
ofloxacin<br />
oxytetracycline<br />
ranitidine<br />
sulfamethoxazole<br />
sulfanilamide<br />
tetracycline<br />
thiabendazole<br />
triclocarban<br />
triclosan<br />
trimethoprim<br />
‡<br />
*<br />
‡<br />
‡<br />
*<br />
‡<br />
water research 44 (2010) 658–668 665<br />
*<br />
‡*<br />
*<br />
*<br />
‡*<br />
*<br />
4-epianhydrotetracycline<br />
4-epichlortetracycline<br />
4-epitetracycline<br />
anhydrotetracycliine<br />
azithromycin<br />
caffeine<br />
carbamazepine<br />
chlortetracycline<br />
cimetidine<br />
ciprofloxacin<br />
clarithromycin<br />
codeine<br />
cotinine<br />
diltiazem<br />
diphenhydramine<br />
doxycycline<br />
enrofloxacin<br />
erythromycin-H2O fluoxetine<br />
gemfibrozil<br />
ibuprofen<br />
isochlortetracycline<br />
lomefloxacin<br />
metformin<br />
miconazole<br />
minocycline<br />
naproxen<br />
norfloxacin<br />
ofloxacin<br />
oxytetracycline<br />
ranitidine<br />
sulfamethoxazole<br />
sulfanilamide<br />
tetracycline<br />
thiabendazole<br />
triclocarban<br />
triclosan<br />
trimethoprim<br />
*<br />
*<br />
‡<br />
‡<br />
*<br />
*<br />
*<br />
*<br />
*<br />
*<br />
100<br />
000<br />
10<br />
000<br />
1000<br />
100<br />
10<br />
1<br />
0.<br />
1<br />
0.<br />
01<br />
0.<br />
001<br />
0.<br />
0001<br />
Fig. 2 – Predicted equilibrium concentrations of PPCPs associated with particulates of soil-biosolids mixtures (top) and<br />
dissolved in porewater (bottom) after land application of biosolids on agricultural soil. Data depict the environmentally<br />
relevant range between pH 7 and 9. Circles represent the lowest ecological effect concentrations contained in the EPA Ecotox<br />
database. Some concentrations were calculated based on estimates only (z) and some analytes were detected inconsistently (*).<br />
Effect concentration [µg L -1 ]
666<br />
antibiotics, specifically drugs belonging to the tetracyclines,<br />
fluoroquinolones and sulfonamides, may be taken up by<br />
crop plants (Migliore et al., 2003; Kumar et al., 2005; Dolliver<br />
et al., 2007). This presents a potential exposure pathway to<br />
humans through ingestion of contaminated food and may<br />
result in the promotion of resistant bacteria in humans<br />
(Shoemaker et al., 2001).<br />
Furthermore, information is lacking to determine risks to<br />
the health of agricultural soils and soil-dwelling organisms.<br />
Few studies have examined the half-lives in soils of PPCPs<br />
sequestered in biosolids. The effect of ppm levels of sanitizing<br />
agents on soil microbial communities has rarely been investigated<br />
to date (Liu et al., 2009). Similarly, the half-life of PPCPs<br />
in biosolids-amended soils will require additional research to<br />
inform risk assessment analyses. Passage through municipal<br />
digesters and chemical aging may reduce the bioavailability of<br />
PPCPs and with it the risk of chemical uptake of and exposure<br />
to soil-dwelling organisms. However, a reduced bioavailability<br />
also may imply a prolonged half-life of these compounds in<br />
the environment, with possible delayed release in sensitive<br />
compartments. Furthermore, studies are lacking on the<br />
potential of biosolids-derived antimicrobials and antibiotics to<br />
exert selective pressure for the enrichment of drug-resistant<br />
microorganisms, a scenario demonstrated in vitro (Braoudaki<br />
and Hilton, 2004). Also unavailable are threshold concentrations<br />
for toxic effects in terrestrial organisms of many PPCPs,<br />
including some that have been detected in biosolids at ppm<br />
levels. The EPA’s Ecotox database (www.epa.gov/ecotox)<br />
provides only some toxicity values in aquatic organisms for<br />
the PPCPs investigated here. Bioaccumulation and biomagnification<br />
are other aspects that will need to be considered<br />
for risk assessment purposes. Although PPCPs are not<br />
typically thought of as representing persistent hydrophobic<br />
pollutants, some may be subject to bioaccumulation and<br />
possibly biomagnification thereafter in both terrestrial and<br />
aquatic environments. Bioaccumulation was demonstrated<br />
for triclosan and triclocarban in lab and field studies that<br />
examined uptake of these compounds from soil, sediment<br />
and water (Coogan and La Point, 2008; Higgins et al., 2009;<br />
Kinney et al., 2008).<br />
5. Conclusions<br />
Overall, this study reemphasizes the significance of biosolids<br />
recycling as a mechanism for the release of PPCPs into the<br />
environment. Based on the mean concentrations of all analytes<br />
detected, it is estimated that the total loading to U.S. soils<br />
from nationwide biosolids recycling is on the order of<br />
210–250 metric tons per year for the 72 PPCPs investigated<br />
here.<br />
It is concluded that, despite large variations found by the<br />
U.S. EPA between different treatment plants, mean concentrations<br />
of PPCPs in U.S. biosolids on a nationwide basis have<br />
remained fairly constant between 2001 and 2007 and possibly<br />
longer. Good agreement between this and the U.S. EPA study<br />
further suggests that the here demonstrated use of mega<br />
composite samples represents a cost-effective approach for<br />
collecting regional and nationwide information on average<br />
concentrations of contaminants in biosolids.<br />
water research 44 (2010) 658–668<br />
Acknowledgements<br />
We thank Rick Stevens and Harry B. McCarty from the U.S. EPA<br />
for providing samples from the 2001 National Sewage Sludge<br />
Survey. We also thank Randhir Deo and Jochen Heidler for<br />
their valuable input. This work was supported in part by the<br />
National Institute of Environmental Health Sciences by NIEHS<br />
grant 1R01ES015445 and by the Johns Hopkins University<br />
Center for a Livable Future.<br />
Appendix.<br />
Supplementary material<br />
Additional details regarding experimental procedures and<br />
results can be found in the Supplementary Material accompanying<br />
this article. This information is available free of<br />
charge via the Internet at www.iwaponline.com.<br />
Supplementary data associated with this article can be<br />
found in the online version at doi:10.1016/j.watres.2009.12.<br />
032.<br />
references<br />
Bantle, J.A., Burton, D.T., Dawson, D.A., Dumont, J.N., Finch, R.A.,<br />
Fort, D.J., Linder, G., Rayburn, J.R., Buchwalter, D.,<br />
Gaudethull, A.M., Maurice, M.A., Turley, S.D., 1994. Fetax<br />
interlaboratory validation study - Phase II testing.<br />
Environmental Toxicology and Chemistry 13 (10), 1629–1637.<br />
Brain, R.A., Johnson, D.J., Richards, S.M., Sanderson, H., Sibley, P.K.,<br />
Solomon, K.R., 2004. Effects of 25 pharmaceutical compounds<br />
to Lemna gibba using a seven-day static-renewal test.<br />
Environmental Toxicology and Chemistry 23 (2), 371–382.<br />
Brain, R.A., Ramirez, A.J., Fulton, B.A., Chambliss, C.K., Brooks, B.<br />
W., 2008. Herbicidal Effects of Sulfamethoxazole in Lemna<br />
gibba: Using p-Aminobenzoic Acid As a Biomarker of Effect.<br />
Environmental Science & Technology 42 (23), 8965–8970.<br />
Braoudaki, M., Hilton, A.C., 2004. Low level of cross-resistance<br />
between triclosan and antibiotics in Escherichia coli K-12 and<br />
E. coli O55 compared to E. coli O157. Fems Microbiology<br />
Letters 235, 305–309.<br />
Cahill, J.D., Furlong, E.T., Burkhardt, M.R., Kolpin, D., Anderson, L.G.,<br />
2004. Determination of pharmaceutical compounds in surfaceand<br />
ground-water samples by solid-phase extraction and highperformance<br />
liquid chromatography-electrospray ionization<br />
mass spectrometry. Journal of Chromatography A 1041, 171–180.<br />
Causarano, H.J., Franzluebbers, A.J., Shaw, J.N., Reeves, D.W.,<br />
Raper, R.L., Wood, C.W., 2008. Soil organic carbon fractions<br />
and aggregation in the Southern Piedmont and Coastal Plain.<br />
Soil Science Society of America Journal 72, 221–230.<br />
Chalew, T., Halden, R., 2008. Environmental exposure of aquatic<br />
and terrestrial biota to triclosan and triclocarban. Journal of<br />
the American Water Resources Association 45 (1), 4–13.<br />
Chu, S.G., Metcalfe, C.D., 2007. Simultaneous determination of<br />
triclocarban and triclosan in municipal biosolids by liquid<br />
chromatography tandem mass spectrometry. Journal of<br />
Chromatography A 1164, 212–218.<br />
Coogan, M.A., La Point, T.W., 2008. Snail bioaccumulation of<br />
triclocarban, triclosan, and methyltriclosan in a North Texas,<br />
USA, stream affected by wastewater treatment plant runoff.<br />
Environmental Toxicology and Chemistry 27, 1788–1793.
Daughton, C.G., Ternes, T.A., 1999. Pharmaceuticals and personal<br />
care products in the environment: agents of subtle change?<br />
Environmental Health Perspectives 107, 907–938.<br />
De Lannoy, G.J.M., Verhoest, N.E.C., Houser, P.R., Gish, T.J., Van<br />
Meirvenne, M., 2006. Spatial and temporal characteristics of<br />
soil moisture in an intensively monitored agricultural field<br />
(OPE3). Journal of Hydrology 331, 719–730.<br />
Dolliver, H., Kumar, K., Gupta, S., 2007. Sulfamethazine uptake by<br />
plants from manure-amended soil. Journal of Environmental<br />
Quality 36, 1224–1230.<br />
Ericson, J.F., 2007. An evaluation of the OECD 308 water/sediment<br />
systems for investigating the biodegradation of<br />
pharmaceuticals. Environmental Science & Technology 41,<br />
5803–5811.<br />
Fent, K., Weston, A.A., Caminada, D., 2006. Ecotoxicology of<br />
human pharmaceuticals. Aquatic Toxicology 76, 122–159.<br />
Fick, J., Soderstrom, H., Lindberg, R.H., Phan, C., Tysklind, M.,<br />
Larsson, D.G.J., 2009. Contamination of surface,<br />
ground, and drinking water from pharmaceutical<br />
production. Environmental Toxicology and Chemistry 28,<br />
2522–2527.<br />
Focazio, M.J., Kolpin, D.W., Barnes, K.K., Furlong, E.T., Meyer, M.T.,<br />
Zaugg, S.D., Barber, L.B., Thurman, M.E., 2008. A national<br />
reconnaissance for pharmaceuticals and other organic<br />
wastewater contaminants in the United States – II untreated<br />
drinking water sources. Science of the Total Environment 402,<br />
201–216.<br />
Gobel, A., Thomsen, A., McArdell, C.S., Alder, A.C., Giger, W.,<br />
Theiss, N., Loffler, D., Ternes, T.A., 2005a. Extraction and<br />
determination of sulfonamides, macrolides, and<br />
trimethoprim in sewage sludge. Journal of Chromatography A<br />
1085, 179–189.<br />
Gobel, A., Thomsen, A., McArdell, C.S., Joss, A., Giger, W., 2005b.<br />
Occurrence and sorption behavior of sulfonamides,<br />
macrolides, and trimethoprim in activated sludge treatment.<br />
Environmental Science & Technology 39, 3981–3989.<br />
Golet, E.M., Strehler, A., Alder, A.C., Giger, W., 2002.<br />
Determination of fluoroquinolone antibacterial agents in<br />
sewage sludge and sludge-treated soil using accelerated<br />
solvent extraction followed by solid-phase extraction.<br />
Analytical Chemistry 74, 5455–5462.<br />
Halden, R.U., Paull, D.H., 2005. Co-occurrence of triclocarban and<br />
triclosan in US water resources. Environmental Science &<br />
Technology 39, 1420–1426.<br />
Halling-Sorensen, B., Nielsen, S.N., Lanzky, P.F., Ingerslev, F.,<br />
Lutzhoft, H.C.H., Jorgensen, S.E., 1998. Occurrence, fate and<br />
effects of pharmaceutical substances in the environment -<br />
a review. Chemosphere 36, 357–394.<br />
Halling-Sorensen, B., 2001. Inhibition of aerobic growth and<br />
nitrification of bacteria in sewage sludge by antibacterial<br />
agents. Archives of Environmental Contamination and<br />
Toxicology 40 (4), 451–460.<br />
Hanson, M.L., Knapp, C.W., Graham, D.W., 2006. Field assessment<br />
of oxytetracycline exposure to the freshwater macrophytes<br />
Egeria densa Planch and Ceratophyllum demersum L.<br />
Environmental Pollution 141 (3), 434–442.<br />
Heberer, T., Fuhrmann, B., Schmidt-Baumler, K., Tsipi, D.,<br />
Koutsouba, V., Hiskia, A., 2000. Occurrence of pharmaceutical<br />
residues in sewage, river, ground-, and drinking water in<br />
Greece and Germany. Abstracts of Papers of the American<br />
Chemical Society 219 U623.<br />
Heidler, J., Halden, R.U., 2007. Mass balance assessment of<br />
triclosan removal during conventional sewage treatment.<br />
Chemosphere 66, 362–369.<br />
Heidler, J., Sapkota, A., Halden, R.U., 2006. Partitioning,<br />
persistence, and accumulation in digested sludge of the<br />
topical antiseptic triclocarban during wastewater treatment.<br />
Environmental Science & Technology 40, 3634–3639.<br />
water research 44 (2010) 658–668 667<br />
Higgins, C.P., Paesani, Z.J., Chalew, T.E.A., Halden, R.U., 2009.<br />
Bioaccumulation of triclocarban in Lumbriculus variegates.<br />
Environmental Toxicology and Chemistry 65, 141–148.<br />
Hirsch, R., Ternes, T., Haberer, K., Kratz, K.L., 1999. Occurrence of<br />
antibiotics in the aquatic environment. Science of the Total<br />
Environment 225, 109–118.<br />
Jones-Lepp, T.L., Stevens, R., 2007. Pharmaceuticals and personal<br />
care products in biosolids/sewage sludge: the interface<br />
between analytical chemistry and regulation. Analytical and<br />
Bioanalytical Chemistry 387, 1173–1183.<br />
Jos, A., Repetto, G., Rios, J.C., Hazen, N., Molero, M.L., del Peso, A.,<br />
Salguero, M., Fernandez-Freire, P., Perez-Martin, J.M.,<br />
Camen, A., 2003. Ecotoxicological evaluation of<br />
carbamazepine using six different model systems with<br />
eighteen endpoints. Toxicology in Vitro 17 (5–6), 525–532.<br />
Kinney, C.A., Furlong, E.T., Kolpin, D.W., Burkhardt, M.R.,<br />
Zaugg, S.D., Werner, S.L., Bossio, J.P., Benotti, M.J., 2008.<br />
Bioaccumulation of pharmaceuticals and other anthropogenic<br />
waste indicators in earthworms from agricultural soil<br />
amended with biosolid or swine manure. Environmental<br />
Science & Technology 42, 1863–1870.<br />
Kolpin, D.W., Furlong, E.T., Meyer, M.T., Thurman, E.M., Zaugg, S.D.,<br />
Barber, L.B., Buxton, H.T., 2002. Pharmaceuticals, hormones, and<br />
other organic wastewater contaminants in US streams,<br />
1999–2000: a national reconnaissance. Environmental Science &<br />
Technology 36, 1202–1211.<br />
Kumar, K., Gupta, S.C., Baidoo, S.K., Chander, Y., Rosen, C.J., 2005.<br />
Antibiotic uptake by plants from soil fertilized with animal<br />
manure. Journal of Environmental Quality 34, 2082–2085.<br />
Kupper, T., Berset, J.D., Etter-Holzer, R., Furrer, R., Tarradellas, J.,<br />
2004. Concentraions and specific loads of polycyclic musks in<br />
sewage sludge originating from a monitoring network in<br />
Switzerland. Chemosphere 54, 1111–1120.<br />
Lindberg, R.H., Wennberg, P., Johansson, M.I., Tysklind, M.,<br />
Andersson, B.A.V., 2005. Screening of human antibiotic<br />
substances and determination of weekly mass flows in five<br />
sewage treatment plants in Sweden. Environmental Science &<br />
Technology 39, 3421–3429.<br />
Lindberg, R.H., Olofsson, U., Rendahl, P., Johansson, M.I., Tysklind, M.<br />
, Andersson, B.A.V., 2006. Behavior of fluoroquinolones and<br />
trimethoprim during mechanical, chemical, and active sludge<br />
treatment of sewage water and digestion of sludge.<br />
Environmental Science & Technology 40, 1042–1048.<br />
Lindberg, R.H., Bjorklund, K., Rendahl, P., Johansson, M.I.,<br />
Tysklind, M., Andersson, B.A.V., 2007. Environmental risk<br />
assessment of antibiotics in the Swedish environment with<br />
emphasis on sewage treatment plants. Water Research 41,<br />
613–619.<br />
Lindsey, M.E., Meyer, M., Thurman, E.M., 2001. Analysis of trace<br />
levels of sulfonamide and tetracycline antimicrobials, in<br />
groundwater and surface water using solid-phase extraction<br />
and liquid chromatography/mass spectrometry. Analytical<br />
Chemistry 73, 4640–4646.<br />
Liu, F., Ying, G.-G., Yang, L.-H., Zhou, Q.-X., 2009. Terrestrial<br />
ecotoxicological effects of the antimicrobial agent triclosan.<br />
Ecotoxicology and Environmental Safety 72, 86–92.<br />
Loos, R., Wollgast, J., Huber, T., Hanke, G., 2007. Polar herbicides,<br />
pharmaceutical products, perfluorooctanesulfonate (PFOS),<br />
perfluorooctanoate (PFOA), and nonylphenol and its<br />
carboxylates and ethoxylates in surface and tap waters<br />
around Lake Maggiore in Northern Italy. Analytical and<br />
Bioanalytical Chemistry 387, 1469–1478.<br />
Loraine, G.A., Pettigrove, M.E., 2006. Seasonal variations in<br />
concentrations of pharmaceuticals and personal care products<br />
in drinking water and reclaimed wastewater in Southern<br />
California. Environmental Science & Technology 40, 687–695.<br />
Luetzhoft, H.C.H., Halling-Sorensen, B., Jorgensen, S.E., 1999.<br />
Algal toxicity of antibacterial agents applied in Danish fish
668<br />
farming. Archives of Environmental Contamination and<br />
Toxicology 36 (1), 1–6.<br />
McAvoy, D.C., Schatowitz, B., Jacob, M., Hauk, A., Eckhoff, W.S.,<br />
2002. Measurement of triclosan in wastewater treatment<br />
systems. Environmental Toxicology and Chemistry 21,<br />
1323–1329.<br />
Migliore, L., Cozzolino, S., Fiori, M., 2003. Phytotoxicity to and<br />
uptake of enrofloxacin in crop plants. Chemosphere 52,<br />
1233–1244.<br />
Moldovan, Z., 2006. Occurrences of pharmaceutical and personal<br />
care products as micropollutants in rivers from Romania.<br />
Chemosphere 64, 1808–1817.<br />
National Research Council, 2002. Biosolids Applied to Land:<br />
Advancing Standards and Practices Washington, D.C.<br />
North East Biosolids and Residuals Association, 2007. A National<br />
Biosolids Regulation, Quality, End Use & Disposal Survey<br />
Tamworth, NH.<br />
Roberts, P.H., Thomas, K.V., 2006. The occurrence of selected<br />
pharmaceuticals in wastewater effluent and surface waters of<br />
the lower Tyne catchment. Science of the Total Environment<br />
356, 143–153.<br />
Robinson, A.A., Belden, J.B., Lydy, M.J., 2005. Toxicity of<br />
fluoroquinolone antibiotics to aquatic organisms.<br />
Environmental Toxicology and Chemistry 24 (2), 423–430.<br />
Sapkota, A., Heidler, J., Halden, R.U., 2007. Detection of<br />
triclocarban and two co-contaminating chlorocarbanilides in<br />
US aquatic environments using isotope dilution liquid<br />
chromatography tandem mass spectrometry. Environmental<br />
Research 103, 21–29.<br />
Shoemaker, N.B., Vlamakis, H., Hayes, K., Salyers, A.A., 2001.<br />
Evidence for extensive resistance gene transfer among<br />
Bacteroides spp. and among Bacteroides and other genera in<br />
the human colon. Applied and Environmental Microbiology<br />
67, 561–568.<br />
Stackelberg, P.E., Furlong, E.T., Meyer, M.T., Zaugg, S.D.,<br />
Henderson, A.K., Reissman, D.B., 2004. Persistence of<br />
pharmaceutical compounds and other organic wastewater<br />
contaminants in a conventional drinking-watertreatment<br />
plant. Science of the Total Environment 329, 99–113.<br />
Tamtam, F., Mercier, F., Le Bot, B., Eurin, J., Dinh, Q.T.,<br />
Clement, M., Chevreuil, M., 2008. Occurrence and fate of<br />
antibiotics in the Seine river in various hydrological<br />
conditions. Science of the Total Environment 393, 84–95.<br />
Ternes, T.A., 1998. Occurrence of drugs in German sewage<br />
treatment plants and rivers. Water Research 32, 3245–3260.<br />
Ternes, T.A., Herrmann, N., Bonerz, M., Knacker, T., Siegrist, H.,<br />
Joss, A., 2004a. A rapid method to measure the solid-water<br />
water research 44 (2010) 658–668<br />
distribution coefficient (K-d) for pharmaceuticals and musk<br />
fragrances in sewage sludge. Water Research 38, 4075–4084.<br />
Ternes, T.A., Joss, A., Siegrist, H., 2004b. Scrutinizing<br />
pharmaceuticals and personal care products in wastewater<br />
treatment. Environmental Science & Technology 38,<br />
392A–399A.<br />
USEPA, 1994. Biosolids Recycling: Beneficial Technologies for<br />
a Better Environment. Office of Water. U.S. Environmental<br />
Protection Agency, Washington, D.C. EPA 832-R-94–009.<br />
USEPA, 2001. Sampling Procedures for the 2001 National Sewage<br />
Sludge Survey. Office of Science and Technology, Washington,<br />
D.C.<br />
USEPA, 2003. Workshop on Emerging Pollutants. Region/ORD<br />
Workshop on Emerging Pollutants. U.S. EPA, Chicago, IL.<br />
USEPA, 2007a. 2001 National Sewage Sludge Survey Report<br />
Washington, D.C. EPA-822-R-07-006.<br />
USEPA, 2007b. Method 1694: Pharmaceuticals and Personal Care<br />
Products in Water, Soil, Sediment, and Biosolids by HPLC/MS/<br />
MS Washington, D.C. EPA-821-R-08–002.<br />
USEPA, 2009. Targeted National Sewage Sludge Survey<br />
Washington, D.C. EPA-822-R-08–014.<br />
Williams, R.R., Bell, T.A., Lightner, D.V., 1992. Shrimp<br />
antimicrobial testing: II. Toxicity testing and safety<br />
determination for twelve antimicrobials with penaeid shrimp<br />
larvae. Journal of Aquatic Animal Health 4 (4), 262–270.<br />
Wilson, B.A., Smith, V.H., Denoyelles, F., Larive, C.K., 2003. Effects<br />
of three pharmaceutical and personal care products on<br />
natural freshwater algal assemblages. Environmental Science<br />
& Technology 37 (9), 1713–1719.<br />
Wu, C.X., Witter, J.D., Spongberg, A.L., Czajkowski, K.P., 2009.<br />
Occurrence of selected pharmaceuticals in an agricultural<br />
landscape, western Lake Erie basin. Water Research 43,<br />
3407–3416.<br />
Yang, L.H., Ying, G.G., Su, H.C., Stauber, J.L., Adams, M.S.,<br />
Binet, M.T., 2008. Growth-inhibiting effects of 12 antibacterial<br />
agents and their mixtures on the freshwater microalga<br />
Pseudokirchneriella subcapitata. Environmental Toxicology and<br />
Chemistry 27 (5), 1201–1208.<br />
Ying, G.G., Yu, X.Y., Kookana, R.S., 2007. Biological<br />
degradationoftriclocarbanandtriclosaninasoilunder<br />
aerobic and anaerobic conditions and comparison with<br />
environmental fate modelling. Environmental Pollution<br />
150, 300–305.<br />
Zurita, J.L., Repetto, G., Jos, A., Salguero, M., Lopez-Artiguez, M.,<br />
Camean, A.M., 2007. Toxicological effects of the lipid regulator<br />
gemfibrozil in four aquatic systems. Aquatic Toxicology 81 (1),<br />
106–115.