School of Engineering and Science - Jacobs University
School of Engineering and Science - Jacobs University
School of Engineering and Science - Jacobs University
Create successful ePaper yourself
Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.
Iron sedimentation <strong>and</strong> the neodymium isotopic composition <strong>of</strong><br />
Archean seawater as inferred from b<strong>and</strong>ed iron-formations<br />
by<br />
Brian W. Alex<strong>and</strong>er<br />
A thesis submitted in partial fulfillment<br />
<strong>of</strong> the requirements for the degree <strong>of</strong><br />
Doctor <strong>of</strong> Philosophy<br />
in Geochemistry<br />
Approved, Thesis Committee<br />
_____________________________________<br />
Pr<strong>of</strong>. Dr. Michael Bau, Chair<br />
<strong>Jacobs</strong> <strong>University</strong> Bremen<br />
_____________________________________<br />
Pr<strong>of</strong>. Dr. Andrea Koschinsky<br />
<strong>Jacobs</strong> <strong>University</strong> Bremen<br />
_____________________________________<br />
Dr. Per Andersson<br />
Swedish Museum <strong>of</strong> Natural History<br />
_____________________________________<br />
Pr<strong>of</strong>. Dr. Peter Möller<br />
GeoForschungsZentrum Potsdam<br />
_____________________________________<br />
Pr<strong>of</strong>. Dr. Jens Gutzmer<br />
Technische Universität Bergakademie Freiberg<br />
Date <strong>of</strong> Defense: February 10th, 2009<br />
<strong>School</strong> <strong>of</strong> <strong>Engineering</strong> <strong>and</strong> <strong>Science</strong>
ABSTRACT<br />
The neodymium (Nd) isotopic composition <strong>of</strong> circa three billion-year-old (Ga)<br />
seawater is inferred from studies <strong>of</strong> b<strong>and</strong>ed iron-formations (IF) from South Africa.<br />
Iron-formations from the 2.9 Ga Pongola Supergroup display negative Є Nd values that<br />
are indistinguishable from local shale, suggesting that Nd derived from weathering <strong>of</strong><br />
continental crust dominated the Nd isotopic composition <strong>of</strong> shallow seawater. In<br />
contrast, Є Nd (t) in IF from the 2.95 Ga Pietersburg greenstone belt is clearly<br />
distinguishable from local clastic material, <strong>and</strong> indicates that where Nd was not sourced<br />
from local crust, seawater could possess more radiogenic Є Nd values <strong>of</strong> +1. The<br />
Pietersburg IF Є Nd (t) values are therefore interpreted to represent bulk Archean seawater<br />
not affected by local crustal Nd signatures. This Є Nd value <strong>of</strong> +1 is very similar to<br />
published Є Nd (t) estimates for Archean seawater, when only those IFs which display rare<br />
earth element ratios (REE) common to seawater are considered. Screening the data in<br />
such a manner clearly identifies seawater precipitates, <strong>and</strong> relatively constant Є Nd (t) <strong>of</strong><br />
+1 to +2 for Archean IFs suggest these values are a good average for 2.5-3.8 Ga<br />
seawater.<br />
If bulk Archean seawater typically displayed Є Nd (t) <strong>of</strong> +1, then hydrothermal<br />
alteration <strong>of</strong> mafic oceanic crust is considered the likely process for delivering relatively<br />
radiogenic Nd to seawater. Data from this study <strong>and</strong> previous work suggest significant<br />
high temperature (T) fluid input to seawater between 2.5-3.8 Ga, as supported by<br />
positive Eu anomalies in IFs, <strong>and</strong> the absence <strong>of</strong> negative Ce anomalies in any Archean<br />
IFs indicates that seawater was more reducing than modern seawater. Modern high-T<br />
fluids altering mafic oceanic crust possess both positive Є Nd values <strong>and</strong> Eu anomalies,<br />
<strong>and</strong> it is possible that these reducing, Fe-rich fluids exerted a greater control on Archean<br />
seawater chemistry than is observed today.<br />
However, the negative Є Nd (t) values for the Pongola IF data indicate that locally,<br />
shallow seawater could be isotopically distinct, <strong>and</strong> possess an Nd isotopic signature<br />
identical to local crust. If world seawater was isotopically homogenous with respect to<br />
Nd between 2.9-3.0 Ga, these data indicate a significant shift from positive Є Nd values in<br />
seawater during deposition <strong>of</strong> the Pietersburg IFs, to negative Є Nd values during<br />
deposition <strong>of</strong> the Pongola IFs. This is considered unlikely, <strong>and</strong> the Nd isotopic data are<br />
rather interpreted to reflect that, similar to modern oceans, ~3.0 Ga world oceans were<br />
composed <strong>of</strong> water masses that could be isotopically distinct with respect to Nd.<br />
i
The process(es) by which IFs were deposited in Archean seawater is also<br />
investigated through a study <strong>of</strong> Fe-rich shales formed in close association with the<br />
Pongola IFs. Major <strong>and</strong> trace element concentration data for these Fe-rich shales<br />
indicate post-depositional mobility <strong>and</strong> extreme loss <strong>of</strong> alkali metals <strong>and</strong> barium (Ba).<br />
Mobility <strong>of</strong> these elements in modern clastic sediments has been observed in the<br />
presence <strong>of</strong> pore water NH + +<br />
4 , <strong>and</strong> a significant presence <strong>of</strong> NH 4 in Fe-rich Archean<br />
+<br />
sediments would strongly suggest a biologic origin for IFs. However, a conclusive NH 4<br />
signature is not observed in the Pongola Fe-rich shales, <strong>and</strong> the behavior <strong>of</strong> the alkali<br />
metals <strong>and</strong> Ba is best explained by the presence <strong>of</strong> significant concentrations <strong>of</strong><br />
dissolved Fe 2+ in sediment pore waters <strong>and</strong>/or bottom seawater. This suggests that<br />
precipitation <strong>of</strong> Fe(III) mineral phases, either in the water column or at the sediment<br />
water interface, was an unlikely mechanism for the deposition <strong>of</strong> the primary Fe-rich<br />
sediment.<br />
While studies <strong>of</strong> seawater prior to ~2.7 Ga are generally constrained to<br />
investigations <strong>of</strong> IFs, cherts, <strong>and</strong> rare fluid inclusions, the occurrence <strong>of</strong> carbonate rocks<br />
is extensive after 2.7 Ga. Many Archean <strong>and</strong> Paleoproterozoic carbonates are dolomitic<br />
<strong>and</strong>/or highly silicified, <strong>and</strong> the retention <strong>of</strong> primary seawater trace metal distributions<br />
in these rocks has not been conclusively demonstrated. A study <strong>of</strong> limestones <strong>and</strong><br />
silicified dolomites from the ~2.25 Ga Mooidrai carbonates indicates that silicified<br />
dolomites retain primary trace metal signatures consistent with seawater <strong>and</strong><br />
indistinguishable from coeval limestones. Therefore, silicified dolomites free <strong>of</strong> clastic<br />
detritus are considered suitable archives for proxies <strong>of</strong> ancient seawater.<br />
The accuracy <strong>and</strong> reproducibility <strong>of</strong> the trace metal concentration data for 32<br />
elements are evaluated, <strong>and</strong> the inductively coupled plasma mass spectrometry (ICPMS)<br />
analytical methods used are described. The ICPMS data for Sm <strong>and</strong> Nd in IF samples<br />
are in excellent agreement with thermal ionization mass spectrometry (TIMS)<br />
measurements, supporting the accuracy <strong>of</strong> the reported REE data. The data quality is<br />
discussed in the context <strong>of</strong> repeated analyses <strong>of</strong> certified reference materials (CRMs)<br />
commonly used in geochemical research. Measured concentrations for IF <strong>and</strong> carbonate<br />
CRMs match reference data well; however, some elements suffer from major element<br />
interferences in particular rock types. This includes Co in carbonate rocks, <strong>and</strong> Nb in<br />
Fe-rich rock types. The concentrations <strong>of</strong> many trace metals in some CRMs are poorly<br />
constrained, <strong>and</strong> early published datasets appear to frequently overestimate the<br />
abundances <strong>of</strong> these elements.<br />
ii
ACKNOWLEDGEMENTS<br />
A special thank you to my family <strong>and</strong> friends for their encouragement <strong>and</strong> support,<br />
<strong>and</strong> particularly to Claudia Nitzschmann, as three years became four, <strong>and</strong> four years<br />
became five. Thanks as well to my advisor, Michael Bau, for introducing me to the<br />
numerous, endlessly fascinating topics regarding the geochemical evolution <strong>of</strong> the early<br />
Earth. I must confess a compelling attraction to a subject where it is so difficult to prove<br />
my assertions wrong…<br />
A debt <strong>of</strong> gratitude also to the many people that I came to know <strong>and</strong> appreciate<br />
within the geochemistry working group at <strong>Jacobs</strong>, including, but not limited to, faculty<br />
(A. Koschinsky), students (K. Schmidt, S. Kulaksiz), <strong>and</strong> staff (D. Meißner, J. Mawick,<br />
A. Moje). The various excursions, meetings, <strong>and</strong> long hours in the lab were much more<br />
enjoyable because <strong>of</strong> their presence.<br />
from Holl<strong>and</strong> (1972)<br />
iii
This thesis represents original <strong>and</strong> independently conducted research that has not<br />
been submitted to any other university for the conferral <strong>of</strong> a degree.<br />
Brian W. Alex<strong>and</strong>er<br />
Bremen, Germany - Jan. 20, 2010<br />
v
TABLE OF CONTENTS<br />
Chapter 1. Introduction..................................................................................................... 1<br />
1.1. Statement <strong>of</strong> research goal .................................................................................... 1<br />
1.2. Structure <strong>of</strong> thesis.................................................................................................. 2<br />
1.3. Study area.............................................................................................................. 3<br />
1.4. Research methods.................................................................................................. 5<br />
1.5. Background <strong>and</strong> review <strong>of</strong> relevant literature....................................................... 5<br />
1.5.1. Rare earth element geochemistry ................................................................... 6<br />
1.5.2. Rare earth elements in seawater .................................................................... 9<br />
1.5.3. Sm-Nd isotopic signatures in the Earth’s crust <strong>and</strong> seawater ..................... 20<br />
1.5.4. Iron-formations as seawater archives .......................................................... 24<br />
Chapter 2. Trace element analyses in geological materials using low resolution<br />
inductively coupled plasma mass spectrometry (ICPMS)............................ 39<br />
(published as <strong>Jacobs</strong> <strong>University</strong> Technical Report No. 18)<br />
Chapter 3. Continentally-derived solutes in shallow Archean seawater: Rare earth<br />
element <strong>and</strong> Nd isotope evidence in iron formation from the 2.9 Ga<br />
Pongola Supergroup, South Africa............................................................. 119<br />
(published in Geochimica et Cosmoschmica Acta 72 (2008) 378-394)<br />
Chapter 4. Neodymium isotopes in Archean seawater <strong>and</strong> implications for the<br />
marine Nd cycle in Earth’s early oceans .................................................... 139<br />
(published in Earth <strong>and</strong> Planetary <strong>Science</strong> Letters 283 (2009) 144-155)<br />
Chapter 5. Anoxygenic photoautotrophs <strong>and</strong> the origin <strong>of</strong> b<strong>and</strong>ed iron-formation ..... 153<br />
Chapter 6. Preservation <strong>of</strong> primary REE patterns without Ce anomaly during<br />
dolomitization <strong>of</strong> Mid-Paleoproterozoic limestone <strong>and</strong> the potential<br />
re-establishment <strong>of</strong> marine anoxia immediately after the “Great<br />
Oxidation Event” ........................................................................................ 165<br />
(published in South African Journal <strong>of</strong> Geology (2006) 81-86)<br />
Chapter 7. Concluding remarks.................................................................................... 173<br />
vii
viii
Chapter 1. Introduction<br />
1.1. Statement <strong>of</strong> research goal<br />
The research presented in this thesis relates to physical <strong>and</strong> chemical processes<br />
occurring in seawater during the first 2 billion years (2 Ga) <strong>of</strong> Earth’s history, <strong>and</strong><br />
particularly addresses seawater chemistry ca. 3.0 Ga. As samples <strong>of</strong> seawater from this<br />
time period are not available, except perhaps as microscopic fluid inclusions in some<br />
geologic samples, the research fundamentally rests upon archives <strong>of</strong> seawater chemistry<br />
that exist in the rock record. For studies <strong>of</strong> the Archean oceans (ca. 2.5-4.0 Ga) archives<br />
<strong>of</strong> seawater chemistry are limited due to the scarcity <strong>of</strong> the rock record. However,<br />
b<strong>and</strong>ed iron-formations (IFs) are common <strong>and</strong> extensive constituents <strong>of</strong> the Archean<br />
geologic record, <strong>and</strong> the work described here exclusively deals with IFs <strong>and</strong> Fe-rich<br />
sediments associated with IF deposition. Iron-formations were formally defined by<br />
James (1954) as “a chemical sediment, typically thin-bedded or laminated, containing<br />
15 per cent or more iron <strong>of</strong> sedimentary origin, commonly but not necessarily<br />
containing layers <strong>of</strong> chert”. These chemical sediments are clearly <strong>of</strong> marine origin (see<br />
Simonson, 2003, for details), <strong>and</strong> due to their ubiquitous occurrence during the<br />
Archean, IFs are likely one <strong>of</strong> the best seawater archives prior to ~2.7 Ga.<br />
One question regarding Archean oceans is the extent to which seawater chemistry<br />
was controlled by weathering <strong>of</strong> oceanic or continental crust. Different conclusions have<br />
been drawn, with some workers supporting weathering <strong>of</strong> continental crust as a<br />
dominant control on seawater chemistry (e.g., Miller <strong>and</strong> O’Nions, 1985), while others<br />
have supported hydrothermal alteration <strong>of</strong> oceanic crust as a primary control (<strong>Jacobs</strong>en<br />
<strong>and</strong> Pimentel-Klose, 1988a, 1988b; Veizer et al., 1989; Alibert <strong>and</strong> McCulloch, 1993;<br />
Bau et al. 1997a). The research presented here attempts to resolve these discrepant<br />
conclusions through new trace metal <strong>and</strong> isotopic data for ca. 3.0 Ga IFs. A secondary<br />
1
goal <strong>of</strong> the research is to discern if the process by which IFs formed might possibly be<br />
constrained. Though not the direct focus <strong>of</strong> this thesis, elucidating the manner in which<br />
IFs precipitated in ancient seawater is attractive, as these sediments remain a<br />
geochemical mystery even after more than 50 years <strong>of</strong> study.<br />
The general approach followed is to measure the isotopic ratios <strong>of</strong> the rare earth<br />
elements Sm <strong>and</strong> Nd, assuming that these measured ratios reflect the ambient seawater<br />
that precipitated the studied IFs. The inferred Sm-Nd isotopic signature <strong>of</strong> seawater<br />
would reflect different sources <strong>of</strong> Sm <strong>and</strong> Nd to ancient oceans, <strong>and</strong> would constrain the<br />
relative importance <strong>of</strong> these sources on seawater chemistry. As ancient oceanic <strong>and</strong><br />
continental crust possess distinctly different Sm-Nd isotopic signatures, it should be<br />
possible to determine if weathering <strong>of</strong> oceanic or continental crust was more important<br />
in controlling seawater chemistry.<br />
1.2. Structure <strong>of</strong> thesis<br />
This thesis is structured as a compilation <strong>of</strong> independent manuscripts, with the<br />
exception <strong>of</strong> this, the introductory first chapter, <strong>and</strong> the final chapter that contains<br />
concluding remarks. Each individual manuscript contains an abstract, background<br />
information, the discussion <strong>of</strong> results, conclusions, <strong>and</strong> references. The thesis in total<br />
consists <strong>of</strong> six chapters that are outlined in the Table <strong>of</strong> Contents.<br />
This, the first chapter, introduces the research goals <strong>and</strong> provides background<br />
information relevant to the thesis which is not covered extensively in the individual<br />
manuscripts. The manuscripts that follow are generally written to facilitate publication<br />
in international academic research journals. One exception is Chapter 2, as it describes<br />
the relevant analytical methods employed within the <strong>Jacobs</strong> <strong>University</strong> Bremen (<strong>Jacobs</strong>)<br />
Geochemistry Lab, <strong>and</strong> which has been published as <strong>Jacobs</strong> Technical Report No. 18.<br />
2
As a result <strong>of</strong> this approach, slight differences will exist from chapter to chapter<br />
regarding organization or format. If already published or submitted for publication in an<br />
academic journal, the manuscripts reflect the individual formatting <strong>and</strong> length<br />
requirements <strong>of</strong> the respective journal, <strong>and</strong> where appropriate, the manuscript/chapter is<br />
the printed version <strong>of</strong> the journal article.<br />
1.3. Study area<br />
The geologic samples used in this thesis are exclusively from South Africa, <strong>and</strong><br />
Figure 1 lists the major relevant geologic features <strong>of</strong> southern Africa. Since the research<br />
goal revolves around the chemistry <strong>of</strong> ancient seawater, this part <strong>of</strong> the world is an ideal<br />
area for study. The Kaapvaal craton is one <strong>of</strong> the oldest <strong>and</strong> most stable cratons known<br />
in the world (Tankard, 1982), <strong>and</strong> the oldest rock yet dated on the Kaapvaal craton, <strong>and</strong><br />
probably the oldest on the African continent (Poujol et al., 2002), is a 3644 ±4 Ma<br />
tonalitic gneiss from northwest Swazil<strong>and</strong> (Compston <strong>and</strong> Kröner, 1988). The Kaapvaal<br />
craton hosts several Archean sequences containing iron-formation deposited in different<br />
tectonic settings, which is advantageous as these represent seawater precipitates from<br />
different marine environments, e.g., isl<strong>and</strong>-arc or stable continental shelf. Examples<br />
include IF within the >3.2 Ga Barberton greenstone belt, considered to have formed in<br />
an oceanic plateau/isl<strong>and</strong>-arc setting (Lowe, 1994), as well as IF deposited on the stable<br />
Kaapvaal craton during the formation <strong>of</strong> the ~2.5 Ga Transvaal Supergroup (Beukes,<br />
1983)<br />
The IFs discussed in this thesis were selected as they represent seawater<br />
precipitates from different tectonic settings, yet are <strong>of</strong> a similar age (ca. 2.9-3.0 Ga).<br />
Samples were obtained from the 2.9 Ga Mozaan Group <strong>of</strong> the Pongola Supergroup,<br />
which was deposited within a relatively shallow, near-shore continental shelf<br />
3
Figure 1. Map <strong>of</strong> southern Africa showing locations <strong>of</strong> major geologic features discussed in thesis. Ironformation<br />
units that provided samples for geochemical analyses were obtained from the Pongola<br />
Supergroup <strong>and</strong> the Pietersburg greenstone belt.<br />
environment, as well as from the 2.95 Ga Pietersburg greenstone belt, which is<br />
considered to represent a more open ocean, isl<strong>and</strong>-arc environment (Fig. 1). The<br />
hypothesis is that Sm-Nd isotopic signatures in IFs <strong>of</strong> similar age, yet from different<br />
depositional environments, might possess different Nd isotopic signatures that reflect<br />
the respective Nd sources to Archean seawater.<br />
4
1.4. Research methods<br />
The research methods employed for this thesis consist <strong>of</strong> various techniques for<br />
obtaining geochemical analyses <strong>of</strong> rock samples. Three analytical techniques are used<br />
for performing elemental <strong>and</strong> isotopic measurements <strong>of</strong> the samples discussed. The first<br />
technique is X-ray fluorescence (XRF), which provided major element concentration<br />
data. The second analytical method is inductively coupled plasma mass spectrometry<br />
(ICPMS), which was used for trace metal concentration measurements. The third<br />
analytical method is thermal ionization mass spectrometry (TIMS), which was used for<br />
measuring Sm-Nd isotopic ratios.<br />
The XRF analyses were performed either at the GeoForschungsZentrum (GFZ) in<br />
Potsdam, Germany, or at the <strong>University</strong> <strong>of</strong> Johannesburg (South Africa) SpectRAU<br />
facility. The trace metal concentrations determined by ICPMS were provided either by<br />
the GFZ, for the initial research conducted for this thesis, or by the author using<br />
equipment within the Geochemistry Lab at <strong>Jacobs</strong> <strong>University</strong> Bremen (JUB). All Sm-<br />
Nd isotopic analyses were performed at the Laboratory for Isotope Geology at the<br />
Swedish Museum <strong>of</strong> Natural History in Stockholm, Sweden. Details <strong>of</strong> the XRF<br />
analyses are not discussed, as XRF techniques are well-developed <strong>and</strong> are only used for<br />
determining major element concentrations (e.g., Si, Al, Fe). A detailed description <strong>of</strong><br />
the ICPMS analytical methods is provided in Chapter 2, <strong>and</strong> is applicable to data<br />
produced at both the GFZ <strong>and</strong> JUB. Descriptions <strong>of</strong> the Sm-Nd isotopic analyses are<br />
provided within the individual chapters where relevant.<br />
1.5. Background <strong>and</strong> review <strong>of</strong> relevant literature<br />
The principle conclusions <strong>of</strong> this thesis rest upon the isotopic ratios <strong>of</strong> Nd in<br />
Archean seawater as inferred from Sm-Nd analyses <strong>of</strong> iron-formations. Therefore, it is<br />
5
appropriate to briefly discuss: 1) basic rare earth element geochemistry, 2) rare earth<br />
element behavior in seawater, 3) Sm-Nd isotopic signatures in the Earth’s crust <strong>and</strong><br />
seawater, <strong>and</strong> 4) the use <strong>of</strong> IFs as archives <strong>of</strong> ancient seawater chemistry.<br />
1.5.1. Rare earth element geochemistry<br />
The rare earth elements include Sc, Y, <strong>and</strong> the fifteen lanthanide elements which<br />
have atomic masses between 57 <strong>and</strong> 71 (La, Ce, Pr, Nd, Pm, Sm, Eu, Gd, Tb, Dy, Ho,<br />
Er, Tm, Yb, Lu). Hereafter in this thesis, the lanthanides will be exclusively referred to<br />
as the rare earth elements (REE), excluding Sc <strong>and</strong> Y. However, as described below,<br />
yttrium is useful when grouped with the lanthanides for studying geochemical<br />
processes, <strong>and</strong> where yttrium data are available discussions <strong>of</strong> the REE <strong>and</strong> Y will be<br />
indicated by the acronym REY.<br />
Among the lanthanides promethium (Pm) is a special case, as it has three<br />
radioactive isotopes ( 145 Pm, 146 Pm, 147 Pm), <strong>of</strong> which 145 Pm has the longest half-life <strong>of</strong><br />
~17.7 years. As such Pm is considered an extinct element in natural samples <strong>and</strong> is not<br />
included in discussions <strong>of</strong> REY behavior in geological studies. The remaining REY<br />
display similar geochemical behavior, primarily as a result <strong>of</strong> their +3 valence state,<br />
though Ce <strong>and</strong> Eu may also exist in the +4 <strong>and</strong> +2 valence states, respectively. This<br />
geochemically coherent behavior among the REY is also a consequence <strong>of</strong> very similar<br />
ionic radii between 0.977 Å (Lu) <strong>and</strong> 1.16 Å (La). The similar ionic radii results from a<br />
progressive filling <strong>of</strong> the 4f electron orbital that produces a monotonic decrease in the<br />
ionic radii with increasing atomic mass (Figure 2). As a result, lanthanide behavior<br />
tends to vary smoothly <strong>and</strong> predictably as a function <strong>of</strong> ionic radii (i.e., atomic mass) in<br />
many geochemical systems.<br />
6
4<br />
Ce<br />
valency<br />
3<br />
Sc<br />
Lu<br />
Y<br />
La<br />
Eu<br />
2<br />
0.8 0.9 1.0 1.1 1.2 1.3<br />
ionic radius (Å)<br />
Figure 2. Valence state <strong>and</strong> ionic radius (coordination number 8) for rare earth elements. Yttrium has an<br />
ionic radius that falls between Dy <strong>and</strong> Ho. Cerium <strong>and</strong> Eu are the only rare earth elements that exist at<br />
valence states other than +3 (ionic radii data from Shannon, 1976).<br />
Rare earth element data are typically normalized for plotting, as REY<br />
concentrations may vary by an order <strong>of</strong> magnitude (or more) between adjacent elements<br />
in the series, due to the greater abundance <strong>of</strong> elements with even atomic numbers<br />
relative to elements with odd atomic numbers (Oddo-Harkins rule). Figure 3 shows data<br />
for mid-ocean ridge basalt (MORB) that has been normalized to C1 chondritic meteorite<br />
(Anders <strong>and</strong> Grevesse, 1989), as well as non-normalized data for comparison, <strong>and</strong><br />
where Y has been inserted between Dy <strong>and</strong> Ho according to its ionic radius (see Figure<br />
2). If none <strong>of</strong> the REY exhibit anomalous behavior then the normalized REY pattern<br />
will vary smoothly across the series as seen in Figure 3. It should be noted that the C1<br />
chondritic meteorite is <strong>of</strong>ten used for normalizing REY data, as it is considered to<br />
possess a REY distribution equivalent to bulk silicate Earth. However, it is sometimes<br />
useful to normalize REY data to other geochemical reservoirs, <strong>and</strong> another commonly<br />
used REY dataset for normalization purposes (<strong>and</strong> one used extensively in this thesis) is<br />
upper continental crust, as represented by the Post Archean Average Shale (PAAS) <strong>of</strong><br />
McLennan (1989).<br />
7
100<br />
10<br />
MORB<br />
1<br />
0.1<br />
raw data (mg/kg)<br />
chondrite-normalized<br />
La Ce Pr Nd<br />
Sm Eu Gd Tb Dy Y Ho Er Tm Yb Lu<br />
Figure 3. Typical rare earth element plot where lanthanide elements are arranged in order <strong>of</strong> increasing<br />
atomic mass <strong>and</strong> decreasing ionic radii, with yttrium inserted between Dy <strong>and</strong> Ho in accordance with its<br />
ionic radius. Normalizing produces smooth REY patterns if none <strong>of</strong> the REY exhibit anomalous<br />
behavior. Mid-ocean ridge basalt (MORB) data from Niu et al. (1999).<br />
Some <strong>of</strong> the rare earth elements may be fractionated from neighboring REY as the<br />
result <strong>of</strong> natural geochemical processes, with Ce <strong>and</strong> Eu <strong>of</strong>fering the best known<br />
examples. The fractionation <strong>of</strong> Ce <strong>and</strong> Eu from the other REY is primarily a<br />
consequence <strong>of</strong> the different valence states that these elements can possess. Cerium is<br />
readily oxidized from Ce 3+ to Ce 4+ at the Earth’s surface <strong>and</strong> in the oxic marine<br />
environment, fractionating it from the other REY by forming Ce(IV) compounds <strong>of</strong> low<br />
solubility. At elevated temperatures (>200 °C) europium can fractionate from the other<br />
REY when Eu is reduced from Eu 3+ to Eu 2+ , e.g., during igneous or hydrothermal<br />
processes, with Eu fractionation occuring due to the different charge <strong>and</strong> significantly<br />
larger ionic radius <strong>of</strong> the Eu 2+ ion (Fig. 2). Examples <strong>of</strong> Ce <strong>and</strong> Eu fractionation are<br />
presented in Figure 4. The fractionation <strong>of</strong> Ce from the neighboring elements La <strong>and</strong> Pr<br />
is shown for average world seawater, <strong>and</strong> Eu fractionation as observed in high<br />
8
10 -2 360 °C hydrothermal fluid<br />
10 -3<br />
10 -4<br />
sample/PAAS<br />
10 -5<br />
10 -6<br />
10 -7<br />
10 -8<br />
10 -9<br />
average world seawater<br />
La Ce Pr Nd PmSm Eu Gd Tb Dy Y Ho Er Tm Yb Lu<br />
Figure 4. REY distributions in average world seawater <strong>and</strong> high-T hydrothermal fluid from the mid-<br />
Atlantic ridge normalized to continental crust as represented by PAAS (McLennan, 1989). Seawater<br />
data are the average <strong>of</strong> 172 worldwide analyses <strong>of</strong> all depths (average depth 1560 m) for which<br />
complete REY datasets are available (Zhang <strong>and</strong> Nozaki, 1996; Alibo <strong>and</strong> Nozaki, 1999; Bau et al.,<br />
1997b; Nozaki et al., 1999; Nozaki <strong>and</strong> Alibo, 2003). Hydrothermal fluid data for sample BS-07-3/3 <strong>of</strong><br />
Bau <strong>and</strong> Dulski (1999). Note strong negative Ce anomaly in seawater <strong>and</strong> strong positive Eu anomaly in<br />
high-T hydrothermal fluid.<br />
temperature (T= 360 °C) hydrothermal fluids emanating from the mid-Atlantic oceanic<br />
ridge (MAR).<br />
1.5.2. Rare earth elements in seawater<br />
The abundances <strong>of</strong> the REY in seawater are extremely low relative to their crustal<br />
averages, with world seawater concentrations ~10 -7<br />
<strong>of</strong> the levels found in upper<br />
continental crust (see Figure 4). Higher REY abundances are typically found in river<br />
waters <strong>and</strong> pore waters <strong>of</strong> marine sediments, though not as high as the concentrations<br />
observed for high-T hydrothermal fluids (Figure 5). It is reasonable that these fluids<br />
(river water, pore water, <strong>and</strong> hydrothermal fluid) represent the primary dissolved REY<br />
sources for the oceans <strong>of</strong> the world, <strong>and</strong> assuming this is true, it is notable that each <strong>of</strong><br />
these REY sources contains several times the total REY concentration that exists in<br />
9
10 -2 hydrothermal fluid<br />
10 -3<br />
10 -4<br />
sample/PAAS<br />
10 -5<br />
10 -6<br />
10 -7<br />
10 -8<br />
10 -9<br />
river water<br />
marine pore water<br />
estuaries<br />
world seawater<br />
La Ce Pr Nd PmSm Eu Gd Tb Dy Y Ho Er Tm Yb Lu<br />
Figure 5. Average REE <strong>and</strong> REY distributions in natural waters. Seawater <strong>and</strong> hydrothermal fluid data<br />
are the same as in Fig. 4. River water is an average <strong>of</strong> 155 filtered (0.2-0.45 μm) samples with pH<br />
between 5-9, <strong>and</strong> represents complete REY datasets for rivers in Venezuela, Australia, USA, <strong>and</strong><br />
Sardinia (Tosiani et al., 2004; Lawrence et al., 2006; Bau et al. 2006; Cidu <strong>and</strong> Biddau, 2007), as well as<br />
complete REE data for rivers in France <strong>and</strong> Argentina (Tricca et al., 1999; Gammons et al., 2005b; Stille<br />
et al., 2006). The marine pore water REE pattern is the average <strong>of</strong> 89 filtered (0.2-0.45 μm) samples<br />
compiled from isotope dilution REE data for the eastern <strong>and</strong> western margins <strong>of</strong> North America<br />
(Elderfield <strong>and</strong> Sholkovitz, 1987; German <strong>and</strong> Elderfield, 1989; Sholkovitz et al., 1989; Sholkovitz et al.,<br />
1992), as well as complete REE datasets for pore waters from the western margins <strong>of</strong> North <strong>and</strong> South<br />
America (Haley <strong>and</strong> Klinkhammer, 2003; Haley et al., 2004). Estuary water is an average <strong>of</strong> 134 filtered<br />
(0.2-0.45 μm) samples with salinities between 1-30 ‰ (mean= 13.8 ‰, median= 14.2 ‰) <strong>and</strong> represents<br />
instrument neutron activation analysis data for samples from France (Martin et al., 1976), as well as<br />
isotope dilution REE data from estuaries along the margins <strong>of</strong> North America (Sholkovitz <strong>and</strong> Elderfeld,<br />
1988; Goldstein <strong>and</strong> <strong>Jacobs</strong>en, 1988; Elderfeld et al., 1990; Sholkovitz et al., 1992), Great Britain<br />
(Elderfeld et al., 1990), South America (Sholkovitz, 1993), <strong>and</strong> Papua New Guinea (Sholkovitz <strong>and</strong><br />
Szymczak, 2000). Additionally, estuarine water average incorporates complete REY data for estuaries<br />
from Australia (Lawrence <strong>and</strong> Kamber, 2006) <strong>and</strong> Germany (Kulaksız <strong>and</strong> Bau, 2007).<br />
modern seawater. Therefore, some process or processes must exist that effectively<br />
remove REY from these source solutions upon mixing with ambient seawater.<br />
The distinctly lower REY concentrations found in estuaries relative to river water<br />
(Fig. 5) indicate that significant REY removal occurs during the mixing <strong>of</strong> freshwater<br />
<strong>and</strong> seawater. This phenomenon has been noted in previous experimental studies (Hoyle<br />
et al., 1984), as well studies <strong>of</strong> natural systems (e.g., Elderfield et al., 1990). The<br />
removal <strong>of</strong> REY in estuaries is due to the coagulation <strong>and</strong> ‘salting-out’ <strong>of</strong> the major<br />
REY-carrying phases present in river water, which are Fe-organic colloids smaller than<br />
10
~0.2 μm (Sholkovitz, 1995). The result <strong>of</strong> this process is that more than ~70% <strong>of</strong> the<br />
10 3 times<br />
more REY than ambient seawater, the metal-oxides that precipitate from them<br />
ultimately act as sinks for seawater REY (German et al., 1990; German et al., 1991a).<br />
Pore waters within marine sediments are less well studied than either river or<br />
estuarine waters, <strong>and</strong> REE data are available for only a h<strong>and</strong>ful <strong>of</strong> sites worldwide (see<br />
Fig. 5 caption for details). Pore waters within marine sediments display strong<br />
variations in both REE concentrations <strong>and</strong> normalized REE patterns as a function <strong>of</strong><br />
depth below the seawater-sediment interface (Elderfield <strong>and</strong> Sholkovitz, 1987; Haley et<br />
al., 2004). Though data are few, it appears that increased REE concentrations in pore<br />
waters are related to redox cycling <strong>of</strong> Fe in the upper few centimeters <strong>of</strong> the sediment<br />
(Elderfield <strong>and</strong> Sholkovitz, 1987), as well the degradation <strong>of</strong> organic matter (Haley et<br />
al., 2004). It also appears that pore water fluxes may be a significant source <strong>of</strong> REE to<br />
seawater; however, the variability in REE patterns <strong>and</strong> higher concentrations found in<br />
pore waters are generally not observed in seawater sampled a few meters above the<br />
seawater-sediment interface (Elderfield <strong>and</strong> Sholkovitz, 1987; Haley et al., 2004),<br />
implying that REE sourced from pore waters may be rapidly scavenged from oxic<br />
11
seawater <strong>and</strong> returned to the sediment column. In consideration <strong>of</strong> the above<br />
observations, river waters are the dominant source <strong>of</strong> the REY to modern seawater. It<br />
should be noted that aeolian dust also delivers REY to the oceans, but the magnitude <strong>of</strong><br />
this flux to dissolved REY in seawater is debated (e.g., Nakai et al., 1993; Tachikawa et<br />
al., 1999; Greaves et al., 1999; Bayon et al., 2004), <strong>and</strong> regardless, the REY distribution<br />
in the aeolian flux will generally reflect continental REY sources. It should be noted<br />
that the Pacific Ocean seems atypical in this regard, as volcanic particles derived from<br />
intraplate <strong>and</strong> circum-Pacific volcanism appear to be a significant REY source, <strong>and</strong> the<br />
reader is referred to Goldstein <strong>and</strong> Hemming (2003) for a more thorough discussion.<br />
Within seawater itself, the REY display consistent behavior. Figure 6 shows rare<br />
earth element abundances <strong>and</strong> distributions from the Atlantic, Pacific, <strong>and</strong> Indian<br />
oceans. All <strong>of</strong> the major world oceans have been sampled, with exceptions being the<br />
Southern Ocean (south <strong>of</strong> 60° latitude) <strong>and</strong> the central Arctic Ocean. The available rare<br />
earth element dataset for seawater is extensive, <strong>and</strong> approximately 900 samples<br />
worldwide have been analyzed, with the Pacific Ocean providing the most samples<br />
(47%), followed by the Indian (29%), Atlantic (15%), Arctic (8%), <strong>and</strong> Mediterranean<br />
(
sample/PAAS<br />
10 -6 10 -7<br />
deep water >2000 m<br />
NW Pacific<br />
NE Indian<br />
SW Pacific<br />
S Atlantic<br />
10 -8<br />
shallow water 2000 m) possesses an<br />
indistinguishable rare earth element pattern <strong>and</strong> very similar concentrations regardless <strong>of</strong> the ocean basin<br />
sampled.<br />
<strong>and</strong> concentrations observed in modern seawater are very similar regardless <strong>of</strong> location<br />
(Fig. 6).<br />
The consistent REY patterns in modern seawater regardless <strong>of</strong> location or depth<br />
reflect the interplay between solution complexation effects <strong>and</strong> particle surface<br />
adsorption processes. The REY are lithophilic <strong>and</strong> generally insoluble elements, <strong>and</strong><br />
precipitation <strong>of</strong> REY mineral phases in equilibrium with seawater does not control REY<br />
concentrations or distributions, as all <strong>of</strong> the lanthanides are undersaturated in seawater<br />
(Brookins, 1989). Dissolved REE in seawater do not generally exist as hydrated free<br />
metal ions, but are predominantly complexed by carbonate anions (CO 2- 3 ) as illustrated<br />
in Figure 7 using the data <strong>of</strong> Cantrell <strong>and</strong> Byrne (1987).<br />
13
100<br />
MCO + 3 + M(CO 3 )- 2<br />
rare earth element<br />
% total<br />
80<br />
60<br />
40<br />
20<br />
MCO + 3<br />
M(CO 3<br />
) - 2<br />
M 3+<br />
0<br />
La Ce Pr Nd Pm Sm Eu Gd Tb Dy Ho Er Tm Yb Lu<br />
Figure 7. Speciation <strong>of</strong> the lanthanides in seawater at 25° C <strong>and</strong> 1 atmosphere pressure with 35.1 ‰<br />
salinity <strong>and</strong> a total carbonate ion concentration [CO 3 2- ] <strong>of</strong> 1.39 x 10 -4 moles/kg. Figure adapted from<br />
Cantrell <strong>and</strong> Byrne (1987), <strong>and</strong> M refers to individual lanthanide elements. Hydroxide, chloride, fluoride,<br />
<strong>and</strong> sulfate REE complexes are not shown, as these are minor chemical species affecting REE behavior<br />
in seawater (Cantrell <strong>and</strong> Byrne, 1987).<br />
The carbonate anions binding to dissolved REY in seawater form either M(CO 3 ) +<br />
or M(CO 3 ) 2<br />
-<br />
complexes, where M refers to the individual element. Monocarbonate<br />
M(CO 3 ) +<br />
complexes are more prevalent for light rare earth elements (LREE), <strong>and</strong><br />
dicarbonate M(CO 3 ) - 2 complexes more so for heavy rare earth elements (HREE), with<br />
the relative proportions <strong>of</strong> these carbonate complexes roughly equal for middle rare<br />
earth elements (MREE) such as Gd <strong>and</strong> Tb. The calculated carbonate complexation<br />
constants are relatively unaffected by changes in temperature (Cantrell <strong>and</strong> Byrne,<br />
1987), <strong>and</strong> it is surmised that the relative lanthanide distributions between mono- <strong>and</strong><br />
dicarbonate complexes remains constant throughout the marine water column<br />
(Brookins, 1989).<br />
The solution complexation behavior <strong>of</strong> the REY strongly affects the process that<br />
controls the distribution <strong>of</strong> these elements in seawater, which is scavenging by<br />
particulate matter. The REY are readily adsorbed onto particle surfaces in natural<br />
14
waters, as evidenced by the dominant association <strong>of</strong> the REY with Fe-organic colloids<br />
in river waters (e.g., Sholkovitz, 1995). The particle reactive nature <strong>of</strong> the rare earths<br />
was well exploited by early workers conducting seawater analyses, in which the REY<br />
were quantitatively removed <strong>and</strong> concentrated from seawater samples by coprecipitation<br />
with Fe-hydroxides (e.g., Goldberg et al., 1963; Høgdahl et al., 1968;<br />
Elderfield <strong>and</strong> Greaves, 1982). This scavenging by particulate matter is considered the<br />
primary removal process for the REY in seawater (e.g., de Baar et al., 1985a; Elderfield,<br />
1988), <strong>and</strong> it appears that particle coatings consisting <strong>of</strong> organic films <strong>and</strong> Fe-Mn<br />
oxides are particularly important in this scavenging process (e.g., Byrne <strong>and</strong> Kim, 1990;<br />
Sholkovitz et al., 1994, <strong>and</strong> references therein). Experimental work indicates these<br />
scavenging reactions approach steady state within minutes (e.g., Byrne <strong>and</strong> Kim, 1990;<br />
Koeppenkastrop <strong>and</strong> De Carlo, 1993; Bau, 1999), suggesting that equilibrium<br />
conditions characterize REY partitioning between seawater <strong>and</strong> particle surfaces.<br />
The general shape <strong>of</strong> the REY patterns shown in Figs. 4-6, with a strong<br />
enrichment <strong>of</strong> the HREE over the LREE in seawater, can be attributed to the<br />
competition between carbonate complexation in solution <strong>and</strong> reactive particle surfaces.<br />
The exact process by which REY are adsorbed to particle surfaces is still a matter <strong>of</strong><br />
study, <strong>and</strong> Piasecki <strong>and</strong> Sverjensky (2008) provide a review <strong>of</strong> proposed sorption<br />
mechanisms in addition to a triple-layer surface complexation model that is consistent<br />
with X-ray studies. However, to a first approximation, particle surface adsorption<br />
constants for individual REY can be described by their first hydrolysis constants (Erel<br />
<strong>and</strong> Morgan, 1991; Bau et al., 1996). Using such an approach, calculated partition<br />
coefficients between particle surfaces <strong>and</strong> seawater predict a preferential LREE<br />
enrichment for the adsorbed component (Erel <strong>and</strong> Stolper; 1993), consistent with LREE<br />
depletion in the solution phase <strong>and</strong> the observed REY patterns for world seawater. The<br />
15
net effect <strong>of</strong> carbonate complexation in solution <strong>and</strong> particle surface adsorption for the<br />
REY is that heavier rare earth elements tend to remain in solution, <strong>and</strong> conversely, the<br />
LREE partition more easily onto particle surfaces. The result is a generally monotonic<br />
increase in shale-normalized rare earth element concentrations in seawater as a function<br />
<strong>of</strong> increasing atomic mass. However, as seen in Figs. 4-6, some elements in the REY<br />
series display anomalous behavior relative to their immediate neighbors.<br />
The most strikingly anomalous element among the lanthanides is Ce, which, as<br />
mentioned previously, can exist in the Ce 4+<br />
valence state in oxic marine systems.<br />
Experimental work has demonstrated that when lanthanides are adsorbed onto Mnoxides<br />
<strong>and</strong> Fe-oxyhydroxides, positive Ce anomalies develop in the solid phases, which<br />
have been attributed to the oxidation <strong>of</strong> Ce(III) to highly insoluble Ce(IV) on mineral<br />
surfaces (e.g., Koeppenkastrop <strong>and</strong> De Carlo, 1992; Bau, 1999). Positive Ce anomalies<br />
have also been noted in marine hydrogenetic Fe-Mn crusts (e.g., Bau et al., 1996, <strong>and</strong><br />
references therein), consistent with the oxidation <strong>of</strong> adsorbed Ce(III) to Ce(IV) <strong>and</strong> the<br />
preferential retention <strong>of</strong> Ce on metal-oxyhydroxide surfaces in the marine environment.<br />
It should be noted that this fractionation <strong>of</strong> Ce is not observed under anoxic conditions,<br />
<strong>and</strong> the REY patterns <strong>of</strong> seawater sampled from anoxic basins do not display negative<br />
Ce anomalies (e.g., de Baar et al., 1988; German et al., 1991b; Bau et al., 1997b).<br />
The other rare earth element that displays strongly anomalous behavior in<br />
seawater is Y, which is highly enriched relative to neighboring Dy <strong>and</strong> Ho (Figs. 4-6).<br />
While the ionic radii <strong>of</strong> Y is similar to Ho, <strong>and</strong> these elements are considered to be<br />
geochemical twins that exhibit similar behavior in most geological processes, Y is<br />
strongly fractionated from Ho (<strong>and</strong> the other lanthanides) in seawater. Relative to Ho,<br />
carbonate complexation constants for Y in seawater are similar for moncarbonate<br />
species, <strong>and</strong> somewhat different for dicarbonate species (Luo <strong>and</strong> Byrne, 2004).<br />
16
10 2 world seawater (x10 8 )<br />
sample/PAAS<br />
10 1<br />
10 0<br />
hydrogenetic Fe-Mn crust<br />
La Ce Pr Nd PmSm Eu Gd Tb Dy Y Ho Er Tm Yb Lu<br />
Figure 8. Comparison <strong>of</strong> REY patterns in average world seawater (data same as Fig. 4) <strong>and</strong> average<br />
(n=12) non-phosphatized hydrogenetic Fe-Mn crusts from the central Pacific Ocean (data from Bau et<br />
al., 1996). Note complementary Ce <strong>and</strong> Y anomalies between seawater <strong>and</strong> precipitated Fe-Mn oxide<br />
particles as represented by Fe-Mn crust.<br />
However, the strong positive Y anomaly observed in seawater results from particle<br />
surface complexation effects, as Y adsorbed to particle surfaces is bound less strongly<br />
than the lanthanide elements (e.g., Bau, 1999). The significant fractionation <strong>of</strong> both Y<br />
<strong>and</strong> Ce from the other REY in seawater due to processes occurring on particle surfaces<br />
is clearly evident in the REY patterns for hydrogenetic Fe-Mn crusts as shown in Figure<br />
8. This does not imply that hydrogenetic Fe-Mn crusts are solely responsible for these<br />
anomalies, as Ce <strong>and</strong> Y fractionation has been observed to develop as a function <strong>of</strong><br />
increasing salinity in estuaries (Lawrence <strong>and</strong> Kamber, 2006). It therefore appears that<br />
significant Ce <strong>and</strong> Y anomalies begin to form during mixing <strong>of</strong> river waters <strong>and</strong><br />
seawater, <strong>and</strong> may be present in the truly dissolved REY fraction <strong>of</strong> river waters (Byrne<br />
<strong>and</strong> Liu, 1998).<br />
Of the remaining REY, La, Gd, <strong>and</strong> Lu display anomalous behavior in seawater to<br />
varying degrees. Strong positive La anomalies are typical <strong>of</strong> seawater, though this<br />
17
feature was not recognized in early studies due to the negative anomaly present for<br />
neighboring Ce (e.g., de Baar et al., 1983). Positive Gd <strong>and</strong> Lu anomalies are also<br />
present in seawater REY patterns, though the magnitude <strong>of</strong> these anomalies are<br />
significantly smaller (see Fig. 6). The presence <strong>of</strong> positive La, Gd, <strong>and</strong> Lu anomalies<br />
was originally postulated to arise from a greater stability <strong>of</strong> the solution complexes for<br />
these elements, which presumably resulted from the exactly empty, half-filled, <strong>and</strong><br />
completely filled 4f electron shell configurations for La, Gd, <strong>and</strong> Lu, respectively (de<br />
Baar et al., 1985b). Early work by Cantrell <strong>and</strong> Byrne (1987), as shown in Fig. 7,<br />
predicted no anomalous behavior for these elements, though this study reported<br />
experimental data for only a few REY (e.g., Ce, Eu, Tb), which were then extrapolated<br />
across the entire REY series. More recent experimental studies (Luo <strong>and</strong> Byrne, 2004)<br />
determined equilibrium formation constants for carbonate complexes <strong>of</strong> all 15 REY,<br />
<strong>and</strong> these results are shown in Figure 9. Equilibrium constants for the formation <strong>of</strong><br />
monocarbonate REY complexes in high ionic strength solutions mimicking seawater do<br />
not vary smoothly across the REY series, with La, Gd, Y, <strong>and</strong> Lu showing the greatest<br />
deviations. Predominance diagrams for mono- <strong>and</strong> dicarbonate REY complexes in<br />
seawater also predict that La, Gd, Y, <strong>and</strong> Lu should fractionate from their immediate<br />
neighbors (Fig. 9).<br />
It is also likely that the observed positive La <strong>and</strong> Gd anomalies in seawater are to<br />
some degree generated during adsorption/desorption processes on scavenging particle<br />
surfaces. The authors Byrne <strong>and</strong> Kim (1990) <strong>and</strong> Lee <strong>and</strong> Byrne (1993) suggested that<br />
organic lig<strong>and</strong>s on particle surfaces (e.g., carboxylate) should discriminate against La<br />
<strong>and</strong> Gd, producing an enrichment <strong>of</strong> these elements in seawater. This would be<br />
consistent with the conclusions <strong>of</strong> Bau et al. (1999) regarding REY surface<br />
complexation on hydrous Fe-Mn oxides in seawater. It therefore appears that the La,<br />
18
6.2<br />
La Ce Pr Nd PmSm Eu Gd Tb Dy Y Ho Er Tm Yb Lu<br />
6.0<br />
5.8<br />
log CO3<br />
β 1<br />
5.6<br />
5.4<br />
5.2<br />
5.0<br />
log CO3<br />
β 1<br />
= [MCO + 3 ][M3+ ] -1 [CO 2-<br />
3 ]-1 TOTAL<br />
9.0<br />
8.5<br />
M(CO 3<br />
) - 2<br />
8.0<br />
7.5<br />
7.0<br />
C M 2<br />
pH<br />
6.5<br />
6.0<br />
5.5<br />
5.0<br />
4.5<br />
MCO + 3<br />
C M 1<br />
M 3+<br />
La Ce Pr Nd PmSm Eu Gd Tb Dy Y Ho Er Tm Yb Lu<br />
Figure 9. Carbonate equilibrium constants <strong>and</strong> REY speciation for seawater as predicted by Luo <strong>and</strong><br />
Byrne (2004). The top diagram represents equilibrium constants (molal concentration) for the formation<br />
<strong>of</strong> monocarbonate complexes in a 0.7 molal NaClO 4 solution at 25° C. The equilibrium constants for the<br />
REY do not vary smoothly, <strong>and</strong> local minimums occur for La, Gd, Y, <strong>and</strong> Lu. The lower diagram<br />
predicts the predominance <strong>of</strong> REY carbonate complexes in seawater as a function <strong>of</strong> pH, <strong>and</strong> the grey<br />
area represents a typical seawater pH range <strong>of</strong> ~7.7-8.1. The curve C M 1 represents the pH values at which<br />
concentrations <strong>of</strong> M + <strong>and</strong> MCO 3 + are equal, <strong>and</strong> C M 2 is the pH at which MCO 3<br />
+<br />
= M(CO 3 ) 2 .<br />
Gd, <strong>and</strong> Y anomalies in seawater are due to the interplay <strong>of</strong> competing solution <strong>and</strong><br />
surface complexation effects. Regarding Lu, it is interesting that the predominance<br />
diagram in Fig. 9 rather suggests that Yb is anomalous. As both Yb <strong>and</strong> Lu are expected<br />
to be primarily complexed by dicarbonate at seawater pH values, the data in Fig. 9<br />
predict that Yb should display a negative anomaly, resulting in an apparent ‘positive’<br />
Lu anomaly. For example, at a seawater pH <strong>of</strong> 7.5 the data <strong>of</strong> Luo <strong>and</strong> Byrne (2004)<br />
predict a greater relative proportion <strong>of</strong> Yb exists as monocarbonate YbCO + 3 than would<br />
be expected from interpolation between Tm <strong>and</strong> Lu. This suggests preferential<br />
scavenging <strong>of</strong> Yb from seawater by reactive particle surfaces; however, there is no<br />
19
sound geochemical basis for anomalous behavior <strong>of</strong> Yb (e.g., electron configuration,<br />
etc.). Considering that the monocarbonate equilibrium constants for Yb are consistent<br />
with those <strong>of</strong> Er <strong>and</strong> Tm, <strong>and</strong> no discussion <strong>of</strong> ‘anomalous’ dicarbonate Yb<br />
complexation is provided by Luo <strong>and</strong> Byrne (2004), it is concluded that the slight<br />
positive Lu anomalies observed in seawater (see Fig. 6) likely represent real excursions<br />
<strong>of</strong> Lu concentrations from those predicted by extrapolation from Tm <strong>and</strong> Yb.<br />
1.5.3. Sm-Nd isotopic signatures in the Earth’s crust <strong>and</strong> seawater<br />
Samarium <strong>and</strong> Nd each have several isotopes, <strong>and</strong> 147 Sm radioactively decays to<br />
stable 143 Nd with a half-life <strong>of</strong> 1.06 × 10 11 years. Even though both Sm <strong>and</strong> Nd have<br />
similar geochemical behavior, like all the lanthanides, fractionation between Sm <strong>and</strong> Nd<br />
does occur during differentiation <strong>of</strong> the Earth’s crust. Figure 10 presents complete REY<br />
patterns for evolved continental crust <strong>and</strong> MORB that have been normalized to<br />
chondrite (i.e., bulk silicate Earth), showing that crustal differentiation leads to<br />
distinctive Sm/Nd ratios. Upper continental crust is enriched in Nd relative to Sm <strong>and</strong><br />
possesses a low Sm/Nd ratio <strong>of</strong> ~0.17, whereas MORB derived from a depleted mantle<br />
source possesses higher Sm/Nd ratios <strong>of</strong> ~0.32. As a result <strong>of</strong> different Sm/Nd ratios<br />
<strong>and</strong> the subsequent decay <strong>of</strong> 147 Sm to 143 Nd, Sm-Nd isotopic ratios will evolve<br />
differently in continental <strong>and</strong> oceanic crust.<br />
The Nd isotopic signature <strong>of</strong> a geologic sample is reported as the 143 Nd/ 144 Nd<br />
ratio, as 144 Nd is a stable isotope whose abundance does not change with time. The Sm-<br />
Nd isotopic system has been used extensively for dating geologic samples <strong>and</strong> a<br />
thorough review <strong>of</strong> the techniques <strong>and</strong> applications may be found in Faure (1986). The<br />
determination <strong>of</strong> radiomentric ages relies upon constructing an isochron using measured<br />
147 Sm/ 144 Nd <strong>and</strong> 143 Nd/ 144 Nd ratios for a suite <strong>of</strong> cogenetic samples (e.g., minerals).<br />
20
1000<br />
chondrite-normalized<br />
100<br />
10<br />
1<br />
upper continental crust<br />
MORB<br />
La Ce Pr Nd PmSm Eu Gd Tb Dy Y Ho Er Tm Yb Lu<br />
Figure 10. Typical REY distributions in upper continental crust (data from Taylor <strong>and</strong> McLennan, 1995)<br />
<strong>and</strong> MORB (Niu et al., 1999) illustrating different Sm/Nd ratios. Light REE are strongly enriched in<br />
continental crust, which possesses negative Eu anomalies due to partitioning <strong>of</strong> Eu 2+ into plagioclase-rich<br />
residual material during intra-crustal partial melting <strong>and</strong> fractional crystallization.<br />
From the isochron an initial 143 Nd/ 144 Nd ratio may be determined, which allows the<br />
calculation <strong>of</strong> a geologic age for the samples.<br />
Alternatively, if the geologic age <strong>of</strong> a group <strong>of</strong> samples is already known, then<br />
measured Sm-Nd isotopic ratios in individual samples allow the calculation <strong>of</strong> initial<br />
143 Nd/ 144 Nd ratios without the need for constructing an isochron. This is the approach<br />
used in this thesis, as the samples investigated are marine sediments, <strong>and</strong> not igneous<br />
rocks which can <strong>of</strong>ten be considered to have crystallized in a geologically instantaneous<br />
time period. The purpose <strong>of</strong> this research is not to date the sediments studied, but rather<br />
to examine any variations in their initial 143 Nd/ 144 Nd ratios, as these variations may shed<br />
light on sources <strong>of</strong> Nd to Archean seawater, as well as any isotopic variability in<br />
Archean oceans.<br />
Measured 143 Nd/ 144 Nd ratios in a geologic sample are typically expressed using<br />
the Є Nd (t) notation <strong>of</strong> DePaolo <strong>and</strong> Wasserburg (1976). The Є Nd (t) value describes the<br />
21
deviation <strong>of</strong> the 143 Nd/ 144 Nd ratio measured in a sample relative to the 143 Nd/ 144 Nd ratio<br />
in a chondritic uniform reservoir (CHUR) in parts per 10 4 :<br />
Є Nd (t)<br />
=<br />
⎡(<br />
⎢<br />
⎢⎣<br />
143<br />
Nd/<br />
I<br />
144<br />
t<br />
CHUR<br />
Nd)<br />
i<br />
⎤<br />
− 1⎥<br />
⎥⎦<br />
x 10<br />
4<br />
where ( 143 Nd/ 144 Nd) i is the initial ratio in the sample (i.e., at the time it formed, t), I t CHUR<br />
is the CHUR 143 Nd/ 144 Nd ratio at the time the sample formed, <strong>and</strong> CHUR is considered<br />
to represent bulk silicate Earth. Since continental crust is characterized by low Sm/Nd<br />
ratios that produce relatively low amounts <strong>of</strong> radiogenic Nd, it displays negative Є Nd (t)<br />
values. Conversely, mafic oceanic crust derived from a depleted upper mantle is<br />
characterized by positive Є Nd (t) values, as MORB has relatively high Sm/Nd ratios <strong>and</strong><br />
excess radiogenic Nd. As the REY in modern seawater are primarily sourced from the<br />
weathering <strong>of</strong> continental crust, modern seawater possesses negative Є Nd (0) values<br />
between -1 <strong>and</strong> -20, though extreme values in this range are restricted to shallow waters<br />
(Goldstein <strong>and</strong> Hemming, 2003). Figure 11 shows average Є Nd (0) values for different<br />
world oceans, as well as simple models for the Nd isotopic evolution in continental<br />
crust <strong>and</strong> a depleted mantle over the past 4.0 Ga.<br />
The significant Nd isotopic variations between ocean basins indicates that the<br />
marine residence time <strong>of</strong> Nd (τ Nd ) is less than the mixing time <strong>of</strong> the oceans (~1000<br />
years, e.g., Broecker <strong>and</strong> Peng, 1982), <strong>and</strong> estimates <strong>of</strong> τ Nd in oxic seawater typically<br />
range from 300-1500 years (Je<strong>and</strong>el et al., 1995; Tachikawa et al., 2003). This is<br />
consistent with the particle reactive nature <strong>of</strong> the REE <strong>and</strong> the efficient Nd scavenging<br />
processes operating in modern oceans, as REE-rich hydrothermal fluid with highly<br />
positive Є Nd (0) has a negligible effect on 143 Nd/ 144 Nd ratios in world seawater (Fig. 11).<br />
22
15<br />
10<br />
TAG hydrothermal fluid<br />
5<br />
depleted mantle<br />
ε Nd<br />
(t)<br />
0<br />
-5<br />
Pacific Ocean<br />
Indian Ocean<br />
-10<br />
Atlantic Ocean<br />
-15<br />
mafic igneous<br />
-20<br />
felsic igneous<br />
0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0<br />
Ga<br />
continental crust<br />
Figure 11. The range <strong>of</strong> Nd isotopic ratios (grey bar) as observed in modern seawater (Goldstein <strong>and</strong><br />
Hemming, 2003), as well as Nd isotopic evolution in a depleted mantle <strong>and</strong> average continental crust.<br />
Average ocean basin seawater data from Piepgras <strong>and</strong> Wasserburg, (1980), <strong>and</strong> Trans-Atlantic<br />
Geotraverse (TAG) mid-ocean ridge hydrothermal fluid data reported by Mills et al. (2001). Isotopic<br />
evolution lines represent simple linear models where Є Nd (4.56 Ga) = 0, depleted mantle Є Nd (0) = +9<br />
(Salters <strong>and</strong> Stracke, 2004), <strong>and</strong> average continental crust Є Nd (0) = -17 (Goldstein et al., 1984). Also<br />
shown are data for mafic <strong>and</strong> felsic igneous rocks from the Kaapvaal craton, illustrating that ca. 3.0 Ga<br />
crust in the study area displayed a wide range <strong>of</strong> Є Nd (t) values. Igneous rock data from Hegner et al.<br />
(1984), Wilson <strong>and</strong> Carlson (1989), Nelson et al. (1992), Kröner <strong>and</strong> Tegtmeyer (1994), Kröner et al.<br />
(1996), <strong>and</strong> Chavagnac (2004).<br />
For the early Earth, the isotopic evolution <strong>of</strong> Nd in continental crust <strong>and</strong> a<br />
depleted mantle reservoir is debated (cf. Nägler <strong>and</strong> Kramers, 1998; Patchett <strong>and</strong><br />
Samson, 2003), <strong>and</strong> the isotopic evolution lines in Fig. 11 represent simple linear<br />
models. Regardless <strong>of</strong> the model favored, however, it appears that by ~3.8 Ga Є Nd values<br />
in the upper mantle were at least +2 (Bennet, 2003), implying that significant crustal<br />
differentiation was occurring in the early Archean. For southern Africa, the range <strong>of</strong><br />
observed Є Nd (t) values observed for >2.7 Ga felsic <strong>and</strong> igneous rocks is large (≥10 Є Nd -<br />
units, Fig. 11), indicating that crustal sources <strong>of</strong> Nd to Archean seawater in the vicinity<br />
<strong>of</strong> the Kaapvaal craton were isotopically diverse. Due to the large range in Є Nd (t) values<br />
for Nd sources, <strong>and</strong> assuming suitable archives for seawater 143 Nd/ 144 Nd ratios are<br />
23
available, it should be possible to distinguish if ca. 3.0 Ga seawater near the Kaapvaal<br />
craton was similar to modern seawater in possessing negative Є Nd (t).<br />
1.5.4. Iron-formations as seawater archives<br />
Reconstructing Nd isotopic ratios in paleoseawater is possible as isotopic<br />
fractionation during incorporation <strong>of</strong> Nd into pure marine chemical sediments has not<br />
been observed (Frank, 2002). Even marine precipitates possessing strongly fractionated<br />
REY patterns relative to seawater, such as hydrogenetic Fe-Mn crusts (Fig. 8),<br />
accurately record the Nd isotopic composition <strong>of</strong> ambient seawater (Goldstein <strong>and</strong><br />
Hemming, 2003, <strong>and</strong> references therein). Therefore, sufficiently pure marine chemical<br />
precipitates from the Archean should represent suitable archives for reconstructing<br />
primary seawater Nd isotopic ratios.<br />
Potential seawater archives for the Archean would be carbonate rocks such as<br />
limestones or dolomites; however, the carbonate record prior to ~2.7 Ga is poor,<br />
consisting primarily <strong>of</strong> thin, discontinuous, <strong>and</strong> poorly preserved occurrences<br />
(Grotzinger, 1989). Prior to 2.7 Ga the best c<strong>and</strong>idates for archives <strong>of</strong> seawater<br />
143 Nd/ 144 Nd ratios are iron-formations, as they are ubiquitous in both time <strong>and</strong> space for<br />
this period <strong>of</strong> Earth’s history. While the exact process by which they formed remains<br />
unknown, the frequent association <strong>of</strong> IFs with marine carbonates <strong>and</strong> sea level rise<br />
(Klein <strong>and</strong> Beukes, 1989; Simonson <strong>and</strong> Hassler, 1996) supports the general consensus<br />
that IFs are chemical sediments precipitated from seawater (cf. Simonson, 2003).<br />
The geochemical evidence used in this thesis to screen IF samples for suitability<br />
as seawater archives primarily rests upon measured concentrations <strong>of</strong> elements such as<br />
Al, Ti, <strong>and</strong> Th. These refractory elements are good indicators <strong>of</strong> clastic detritus that may<br />
contaminate pure marine precipitates. Data presented in the following chapters suggests<br />
24
10 0<br />
10 -1<br />
Isua IF (3.8 Ga)<br />
Kuruman IF (2.5 Ga)<br />
10 -4<br />
Campbellr<strong>and</strong> dolomite (2.5 Ga)<br />
Holocene carbonate reef<br />
La Ce Pr Nd PmSm Eu Gd Tb Dy Y Ho Er Tm Yb Lu<br />
sample/PAAS<br />
10 -2<br />
10 -3<br />
seawater x10 4<br />
Figure 12. Distributions <strong>of</strong> the REY in average world seawater (same as Fig. 4), <strong>and</strong> various chemical<br />
sediments that are free from contamination with Al-rich clastic detritus. Data representing the Isua IF<br />
(Greenl<strong>and</strong>) are for the IF-G reference st<strong>and</strong>ard analyzed at <strong>Jacobs</strong> <strong>University</strong> Bremen as discussed in<br />
Chapter 5. Kuruman (South Africa) IF data represent the average <strong>of</strong> two analyses as reported in Bau et al.<br />
(1997a). Campbellr<strong>and</strong> dolomite represents unpublished data from GKP-01 drillcore (Agouron-<br />
Griqual<strong>and</strong> Paleoproterozoic Drilling Project, administered by <strong>University</strong> <strong>of</strong> Johannesburg). Holocene<br />
(≤10,000 years old) carbonate reef data from Webb <strong>and</strong> Kamber (2000). The REY patterns <strong>of</strong> IFs <strong>and</strong><br />
both ancient <strong>and</strong> modern carbonates are very similar to each other <strong>and</strong> modern seawater. These<br />
observations suggest that Archean IFs possessing seawater-like REY distributions are likely to be<br />
excellent archives for the Nd isotopic ratios present in contemporaneous seawater.<br />
that, as a general rule, IF samples containing more than 0.5-0.7% Al 2 O 3 are unsuitable<br />
for reconstructing seawater Nd isotopic ratios. Additionally, the distribution <strong>of</strong> the REY<br />
is very useful at screening potential seawater archives, as Archean marine precipitates<br />
frequently display REY patterns that are very similar to those observed in modern<br />
seawater.<br />
Figure 12 shows REY patterns observed in Archean IFs <strong>and</strong> both modern <strong>and</strong><br />
ancient marine carbonates. The distribution <strong>of</strong> the REY is generally indistinguishable<br />
between modern seawater <strong>and</strong> carbonate reefs, as well as Archean dolomite <strong>and</strong> IFs.<br />
The general enrichment <strong>of</strong> the HREE over the LREE <strong>and</strong> positive La <strong>and</strong> Y anomalies<br />
found in seawater are common to all marine precipitates shown in Fig. 12 regardless <strong>of</strong><br />
age. The primary differences are for Ce, which reflects the different oxidation states <strong>of</strong><br />
25
modern <strong>and</strong> Archean oceans, as well as for Eu, which indicates a greater contribution to<br />
Archean seawater from high-T hydrothermal fluids (e.g., Figs. 4,5). It should be noted<br />
that the small negative Ce anomaly in the shallow water Holocene reef carbonates<br />
relative to world seawater is not inconsistent, as very shallow seawater <strong>of</strong>ten displays<br />
smaller negative Ce anomalies compared to bulk seawater (e.g., de Baar et al., 1983).<br />
The fact that Archean IFs commonly have seawater-like REY distributions is not<br />
consistent with REY scavenging by Fe-oxyhydroxides in a manner similar to that<br />
observed in modern oceans. As shown in Fig. 8, adsorption-desorption reactions at<br />
equilibrium strongly fractionate the REY between metal-oxides <strong>and</strong> seawater. If<br />
Archean IFs precipitated from seawater as Fe(III) minerals, then their observed REY<br />
distributions would require quantitative scavenging <strong>of</strong> all REY in solution, perhaps in a<br />
manner analogous to the laboratory co-precipitation <strong>of</strong> REE-Fe-oxide phases from<br />
seawater in early studies (e.g., Goldberg et al., 1963). However, the exact nature <strong>of</strong> the<br />
primary IF precipitate remains unresolved, though data presented in Chapter 4 suggests<br />
that initial deposition <strong>of</strong> Fe(II) mineral phases may have been important. Regardless, the<br />
seawater-like REY distributions in Archean IFs, <strong>and</strong> the lack <strong>of</strong> other suitable archives,<br />
indicates that IFs <strong>of</strong>fer the best opportunity for discerning the Nd isotopic composition<br />
<strong>of</strong> >2.7 Ga seawater.<br />
26
References<br />
Alibert, C., McCulloch, M.T., 1993. Rare earth element <strong>and</strong> Nd isotopic compositions<br />
<strong>of</strong> the b<strong>and</strong>ed iron-formations <strong>and</strong> associated shales from Hamersley, Western<br />
Australia. Geochim. Cosmochim. Acta 57, 187-204.<br />
Alibo, D.S., Nozaki, Y., 1999. Rare earth elements in seawater: Particle association,<br />
shale normalization, <strong>and</strong> Ce-oxidation. Geochim. Cosmochim. Acta 63, 363-372.<br />
Anders, E., Grevesse, N., 1989. Abundances <strong>of</strong> the elements: meteoric <strong>and</strong> solar.<br />
Geochim. Cosmochim. Acta 53, 197-214.<br />
Bau M. (1999) Scavenging <strong>of</strong> dissolved yttrium <strong>and</strong> rare earths by precipitating iron<br />
oxyhydroxide: Experimental evidence for Ce oxidation, Y-Ho fractionation, <strong>and</strong><br />
lanthanide tetrad effect. Geochim. Cosmochim. Acta, 63, 67–77.<br />
Bau M. <strong>and</strong> Dulski P. (1999) Comparing yttrium <strong>and</strong> rare earths in hydrothermal fluids<br />
from the Mid-Atlantic Ridge: implications for Y <strong>and</strong> REE behaviour during nearvent<br />
mixing <strong>and</strong> for the YrHo ratio <strong>of</strong> Proterozoic seawater. Chem. Geol., 155,<br />
77–90.<br />
Bau M., Koschinsky A., Dulski P., <strong>and</strong> Hein J.R. (1996) Comparison <strong>of</strong> the partitioning<br />
behaviours <strong>of</strong> yttrium, rare earth elements, <strong>and</strong> titanium between hydrogenetic<br />
marine ferromanganese crusts <strong>and</strong> seawater. Geochim. Cosmochim. Acta, 60,<br />
1709-1725.<br />
Bau M., Höhndorf A., Dulski P., <strong>and</strong> Beukes, N.J. (1997a). Sources <strong>of</strong> rare-earth<br />
elements <strong>and</strong> iron in Paleoproterozoic iron-formations from the Transvaal<br />
Supergroup, South Africa: Evidence from neodymium isotopes. J. Geol. 105, 121-<br />
129.<br />
27
Bau M., Möller P., <strong>and</strong> Dulski P. (1997b) Yttrium <strong>and</strong> lanthanides in eastern<br />
Mediterranean seawater <strong>and</strong> their fractionation during redox-cycling. Mar. Chem.<br />
56, 123-131.<br />
Bau M., Knappe A., Dulski P. (2006) Anthropogenic gadolinium as a micropollutant in<br />
river waters in Pennsylvania <strong>and</strong> in Lake Erie, northeastern United States. Chemie<br />
der Erde, (66), 143-152.<br />
Germain Bayon G., German C.R., Burton K.W., Nesbitt R.W., Rogers N. (2004)<br />
Sedimentary Fe–Mn oxyhydroxides as paleoceanographic archives <strong>and</strong> the role <strong>of</strong><br />
aeolian flux in regulating oceanic dissolved REE. Earth Planet. Sci. Lett., 224,<br />
477-492.<br />
Bennet V.C. (2003) Compositional evolution <strong>of</strong> the mantle, pp. 493-519. In The Mantle<br />
<strong>and</strong> Core (ed. R.W. Carlson) Vol. 2 Treatise on Geochemistry (eds. H.D. Holl<strong>and</strong><br />
<strong>and</strong> K.K. Turekian), Elsevier-Pergamon, Oxford.<br />
Beukes, N.J., 1983. Palaeoenvironmental setting <strong>of</strong> iron-formations in the depositional<br />
basin <strong>of</strong> the Transvaal Supergroup, South Africa, in: Trendall, A.F., Morris, R.C.<br />
(Eds.), Iron-formation Facts <strong>and</strong> Problems, 131-198, Vol. 6 Developments in<br />
Precambrian Geology (Ed. B.F. Windley), Elsevier, Amsterdam.<br />
Brookins, D.G (1989) Aqueous geochemistry <strong>of</strong> rare earth elements, in: Lipin, B.R.,<br />
McKay, G.A. (Eds.), Geochemistry <strong>and</strong> Mineralogy <strong>of</strong> Rare Earth Elements.<br />
Reviews in Mineralogy 21, Mineral. Soc. Amer., 201-225.<br />
Byrne R.H. <strong>and</strong> Cantrell K.J. (1987) Rare earth element complexation by carbonate <strong>and</strong><br />
oxalate ions. Geochim. Cosmochim. Acta, 51, 597-605.<br />
Byrne R.H. <strong>and</strong> Kim K. (1990) Rare earth element scavenging in seawater. Geochim.<br />
Cosmochim. Acta, 54, 2645-2656.<br />
28
Byrne R.H. <strong>and</strong> Liu X. (1998) A coupled riverine-marine fractionation model for<br />
dissolved rare earths <strong>and</strong> yttrium. Aquatic Geochem., 4, 103–121.<br />
Chavagnac V. (2004) A geochemical <strong>and</strong> Nd isotopic study <strong>of</strong> Barberton komatiites<br />
(South Africa): implication for the Archean mantle. Lithos, 75, 253– 281.<br />
Cidu R. <strong>and</strong> Biddau R. (2007) Transport <strong>of</strong> trace elements under different seasonal<br />
conditions: Effects on the quality <strong>of</strong> river water in a Mediterranean area. Applied<br />
Geochemistry 22, 2777–2794.<br />
Compston W. <strong>and</strong> Kröner A. (1988) Multiple zircon growth within early Archaean<br />
tonalitic gneiss from the Ancient Gneiss Complex, Swazil<strong>and</strong>. Earth Planet. Sci.<br />
Lett., 87, 13-28.<br />
de Baar, H. J. W., M. P. Bacon <strong>and</strong> P. G. Brewer (1983) Rare earth distributions with a<br />
positive cerium anomaly in the western North Atlantic Ocean. Nature, 301, 324-<br />
327.<br />
de Baar H.J.W., Bacon M.P., Brewer P.G. <strong>and</strong> Brul<strong>and</strong> K.W. (1985a). Rare earth<br />
elements in the Atlantic <strong>and</strong> Pacific Oceans. Geochim. Cosmochim. Acta, 49,<br />
1943-1959.<br />
de Baar H.J.W., Brewer P.G., <strong>and</strong> Bacon M.P. (1985b) Anomalies in rare earth<br />
distributions in seawater: Gd <strong>and</strong> Tb. Geochim. Cosmochim. Acta, 49, 1961-1969.<br />
de Baar H.J.W., German C.R., Elderfield H., <strong>and</strong> van Gaans P. (1988) Rare earth<br />
element distributions in anoxic waters <strong>of</strong> the Cariaco Trench. Geochim.<br />
Cosmochim. Acta, 52, 1203-1219.<br />
DePaolo, D.J., Wasserburg, G.J., 1976. Nd isotopic variations <strong>and</strong> petrogenetic models.<br />
Geophys. Res. Lett. 3, 249-252.<br />
Elderfield H. (1988) The oceanic chemistry <strong>of</strong> the rare-earth elements. Phil. Trans. R.<br />
Soc. Lond., A325, 105-126.<br />
29
Elderfield H. <strong>and</strong> Greaves M.J. (1982) The rare earth elements in seawater. Nature, 296,<br />
214-219.<br />
Elderfield, H. <strong>and</strong> E. R. Sholkovitz (1987). Rare earth elements in the pore waters <strong>of</strong><br />
reducing nearshore sediments. Earth Planet. Sci. Lett., 82, 280-290.<br />
Elderfeld H., Upstil-Goddard R., <strong>and</strong> Sholkovitz E.R. (1990) The rare earth elements in<br />
rivers, estuaries <strong>and</strong> coastal sea waters: processes affecting crustal input <strong>of</strong><br />
elements to the ocean <strong>and</strong> their significance to the composition <strong>of</strong> seawater.<br />
Geochim. Cosmochim. Acta, 54, 971-991.<br />
Erel Y. <strong>and</strong> Morgan J.J. (1991) The effect <strong>of</strong> surface reactions on the relative<br />
abundances <strong>of</strong> trace metals in deep-ocean water. Geochim. Cosmochim. Acta, 55,<br />
1807-1813.<br />
Erel Y. <strong>and</strong> Stolper E.M. (1993) Modeling <strong>of</strong> rare-earth element partitioning between<br />
particles <strong>and</strong> solution in aquatic environments. Geochim. Cosmochim. Acta, 57,<br />
513-518.<br />
Faure G. (1986) Principles <strong>of</strong> Isotope Geology. John Wiley <strong>and</strong> Sons, New York,<br />
Gammons C.H. , Wood S.A.., Pedrozo F., Varekamp J. C., Nelson B.J., Shope C.L., <strong>and</strong><br />
Baffico G. (2005b) Hydrogeochemistry <strong>and</strong> rare earth element behavior in a<br />
volcanically acidified watershed in Patagonia, Argentina. Chem. Geol., 222, 249-<br />
267.<br />
German, C. R. <strong>and</strong> H. Elderfield (1989). Rare earth elements in Saanich Inlet, British<br />
Columbia, a seasonally anoxic basin. Geochim. Cosmochim. Acta, 53, 2561-2571.<br />
German C. R., Klinkhammer G.P., Edmond J.M., Mitra A., <strong>and</strong> Elderfield H. (1990)<br />
Hydrothermal scavenging <strong>of</strong> rare-earth elements in the ocean. Nature, 345, 516-<br />
518.<br />
30
German C. R., Campbell A.C., <strong>and</strong> Edmond J.M. (1991a) Hydrothermal scavenging at<br />
the Mid-Atlantic Ridge: Modification <strong>of</strong> trace element dissolved fluxes. Earth<br />
Planet. Sci. Lett., 107, 101-114.<br />
German C.R., Holliday B.P., <strong>and</strong> Elderfeld H. (1991b) Redox cycling <strong>of</strong> rare earth elements<br />
in the suboxic zone <strong>of</strong> the Black Sea. Geochim. Cosmochim. Acta, 55, 3553-3558.<br />
German C.R., Masuzawa T., Greaves M.J., Elderfeld H. <strong>and</strong> Edmond J.M. (1995)<br />
Dissolved rare earth elements in the Southern Ocean: cerium oxidation <strong>and</strong> the<br />
influence <strong>of</strong> hydrography. Geochim. Cosmochim. Acta 59, 1551-1558.<br />
Goldberg E.D., Koide M., Schmitt R.A., <strong>and</strong> Smith R.H. (1963) Rare-earth distributions<br />
in the marine environment. J. Geophys. Res., 68, 4209-4217.<br />
Goldstein S.L. <strong>and</strong> <strong>Jacobs</strong>en S.B. (1988) REE in the Great Whale River estuary,<br />
northwest Quebec. Earth Planet. Sci. Lett., 88, 241-252.<br />
Goldstein S.L. <strong>and</strong> Hemming S.R. (2003) Long-lived isotopic tracers in oceanography,<br />
paleoceanography, <strong>and</strong> ice-sheet dynamics, pp. 453-489. In The Oceans <strong>and</strong><br />
Marine Geochemistry (ed. H. Elderfield) Vol. 6 Treatise on Geochemistry (eds.<br />
H.D. Holl<strong>and</strong> <strong>and</strong> K.K. Turekian), Elsevier-Pergamon, Oxford.<br />
Goldstein S.L., O'Nions R.K., <strong>and</strong> Hamilton P.J. (1984) A Sm-Nd isotopic study <strong>of</strong><br />
atmospheric dusts <strong>and</strong> particulates from major river systems. Earth Planet. Sci.<br />
Lett., 70, 221-236.<br />
Greaves M.J., Elderfield H., <strong>and</strong> Sholkovitz E.R. (1999) Aeolian sources <strong>of</strong> rare earth<br />
elements to the Western Pacific Ocean. Marine Chem., 68, 31-38.<br />
Grotzinger, J.P., 1989. Facies <strong>and</strong> evolution <strong>of</strong> Precambrian carbonate depositional<br />
systems: emergence <strong>of</strong> the modern platform archetype. controls on carbonate<br />
platform <strong>and</strong> basin development, SEPM (Soc. Econ. Paleont. Mineral.) Spec. Pub.<br />
44. 79-106.<br />
31
Haley B.A., Klinkhammer G.P. (2003) Complete separation <strong>of</strong> rare earth elements from<br />
small volume seawater samples by automated ion chromatography: method<br />
development <strong>and</strong> application to benthic flux. Mar. Chem., 82, 197–220.<br />
Haley B.A., Klinkhammer G.P. <strong>and</strong> McManus J. (2004) Rare earth elements in pore<br />
waters <strong>of</strong> marine sediments. Geochim. Cosmochim. Acta, 68, 1265–1279.<br />
Hegner E., Kroner A., <strong>and</strong> H<strong>of</strong>mann A.W. (1984) Age <strong>and</strong> isotope geochemistry <strong>of</strong> the<br />
Archaean Pongola <strong>and</strong> Usushwana suites in Swazil<strong>and</strong>, southern Africa: a case for<br />
crustal contamination <strong>of</strong> mantle-derived magma. Earth Planet. Sci. Lett., 70, 267-<br />
279.<br />
Høgdahl O.T., Melsom S., <strong>and</strong> Bowen V.T. (1968) Neutron activation analysis <strong>of</strong><br />
lanthanide elements in sea water. in Trace Inorganics in Water, pp. 308-325,<br />
Advan. Chem. Series 73.<br />
Holl<strong>and</strong> H.D. (1972) The geologic history <strong>of</strong> seawater – an attempt to solve the<br />
problem. Geochim. Cosmochim. Acta, 36, 637-651.<br />
Hoyle J., Elderfield H., Gledhill A., <strong>and</strong> Greaves M. (1984) The behaviour <strong>of</strong> the rare<br />
earth elements during mixing <strong>of</strong> river <strong>and</strong> sea waters. Geochim. Cosmochim.<br />
Acta, 48, 143-149.<br />
<strong>Jacobs</strong>en, S.B., Pimentel-Klose, M.R., 1988a. A Nd isotopic study <strong>of</strong> the Hamersley <strong>and</strong><br />
Michipicoten b<strong>and</strong>ed iron formations: the source <strong>of</strong> REE <strong>and</strong> Fe in Archean<br />
oceans. Earth Planet. Sci. Lett. 87, 29-44.<br />
<strong>Jacobs</strong>en, S.B., Pimentel-Klose, M.R., 1988b. Nd isotopic variations in Precambrian<br />
b<strong>and</strong>ed iron formations. Geophys. Res. Lett. 15, 393-396.<br />
James H.L. (1954) Sedimentary facies <strong>of</strong> iron-formation. Econ. Geol., 49, 235-293.<br />
32
Je<strong>and</strong>el, C., Bishop, K., Zindler, A., 1995. Exchange <strong>of</strong> neodymium <strong>and</strong> its isotopes<br />
between seawater <strong>and</strong> small <strong>and</strong> large particles in the Sargasso Sea. Geochim.<br />
Cosmochim. Acta 59, 535-547.<br />
Klein, C., Beukes, N.J., 1989. Geochemistry <strong>and</strong> sedimentology <strong>of</strong> a facies transition<br />
from limestone to iron-formation in the Early Proterozoic Transvaal Supergroup,<br />
South Africa. Econ. Geol. 84, 1733-1774.<br />
Koeppenkastrop D. <strong>and</strong> De Carlo E.H. (1992) Sorption <strong>of</strong> rare-earth elements from<br />
seawater onto synthetic mineral particles: An experimental approach. Chem.<br />
Geol., 95, 251-263.<br />
Koeppenkastrop D. <strong>and</strong> De Carlo E.H. (1993) Uptake <strong>of</strong> rare earth elements from<br />
solution by metal oxides. Environ. Sci. Technol., 27, 1796-1802.<br />
Kröner A., <strong>and</strong> Tegtmeyer A. (1994) Gneiss-greenstone relationships in the Ancient<br />
Gneiss Complex <strong>of</strong> southwestern Swazil<strong>and</strong>, southern Africa, <strong>and</strong> implications for<br />
early crustal evolution. Precam. Res., 67, 109-139.<br />
Kröner A., Hegner E., Wendt J.I., <strong>and</strong> Byerly G.R. (1996) The oldest part <strong>of</strong> the<br />
Barberton granitoid-greenstone terrain, South Africa: evidence for crust formation<br />
between 3.5 <strong>and</strong> 3.7 Ga. Precam. Res. 78, 105-124.<br />
Kulaksız S. <strong>and</strong> Bau M. (2007) Contrasting behaviour <strong>of</strong> anthropogenic gadolinium <strong>and</strong><br />
natural rare earth elements in estuaries <strong>and</strong> the gadolinium input into the North<br />
Sea. Earth Planet. Sci. Lett., 260, 361-371.<br />
Lawrence M.G., <strong>and</strong> Kamber B.S. (2006) The behaviour <strong>of</strong> the rare earth elements<br />
during estuarine mixing-revisited. Marine Chemistry, 100, 147-161.<br />
Lawrence M.G., Greig A., Collerson K.D., <strong>and</strong> Kamber B.S. (2006) Rare Earth Element<br />
<strong>and</strong> Yttrium Variability in South East Queensl<strong>and</strong> Waterways. Aqua. Geochem.,<br />
12, 39–72.<br />
33
Lee J.H. <strong>and</strong> Byrne R.H. (1993) Complexation <strong>of</strong> trivalent rare earth elements (Ce, Eu,<br />
Gd, Tb, Yb) by carbonate ions. Geochim. Cosmochim. Acta, 57, 295-302.<br />
Liu X. <strong>and</strong> Byrne R.H. (1998) Comprehensive investigation <strong>of</strong> yttrium <strong>and</strong> rare earth<br />
element complexation by carbonate ions using ICP-mass spectrometry. J. Solution<br />
Chem., 27, 803-815.<br />
Lowe D.R. (1994) Accretionary history <strong>of</strong> the Archean Barberton Greenstone Belt<br />
(3.55-3.22 Ga), southern Africa. Geology, 200, 1099-1102.<br />
Luo Y. <strong>and</strong> Byrne R.H. (2004) Carbonate complexation <strong>of</strong> yttrium <strong>and</strong> the rare earth<br />
elements in natural waters. Geochim. Cosmochim. Acta, 68, 691–699.<br />
Martin J.M., Hogdahl 0., <strong>and</strong> Phillppot L.C. (1976) Rare earh element supply to the<br />
oceans. J. Geophys. Res., 81, 3119-3124.<br />
McLennan, S.M., 1989. Rare earth elements in sedimentary rocks: influence <strong>of</strong><br />
provenance <strong>and</strong> sedimentary processes, in: Lipin, B.R., McKay, G.A. (Eds.),<br />
Geochemistry <strong>and</strong> Mineralogy <strong>of</strong> Rare Earth Elements. Reviews in Mineralogy<br />
21, Mineral. Soc. Amer., 169-200.<br />
Miller, R.G., O'Nions, R.K., 1985. Sources <strong>of</strong> Precambrian chemical <strong>and</strong> clastic<br />
sediments. Nature 314, 325-330.<br />
Mills R.A., Wells D.M., Roberts S. (2001) Genesis <strong>of</strong> ferromanganese crusts from the<br />
TAG hydrothermal field. Chem. Geol., 176, 283-293.<br />
Nägler T.F <strong>and</strong> Kramers J.D. (1998) Nd isotopic evolution <strong>of</strong> the upper mantle during<br />
the Precambrian: models, data <strong>and</strong> the uncertainty <strong>of</strong> both. Precam. Res., 91, 233-<br />
252.<br />
Nakai S., Halliday A.N., <strong>and</strong> Rea D.K. (1993) Provenance <strong>of</strong> dust in the Pacific Ocean.<br />
Earth Planet. Sci. Lett., 119, 143-157.<br />
34
Nelson D.R., Trendall A.F., de Laeter J.R., Grobler N.J., Fletcher I.R. (1992) A<br />
comparative study <strong>of</strong> the geochemical <strong>and</strong> isotopic systematics <strong>of</strong> late Archaean<br />
flood basalts from the Pilbara <strong>and</strong> Kaapvaal Cratons. Precam. Res., 54, 231-256.<br />
Niu, Y., Collerson, K., Batiza, R., Wendt, J. <strong>and</strong> Regelous, M. (1999). Origin <strong>of</strong><br />
enriched-type mid-ocean ridge basalt at ridges far from mantle plumes: The East<br />
Pacific Rise at 11°20'N. J. Geophys. Res., 104(B4): doi: 10.1029/1998JB900037.<br />
Nozaki, Y., Alibo, D.S., Amakawa, H., Gamo, T., Hasumoto, H., 1999. Dissolved rare<br />
earth elements <strong>and</strong> hydrography in the Sulu Sea. Geochim. Cosmochim. Acta 63,<br />
2171–2181.<br />
Nozaki Y., Lerche D., Alibo D.S., <strong>and</strong> Tsutsumi M. (2000a) Dissolved indium <strong>and</strong> rare<br />
earth elements in three Japanese rivers <strong>and</strong> Tokyo Bay: Evidence for<br />
anthropogenic Gd <strong>and</strong> In. Geochim. Cosmochim. Acta, 64, 3975–3982.<br />
Nozaki Y., Lerche D., Alibo D.S., <strong>and</strong> Snidvongs A. (2000b) The estuarine<br />
geochemistry <strong>of</strong> rare earth elements <strong>and</strong> indium in the Chao Phraya River,<br />
Thail<strong>and</strong>. Geochim. Cosmochim. Acta, 64, 3983–3994.<br />
Nozaki, Y., Alibo, D.S., 2003. Importance <strong>of</strong> vertical geochemical processes in<br />
controlling the oceanic pr<strong>of</strong>iles <strong>of</strong> dissolved rare earth elements in the<br />
northeastern Indian Ocean. Earth Planet. Sci. Lett., 205, 155-172.<br />
Patchett P.J. <strong>and</strong> Samson S.D. (2003) Ages <strong>and</strong> growth <strong>of</strong> the continental crust from<br />
radiogenic isotopes, pp. 321-348. In The Crust (ed. R.L. Rudnick) Vol. 3 Treatise<br />
on Geochemistry (eds. H.D. Holl<strong>and</strong> <strong>and</strong> K.K. Turekian), Elsevier-Pergamon,<br />
Oxford.<br />
Piasecki W. And Sverjensky D.A. (2008) Speciation <strong>of</strong> adsorbed yttrium <strong>and</strong> rare earth<br />
elements on oxide surfaces. Geochim. Cosmochim. Acta, 72, 3964–3979.<br />
35
Piepgras D.J. <strong>and</strong> Wasserburg G.J. (1980) Neodymium isotopic variations in seawater.<br />
Earth Planet. Sci. Lett., 50, 128-138.<br />
Poujol M., Robb L.J., Anhaeusser C.R., <strong>and</strong> Gericke B. (2002) Geochronological<br />
constraints on the evolution <strong>of</strong> the Kaapvaal craton, South Africa. Econ. Geol.<br />
Res. Institute, Information Circular No. 360, Univ. Witwatersr<strong>and</strong>, pp. 37.<br />
Shannon R.D. (1976) Revised effective ionic radii <strong>and</strong> systematic studies <strong>of</strong> interatomic<br />
distances in halides <strong>and</strong> chalcogenides. Acta Crystallograph., A32, 751-767.<br />
Salters V., <strong>and</strong> Stracke A. (2004) Composition <strong>of</strong> the depleted mantle. Geochemistry,<br />
Geophysics, Geosystems, 5, doi: 10.1029/2003GC000597.<br />
Sholkovitz E.R. (1993) The geochemistry <strong>of</strong> rare earth elements in the Amazon River<br />
Estuary. Geochim. Cosmochim. Acta, 57, 2181-2190.<br />
Sholkovitz E.R. (1995) The aquatic chemistry <strong>of</strong> rare earh elements in rivers <strong>and</strong><br />
estuaries. Aquatic Geochem. 1, 1-34.<br />
Sholkovitz E.R. (1996) A compilation <strong>of</strong> the rare earth element composition <strong>of</strong> rivers,<br />
estuares <strong>and</strong> the oceans. Woods Hole Oceanog. Inst. Tech. Rept., WHOI-96-13.<br />
pp. 78.<br />
Sholkovitz E.R. <strong>and</strong> Elderfeld H. (1988) The cycling <strong>of</strong> dissolved rare earth elements in<br />
Chesapeake Bay. Global Biogeochem. Cycles, 2, 157-176.<br />
Sholkovitz E.R. <strong>and</strong> Szymczak R. (2000) The estuarine chemistry <strong>of</strong> rare earth<br />
elements: comparison <strong>of</strong> the Amazon, Fly, Sepik <strong>and</strong> the Gulf <strong>of</strong> Papua systems.<br />
Earth Planet. Sci. Lett., 179, 299-309.<br />
Sholkovitz, E. R, D. J. Piepgras <strong>and</strong> S. B. <strong>Jacobs</strong>en (1989). The pore water chemistry <strong>of</strong><br />
rare earth elements in Buzzards Bay sediments. Geochim. Cosmochim. Acta, 53,<br />
2847-2856.<br />
36
Sholkovitz E.R., Shaw T.J., <strong>and</strong> Schneider D.L. (1992) The geochemistry <strong>of</strong> rare earth<br />
elements in the seasonally anoxic water column <strong>and</strong> pore waters <strong>of</strong> Chesapeake<br />
Bay. Geochim. Cosmochim. Acta, 56, 3389-3402.<br />
Sholkovitz E.R., L<strong>and</strong>ing W.M., <strong>and</strong> Lewis B.L. (1994) Ocean particle chemistry: The<br />
fractionation <strong>of</strong> rare earth elements between suspended particles <strong>and</strong> seawater.<br />
Geochim. Cosmochim. Acta, 58, 1567-1579.<br />
Simonson, B.M., 2003. Origin <strong>and</strong> evolution <strong>of</strong> large Precambrian iron formations.<br />
Geol. Soc. Amer. Spec. Paper 370, 231-243.<br />
Simonson, B.M., Hassler, S.W., 1996. Was the deposition <strong>of</strong> large Precambrian iron<br />
formations linked to major marine transgressions?. J. Geol. 104, 665-676.<br />
Stille P., Steinmann M., Pierret M.-C., Gauthier-Lafaye F., Chabaux F., Viville D.,<br />
Pourcelot L., Matera V., Aouad G., <strong>and</strong> Aubert D. (2006) The impact <strong>of</strong><br />
vegetation on REE fractionation in stream waters <strong>of</strong> a small forested catchment<br />
(the Strengbach case). Geochimica et Cosmochimica Acta 70, 3217–3230.<br />
Tachikawa, K., Je<strong>and</strong>el C., <strong>and</strong> Roy-Barman M. (1999) A new approach to the Nd<br />
residence time in the ocean: the role <strong>of</strong> atmospheric inputs. Earth Planet. Sci. Lett.,<br />
170, 433-446.<br />
Tachikawa, K., Athias, V., Je<strong>and</strong>el, C., 2003. Neodymium budget in the modern ocean<br />
<strong>and</strong> paleo-oceanographic implications. J. Geophys. Res. 108, 1-13.<br />
Tankard A. J., Jackson M. P. A., Eriksson K. A., Hobday D. K., Hunter D. R., <strong>and</strong><br />
Minter W. E. L. (1982) Crustal Evolution <strong>of</strong> Southern Africa. Springer Verlag,<br />
New York.<br />
Taylor S.R. <strong>and</strong> McLennan S.M. (1995) The geochemical evolution <strong>of</strong> the continental<br />
crust. Reviews <strong>of</strong> Geophysics, 33, 241-265.<br />
37
Tosiani T., Loubet M., Viers J., Valladon M., Tapia J., S. Marrero S., Yanes C.,<br />
Ramirez A., Dupre B. (2004) Major <strong>and</strong> trace elements in river-borne materials<br />
from the Cuyuni basin (southern Venezuela): evidence for organo-colloidal<br />
control on the dissolved load <strong>and</strong> element redistribution between the suspended<br />
<strong>and</strong> dissolved load. Chem. Geol., 211, 305–334.<br />
Tricca A., Stille P., Steinmann M.,Kiefel B., Samuel J., <strong>and</strong> Eikenberg J. (1999) Rare<br />
earth elements <strong>and</strong> Sr <strong>and</strong> Nd isotopic compositions <strong>of</strong> dissolved <strong>and</strong> suspended<br />
loads from small river systems in the Vosges mountains France/, the river Rhine<br />
<strong>and</strong> groundwater. Chemical Geology 160, 139–158.<br />
Veizer, J., Hoefs, J., Lowe, D.R., Thurston, P.C. 1989. Geochemistry <strong>of</strong> Precambrian<br />
carbonates: II. Archean greenstone belts <strong>and</strong> Archean sea water. Geochim.<br />
Cosmochim. Acta, 53, 859-871.<br />
Wilson A.H., <strong>and</strong> Carlson R.W. (1989) A Sm-Nd <strong>and</strong> Pb isotope study <strong>of</strong> Archaean<br />
greenstone belts in the southern Kaapvaal Craton, South Africa. Earth Planet. Sci.<br />
Lett., 96, 89-105.<br />
Zhang, J., Nozaki, Y., 1996. Rare earth elements <strong>and</strong> yttrium in seawater: ICP-MS<br />
determinations in the East Caroline, Coral Sea, <strong>and</strong> South Fiji basins <strong>of</strong> the<br />
western South Pacific Ocean. Geochim. Cosmochim. Acta 60, 4631-4644.<br />
38
Chapter 2. Trace element analyses in geological materials using low resolution<br />
inductively coupled plasma mass spectrometry (ICPMS)<br />
39
THIS PAGE INTENTIONALLY LEFT BLANK<br />
40
Brian W. Alex<strong>and</strong>er<br />
Trace element analyses in geological materials<br />
using low resolution inductively coupled plasma<br />
mass spectrometry (ICPMS)<br />
Technical Report No. 18<br />
August 2008<br />
<strong>School</strong> <strong>of</strong> <strong>Engineering</strong> <strong>and</strong> <strong>Science</strong>
Abstract<br />
The benefits <strong>of</strong> inductively coupled plasma mass spectrometry (ICPMS) for<br />
geochemical studies include excellent instrument sensitivity for determining a large<br />
number <strong>of</strong> elements (10 mg/kg). However, some<br />
elements suffer from interferences due to the major element content <strong>of</strong> particular rock<br />
types, <strong>and</strong> are unlikely to be quantifiable when present at low concentrations. This<br />
includes Co in carbonate rocks, <strong>and</strong> Nb in Fe-rich rock types. The concentrations <strong>of</strong><br />
many trace metals in some CRMs are poorly constrained, <strong>and</strong> early published datasets<br />
appear to frequently overestimate the abundances <strong>of</strong> these elements.<br />
ii
Table <strong>of</strong> Contents<br />
List <strong>of</strong> Figures .......................................................................................................................... iv<br />
List <strong>of</strong> Tables............................................................................................................................. v<br />
1. Introduction .......................................................................................................................... 1<br />
2. Sample preparation............................................................................................................... 2<br />
3. Sample decomposition ......................................................................................................... 2<br />
4. ICPMS methods ................................................................................................................... 5<br />
4.1. Analyzed elements <strong>and</strong> mass interferences .................................................................. 5<br />
4.2. External calibration ...................................................................................................... 9<br />
4.3. Data collection............................................................................................................ 13<br />
4.4. Internal st<strong>and</strong>ardization .............................................................................................. 14<br />
5. Analytical precision............................................................................................................ 16<br />
6. Analytical recovery ............................................................................................................ 24<br />
7. Analytical accuracy ............................................................................................................ 27<br />
7.1. Reference values <strong>and</strong> major element interferences .................................................... 27<br />
7.2. High Fe content rocks................................................................................................. 33<br />
7.3. Shales <strong>and</strong> clastic sediments....................................................................................... 39<br />
7.4. High Si content rocks (cherts) .................................................................................... 42<br />
7.5. Carbonate rocks.......................................................................................................... 44<br />
7.6. Marine ferromanganese nodules <strong>and</strong> crusts................................................................ 47<br />
7.7. Basalts ........................................................................................................................ 49<br />
7.8. Rare earth element ratios............................................................................................ 51<br />
8. Summary <strong>and</strong> conclusions.................................................................................................. 53<br />
References ............................................................................................................................... 56<br />
Appendix 1. Analytical data .................................................................................................... 59<br />
Appendix 2. Interferences due to major elements ................................................................... 64<br />
iii
List <strong>of</strong> Figures<br />
Figure 1. Sample decomposition methods………………………………………………….......3<br />
Figure 2. Preparation <strong>of</strong> multi-element st<strong>and</strong>ards…………..………………………………... 10<br />
Figure 3. Solutions used <strong>and</strong> analysis order for ICPMS measurements……............................12<br />
Figure 4. ICPMS instrument quantification limits……………...……………………………. 17<br />
Figure 5. Characterization <strong>of</strong> measures <strong>of</strong> analytical precision……………………………… 19<br />
Figure 6. Average sample precision for different reference materials……...….…….............. 20<br />
Figure 7. Average run precision for different reference materials…..………………….......... 21<br />
Figure 8. Comparison <strong>of</strong> different measures <strong>of</strong> precision…………………............................. 22<br />
Figure 9. Sample <strong>and</strong> method precision for HNO 3 carbonate decomposition……………….. 24<br />
Figure 10. Analytical recovery for HNO 3 carbonate <strong>and</strong> HF-HClO 4 decompositions….......... 26<br />
Figure 11. Mg <strong>and</strong> Ca interferences on 59 Co………………………….…………………….... 30<br />
Figure 12. Analytical results for iron-formation FeR-2…………………………………….... 34<br />
Figure 13. Analytical results for iron-formation FeR-4……………………………….……... 35<br />
Figure 14. REY data for second lot <strong>of</strong> iron-formation IF-G………………………………..... 37<br />
Figure 15. Analytical results for iron-formation IF-G………………………...……………....38<br />
Figure 16. Analytical results for shale SCo-1………………………………………………... 40<br />
Figure 17. Analytical results for shale SGR-1b……………………………….………............41<br />
Figure 18. Analytical results for chert JCh-1…………………………………….…………... 43<br />
Figure 19. Analytical results for dolomite JDo-1…….………………………………….........45<br />
Figure 20. Comparison <strong>of</strong> JDo-1 HNO 3 carbonate <strong>and</strong> HF-HClO 4 decompositions………….46<br />
Figure 21. Analytical results for Fe-Mn nodule JMn-1……………………………….............49<br />
Figure 22. Analytical results for basalt BHVO-2……………………………………..……....50<br />
iv
List <strong>of</strong> Tables<br />
Table 1. List <strong>of</strong> certified reference materials (CRMs)…………………………………….........4<br />
Table 2. Isotopes monitored in ICPMS measurements <strong>and</strong> IQLs….……………………….... ..6<br />
Table 3. Long term stability <strong>of</strong> HFSE in iron-formation IF-G….............................................. 13<br />
Table 4. Major element interferences for low mass elements (
1. Introduction<br />
The advent <strong>of</strong> inductively coupled plasma mass spectrometry (ICPMS) has<br />
proven enormously beneficial to geochemical studies. Since commercialization <strong>of</strong><br />
ICPMS technology in the early 1980’s, ICPMS has become the method <strong>of</strong> choice for<br />
geochemical analyses, <strong>and</strong> is ideal for trace metal determinations <strong>and</strong> isotopic studies<br />
<strong>of</strong> many elements. The widespread implementation <strong>of</strong> ICPMS technology results from<br />
its ability to quickly quantify a large number <strong>of</strong> elements (>40) in geological samples<br />
on a routine basis. Additionally, ICPMS <strong>of</strong>fers instrument sensitivity that permits<br />
quantification <strong>of</strong> ng/kg (parts-per-trillion, ppt) concentrations <strong>of</strong> many elements<br />
regardless <strong>of</strong> sample matrix, <strong>and</strong> as such is ideally suited for studies <strong>of</strong> geochemically<br />
diverse samples. This report describes the ICPMS analytical methods employed<br />
within the Geochemistry Lab at <strong>Jacobs</strong> <strong>University</strong> Bremen (JUB) <strong>and</strong> includes a<br />
critical discussion <strong>of</strong> the precision <strong>and</strong> accuracy <strong>of</strong> these methods.<br />
The Geochemistry Lab at JUB is relatively new, with construction <strong>of</strong> the lab<br />
completed in the Summer <strong>of</strong> 2004. In the Fall <strong>of</strong> 2004 a PerkinElmer DRC-e<br />
quadrupole inductively coupled plasma mass spectrometer (ICPMS) was installed,<br />
which is the principal analytical instrument for determinations <strong>of</strong> trace metal<br />
concentrations. This report is not intended to provide detailed information regarding<br />
basic ICPMS principles <strong>and</strong> technology, <strong>and</strong> the reader is referred to Thomas (2003)<br />
for a thorough review. When samples are decomposed into liquid form, ICP<br />
instruments are ideal for multi-element geochemical analyses, as they allow relatively<br />
fast sample introduction into mass spectrometers, <strong>and</strong> are readily automated for the<br />
processing <strong>of</strong> large numbers <strong>of</strong> samples.<br />
This report describes the steps utilized at JUB to obtain accurate trace metal<br />
concentration data in a variety <strong>of</strong> rock types, <strong>and</strong> proceeds from the processing <strong>of</strong><br />
h<strong>and</strong> samples to the decomposition <strong>of</strong> sample powders, <strong>and</strong> finally to the methods<br />
employed to ensure the accuracy <strong>of</strong> the reported data. Sample preparation techniques<br />
are not the primary focus <strong>of</strong> this paper, <strong>and</strong> are only briefly discussed. Rather,<br />
considering the relatively short existence <strong>of</strong> the Geochemistry Lab at JUB <strong>and</strong> the<br />
necessary development <strong>of</strong> accurate <strong>and</strong> routine geochemical analytical techniques,<br />
most <strong>of</strong> the discussion will focus on a critical assessment <strong>of</strong> the methods employed in<br />
ICPMS trace metal determinations in a variety <strong>of</strong> geologic materials, <strong>and</strong> the relative<br />
precision <strong>and</strong> accuracy <strong>of</strong> these measurements.<br />
1
2. Sample preparation<br />
The majority <strong>of</strong> this report focuses on geochemical analyses <strong>of</strong> rock reference<br />
st<strong>and</strong>ards issued by governmental or research organizations. However, a brief<br />
description is given <strong>of</strong> the sample preparation methods used for natural rock samples<br />
that are collected as part <strong>of</strong> research studies at JUB. Rock samples, whether from<br />
outcrops or drillcores, are first processed by h<strong>and</strong>-trimming to avoid weathering rinds,<br />
quartz/calcite veining, or obvious alteration features. A geologist’s hammer is used to<br />
produce roughly 50 g <strong>of</strong> small, 1-3 cm pieces that are crushed in a Fritsch Pulverisette<br />
1 jaw crusher to obtain chips approximately 0.5 cm in size. Harder rock types (e.g.,<br />
chert) are processed more finely to produce smaller sizes (≤0.3 cm chips). Sample<br />
chips are rinsed thoroughly with deionized water to remove dust <strong>and</strong> dried overnight<br />
in a laboratory oven at ~110 °C, after which approximately 20 g <strong>of</strong> chips are h<strong>and</strong>picked<br />
to produce homogeneous samples devoid <strong>of</strong> secondary mineral veins. The<br />
sample chips are then powdered in a Fritsch Pulverisette 6 planetary mill using agate<br />
balls in a sealed agate mortar.<br />
3. Sample decomposition<br />
The ICPMS used in the JUB Geochemistry Lab is configured to accept<br />
samples in liquid form only, though ICP instruments with laser-ablation attachments<br />
for the analyses <strong>of</strong> solid samples are increasingly common. One advantage <strong>of</strong><br />
converting samples to liquid form is that any effects due to heterogeneities within the<br />
sample (e.g., inclusions or minor mineral phases) are removed, <strong>and</strong> this approach is<br />
typically termed a ‘whole-rock’ analysis. The common method for rocks <strong>and</strong> minerals<br />
is to decompose the sample in strong mineral acids at elevated temperatures <strong>and</strong><br />
pressures. The decomposition procedures used at JUB, <strong>and</strong> in fact, many <strong>of</strong> the<br />
ICPMS techniques as well, have been adapted from methods developed by P. Dulski<br />
at the GeoForschungsZentrum (GFZ) in Potsdam, Germany (see Dulski, 1994;<br />
Dulski, 2001).<br />
Two sample decomposition methods are currently employed: the first is a<br />
high-temperature, high-pressure decomposition utilizing concentrated hydr<strong>of</strong>luoric<br />
(HF) <strong>and</strong> perchloric (HClO 4 ) acids that is suitable for the total dissolution <strong>of</strong> silicatebearing<br />
samples (e.g., iron-formations, basalts), while the second method is a lowtemperature<br />
nitric acid (HNO 3 ) decomposition used for dissolving carbonate samples<br />
such as limestones <strong>and</strong> dolomites. It is important to note that the HF-HClO 4<br />
2
Figure 1. Flow-chart diagram <strong>of</strong> the two sample decomposition methods employed within the<br />
Geochemistry Lab at JUB for silicate-bearing samples (HF-HClO 4 pressure decomposition) <strong>and</strong><br />
carbonate samples such as limestones <strong>and</strong> dolomites (HNO 3 carbonate decomposition). All reagents<br />
used are super-, ultra-pure grade <strong>and</strong> dilutions are performed with deionized water (18.2 MΩ). Modified<br />
after Dulski (2001).<br />
decomposition provides complete dissolution <strong>of</strong> the sample powder, primarily due to<br />
the ability <strong>of</strong> HF to dissolve silicate minerals, whereas the low-temperature HNO 3<br />
decomposition is considered to dissolve only the carbonate mineral fraction <strong>of</strong> the<br />
sample powder. The HNO 3 decomposition method, referred to as the ‘carbonate<br />
decomposition’, will not dissolve refractory organic carbon <strong>and</strong>/or primary silicate<br />
minerals. The carbonate decomposition will attack <strong>and</strong> leach secondary minerals such<br />
as clays, though the extent <strong>of</strong> this effect will vary as a function <strong>of</strong> sample<br />
composition. The two methods are schematically described in Figure 1, <strong>and</strong> most <strong>of</strong><br />
the data <strong>and</strong> discussion presented here will focus on the HF-HClO 4 decomposition<br />
method. Regardless <strong>of</strong> the method chosen, every sample decomposition contains at<br />
least one certified reference material (CRM) which is treated analytically as an<br />
unknown sample, <strong>and</strong> which is considered to provide the best estimate <strong>of</strong> the accuracy<br />
<strong>of</strong> the complete decomposition method <strong>and</strong> ICPMS analysis. The CRMs commonly<br />
used in the Geochemistry Lab are listed in Table 1 <strong>and</strong> represent a variety <strong>of</strong> rock<br />
types, <strong>and</strong> for any individual sample decomposition a CRM is chosen that most<br />
closely resembles the rock type <strong>of</strong> the samples.<br />
3
Table 1. Certified reference materials (CRM) used as quality assurance st<strong>and</strong>ards for routine sample<br />
decomposition <strong>and</strong> ICPMS measurements within the JUB Geochemistry Lab.<br />
number <strong>of</strong> sample<br />
CRM<br />
description decompositions<br />
issuing organization<br />
FeR-2 Al-rich iron-formation 3 CCRMP Canadian Certified Reference Materials<br />
Project<br />
FeR-4 iron-formation 2 555 Booth Street Ottawa, Ontario K1A 0G1 Canada<br />
IWG-GIT International Working Group - Groupe<br />
2 first lot International de Travail<br />
IF-G a Isua iron-formation<br />
Service d’Analyse des Roches et des Minéraux,<br />
4 second lot<br />
CPRG-CNRS,<br />
B.P. 20 V<strong>and</strong>oeuvre-lés-Nancy Cedex, France<br />
SCo-1 silty marine shale 4<br />
USGS United States Geological Survey<br />
SGR-1b carbon-rich shale 3 U.S. Geological Survey, Box 25046, MS 973<br />
BHVO-2 Hawaiian basalt 7<br />
Denver, CO 80225, USA<br />
JMn-1 manganese nodule 4<br />
JCh-1 chert 3<br />
JDo-1<br />
dolomite<br />
12<br />
(5 HF-HClO 4 ) b<br />
(7 HNO 3 ) b<br />
GSJ Geological Survey <strong>of</strong> Japan<br />
available from: Seishin Trading Co., 1-2-1 Sanshin<br />
Bulding, Sannomiya, Tyuo-ku, Kobe, 650-0021,<br />
Japan<br />
a original IF-G powder (first lot) no longer available, <strong>and</strong> newly processed IF-G powder (second lot) available 2006.<br />
b refers to decomposition method used.<br />
Sample decompositions were performed in 30 ml polytetrafluoroethylene<br />
(PTFE) vessels using a PicoTrace DAS acid digestion system (Bovenden, Germany).<br />
As the DAS system contains 16 PTFE vessels, routine sample decompositions consist<br />
<strong>of</strong> 14 samples, one CRM, <strong>and</strong> one method blank. Typically, these decompositions will<br />
convert 100 mg <strong>of</strong> sample powder into 50 g <strong>of</strong> a clear sample liquid, in a matrix <strong>of</strong><br />
either 0.5 molar (moles/l, M) HCl or 0.5 M HNO 3 . This dilution factor <strong>of</strong> 500<br />
produces a total dissolved solid (TDS) content <strong>of</strong> 0.2% in the sample liquid (assuming<br />
no loss <strong>of</strong> volatiles such SiF 4 or CO 2 ), which is generally considered the maximum<br />
TDS content practical for routine ICPMS measurements. The ideal dilution factor for<br />
a particular rock or mineral reflects the balance between two competing effects; the<br />
minimization <strong>of</strong> matrix effects in the sample by utilizing high dilution factors, versus<br />
the detection <strong>and</strong> quantification limits imposed by the method <strong>and</strong> instrumentation.<br />
Typical dilution factors employed within the JUB Geochemistry Lab for ICPMS<br />
measurements range from 250 for very pure, trace metal-poor limestones <strong>and</strong> cherts,<br />
to ≥5000 for trace metal-rich samples such as marine ferromanganese crusts <strong>and</strong><br />
nodules. The use <strong>of</strong> dilution factors
during decomposition (SiF 4 <strong>and</strong> CO 2 , respectively), thereby reducing the total<br />
dissolved solid content <strong>of</strong> the ICPMS sample solution.<br />
Cleaning <strong>of</strong> the DAS system between sample batches is accomplished by<br />
heating an acid mixture <strong>of</strong> 5 ml <strong>of</strong> 15 M HNO 3 <strong>and</strong> 1 ml 23 M HF in the sealed PTFE<br />
vessels for 12 hours at ~165 °C. Memory effects within the DAS system for the<br />
elements considered in this research have never been observed, <strong>and</strong> decomposition<br />
method blanks for these elements are essentially controlled by good laboratory<br />
practices <strong>and</strong> the purity <strong>of</strong> the reagents used in the decomposition.<br />
4. ICPMS methods<br />
4.1. Analyzed elements <strong>and</strong> mass interferences<br />
Currently, the concentrations <strong>of</strong> 32 elements are determined during ICPMS<br />
analyses conducted within the JUB Geochemistry Lab. Initially, <strong>and</strong> until June 2006,<br />
concentrations <strong>of</strong> only 24 <strong>of</strong> these 32 elements were determined during routine<br />
ICPMS analyses, <strong>and</strong> the current element list <strong>and</strong> the scanned isotopic masses are<br />
presented in Table 2. Wherever possible, multiple isotopes are monitored for any<br />
given element, as the well-known abundances <strong>of</strong> natural isotopes provide a method<br />
for checking the quality <strong>of</strong> the analysis. This is accomplished by normalizing the<br />
signal intensity in cps (counts-per-second) for an isotope by its abundance. If no<br />
isotopic fractionation has occurred in the sample, <strong>and</strong> the measured intensities are not<br />
affected by interferences (e.g., interferences from isotopes <strong>of</strong> other elements), then<br />
these normalized intensities should be equivalent for all isotopes <strong>of</strong> an element.<br />
For example, the rare earth element (REE) Yb is monitored at three isotopic<br />
masses, 171 Yb, 172 Yb, <strong>and</strong> 174 Yb, which have natural isotopic abundances <strong>of</strong> 14.3%,<br />
21.9%, <strong>and</strong> 31.8%, respectively. If the following holds true:<br />
⎛<br />
⎜<br />
⎝<br />
171<br />
Yb<br />
cps<br />
0.143<br />
⎞<br />
⎟<br />
⎠<br />
=<br />
⎛<br />
⎜<br />
⎝<br />
172<br />
Yb<br />
0.219<br />
cps<br />
⎞<br />
⎟<br />
⎠<br />
=<br />
⎛<br />
⎜<br />
⎝<br />
174<br />
Yb<br />
cps<br />
0.318<br />
⎞<br />
⎟<br />
⎠<br />
(1)<br />
then these isotopes <strong>of</strong> are suitable for quantifying the concentration <strong>of</strong> Yb in the<br />
analyzed solution. The approach described by Eq. 1 is not suitable if some process,<br />
whether natural or analytical, has fractionated the monitored isotopes for a given<br />
element. Of the 32 elements quantified by ICPMS measurement, only Pb may be<br />
fractionated in such a manner due to natural processes occurring within the sample.<br />
5
Table 2. Elements analyzed in routine ICPMS determinations, specific isotopes measured, interfering<br />
polyatomic species, <strong>and</strong> instrument quantification limits (IQLs) in HCl <strong>and</strong> HNO 3 acid matrices. See text for<br />
details. The eight elements in bold type were added to the routine ICPMS analysis in June 2006.<br />
average instrument quantification limit<br />
element<br />
1 Sc<br />
monitored<br />
masses<br />
45 Sc<br />
12 C 16 O 2 1 H + , 13 C 16 O 2<br />
+<br />
observed/corrected interferences<br />
in rock/mineral samples (mg/kg)<br />
0.5 M HCl matrix<br />
(n = 22)*<br />
0.5 M HNO 3 matrix<br />
(n=4)*<br />
0.758 0.233<br />
2 Ti<br />
3 Co<br />
4 Ni<br />
5 Rb<br />
6 Sr<br />
7 Y<br />
8 Zr<br />
9 Nb<br />
10 Mo<br />
47 Ti, 49 Ti<br />
59 Co<br />
60 Ni, 62 Ni<br />
85 Rb<br />
88 Sr<br />
89 Y<br />
90 Zr, 91 Zr<br />
93 Nb<br />
95 Mo, 97 Mo<br />
14 N 16 O 1 2 H + , 12 C 35 Cl + , 14 N 35 Cl + 1.94 0.284<br />
42 Ca 16 O 1 H + , 43 Ca 16 O + , 24 Mg 35 Cl + 0.009 0.020<br />
14 N 16 2 O + 2 , 44 Ca 16 O + , 25 Mg 35 Cl + , 25 Mg 37 Cl + , 27 Al 35 Cl + 0.077 0.168<br />
48 Ca 37 Cl + 0.019 0.009<br />
48 Ca 40 Ar + 0.047 0.280<br />
54 Fe 35 Cl + 0.003 0.004<br />
55 Mn 35 Cl + , 40 Ar 16 O 35 Cl + , 40 Ca 16 O 35 Cl + , 56 Fe 35 Cl + 0.018 0.012<br />
40 Ar 16 O 37 Cl + , 40 Ca 16 O 37 Cl + , 56 Fe 37 Cl + 0.014 0.015<br />
55 Mn 40 Ar + , 44 Ca 16 O 35 Cl + , 44 Ca 16 O 37 Cl + 0.069 0.076<br />
11 Cs<br />
12 Ba<br />
13 La<br />
14 Ce<br />
15 Pr<br />
16 Nd<br />
17 Sm<br />
133 Cs 0.004 0.004<br />
135 Ba, 137 Ba 0.017 0.055<br />
139 La 0.003 0.006<br />
140 Ce 0.003 0.005<br />
141 Pr 0.002 0.004<br />
145 Nd, 146 Nd 0.005 0.007<br />
147 Sm, 149 Sm 0.005 0.005<br />
18 Eu<br />
19 Gd<br />
20 Tb<br />
21 Dy<br />
22 Ho<br />
23 Er<br />
24 Tm<br />
25 Yb<br />
26 Lu<br />
27 Hf<br />
28 Ta<br />
29 W<br />
151 Eu, 153 Eu<br />
158 Gd, 160 Gd<br />
159 Tb<br />
161 Dy, 163 Dy<br />
165 Ho<br />
166 Er, 167 Er<br />
169 Tm<br />
171 Yb, 172 Yb, 174 Yb<br />
175 Lu<br />
178 Hf, 179 Hf<br />
181 Ta<br />
182 W, 183 W<br />
135 Ba 16 O, 134 Ba 17 (OH), 137 Ba 16 O, 136 Ba 17 (OH) 0.002 0.004<br />
142 Nd 16 O, 142 Ce 16 O, 141 Pr 17 (OH), 144 Nd 16 O, 144 Sm 16 O, 160 Dy,<br />
Pr 19 (OH), 143 Nd 17 (OH)<br />
0.003 0.005<br />
141 Pr 18 O, 143 Nd 16 O, 142 Nd 17 (OH), 142 Ce 17 (OH) 0.002 0.004<br />
145 Nd 16 O, 144 Nd 17 (OH), 144 Sm 17 (OH), 147 Sm 16 O, 146 Nd 17 (OH) 0.003 0.004<br />
149 Sm 16 O, 148 Nd 17 (OH), 148 Sm 17 (OH) 0.002 0.004<br />
150 Nd 16 O, 150 Sm 16 O, 149 Sm 17 (OH), 151 Eu 16 O, 150 Nd 17 (OH),<br />
Sm 17 (OH)<br />
0.002 0.006<br />
153 Eu 16 O, 152 Sm 17 (OH), 134 Ba 35 Cl 0.002 0.003<br />
155 Gd 16 O, 154 Gd 17 (OH), 154 Sm 17 (OH), 134,136 Ba 37,35 Cl, 135,137 Ba 37,35 Cl,<br />
Gd 16 O, 155 Gd 17 (OH), 158 Gd 16 O, 157 Gd 17 (OH), 137 Ba 37 Cl<br />
0.002 0.005<br />
158 Gd 17 (OH), 159 Tb 16 O, 138 Ba 37 Cl 0.002 0.004<br />
162 Dy 16 O, 161 Dy 17 (OH), 163 Dy 16 O, 162 Dy 17 (OH) 0.004 0.005<br />
165 Ho 16 O 0.008 0.011<br />
166 Er 16 O, 167 Er 16 O, 166 Er 17 (OH) 0.196 0.067<br />
30 Pb<br />
31 Th<br />
206 Pb, 207 Pb, 208 Pb 0.022 0.035<br />
232 Th 0.003 0.005<br />
32 U<br />
238 U 0.002 0.004<br />
* n equals the number <strong>of</strong> separate ICPMS runs which contained at least one Certified Reference Material (CRM), with each run consisting<br />
<strong>of</strong> a separate acid decomposition <strong>of</strong> a mixed batch <strong>of</strong> samples <strong>and</strong> CRMs, <strong>and</strong> from which the average instrument quantification limits (IQL)<br />
for solid samples were calculated. See text for details regarding the calculation <strong>of</strong> detection <strong>and</strong> quantification limits.<br />
6
Fractionation between the three isotopes <strong>of</strong> Pb monitored during ICPMS<br />
analyses, 206 Pb, 207 Pb, <strong>and</strong> 208 Pb, will occur due to radioactive decay <strong>of</strong> U <strong>and</strong> Th<br />
within the sample. The half-lives <strong>of</strong> the respective decay series, 238 U → 206 Pb, 235 U →<br />
207 Pb, <strong>and</strong> 232 Th → 208 Pb, range between ~700 million to ~14 billion years (Faure,<br />
1986). Therefore, changes in the natural modern abundances <strong>of</strong> the Pb isotopes ( 206 Pb<br />
= 24.1%, 207 Pb = 22.1%, 208 Pb = 52.4%), due to the generation <strong>of</strong> radiogenic Pb<br />
within the sample may be significant, particularly in the geologically old (>2.5 Ga)<br />
samples commonly analyzed at JUB. Therefore, only in the case <strong>of</strong> Pb is Eq. 1 not<br />
used to monitor agreement between the concentrations determined from individual<br />
isotopes for a given element. For the remaining elements with more than one atomic<br />
mass available for monitoring, the application <strong>of</strong> Eq. 1 also provides a method for<br />
identifying interferences that may be affecting isotopes <strong>of</strong> a particular element.<br />
The issue <strong>of</strong> ions with masses similar to the isotopes <strong>of</strong> interest producing<br />
undesirable interferences in ICPMS analyses is common. The elements in geological<br />
materials most suitable for routine ICPMS analyses have masses greater than 80<br />
atomic mass units (amu). This primarily results from the fact that lower mass<br />
elements (e.g., many transition metals) may suffer severe interferences from<br />
polyatomic species generated within the plasma during ionization <strong>of</strong> the sample.<br />
Many <strong>of</strong> these polyatomic interferences result from the use <strong>of</strong> argon as the plasma<br />
source gas in most ICPMS instruments, <strong>and</strong> the subsequent formation <strong>of</strong> Ar-oxides<br />
<strong>and</strong> Ar-hydroxides (ArO + <strong>and</strong> ArOH + ), though other interfering species may be<br />
significant due to the choice <strong>of</strong> acid used for decomposing <strong>and</strong> diluting samples (e.g.,<br />
chloride <strong>and</strong> nitrogen species).<br />
A common example <strong>of</strong> the difficulties such polyatomic interferences present<br />
is evident in determinations <strong>of</strong> Fe, in which the isotope <strong>of</strong> choice for analysis is the<br />
most abundant one, 56 Fe (91.72%). However, within the plasma large numbers <strong>of</strong><br />
40 Ar 16 O + ions are produced which are indistinguishable from 56 Fe in low-resolution<br />
ICPMS analyses. High-resolution ICPMS instruments are capable <strong>of</strong> distinguishing<br />
between the 40 Ar 16 O + <strong>and</strong> 56 Fe peaks, but can only achieve this peak resolution at the<br />
expense <strong>of</strong> ion-beam transmission efficiency <strong>and</strong> a consequent reduction in<br />
instrument sensitivity. As a result high-resolution ICPMS methods are generally not<br />
suitable for element determinations in the sub-ppb (parts-per-billion, μg/kg) range<br />
necessary for analyses <strong>of</strong> many trace metals in geological samples, unless intensive<br />
7
sample preparation methods are employed which isolate <strong>and</strong> concentrate the elements<br />
<strong>of</strong> interest from undesirable matrix elements.<br />
Table 2 contains interferences commonly observed in ICPMS analyses<br />
performed within the JUB Geochemistry Lab. The inability <strong>of</strong> the low-resolution<br />
DRC-e ICPMS to resolve interferences from the isotopes <strong>of</strong> interest does not<br />
significantly inhibit the accurate determination <strong>of</strong> the 32 elements routinely analyzed,<br />
as correcting for these interferences is possible through careful characterization <strong>and</strong><br />
quantification <strong>of</strong> the interfering polyatomic species. The majority <strong>of</strong> these corrected<br />
interferences are for the REE (Table 2), as the REE can form significant amounts <strong>of</strong><br />
oxide <strong>and</strong> hydroxide species during ionization. For example, at typical instrument<br />
settings as much as 3% <strong>of</strong> the Ce present in the sample solution will be ionized to<br />
140 Ce 16 O + , <strong>and</strong> the<br />
140 Ce 16 O + ion will consequently contribute to any Gd<br />
determinations that utilize the<br />
156 Gd isotope. The isotopes monitored for<br />
quantification <strong>of</strong> the REE are therefore carefully selected to minimize these effects<br />
<strong>and</strong> maximize analytical accuracy, though as seen in Table 2 a significant number <strong>of</strong><br />
corrections are necessary. While implementing such corrections is not trivial,<br />
uncorrected values may result in reported concentrations for REE elements that are<br />
erroneously high by as much as several percent, as illustrated by the example with<br />
CeO + .<br />
Corrections for polyatomic interferences are performed mathematically,<br />
typically <strong>of</strong>fline using commercially available spreadsheet s<strong>of</strong>tware such as Micros<strong>of</strong>t<br />
Excel (the approach followed at JUB). Interferences are identified <strong>and</strong> quantified by<br />
analyzing solutions that contain relatively high concentrations (50-1000 μg/kg) <strong>of</strong> a<br />
single element, <strong>and</strong> surveying the range <strong>of</strong> masses that might be affected by<br />
interferences produced by oxide, hydroxide, chloride, or nitrogen species <strong>of</strong> that<br />
particular element. For the REE, the most significant interferences are generated by<br />
REE-oxides or -hydroxides, <strong>and</strong> therefore are found at 16 <strong>and</strong> 17 amu above the<br />
interfering element (i.e., REE 16 O + <strong>and</strong> REE 16 O 1 H + ). The measured intensities <strong>of</strong> the<br />
interfering element, <strong>and</strong> the various interferences it produces, allow calculation <strong>of</strong><br />
ratios expressing the relative amount <strong>of</strong> interfering species generated as a function <strong>of</strong><br />
the concentration <strong>of</strong> the interfering element. For example, if as mentioned above, 3%<br />
<strong>of</strong> 140 Ce in the sample is ionized to 140 Ce 16 O + (i.e., 140 Ce 16 O + / 140 Ce equals 0.03), <strong>and</strong> if<br />
the Ce concentration in the sample corresponds to a measured intensity <strong>of</strong> 100,000<br />
8
cps, then 3000 cps must be subtracted from any signal intensity measured at mass 156<br />
(e.g., 156 Gd or 156 Dy). For a detailed treatment <strong>of</strong> possible REE interfering species <strong>and</strong><br />
examples <strong>of</strong> calculations <strong>of</strong> the appropriate correction ratios the reader is referred to<br />
Dulski (1994). Considering the number <strong>of</strong> isotopically different REE-oxyhydroxide<br />
species that may be generated (Table 2), such corrections may seem unwieldy, but<br />
determination <strong>of</strong> these correction ratios is not necessary with each individual ICPMS<br />
analysis. The approach adopted here, <strong>and</strong> following the methods <strong>of</strong> Dulski (1994),<br />
requires periodic determination <strong>of</strong> these correction ratios at very specific ICPMS<br />
instrument settings which consistently produce similar REEO(H) + /REE + values, <strong>and</strong><br />
then ensuring that sample analyses are always performed at identical ICPMS<br />
instrument settings. For the analyses performed within the JUB Geochemistry Lab<br />
this entails tuning <strong>and</strong> optimization <strong>of</strong> the ICPMS so that CeO + /Ce + does not deviate<br />
from 0.029 ±0.001.<br />
4.2. External calibration<br />
The determination <strong>of</strong> trace metal concentrations in sample solutions is<br />
performed using laboratory prepared external calibration st<strong>and</strong>ards. This procedure<br />
consists <strong>of</strong> using commercially available 1000 mg/l single element st<strong>and</strong>ards<br />
(Inorganic Ventures, USA) that are combined to form 1 mg/kg (parts-per-million,<br />
ppm) multi-element stock solutions. The 1 mg/kg stock solutions are prepared in acid<br />
matrices that are similar to those present in the 1000 mg/l single element st<strong>and</strong>ards,<br />
which are typically ~0.5 M HNO 3 (2.2% v/v). These multi-element stock solutions are<br />
further diluted immediately prior to an individual ICPMS analysis to produce 10 <strong>and</strong><br />
20 μg/kg calibration st<strong>and</strong>ards, which are analyzed in conjunction with the sample<br />
solutions. Preparation <strong>of</strong> these external calibration st<strong>and</strong>ards is presented<br />
schematically in Figure 2.<br />
Ideally, the determination <strong>of</strong> elemental concentrations in sample solutions<br />
would proceed using the highly accurate method <strong>of</strong> st<strong>and</strong>ard addition, in which the<br />
sample solutions are split into a minimum <strong>of</strong> three aliquots, two <strong>of</strong> which would be<br />
spiked with the elements <strong>of</strong> interest at different concentrations. By comparing the<br />
ICPMS response for these three solutions it is possible to calculate the concentrations<br />
<strong>of</strong> the spiked elements in the non-spiked sample solution. The greatest advantage <strong>of</strong><br />
the st<strong>and</strong>ard addition method is that it minimizes effects caused by matrix differences<br />
9
Figure 2. Diagram showing preparation <strong>of</strong> multi-element st<strong>and</strong>ards used for ICPMS external calibration<br />
<strong>and</strong> internal st<strong>and</strong>ardization. Note that immediately prior to an ICPMS analysis, all solutions receive 10<br />
μg/kg <strong>of</strong> the Ru, Re, Bi internal st<strong>and</strong>ard (see text for details). Trace amounts <strong>of</strong> HF (0.05 M) are used<br />
to stabilize high field strength elements (Ti, Zr, Nb, Ta, etc.).<br />
between samples <strong>and</strong> calibration st<strong>and</strong>ards, as the st<strong>and</strong>ards used for calibration are<br />
incorporated directly into aliquots <strong>of</strong> the sample solution itself. The greatest drawback<br />
to the st<strong>and</strong>ard addition method is that it does not lend itself to multi-element<br />
analyses, as it is most accurate when the spike concentrations for an individual<br />
element are similar to the unknown concentration in the sample. For geological<br />
samples, in which concentrations <strong>of</strong> trace metals routinely range over three orders <strong>of</strong><br />
magnitude, this would necessitate the creation <strong>of</strong> calibration st<strong>and</strong>ards containing 32<br />
potentially different, customized element concentrations, which assumes some preexisting<br />
knowledge <strong>of</strong> these elemental concentrations in the unknown sample.<br />
Combined with the fact that every sample would require at least triple the ICPMS<br />
analysis time, the st<strong>and</strong>ard addition calibration method is generally not recommended<br />
for routine, multi-element ICPMS analyses <strong>of</strong> geological samples.<br />
However, the external st<strong>and</strong>ard calibration method suffers from the<br />
aforementioned ‘matrix effects’. These matrix effects are particularly troublesome for<br />
analyses <strong>of</strong> geological materials which contain significant amounts <strong>of</strong> dissolved solids<br />
at typical ICPMS dilution factors. For example, an iron-formation with 90% Fe 2 O 3<br />
that is diluted 1000x would produce a solution containing more than 600 ppm Fe, <strong>and</strong><br />
this is generally the maximum dilution suitable for pure iron-formation samples. High<br />
dissolved metal concentrations tend to suppress ICPMS instrument sensitivity in a<br />
10
variety <strong>of</strong> ways. These range from reductions in ionization efficiency within the<br />
plasma to the deposition <strong>of</strong> ion-beam inhibiting sample material (e.g., salts or metaloxides)<br />
in the interface region between the plasma source <strong>and</strong> mass spectrometer.<br />
Therefore, there are two sample dependent matrix effects which need to be addressed<br />
for routine geochemical analyses, one <strong>of</strong> which may suppress analyte intensities<br />
within an individual sample (e.g., ionization efficiency), <strong>and</strong> a second effect which<br />
may induce a time-dependent signal drift due to high TDS content in the samples.<br />
Correcting for these sample matrix effects is achieved by using internal<br />
st<strong>and</strong>ardization, in which all ICPMS solutions (blanks, external calibration st<strong>and</strong>ards,<br />
CRMs, <strong>and</strong> samples) are spiked with an equivalent amount <strong>of</strong> an internal st<strong>and</strong>ard.<br />
The applicability <strong>of</strong> this method is critically dependent upon the condition that the<br />
chosen internal st<strong>and</strong>ard must not be present in appreciable amounts in any <strong>of</strong> the<br />
ICPMS solutions (blanks, st<strong>and</strong>ards, or samples). For this purpose three elements are<br />
used for internal st<strong>and</strong>ardization following the methods <strong>of</strong> Doherty (1989); Ru, Re,<br />
<strong>and</strong> Bi. These three elements are ideal as they have low blank values for the reagents<br />
used in the decomposition <strong>and</strong> ICPMS methods, are present at very low<br />
concentrations in the majority <strong>of</strong> geological materials, produce no significant<br />
interference effects for the elements <strong>of</strong> interest, <strong>and</strong> are themselves free <strong>of</strong> significant<br />
interferences for their monitored isotopes ( 101 Ru, 187 Re, <strong>and</strong> 209 Bi).<br />
The default ICPMS analysis consists <strong>of</strong> running samples in small batches<br />
bracketed by calibration st<strong>and</strong>ards <strong>and</strong> blanks, <strong>and</strong> is described in Figure 3. A typical<br />
ICPMS analysis consists <strong>of</strong> four to six <strong>of</strong> these small sample batches run<br />
consecutively in a single day, <strong>and</strong> is called an ICPMS run. Prior to analysis,<br />
appropriate sample dilution factors are determined based upon the sample type, <strong>and</strong><br />
samples are diluted either in 0.5 M HCl for HF-HClO 4 pressure decompositions, or in<br />
0.5 M HNO 3 for carbonate decompositions. Acid <strong>and</strong> method blanks, calibration<br />
st<strong>and</strong>ards, as well as CRMs are diluted in the same acid matrix as the samples, <strong>and</strong> all<br />
solutions are spiked with 10 μg/kg <strong>of</strong> the Ru, Re, Bi internal st<strong>and</strong>ard. Matching the<br />
acid matrix <strong>of</strong> the samples <strong>and</strong> calibration st<strong>and</strong>ards is crucial, as calibration<br />
st<strong>and</strong>ards prepared in HNO 3 typically display ~10% greater instrument response than<br />
calibration st<strong>and</strong>ards prepared in HCl. While HNO 3 is more commonly used for<br />
preparation <strong>of</strong> ICPMS solutions as a result <strong>of</strong> this improved instrument response (as<br />
well as fewer interfering polyatomic species at low masses), it does not produce<br />
11
Figure 3. Diagram depicting the order in which various blanks, calibrations st<strong>and</strong>ards, <strong>and</strong> samples<br />
are analyzed for a single ICPMS sample batch. A typical ICPMS analysis consists <strong>of</strong> four or more<br />
sample batches run consecutively. The acid blank refers to the 0.5 M HCl or HNO 3 acid matrix used<br />
for diluting all ICPMS solutions for a given analysis. Note that all solutions are spiked with 10 μg/kg<br />
<strong>of</strong> the internal st<strong>and</strong>ard, <strong>and</strong> that Ru, Re, <strong>and</strong> Bi in the second 10 μg/kg 32 element calibration<br />
st<strong>and</strong>ard within a batch are used for application <strong>of</strong> internal st<strong>and</strong>ard corrections for that batch. The<br />
certified reference material is a commercially available geost<strong>and</strong>ard that is included within every<br />
sample decomposition <strong>and</strong> run repeatedly with every sample batch. See text for details.<br />
stable solutions with respect to high field strength elements (HFSE) such as Zr, Nb,<br />
Hf, <strong>and</strong> Ta (Münker, 1998), <strong>and</strong> therefore these elements are stabilized using trace<br />
amounts <strong>of</strong> HF in the stock 1 mg/kg calibration solutions. This stability issue for<br />
HFSE in dilute HNO 3 (0.5 M) means that samples processed using the carbonate<br />
decomposition method should be analyzed by ICPMS within 24 hours following<br />
completion <strong>of</strong> the sample decomposition. For samples processed using the HF-HClO 4<br />
pressure decomposition <strong>and</strong> diluted in 0.5 M HCl, samples are also analyzed<br />
immediately, though this acid matrix appears sufficient for stabilizing HFSE on<br />
12
Table 3. Change in HFSE concentration in IF-G<br />
iron-formation stored in 0.5 M HCl over a period<br />
<strong>of</strong> ~11 months.<br />
mg/kg<br />
IF-G<br />
27-Oct-06*<br />
IF-G<br />
16-Sep-07<br />
% difference<br />
2007/2006<br />
Ti 23.9 23.5 -1.88<br />
Zr 0.903 0.865 -4.24<br />
Nb 0.914 0.876 -4.12<br />
Mo 0.562 0.604 7.50<br />
Hf 0.0246 0.0227 -7.57<br />
Ta 0.168 0.157 -6.71<br />
W 251 248 -1.35<br />
* date <strong>of</strong> ICPMS analysis.<br />
longer time scales, as measured HFSE values in the IF-G geost<strong>and</strong>ard change by less<br />
than 8% over a period <strong>of</strong> ~11 months (Table 3).<br />
4.3. Data collection<br />
Unknown samples that are analyzed within an ICPMS run (i.e., in a single<br />
day, as part <strong>of</strong> one or more sample batches that are analyzed consecutively), are<br />
generally measured once, whereas blanks, calibration st<strong>and</strong>ards, <strong>and</strong> CRMs are<br />
measured repeatedly. Prior to beginning any ICPMS measurements, the instrument is<br />
prepared by performing three consecutive analyses <strong>of</strong> the CRM solution to be used<br />
during that ICPMS run. This has the primary effect <strong>of</strong> conditioning the cones located<br />
in the interface region between the plasma <strong>and</strong> the mass spectrometer. These cones<br />
are instrumental in shaping the ion beam <strong>and</strong> must be periodically cleaned, <strong>and</strong> this<br />
pre-treatment significantly stabilizes signal intensities. A rinse solution is pumped<br />
through the sample introduction system before the measurement <strong>of</strong> any solution, be it<br />
a blank, st<strong>and</strong>ard, or sample. Nitric acid (0.5 M) is the typical rinsing agent, <strong>and</strong> a six<br />
minute rinsing step has been found effective for minimizing memory effects. An<br />
ICPMS measurement <strong>of</strong> any single solution consists <strong>of</strong> 60 individual scans through<br />
the mass range 45 Sc to 238 U using peak hopping mode <strong>and</strong> a peak dwell time <strong>of</strong> 50 ms.<br />
For any individual element, these 60 scans are then averaged to produce the raw<br />
ICPMS data in cps. Scanning time for a single solution is approximately 3.5 minutes,<br />
<strong>and</strong> when rinsing time, the time required for the autosampler to move to a different<br />
13
solution, etc., are considered, the total amount <strong>of</strong> time necessary to measure any<br />
single solution is ~10 minutes.<br />
4.4. Internal st<strong>and</strong>ardization<br />
After an ICPMS analysis, the raw data in cps are transferred to an Excel<br />
spreadsheet for the application <strong>of</strong> three separate mathematical corrections. These<br />
corrections are applied to all analytical solutions (blanks, calibration st<strong>and</strong>ards,<br />
samples) in the following order: 1) internal st<strong>and</strong>ard corrections; 2) corrections for<br />
interfering polyatomic species; <strong>and</strong> 3) blank corrections. The internal st<strong>and</strong>ard (IS)<br />
correction using 101 Ru, 187 Re, <strong>and</strong> 209 Bi follows the method <strong>of</strong> Doherty (1989). The<br />
general application <strong>of</strong> the internal st<strong>and</strong>ard correction method is illustrated by the<br />
following equation, where Ru is used to correct the signal intensity <strong>of</strong> Sr:<br />
I<br />
⎛ ⎞<br />
⎜ I Ru,<br />
std<br />
= ∗ ⎟<br />
, sample I Sr,<br />
sample<br />
(2)<br />
⎜ ⎟<br />
⎝ I Ru,<br />
sample ⎠<br />
corr<br />
Sr<br />
where I is raw intensities in cps, I Ru,std is the measured intensity for the second 10<br />
μg/kg calibration st<strong>and</strong>ard within an ICPMS sample batch (see Fig. 3), <strong>and</strong> I corr is the<br />
corrected intensity. The term in parentheses in (2) is the IS correction factor. This<br />
approach assumes that any matrix or instrument drift effects that suppress or enhance<br />
the signal intensity for Sr proportionally affect the Ru signal intensity, <strong>and</strong> internal<br />
st<strong>and</strong>ard corrections are most effective when the element <strong>of</strong> interest <strong>and</strong> the IS are <strong>of</strong><br />
similar mass (Thompson <strong>and</strong> Houk, 1987). Therefore, 101 Ru is used as an internal<br />
st<strong>and</strong>ard for all elements with atomic masses lower than 101 (Sc, Ti, Co, Ni, Rb, Sr,<br />
Y, Zr, Nb, <strong>and</strong> Mo), <strong>and</strong> 209 Bi is used for elements with masses greater than 209 (Th<br />
<strong>and</strong> U). For analytes with masses between Ru <strong>and</strong> Re, the IS correction factor term in<br />
(2) is exp<strong>and</strong>ed to a mass-dependent linear function between 101 Ru <strong>and</strong> 187 Re:<br />
IS correction factor =<br />
⎛<br />
⎜ I<br />
⎜<br />
⎝ I<br />
Re, std<br />
Re, sample<br />
Ru, sample<br />
( M analyte − M Ru ) I Ru, std<br />
+<br />
( M Re − M Ru ) I Ru, sample<br />
Ru, std<br />
− * (3)<br />
I<br />
I<br />
⎞<br />
⎟<br />
⎟<br />
⎠<br />
where M is the atomic mass number for Ru, Re, <strong>and</strong> the analyte <strong>of</strong> interest. This<br />
treatment provides an IS correction factor that varies smoothly as a function <strong>of</strong> mass<br />
between 101 Ru <strong>and</strong> 187 Re. Internal st<strong>and</strong>ard correction factors typically range from<br />
14
0.90, corresponding to an anomalous increase in measured analyte intensities in the<br />
sample, to 1.10, indicating an anomalous decrease in measured analyte intensities.<br />
Experience has shown that analytical accuracy is <strong>of</strong>ten compromised when IS<br />
correction factors deviate from unity by more than ~15% (i.e., 1.15), <strong>and</strong> it<br />
may be necessary to re-analyze the samples using a different dilution factor to<br />
minimize matrix <strong>and</strong>/or drift effects.<br />
Following internal st<strong>and</strong>ard corrections, interference corrections for<br />
polyatomic species are applied to the sample data. These interference corrections are<br />
applied only to elements with atomic masses between 151 (Eu) <strong>and</strong> 183 (W). A<br />
detailed treatment <strong>of</strong> the determination <strong>and</strong> application <strong>of</strong> these interference<br />
corrections is beyond the scope <strong>of</strong> this discussion, <strong>and</strong> the reader is referred to Dulski<br />
(1994). The last correction applied to the sample data before calculating elemental<br />
concentrations involves subtracting the internal st<strong>and</strong>ard <strong>and</strong> interference corrected<br />
blank contribution, which would be due to the reagents used in preparing the samples,<br />
<strong>and</strong>/or from any laboratory procedures.<br />
4.5. Analytical blanks <strong>and</strong> quantification limits<br />
Two blank contributions may be defined for the described ICPMS methods.<br />
The first is the acid blank value, which is used for calculating the instrument<br />
detection <strong>and</strong> quantification limits (IDLs <strong>and</strong> IQLs). As all ICPMS solutions are<br />
analyzed in a 0.5 M acid matrix (either HCl or HNO 3 ), the acid blank is simply the<br />
purest 0.5 M acid solution that is capable <strong>of</strong> being produced within the JUB<br />
Geochemistry Lab. Analyses <strong>of</strong> the acid blank allows the calculation <strong>of</strong> IDLs <strong>and</strong><br />
IQLs for any given element, which represent the best achievable (ideal) ICPMS<br />
sensitivity using the available laboratory reagents. The IDL <strong>and</strong> IQL equal 3x <strong>and</strong> 10x<br />
the st<strong>and</strong>ard deviation <strong>of</strong> the acid blank, respectively (MacDougall <strong>and</strong> Crummett,<br />
1980), <strong>and</strong> are best characterized through long term, repeated analyses <strong>of</strong> the acid<br />
blank, as daily fluctuations in instrument sensitivity are common. It is important to<br />
note that IDLs <strong>and</strong> IQLs calculated from the acid blank are not equivalent to detection<br />
<strong>and</strong> quantification limits in a solid sample, that is, in a rock or a mineral. As all solid<br />
samples are highly diluted before ICPMS analyses, IDLs <strong>and</strong> IQLs determined for<br />
solid samples equal the acid blank multiplied by the appropriate dilution factor. Table<br />
2 lists long term average IQLs in solid samples for 0.5 M HCl <strong>and</strong> 0.5 M HNO 3 acid<br />
matrices.<br />
15
The second blank contribution to ICPMS analyses is the method blank. The<br />
method blank is a solution that has been processed in exactly the same manner as a<br />
sample during the decomposition method, <strong>and</strong> represents the contamination added to<br />
a sample as a result <strong>of</strong> the decomposition method <strong>and</strong> subsequent dilution in acid.<br />
While significant amounts <strong>of</strong> concentrated acid are used in the decomposition<br />
methods, these acids (in combination with contamination from labware) contribute<br />
insignificant amounts <strong>of</strong> trace metals to rocks typically analyzed, even for trace<br />
metal-poor samples like iron-formations <strong>and</strong> dolomites (Fig. 4). Method blank values<br />
are determined for every ICPMS analysis to monitor possible contamination sources,<br />
but as a result <strong>of</strong> the above observations, ICPMS data are blank corrected by<br />
subtracting the acid blank intensities. Only in instances where method blank values<br />
are significantly higher than those observed in Fig. 4 are blank corrections performed<br />
using the method blank intensities. The reasoning for this approach is as follows;<br />
since the method blanks are typically below the IQLs determined from the acid<br />
blanks, the method blank is not quantifiable. Only when method blanks exceed the<br />
acid blank IQL are they used for the blank correction. It should be noted that with<br />
regard to the carbonate decomposition method, many refractory elements (e.g., Ti,<br />
Nb, Ta) are not suitable for quantification, as they are expected to be primarily hosted<br />
in silicate-bearing phases that are resistant to dissolution with nitric acid, <strong>and</strong> in<br />
particular the poor IQLs for Sc mean that it is generally not quantifiable in trace<br />
metal-poor samples regardless <strong>of</strong> the decomposition method used (Fig. 4).<br />
5. Analytical precision<br />
For the analyses conducted within the Geochemistry Lab at JUB, analytical<br />
precision may be defined in different ways. The highest degree <strong>of</strong> precision is<br />
expected when considering only the 60 mass scans performed during ICPMS analysis<br />
<strong>of</strong> a single sample solution. If this sample solution is periodically re-measured with<br />
time during an ICPMS run, i.e., by occasionally returning the autosampler probe to<br />
the vial containing the solution, then analytical precision would be expected to<br />
decrease. Further decreases in analytical precision should occur if one considers<br />
repeated analyses <strong>of</strong> the same sample solution on different days, or repeated acid<br />
decompositions <strong>of</strong> the same sample powder, with the decreasing precision reflecting<br />
16
10 2 JDo-1 reference values<br />
10 1<br />
IF-G reference values<br />
IQL HCl (n=21)<br />
method blank HCl (n=19)<br />
10 0<br />
10 -1<br />
mg/kg<br />
10 -2<br />
10 -3<br />
10 -4<br />
10 -5<br />
10 -6<br />
10 -7<br />
10 2 Sc Ti Co Ni Rb Sr Y Zr Nb Mo Cs Ba La Ce Pr Nd Sm Eu Gd Tb Dy Ho Er Tm Yb Lu Hf Ta W Pb Th U<br />
10 1<br />
0.5 M HCl<br />
IQL HNO3 (n=3)<br />
method blank HNO3 (n=3)<br />
10 0<br />
10 -1<br />
mg/kg<br />
10 -2<br />
10 -3<br />
10 -4<br />
10 -5<br />
10 -6<br />
10 -7<br />
0.5 M HNO3<br />
Sc Ti Co Ni Rb Sr Y Zr Nb Mo Cs Ba La Ce Pr Nd Sm Eu Gd Tb Dy Ho Er Tm Yb Lu Hf Ta W Pb Th U<br />
Figure 4. Long term average instrument quantification limits in mg/kg for solid samples (i.e.,<br />
accounting for dilution factors) as determined from acid blanks, <strong>and</strong> average method blanks. Data for<br />
acid blank IQLs are also presented in Table 2. The top figure corresponds to a 0.5 M HCl acid matrix,<br />
<strong>and</strong> includes average method blanks for the HF-HClO 4 pressure decomposition method as well as<br />
reference data for the iron-formation IF-G, which represents a rock type that contains very low trace<br />
metal contents typically dissolved using the HF-HClO 4 method. The bottom figure corresponds to a 0.5<br />
M HNO 3 acid matrix <strong>and</strong> the carbonate decomposition method, <strong>and</strong> includes reference data for the<br />
dolomite JDo-1, which represents a very low trace metal content rock typically dissolved using HNO 3 .<br />
Regardless <strong>of</strong> decomposition method, the method blanks are at least 1 order <strong>of</strong> magnitude lower than<br />
IQLs for all elements, <strong>and</strong> generally 3-6 orders <strong>of</strong> magnitude lower than concentrations observed in the<br />
very trace metal-poor rocks typically analyzed. Method blanks <strong>and</strong> IQLs are significantly lower in 0.5<br />
M HNO 3 . Note that some elements are likely to be unquantifiable in many trace metal-poor rock<br />
samples due to concentrations close to or below the IQL (e.g., Sc, Nb, Ta).<br />
17
error propagation during the complete sample decomposition <strong>and</strong> ICPMS analytical<br />
procedure. However, for many elements, in particular Y, Ba, the REE, <strong>and</strong> U, limits<br />
<strong>of</strong> precision appear to be controlled by the ICPMS measurement, <strong>and</strong> not by the<br />
sample decomposition method.<br />
For the discussion <strong>of</strong> precision, three types <strong>of</strong> analytical precision are defined:<br />
1) sample precision, calculated from the 60 mass scans <strong>of</strong> a single solution <strong>and</strong><br />
representing raw, uncorrected data; (2) run precision, calculated from element<br />
concentrations determined by repeated analyses <strong>of</strong> a single solution over the duration<br />
<strong>of</strong> an ICPMS run (typically 6-12 hours); <strong>and</strong> (3) method precision, calculated from<br />
element concentrations determined from repeated acid decompositions <strong>and</strong> ICPMS<br />
analyses <strong>of</strong> a sample powder. The three types <strong>of</strong> precision <strong>and</strong> their relationships are<br />
illustrated in Figure 5. Note that sample precision reflects uncorrected intensities<br />
(cps), whereas run <strong>and</strong> method precision incorporate internal st<strong>and</strong>ard, interference,<br />
<strong>and</strong> blank corrections. To provide the best estimate <strong>of</strong> precision for analyses <strong>of</strong> rocks<br />
<strong>and</strong> minerals, these three types <strong>of</strong> precision are calculated using data obtained for the<br />
certified reference materials listed in Table 1. The discussion <strong>of</strong> sample, run, <strong>and</strong><br />
method precision is limited to those analyses where the CRM was run repeatedly as<br />
part <strong>of</strong> every sample batch (see Fig. 3). Data are discussed as percent relative st<strong>and</strong>ard<br />
deviation (%RSD), though it must be stressed that st<strong>and</strong>ard deviations themselves<br />
usually vary significantly (±50% relative), so that an average RSD <strong>of</strong> 2% is expected<br />
to typically range as low as 1% <strong>and</strong> as high as 3%.<br />
Sample precision in ICPMS determinations for common rock types is<br />
presented in Figure 6, along with the average sample precision for the 10 μg/kg<br />
calibration st<strong>and</strong>ard. As the 10 μg/kg calibration st<strong>and</strong>ard represents a solution free <strong>of</strong><br />
the matrix effects typical <strong>of</strong> dissolved rock samples, it is expected to display the best<br />
RSD values, which are ~2% for almost all monitored isotopes. However, the 10 μg/kg<br />
calibration st<strong>and</strong>ard sample precision is indistinguishable from the sample precision<br />
determined for dissolved rock solutions, particularly for rock types that are not tracemetal<br />
poor, such as basalt <strong>and</strong> shale. Only for rock types that are very low in trace<br />
metals (e.g., JDo-1 dolomite), are sample precisions worse than those observed in the<br />
10 μg/kg calibration st<strong>and</strong>ard, <strong>and</strong> only then for certain elements (e.g., Rb, Mo, Cs,<br />
Hf, Th, U). The run precision, where measurements <strong>of</strong> a CRM solution are<br />
periodically repeated over the course <strong>of</strong> an ICPMS run, is generally ~2%, similar to<br />
18
Figure 5. Diagram depicting various types <strong>of</strong> precision that may be defined for sample decompositions<br />
<strong>and</strong> ICPMS analyses. Note that sample precision reflects raw uncorrected intensities (cps), while run<br />
<strong>and</strong> method precision are calculated from corrected concentration data.<br />
that <strong>of</strong> the sample precision for many elements, <strong>and</strong> also varies as function <strong>of</strong> rock<br />
type (Fig. 7). Shales display slightly poorer run precision (3-5%), though this might<br />
reflect the limited number <strong>of</strong> analyses where these CRMs were measured repeatedly.<br />
For purposes <strong>of</strong> long term reproducibility, the method precision is the most suitable<br />
measure <strong>of</strong> the st<strong>and</strong>ard deviation expected for a complete sample decomposition <strong>and</strong><br />
ICPMS analyses, as it describes the variability expected in concentration data when a<br />
sample powder is dissolved numerous times over a period <strong>of</strong> months or years.<br />
Comparisons <strong>of</strong> the three types <strong>of</strong> precision discussed here (sample, run, <strong>and</strong> method)<br />
are presented in Figure 8 for CRMs <strong>of</strong> four different rock types. Of the 32 elements<br />
analyzed, the method precision RSD is better than 5% for 25 elements in BHVO-2<br />
(basalt), 31 elements in SGR-1b, 26 elements in FeR-2 (iron-formation), <strong>and</strong> 22<br />
elements in JDo-1 (dolomite). The method precision is also comparable to the sample<br />
<strong>and</strong> run precision, suggesting that any error associated with the HF-HClO 4 sample<br />
decomposition method (e.g., weighing <strong>and</strong>/or dilution errors) is small compared to the<br />
error inherent in the ICPMS analysis.<br />
19
sample precision (%RSD)<br />
20<br />
19<br />
18<br />
17<br />
16<br />
15<br />
14<br />
13<br />
12<br />
11<br />
10<br />
9<br />
8<br />
7<br />
6<br />
5<br />
4<br />
3<br />
2<br />
1<br />
0<br />
BHVO-2 (n=5)<br />
10 μg/kg calibration st<strong>and</strong>ard<br />
20<br />
19<br />
18<br />
17<br />
16<br />
15<br />
14<br />
13<br />
12<br />
11<br />
10<br />
9<br />
8<br />
7<br />
6<br />
5<br />
4<br />
3<br />
2<br />
1<br />
0<br />
Sc-45<br />
Ti-47<br />
Ti-49<br />
Co-59<br />
Ni-60<br />
Ni-62<br />
Rb-85<br />
Sr-88<br />
Y-89<br />
Zr-90<br />
Zr-91<br />
Nb-93<br />
Mo-95<br />
Mo-97<br />
Cs-133<br />
Ba-135<br />
Ba-137<br />
La-139<br />
Ce-140<br />
Pr-141<br />
Nd-145<br />
Nd-146<br />
Sm-147<br />
Sm-149<br />
Eu-151<br />
Eu-153<br />
Gd-158<br />
Gd-160<br />
Tb-159<br />
Dy-161<br />
Dy-163<br />
Ho-165<br />
Er-166<br />
Er-167<br />
Tm-169<br />
Yb-171<br />
Yb-172<br />
Yb-174<br />
Lu-175<br />
Hf-178<br />
Hf-179<br />
Ta-181<br />
W-182<br />
W-183<br />
Pb-206<br />
Pb-207<br />
Pb-208<br />
Th-232<br />
U-238<br />
Sc-45<br />
Ti-47<br />
Ti-49<br />
Co-59<br />
Ni-60<br />
Ni-62<br />
Rb-85<br />
Sr-88<br />
Y-89<br />
Zr-90<br />
Zr-91<br />
Nb-93<br />
Mo-95<br />
Mo-97<br />
Cs-133<br />
Ba-135<br />
Ba-137<br />
La-139<br />
Ce-140<br />
Pr-141<br />
Nd-145<br />
Nd-146<br />
Sm-147<br />
Sm-149<br />
Eu-151<br />
Eu-153<br />
Gd-158<br />
Gd-160<br />
Tb-159<br />
Dy-161<br />
Dy-163<br />
Ho-165<br />
Er-166<br />
Er-167<br />
Tm-169<br />
Yb-171<br />
Yb-172<br />
Yb-174<br />
Lu-175<br />
Hf-178<br />
Hf-179<br />
Ta-181<br />
W-182<br />
W-183<br />
Pb-206<br />
Pb-207<br />
Pb-208<br />
Th-232<br />
sample precision (%RSD)<br />
U-238<br />
20<br />
19<br />
18<br />
17<br />
16<br />
15<br />
14<br />
13<br />
12<br />
11<br />
10<br />
9<br />
8<br />
7<br />
6<br />
5<br />
4<br />
3<br />
2<br />
1<br />
0<br />
SGR-1b (n=2)<br />
10 μg/kg calibration st<strong>and</strong>ard<br />
sample precision (%RSD)<br />
FER-2 (n=3)<br />
10 μg/kg calibration st<strong>and</strong>ard<br />
20<br />
19<br />
18<br />
17<br />
16<br />
15<br />
14<br />
13<br />
12<br />
11<br />
10<br />
9<br />
8<br />
7<br />
6<br />
5<br />
4<br />
3<br />
2<br />
1<br />
0<br />
Sc-45<br />
Ti-47<br />
Ti-49<br />
Co-59<br />
Ni-60<br />
Ni-62<br />
Rb-85<br />
Sr-88<br />
Y-89<br />
Zr-90<br />
Zr-91<br />
Nb-93<br />
Mo-95<br />
Mo-97<br />
Cs-133<br />
Ba-135<br />
Ba-137<br />
La-139<br />
Ce-140<br />
Pr-141<br />
Nd-145<br />
Nd-146<br />
Sm-147<br />
Sm-149<br />
Eu-151<br />
Eu-153<br />
Gd-158<br />
Gd-160<br />
Tb-159<br />
Dy-161<br />
Dy-163<br />
Ho-165<br />
Er-166<br />
Er-167<br />
Tm-169<br />
Yb-171<br />
Yb-172<br />
Yb-174<br />
Lu-175<br />
Hf-178<br />
Hf-179<br />
Ta-181<br />
W-182<br />
W-183<br />
Pb-206<br />
Pb-207<br />
Pb-208<br />
Th-232<br />
U-238<br />
Sc-45<br />
Ti-47<br />
Ti-49<br />
Co-59<br />
Ni-60<br />
Ni-62<br />
Rb-85<br />
Sr-88<br />
Y-89<br />
Zr-90<br />
Zr-91<br />
Nb-93<br />
Mo-95<br />
Mo-97<br />
Cs-133<br />
Ba-135<br />
Ba-137<br />
La-139<br />
Ce-140<br />
Pr-141<br />
Nd-145<br />
Nd-146<br />
Sm-147<br />
Sm-149<br />
Eu-151<br />
Eu-153<br />
Gd-158<br />
Gd-160<br />
Tb-159<br />
Dy-161<br />
Dy-163<br />
Ho-165<br />
Er-166<br />
Er-167<br />
Tm-169<br />
Yb-171<br />
Yb-172<br />
Yb-174<br />
Lu-175<br />
Hf-178<br />
Hf-179<br />
Ta-181<br />
W-182<br />
W-183<br />
Pb-206<br />
Pb-207<br />
Pb-208<br />
Th-232<br />
U-238<br />
sample precision (%RSD)<br />
JDo-1 (n=3)<br />
10 μg/kg calibration st<strong>and</strong>ard 23.0%<br />
Figure 6. Average sample precision (see text for definition) for different rock types dissolved using the<br />
HF-HClO 4 decomposition, as calculated from the number (n) <strong>of</strong> decompositions where the CRM was<br />
measured repeatedly. The average 10 μg/kg calibration st<strong>and</strong>ard is presented for comparison (n=19).<br />
Precision is generally very similar between the rock solutions <strong>and</strong> the calibration st<strong>and</strong>ard, except for<br />
elements that are present at low concentrations in trace-metal poor rocks such as the JDo-1 dolomite<br />
(e.g., Rb, Mo, Cs, Hf, Th, U). This suggests that the st<strong>and</strong>ard deviation <strong>of</strong> ICPMS measurements is<br />
controlled by inherent instrument precision rather than by sample matrix effects.<br />
20
20<br />
18<br />
basalt<br />
BHVO-2 (n=5)<br />
run precision (%RSD)<br />
16<br />
14<br />
12<br />
10<br />
8<br />
6<br />
4<br />
2<br />
0<br />
20<br />
18<br />
Sc Ti Co Ni Rb Sr Y Zr Nb Mo Cs Ba La Ce Pr Nd Sm Eu Gd Tb Dy Ho Er Tm Yb Lu Hf Ta W Pb Th U<br />
shales<br />
SCo-1 (n=2)<br />
SGR-1b (n=2)<br />
run precision (%RSD)<br />
16<br />
14<br />
12<br />
10<br />
8<br />
6<br />
4<br />
2<br />
0<br />
20<br />
18<br />
Sc Ti Co Ni Rb Sr Y Zr Nb Mo Cs Ba La Ce Pr Nd Sm Eu Gd Tb Dy Ho Er Tm Yb Lu Hf Ta W Pb Th U<br />
iron-formations<br />
FER-2 (n=3)<br />
IF-G (n=3)<br />
run precision (%RSD)<br />
16<br />
14<br />
12<br />
10<br />
8<br />
6<br />
4<br />
2<br />
run precision (%RSD)<br />
0<br />
30<br />
28<br />
26<br />
24<br />
22<br />
20<br />
18<br />
16<br />
14<br />
12<br />
10<br />
8<br />
6<br />
4<br />
2<br />
0<br />
Sc Ti Co Ni Rb Sr Y Zr Nb Mo Cs Ba La Ce Pr Nd Sm Eu Gd Tb Dy Ho Er Tm Yb Lu Hf Ta W Pb Th U<br />
dolomite<br />
JDo-1 HCl (n=3)<br />
JDo-1 HNO3 (n=3)<br />
Sc Ti Co Ni Rb Sr Y Zr Nb Mo Cs Ba La Ce Pr Nd Sm Eu Gd Tb Dy Ho Er Tm Yb Lu Hf Ta W Pb Th U<br />
Figure 7. Average run precision for multiple (n) sample decompositions <strong>of</strong> certified reference materials<br />
that were analyzed repeatedly within a single ICPMS run (see text for details). All data for HF-HClO 4<br />
sample decompositions, except for JDo-1 dolomite, which includes data for three carbonate (HNO 3 )<br />
decompositions. Note bottom figure (dolomite) uses different scale. Poor RSD values are generally due<br />
to element concentrations approaching the instrument quantification limit (e.g., Sc, see Fig. 4).<br />
21
20<br />
15<br />
BHVO-2 (n=5)<br />
sample precision<br />
run precision<br />
method precision<br />
%RSD<br />
10<br />
5<br />
0<br />
20<br />
Sc Ti Co Ni Rb Sr Y Zr Nb Mo Cs Ba La Ce Pr Nd Sm Eu Gd Tb Dy Ho Er Tm Yb Lu Hf Ta W Pb Th U<br />
SGR-1b (n=2)<br />
15<br />
%RSD<br />
10<br />
5<br />
20<br />
0<br />
Sc Ti Co Ni Rb Sr Y Zr Nb Mo Cs Ba La Ce Pr Nd Sm Eu Gd Tb Dy Ho Er Tm Yb Lu Hf Ta W Pb Th U<br />
FER-2 (n=3)<br />
70.5% 34.9%<br />
15<br />
%RSD<br />
10<br />
5<br />
0<br />
30<br />
25<br />
Sc Ti Co Ni Rb Sr Y Zr Nb Mo Cs Ba La Ce Pr Nd Sm Eu Gd Tb Dy Ho Er Tm Yb Lu Hf Ta W Pb Th U<br />
JDo-1 (n=3)<br />
36.3% 59.3%<br />
%RSD<br />
20<br />
15<br />
10<br />
5<br />
0<br />
Sc Ti Co Ni Rb Sr Y Zr Nb Mo Cs Ba La Ce Pr Nd Sm Eu Gd Tb Dy Ho Er Tm Yb Lu Hf Ta W Pb Th U<br />
Figure 8. Comparison <strong>of</strong> average sample, run, <strong>and</strong> method precision for various CRMs, where n refers<br />
to the number <strong>of</strong> HF-HClO 4 sample decompositions <strong>and</strong> ICPMS runs in which the CRM was measured<br />
repeatedly. For most elements method precision is better than 5%, <strong>and</strong> is similar to, or even lower than,<br />
the sample or run precision. This suggests that for these elements the st<strong>and</strong>ard deviation <strong>of</strong> the data is<br />
primarily controlled by the inherent precision <strong>of</strong> the ICPMS instrument, <strong>and</strong> not by the decomposition<br />
method. Poor method precision is also due to low abundances for some elements that approach the IQL,<br />
e.g., Nb, Ta, <strong>and</strong> W in FeR-2 <strong>and</strong> JDo-1, as well as Co in JDo-1.<br />
22
The method precision for elements present at low concentrations, particularly<br />
in FeR-2 <strong>and</strong> JDo-1, is typically poorer (e.g., Nb, Ta, W, as well as Co in JDo-1).<br />
However, some elements that are abundantly present at concentrations ranging from<br />
several mg/kg to several percent also display poor RSD values. The BHVO-2 basalt<br />
contains >1.5% Ti, yet suffers from a method precision <strong>of</strong> >10%, whereas Ba in JDo-<br />
1 is present at a concentration (6.14 mg/kg) well above the quantification limit <strong>and</strong><br />
has an RSD above 20%. In the case <strong>of</strong> Ti, it is suspected that the poor ionization<br />
efficiency <strong>of</strong> this metal may mean that the 10 μg/kg calibration st<strong>and</strong>ard is not<br />
appropriate for quantification, as it produces a signal intensity that is not sufficiently<br />
greater than that observed for the background acid matrix, <strong>and</strong> this will be discussed<br />
in greater detail in the section regarding analytical accuracy. The poor method<br />
precision observed for Ba in the JDo-1 dolomite (22.7%) results from an ‘outlier’ for<br />
one <strong>of</strong> the three decompositions used in the calculation <strong>of</strong> the RSD value, as<br />
measured Ba in two decompositions is 5.32 <strong>and</strong> 5.63 mg/kg, while Ba measured in<br />
the third decomposition is 8.61 mg/kg.<br />
To this point the discussion <strong>of</strong> analytical precision has been limited to the HF-<br />
HClO 4 decomposition method, as much <strong>of</strong> the research conducted within the JUB<br />
Geochemistry Lab focuses on whole-rock trace metal analyses, which necessitate<br />
complete dissolution <strong>of</strong> all mineral phases within the sample powder. However, with<br />
regard to analytical precision it is necessary to discuss the HNO 3 carbonate<br />
decomposition as well. As the carbonate decomposition is used for limestone <strong>and</strong><br />
dolomite samples, the JDo-1 dolomite is the typical CRM included in decompositions<br />
<strong>and</strong> analyses <strong>of</strong> these rock types.<br />
Sample <strong>and</strong> method precision for JDo-1 using the carbonate decomposition<br />
method are presented in Figure 9, <strong>and</strong> these measures <strong>of</strong> precision are comparable for<br />
most elements. This suggests that the carbonate decomposition method does not<br />
significantly increase the error budget for these elements, similar to that observed for<br />
the HF-HClO 4 decomposition. The method precision is better than 5% (RSD) for 25<br />
<strong>of</strong> the 32 elements analyzed, <strong>and</strong> several elements displaying poor method precision<br />
(e.g., Rb, Cs, <strong>and</strong> Hf) are expected to be hosted in refractory mineral phases that<br />
would be resistant to dissolution by nitric acid. Elements at or near the IQLs <strong>and</strong><br />
unlikely to be quantifiable in carbonate rocks at concentrations observed in JDo-1<br />
include Sc, Nb, <strong>and</strong> Ta (see Fig. 4). For many elements the method precision as<br />
23
30<br />
25<br />
JDo-1 (n=3)<br />
sample precision HNO 3<br />
36.3% method precision HNO 3<br />
59.3%<br />
method precision HF-HClO 4<br />
20<br />
%RSD<br />
15<br />
10<br />
5<br />
0<br />
Sc Ti Co Ni Rb Sr Y Zr Nb Mo Cs Ba La Ce Pr Nd Sm Eu Gd Tb Dy Ho Er Tm Yb Lu Hf Ta W Pb Th U<br />
Figure 9. Average sample <strong>and</strong> method precision for the HNO 3 carbonate decomposition <strong>of</strong> the JDo-1<br />
dolomite CRM (for run precision see Fig. 7). Method precision for the carbonate decomposition is<br />
better than 5% for 25 <strong>of</strong> the 32 elements analyzed, <strong>and</strong> poorer precision is generally restricted to<br />
elements typically hosted in refractory silicate minerals (Rb, Cs, Hf).<br />
determined for the carbonate decomposition is quite similar to that observed for the<br />
HF-HClO 4 decomposition method, though the HF-HClO 4 decomposition results in<br />
significantly better precision for elements typically hosted by silicate minerals (e.g.,<br />
Rb, Zr, Cs, Hf), due to the complete dissolution <strong>of</strong> refractory Si-bearing phases.<br />
6. Analytical recovery<br />
The relative precision for most elements analyzed is quite good, but this<br />
provides no information regarding potential loss <strong>of</strong> these elements as a consequence<br />
<strong>of</strong> the sample decomposition or ICPMS measurement. Some elements can exist as a<br />
volatile chemical species at some point during the decomposition process, <strong>and</strong><br />
therefore may exhibit non-conservative behavior during sample preparation. The best<br />
example <strong>of</strong> this is Si, which is converted to volatile SiF 4 during HF dissolution <strong>of</strong><br />
silicate minerals, <strong>and</strong> is consequently lost from the sample during evaporation <strong>of</strong> HF<br />
early in the HF-HClO 4 decomposition method. The ability <strong>of</strong> the sample treatment<br />
<strong>and</strong> ICPMS measurements described here to conserve the elements <strong>of</strong> interest<br />
throughout the decomposition <strong>and</strong> analytical procedure is termed analytical recovery.<br />
24
Two approaches may be utilized to determine <strong>and</strong> quantify non-conservative<br />
behavior <strong>of</strong> the elements <strong>of</strong> interest. The first involves preparing artificial laboratory<br />
solutions that contain known quantities <strong>of</strong> the elements <strong>of</strong> interest, <strong>and</strong> then simply<br />
treating these artificial solutions as if they were unknown samples by subjecting them<br />
to a full sample decomposition <strong>and</strong> ICPMS analysis. This characterization <strong>of</strong><br />
analytical recovery is discussed here, while the second approach, which involves<br />
determining elemental concentrations in a CRM <strong>and</strong> comparing these data with the<br />
reference values, is discussed below in the section regarding analytical accuracy.<br />
Analytical recoveries <strong>of</strong> multi-element spikes (10 μg/kg) for the carbonate <strong>and</strong><br />
HF-HClO 4 decomposition methods are presented in Figure 10. Recoveries for most<br />
elements are excellent, particularly for the carbonate decomposition, in which 27 <strong>of</strong><br />
32 elements have measured concentrations between 97-103% <strong>of</strong> the expected values.<br />
However, measured Ni in the carbonate digestion spike solution is ~30% too high,<br />
whereas Ta <strong>and</strong> W exhibit quite low recoveries <strong>of</strong> ~24% <strong>and</strong> ~80%, respectively. The<br />
anomalously high Ni recovery is attributed to contamination, as method blanks for the<br />
carbonate decomposition typically contain 2-5 μg/kg Ni. The significant loss<br />
observed for Ta <strong>and</strong> W is likely due to the filtration procedure performed for the<br />
carbonate decomposition. During the filtration step, the sample solution (10 ml <strong>of</strong> 5<br />
M HNO 3 ) is passed over a 0.2 μm cellulose acetate filter (Fig. 1), which is<br />
subsequently rinsed with ~20 ml <strong>of</strong> 0.5 M HNO 3 . Whereas this rinse step appears<br />
sufficient for most elements, Ta <strong>and</strong> W (<strong>and</strong> to a lesser extent Nb) are apparently<br />
retained, <strong>and</strong> may only be quantitatively rinsed from the filter by adding trace<br />
amounts <strong>of</strong> HF (0.05 M) to the rinsing solution (K. Schmidt, personal<br />
communication). The addition <strong>of</strong> HF to the rinsing solution should only be utilized if<br />
information regarding W is critical, as HF can form unstable complexes with other<br />
trace metals, particularly the rare earths, potentially limiting quantitative recovery <strong>of</strong><br />
these elements. With regard to Nb <strong>and</strong> Ta, the use <strong>of</strong> HF during the filter rinse step is<br />
not particularly informative , as these are immobile elements expected to be hosted in<br />
refractory aluminosilicate phases, which are resistant to dissolution by the carbonate<br />
decomposition method.<br />
For the HF-HClO 4 decomposition, analytical recovery for the 32 elements in<br />
the spike solution is more variable than that observed for the carbonate<br />
decomposition. Low mass elements tend to have slightly low recoveries around 95%,<br />
25
1.35<br />
1.30<br />
1.25<br />
1.20<br />
1.15<br />
HNO 3 carbonate decomp.<br />
HF-HClO 4 decomp.<br />
HF-HClO 4 decomp. pre- June 2006<br />
1.35<br />
1.30<br />
1.25<br />
1.20<br />
1.15<br />
fraction recovered<br />
1.10<br />
1.05<br />
1.00<br />
0.95<br />
0.90<br />
0.85<br />
0.80<br />
0.75<br />
0.30<br />
0.20<br />
0.10<br />
1.10<br />
1.05<br />
1.00<br />
0.95<br />
0.90<br />
0.85<br />
0.80<br />
0.75<br />
0.30<br />
0.20<br />
0.10<br />
Sc Ti Co Ni Rb Sr Y Zr NbMoCs Ba La Ce Pr NdSmEu Gd Tb Dy Ho Er TmYb Lu Hf Ta W Pb Th U<br />
Figure 10. Multi-element analytical recoveries for the carbonate <strong>and</strong> HF-HClO 4 decomposition<br />
methods, expressed as a ratio <strong>of</strong> the measured element concentration divided by the spike concentration.<br />
Note break in the y-axis between 0.35-0.75. Spike solutions contained 10 μg/kg <strong>of</strong> the st<strong>and</strong>ard 32<br />
elements analyzed, except for the pre-June 2006 HF-HClO 4 decomposition, which contained 1 μg/kg <strong>of</strong><br />
the 24 elements routinely measured prior to June, 2006. Recoveries for most elements are between 95-<br />
105%, particularly for the carbonate <strong>and</strong> pre-June 2006 decompositions. Notable exceptions are Ni, Ta,<br />
<strong>and</strong> W for the carbonate digestion, which suggests contamination/interferences (Ni) <strong>and</strong> loss (Ta <strong>and</strong><br />
W) during the decomposition, perhaps as a result <strong>of</strong> the filtering step. The large spread <strong>and</strong> general<br />
increase as a function <strong>of</strong> mass, in the fraction recovered for the 32 element HF-HClO 4 decomposition is<br />
attributed to effects arising from the utilization <strong>of</strong> the internal st<strong>and</strong>ards 101 Ru <strong>and</strong> 187 Re (see text for<br />
details).<br />
<strong>and</strong> analytical recovery generally increases with increasing mass, though Ta does not<br />
follow this trend (Fig. 10). However, a test <strong>of</strong> analytical recovery performed prior to<br />
June 2006, conducted with the 24 elements originally analyzed by ICPMS within the<br />
JUB Geochemistry Lab, resulted in recovery percentages significantly better for many<br />
elements, particularly for elements with atomic masses below 138 (Ba). The exact<br />
reason for the more variable recovery observed for the full 32 element analysis<br />
currently conducted is not clear, though it is likely related to the internal st<strong>and</strong>ard<br />
correction. The generally constant analytical recovery <strong>of</strong> ~95% for elements with<br />
masses below 100, followed by the monotonic increase in recovery between masses<br />
133 (Cs) <strong>and</strong> 183 (W), is mirrored by the IS correction factor for this analysis, which<br />
was 0.95 for all masses below 133 Cs, before linearly increasing to 0.98 for 183 W, <strong>and</strong><br />
then decreasing to 0.96 for Pb, Th, <strong>and</strong> U.<br />
26
For the more recent, full 32 element test <strong>of</strong> analytical recovery, the 10 μg/kg<br />
spike solution was analyzed in the middle <strong>of</strong> an ICPMS run containing Fe-rich<br />
carbonates that had a dilution factor <strong>of</strong> only 250, <strong>and</strong> these carbonates displayed<br />
extreme IS correction factors as high as 2.0, reflective <strong>of</strong> the high TDS content <strong>of</strong><br />
these solutions. In contrast, the analytical recovery as determined from the pre-June<br />
2006 analysis was not included within part <strong>of</strong> an ICPMS run, <strong>and</strong> was conducted<br />
following routine cleaning <strong>and</strong> optimization <strong>of</strong> the ICPMS. It therefore seems that the<br />
discrepancy between the two tests <strong>of</strong> analytical recovery for the HF-HClO 4<br />
decomposition method, one using a 24 element spike <strong>and</strong> the other a 32 element<br />
spike, reflects details <strong>of</strong> the individual tests, including such factors as sample<br />
deposition within the ICPMS interface region affecting signal intensities (i.e., signal<br />
suppression or enhancement).<br />
Based on the few data, it is therefore concluded that for the HF-HClO 4<br />
decomposition analytical recoveries for the majority <strong>of</strong> elements are typically within<br />
2-3% <strong>of</strong> their predicted concentrations, similar to that observed for the carbonate<br />
decomposition method. This variability may represent the optimum analytical<br />
recovery possible, as it is comparable to the inherent precision <strong>of</strong> ~2% observed for<br />
most element measurements (see above discussion regarding precision). However, in<br />
view <strong>of</strong> the limited data, more tests <strong>of</strong> analytical recovery are warranted, particularly<br />
using the HF-HClO 4 decomposition method <strong>and</strong> complete, 32 element spike solutions.<br />
7. Analytical accuracy<br />
7.1. Reference values <strong>and</strong> major element interferences<br />
The best measure <strong>of</strong> the suitability <strong>of</strong> the analytical methods employed within<br />
the JUB Geochemistry Lab is if these methods can accurately (<strong>and</strong> reproducibly)<br />
quantify the elements <strong>of</strong> interest in certified reference materials. As mentioned<br />
previously, the CRMs used as routine quality assurance st<strong>and</strong>ards are chosen to match<br />
the rock types most commonly analyzed within the JUB Geochemistry Lab. If the<br />
measured CRM concentration data for a given sample decomposition is consistent<br />
with the certified data, then this provides the best measure <strong>of</strong> conservative element<br />
behavior <strong>and</strong> overall accuracy for the ICPMS methods employed.<br />
The concentration data for CRMs are generally reported as either<br />
recommended values or preferable/informational values, with recommended values<br />
27
considered more accurate <strong>and</strong> possessing lower degrees <strong>of</strong> uncertainty. Concentration<br />
data for most major <strong>and</strong> some trace elements in CRMs are typically provided by the<br />
issuing organization (Table 1). However, for many trace metals (particularly the REE,<br />
Nb, Mo, Ta, <strong>and</strong> W), values are either not provided by the issuing organization or are<br />
considered informational or preferred values, indicating that higher degrees <strong>of</strong><br />
uncertainty surround these data. Therefore, for a significant number <strong>of</strong> the 32<br />
elements analyzed at JUB, concentration data in CRMs may only be obtained from<br />
combinations <strong>of</strong> separate studies <strong>and</strong> compilations.<br />
One <strong>of</strong> the most comprehensive <strong>and</strong> widely referenced compilations <strong>of</strong><br />
geoanalytical data was assembled by Govindaraju (1994), which combines data<br />
produced using different analytical methods from a multitude <strong>of</strong> sources. However,<br />
for the CRMs listed in Table 1 much <strong>of</strong> the minor <strong>and</strong> trace metal data in Govindaraju<br />
(1994) is highly variable, particularly for the REE <strong>and</strong> other metals present at low<br />
concentrations such as Ti, Nb, Ta, Hf, Th, <strong>and</strong> U. Additionally, for some <strong>of</strong> the CRMs<br />
data has been published since 1994 using newer analytical methods such as laser<br />
ablation <strong>and</strong>/or high resolution multi-collector ICPMS. A thorough study <strong>of</strong> numerous<br />
CRMs, including the ones relevant to this discussion, was conducted by Dulski<br />
(2001), whose data closely match results from this study. Newer data (e.g., Baker et<br />
al., 2002; Bohlar et al., 2004) are consistent with the results presented here <strong>and</strong> the<br />
values <strong>of</strong> Dulski (2001), suggesting that the highly variable trace metal concentrations<br />
observed in older compilations <strong>of</strong> CRM data do not result from sample heterogeneity,<br />
but rather from limitations <strong>of</strong> older analytical methods.<br />
The reference values for the CRMs used in this study are presented in<br />
Appendix 1, along with average concentration data for these CRMs as measured<br />
within the JUB Geochemistry Lab. The literature data presented are not intended to<br />
reflect an exhaustive or comprehensive survey <strong>of</strong> all available data for the various<br />
CRMs, <strong>and</strong> were selected according to three criteria: 1) data published by the CRM<br />
issuing organization, which usually are compilations <strong>of</strong> data produced by different<br />
laboratories using different methods; 2) studies that provided data for a large number<br />
<strong>of</strong> the 32 elements considered for this study; <strong>and</strong>/or 3) work that utilized similar<br />
analytical methods, particularly ICPMS techniques. This approach is intended to<br />
facilitate comparison between the CRM values measured at JUB with worldwide<br />
laboratories using similar analytical methods. Recently, Jochum et al. (2005)<br />
produced a thorough compilation <strong>of</strong> published major element, trace element, <strong>and</strong><br />
28
isotopic data for geologic CRMs, available as an electronic database (GeoReM), <strong>and</strong><br />
the reader is referred to this resource for a complete survey <strong>of</strong> available CRM data.<br />
Analytical accuracy will be discussed individually for each CRM. However,<br />
some general observations can be made regarding the JUB data <strong>and</strong> the reference<br />
values. The first is that the JUB data typically display greater precision compared to<br />
the average reference values, which is ascribed to the greater number <strong>of</strong> different<br />
methods <strong>and</strong> analytical techniques employed in the production <strong>of</strong> the reference<br />
values. The second is that poor precision <strong>and</strong>/or accuracy in the JUB data for specific<br />
elements, particularly those with masses
16000<br />
0.5 M HCl matrix<br />
a<br />
1.6<br />
14000<br />
1.4<br />
interference on 59 Co in sample solution (cps)<br />
12000<br />
10000<br />
8000<br />
6000<br />
4000<br />
2000<br />
MgCl +<br />
CaO(H) +<br />
MgCl + interference in JDo-1<br />
CaO(H) + interference in JDo-1<br />
1.2<br />
1.0<br />
0.8<br />
0.6<br />
0.4<br />
0.2<br />
interference on 59 Co in sample powder (mg/kg)<br />
0<br />
0.0<br />
0 10 20 30 40 50 60 70 80 90 100<br />
MgO, CaO in sample powder (wt.%)<br />
16000<br />
0.5 M HNO 3<br />
matrix<br />
b<br />
1.6<br />
14000<br />
1.4<br />
interference on 59 Co in sample solution (cps)<br />
12000<br />
10000<br />
8000<br />
6000<br />
4000<br />
2000<br />
CaO(H) +<br />
CaO(H) + interference in JDo-1<br />
1.2<br />
1.0<br />
0.8<br />
0.6<br />
0.4<br />
0.2<br />
interference on 59 Co in sample powder (mg/kg)<br />
0<br />
MgCl +<br />
0.0<br />
0 10 20 30 40 50 60 70 80 90 100<br />
MgO, CaO in sample powder (wt.%)<br />
Figure 11. Mg <strong>and</strong> Ca interferences on 59 Co as functions <strong>of</strong> MgO <strong>and</strong> CaO content in sample powders.<br />
MgO <strong>and</strong> CaO are calculated as wt.% assuming the sample powders have been diluted by a factor <strong>of</strong><br />
1000, which is a typical dilution factor for Mg- <strong>and</strong> Ca-rich carbonate rocks (primarily dolomites). Top<br />
figure (a) shows interferences observed in HCl matrix, <strong>and</strong> the MgCl interference on mass 59 predicted<br />
for JDo-1 (18.4% MgO) is >0.6 mg/kg, significantly greater than literature Co values <strong>of</strong> ~0.2 mg/kg<br />
(see App. 1). Bottom figure (b) illustrates that MgCl interferences are eliminated when samples are<br />
analyzed in HNO 3 , yet significant CaO(H) + interferences predicted for JDo-1 (34.0% CaO) still<br />
preclude accurate determinations <strong>of</strong> Co. It appears that reasonable Co determinations may only be<br />
possible for typical carbonate samples (MgO, CaO <strong>of</strong> 5-40%), when Co contents in the rock are at<br />
least one order <strong>of</strong> magnitude higher than the predicted interference concentrations (i.e., >10 mg/kg).<br />
30
Table 4. Interferences observed for elements <strong>of</strong> interest with atomic masses 1000<br />
44 Ca 16 O 37 Cl insignificant interference<br />
* all interfering molecular species presumed to exist in a +1 valence state, <strong>and</strong> bold text indicates dominant species.<br />
interferences must be inferred, as the total interference at a given mass may represent<br />
combinations <strong>of</strong> multiple interfering species, <strong>and</strong> examples would include CO 2,<br />
CO 2 H, NO 2, CCl, <strong>and</strong> NCl. For the major elements Mg, Ca, Al, Mn, <strong>and</strong> Fe, high<br />
concentrations in rock samples are not observed to produce significant interferences<br />
for atomic masses >100. The practice <strong>of</strong> diluting samples with HCl following<br />
decomposition <strong>and</strong> the subsequent formation <strong>of</strong> metal-chloride molecular species<br />
(MCl(O) + ) is responsible for many <strong>of</strong> the interferences listed in Table 4, <strong>and</strong> these<br />
MCl(O) + interferences are insignificant for samples decomposed using the carbonate<br />
decomposition method. Samples decomposed using the HF-HClO 4 method may be<br />
diluted with 0.5 M HNO 3 to reduce or eliminate MCl(O) + , assuming that all HCl used<br />
during the HF-HClO 4 decomposition is evaporated prior to diluting with HNO 3 .<br />
This suggests that use <strong>of</strong> HCl should be avoided in order to minimize MCl(O) +<br />
interferences. However, some elements, particularly the HFSE Nb <strong>and</strong> Ta, are not<br />
stable in low molarity HNO 3 solutions, <strong>and</strong> require HCl or HF to prevent apparent<br />
31
‘loss’ <strong>of</strong> these elements in dissolved rock solutions due to adsorption/precipitation<br />
reactions in sample containers (Münker, 1998). For trace concentrations <strong>of</strong> Nb <strong>and</strong><br />
Ta, Münker (1998) suggested diluting dissolved rock samples in 0.3 M HNO 3 <strong>and</strong><br />
0.06 M HCl for stabilizing these metals for ≥24 h, <strong>and</strong> the addition <strong>of</strong> 0.02 M HF<br />
when Nb <strong>and</strong> Ta in the diluted solution exceeded approximately 100 μg/kg <strong>and</strong> 4<br />
μg/kg, respectively. The consistent use <strong>of</strong> 0.5 M HCl <strong>and</strong> the formation <strong>of</strong> MCl(O) +<br />
interferences does not appear to have adversely affected analytical precision or<br />
accuracy for the majority <strong>of</strong> the ICPMS analyses discussed here. However, for<br />
samples rich in Mg, Mn, <strong>and</strong>/or Fe that also have very low trace metal concentrations,<br />
it would be useful to more fully investigate mixtures <strong>of</strong> HNO 3 <strong>and</strong> HCl in order to<br />
maximize the sensitivity <strong>of</strong> the ICPMS method.<br />
Some interferences persist regardless <strong>of</strong> the acid matrix used <strong>and</strong> result from<br />
analyzing aqueous solutions (e.g., CO 2 on 45 Sc), or from sample ionization in an Ar<br />
plasma. The latter effect is illustrated by presumed 44 Ca + 2 <strong>and</strong> 48 Ca 40 Ar + interferences<br />
on 88 Sr, which may erroneously increase measured Sr concentrations by 1-3 mg/kg as<br />
CaO contents in sample powders approach 40%. Another example is 55 Mn 40 Ar + ,<br />
which produces a 2-5 mg/kg interference on 95 Mo in sample powders that contain 10-<br />
15% MnO. Figures illustrating the significant interferences observed for Ni, Sr, Y, Zr,<br />
Nb, <strong>and</strong> Mo are presented in Appendix 2, <strong>and</strong> permit estimates <strong>of</strong> the magnitude <strong>of</strong><br />
these interferences in various rock samples. It should be stated, however, that<br />
interferences observed in this study may not significantly affect analytical accuracy,<br />
provided the magnitude <strong>of</strong> the interference is small relative to the concentration <strong>of</strong> the<br />
trace metal <strong>of</strong> interest. An example is Nb, which cannot be determined accurately in<br />
Fe-rich, Nb-poor samples (20 mg/kg Nb).<br />
Analytical accuracy, as defined by the agreement between the average<br />
reference value <strong>and</strong> the measured JUB concentrations (App. 1), is presented in the<br />
following figures as a ratio <strong>of</strong> (JUB data/reference value), with a ratio <strong>of</strong> one<br />
indicating perfect agreement between the JUB data <strong>and</strong> the average CRM reference<br />
value. However, significant uncertainties in reference values exist for some elements,<br />
<strong>and</strong> some CRMs are more poorly characterized than others. This uncertainty in the<br />
CRM reference values can produce a wide range in the calculated ratio <strong>of</strong> (JUB<br />
data/reference value). Therefore, the range in calculated ratios due solely to reference<br />
32
value uncertainty (as %RSD) is represented graphically in the following figures by<br />
means <strong>of</strong> vertical grey bars. A cursory examination indicates this range can be quite<br />
large, as illustrated in Fig. 12 for Mo (0.63–1.33) in FeR-2.<br />
7.2. High Fe content rocks<br />
B<strong>and</strong>ed iron-formations (IFs) are a rock type frequently analyzed within the<br />
JUB Geochemistry Lab, <strong>and</strong> these samples are typically more than 90% SiO 2 <strong>and</strong><br />
Fe 2 O 3 in varying proportions, with NaO, MgO, Al 2 O 3 , <strong>and</strong> CaO all commonly present<br />
at low concentrations (
1.50 FeR-2 (n=3)<br />
1.25<br />
JUB / reference average<br />
1.00<br />
0.75<br />
0.50<br />
0.25<br />
0.00<br />
Sc Ti Co Ni Rb Sr Y Zr Nb Mo Cs Ba La Ce Pr Nd Sm Eu Gd Tb Dy Ho Er Tm Yb Lu Hf Ta W Pb Th U<br />
Figure 12. Accuracy estimation for ICPMS analysis <strong>of</strong> iron-formation FeR-2, where n equals the<br />
number <strong>of</strong> separate HF-HClO 4 sample decompositions <strong>and</strong> ICPMS analyses in 0.5 M HCl. Grey bars<br />
represent variability in the calculated ratio that is due solely to uncertainty in the reference average.<br />
Vertical lines represent the variability in the calculated ratio due to the uncertainty <strong>of</strong> the JUB data, <strong>and</strong><br />
all data are presented in App. 1. Precision <strong>of</strong> JUB data is generally better than that observed for the<br />
average reference values, particularly for elements with atomic masses >100 amu (i.e., heavier than<br />
Mo). Measured JUB data for higher mass elements are generally lower than the reference average,<br />
though this reference average is skewed to higher values by the data <strong>of</strong> Yu et al. (2001) (see text for<br />
details <strong>and</strong> App. 1). Data for Nb not included due to probable FeCl interferences <strong>and</strong> large uncertainty<br />
in JUB data for Nb measurements (70% RSD).<br />
FeR-2 compared to very Al-poor IFs. For FeR-2, the st<strong>and</strong>ard deviation <strong>of</strong> the JUB<br />
data is typically 1-3%, considerably better than the range observed in the literature<br />
data (App. 1). The JUB data tends to be somewhat lower than the reference value<br />
average, with 25 <strong>of</strong> the 32 elements analyzed displaying concentrations between 85–<br />
100% <strong>of</strong> the literature data (Fig. 12). However, the average reference values for many<br />
elements are skewed towards higher numbers by the data <strong>of</strong> a single study (Yu et al.,<br />
2001), <strong>and</strong> if the data from Yu et al. (2001) are not considered, then measured JUB<br />
concentrations for 24 <strong>of</strong> 32 elements are within 90-110% <strong>of</strong> reference values (see<br />
App. 1).<br />
FeR-4<br />
The FeR-4 iron-formation has very similar SiO 2 (50.1%) <strong>and</strong> Fe 2 O 3 (39.2%)<br />
contents compared to FeR-2, but only one-third as much Al 2 O 3 (1.70%).<br />
34
1.50 FeR-4 (n=2)<br />
1.25<br />
JUB / reference average<br />
1.00<br />
0.75<br />
0.50<br />
0.25<br />
0.00<br />
Sc Ti Co Ni Rb Sr Y Zr Nb Mo Cs Ba La Ce Pr Nd Sm Eu Gd Tb Dy Ho Er Tm Yb Lu Hf Ta W Pb Th U<br />
Figure 13. Accuracy estimation for ICPMS analysis <strong>of</strong> iron-formation FeR-4, where n equals the<br />
number <strong>of</strong> separate HF-HClO 4 sample decompositions <strong>and</strong> ICPMS analyses in 0.5 M HCl. Grey bars<br />
represent variability in the calculated ratio that is due solely to uncertainty in the reference average.<br />
Vertical lines represent the variability in the calculated ratio due to the uncertainty <strong>of</strong> the JUB data, <strong>and</strong><br />
all data are presented in App. 1. Good agreement is observed for most elements, with exceptions being<br />
Sc, Ti, Co, Ni <strong>and</strong> U. Reference data for Nb, Mo, <strong>and</strong> Ta are not shown, <strong>and</strong> are available only from a<br />
single published study by Yu et al. (2001), who found highly variable results for these elements when<br />
comparing two different sample decomposition methods (HF-HClO 4 versus HF-H 2 SO 4 ). See text for<br />
details.<br />
Consequently, many trace metals have significantly lower concentrations in FeR-4<br />
(e.g., Sc, Zr, Nb, the REE, Hf, Th). The precision <strong>of</strong> the JUB data is poorly<br />
constrained by the limited number <strong>of</strong> analyses, though it seems comparable to, if not<br />
significantly better than, the variation observed in literature reference values. The<br />
relative accuracy <strong>of</strong> the FeR-4 data is presented in Figure 13, <strong>and</strong> is somewhat better<br />
than that observed for FeR-2, except for elements with the lowest <strong>and</strong> highest masses<br />
(Sc, Ti, Co, Ni, <strong>and</strong> U). Poor results for Sc are not unexpected as Sc literature values<br />
for FeR-4 range from 1.1-1.5 mg/kg, which is similar in magnitude to the IQL<br />
observed for Sc (~0.8 mg/kg, Table 2).<br />
However, Ti, Co, <strong>and</strong> Ni are all present in FeR-4 at concentrations well above<br />
their respective IQLs using the ICPMS methods described here (Fig. 4), so low<br />
signal-to-noise ratios are not expected to adversely affect determinations <strong>of</strong> these<br />
elements. For Co <strong>and</strong> Ni, the literature data are reported to only one significant digit<br />
<strong>and</strong> concentrations <strong>of</strong> these two elements are apparently low (2 <strong>and</strong> 8 mg/kg,<br />
35
espectively). Previous workers have observed that many metals with very low<br />
abundances in CRMs are likely present at concentrations lower than initially reported,<br />
<strong>and</strong> that early data compilations frequently overestimate the true abundance <strong>of</strong> these<br />
elements (e.g., Dulski, 2001; Yu et al., 2001), perhaps due to older, less sensitive<br />
analytical techniques. It is therefore suggested that the apparent ‘low’ concentrations<br />
for Co <strong>and</strong> Ni determined at JUB may reflect this observation, though the limited<br />
published data <strong>and</strong> few JUB analyses <strong>of</strong> FeR-4 indicate that more work regarding<br />
these elements is necessary.<br />
The Ti concentration determined in this study <strong>and</strong> the literature reference<br />
value (327 mg/kg <strong>and</strong> 420 mg/kg, respectively), both seem high enough to suggest<br />
that accurate Ti analyses in FeR-4 are possible. However the Ti reference value is<br />
compiled from analyses utilizing sample fusion <strong>and</strong> X-ray fluorescence (XRF)<br />
measurements (Abbey et al., 1983), which may suffer from relatively high blanks<br />
from the flux used during sample fusion, <strong>and</strong>/or poor detection/quantification limits<br />
compared to ICPMS measurements. It is therefore possible that the true Ti abundance<br />
in FeR-4 is lower than 420 mg/kg, <strong>and</strong> the discrepancy between the JUB data <strong>and</strong> the<br />
compiled reference average rather reflects the limited sensitivity <strong>and</strong> precision <strong>of</strong><br />
many <strong>of</strong> the Ti analyses <strong>of</strong> FeR-4.<br />
IF-G<br />
The IF-G iron-formation is the most Fe-rich <strong>and</strong> Al-poor IF utilized as a CRM<br />
within the Geochemistry Lab, with 55.9% Fe 2 O 3 , 41.2% SiO 2 , <strong>and</strong> 0.15% Al 2 O 3 .<br />
Unfortunately, IF-G as prepared by the issuing organization (IWG-GIT, Table 1)<br />
during the original processing run (first lot) is no longer available, <strong>and</strong> subsequently a<br />
second lot <strong>of</strong> IF-G was prepared by IWG-GIT. For many <strong>of</strong> the earliest IF analyses<br />
performed within the Geochemistry Lab, IF-G was the preferred CRM, <strong>and</strong> as a result<br />
the aliquot <strong>of</strong> the original first lot IF-G powder was completely consumed. In June<br />
2007 an aliquot <strong>of</strong> the second lot <strong>of</strong> IF-G powder was obtained from IWG-GIT, which<br />
subsequently has been analyzed several times. As reported by IWG-GIT, the only<br />
differences between the two IF-G lots regard the concentrations <strong>of</strong> Cr, Co, <strong>and</strong> W. As<br />
Co <strong>and</strong> W are relevant to this discussion, it is noted that Co <strong>and</strong> W concentrations in<br />
the second IF-G lot are significantly lower than those originally reported for the first<br />
IF-G lot, due to different crushing methods utilized during preparation <strong>of</strong> the two lots.<br />
36
1<br />
IF-G/shale<br />
0.1<br />
0.01<br />
La Ce Pr Nd Pm Sm Eu Gd Tb Dy Y Ho Er Tm Yb Lu<br />
Fig 14. Rare earth <strong>and</strong> element <strong>and</strong> yttrium data from four separate HF-HClO 4 decompositions <strong>of</strong> the<br />
new, second lot <strong>of</strong> the iron-formation IF-G as prepared <strong>and</strong> issued by IWG-GIT. Data normalized to<br />
Post Archean Australian Shale (PAAS, McLennan, 1989). The majority <strong>of</strong> the REE data are consistent,<br />
with exceptions for the light REE (La, Ce, Pr, <strong>and</strong> Nd).<br />
As a consequence <strong>of</strong> the above events, most analyses performed at JUB <strong>of</strong> the<br />
original IF-G lot used the pre-June 2006 ICPMS method, which quantified 24 trace<br />
metals. Only two analyses <strong>of</strong> this original IF-G lot were performed using the current<br />
32 element ICPMS method, whereas the second IF-G lot has been analyzed several<br />
times using the 32 element method. Unfortunately, results for analyses <strong>of</strong> the second<br />
IF-G lot are not entirely consistent, <strong>and</strong> suggest that heterogeneities may exist within<br />
the sample powder. This is illustrated in Figure 14, which is a REE-yttrium (REY)<br />
diagram depicting data from four separate HF-HClO 4 decompositions for the second<br />
lot <strong>of</strong> IF-G powder. For the majority <strong>of</strong> the REE (Sm through Lu), the data for the<br />
separate decompositions are quite consistent, whereas La, Ce, Pr, <strong>and</strong> Nd are much<br />
more variable. This behavior is not well understood at this time. Fractionation across<br />
the REE series for small aliquots <strong>of</strong> rock samples is <strong>of</strong>ten a function <strong>of</strong> mineralogical<br />
controls on REE distributions within the bulk rock, <strong>and</strong> the data in Fig. 14 are<br />
preliminarily interpreted to reflect small mineralogical heterogeneities in the second<br />
lot <strong>of</strong> IF-G.<br />
In consideration <strong>of</strong> the above observations, the discussion <strong>of</strong> analytical<br />
accuracy is restricted to the two decompositions <strong>of</strong> the original lot <strong>of</strong> IF-G that were<br />
37
1.50 IF-G (n=2)<br />
1.25<br />
JUB / reference average<br />
1.00<br />
0.75<br />
0.50<br />
0.25<br />
0.00<br />
Sc Ti Co Ni Rb Sr Y Zr Nb Mo Cs Ba La Ce Pr Nd Sm Eu Gd Tb Dy Ho Er Tm Yb Lu Hf Ta W Pb Th U<br />
Figure 15. Accuracy estimation for ICPMS analysis <strong>of</strong> iron-formation IF-G, where n equals the number<br />
<strong>of</strong> separate HF-HClO 4 sample decompositions <strong>and</strong> ICPMS analyses in 0.5 M HCl. Grey bars represent<br />
variability in the calculated ratio that is due solely to uncertainty in the reference average. Vertical lines<br />
represent the variability in the calculated ratio due to the uncertainty <strong>of</strong> the JUB data, <strong>and</strong> all data are<br />
presented in App. 1. Data for Sc not reported due to high IQL, <strong>and</strong> Nb not reported due to significant<br />
FeCl interference. Reasons for poor observed Ti accuracy are considered to be similar to those proposed<br />
for FeR-4 (see text for details). Lower JUB concentrations for Hf, Ta, Pb, <strong>and</strong> Th are not considered<br />
anomalous, as it is likely that these elements are present at concentrations lower than originally reported<br />
for IF-G, an observation supported by the data <strong>of</strong> Dulski et al. (2001) <strong>and</strong> Bohlar et al. (2004).<br />
performed using the current 32 element ICPMS analysis (see App. 1). Results from<br />
these analyses are presented in Figure 15, <strong>and</strong> JUB data for 23 <strong>of</strong> 32 elements are<br />
between 90–100% <strong>of</strong> the average reference values. Data for Sc <strong>and</strong> Nb are not shown,<br />
as Sc concentrations in IF-G (~0.3 mg/kg) are below the IQL for Sc (Fig. 4), <strong>and</strong><br />
literature Nb concentrations <strong>of</strong> ~0.1 mg/kg are significantly lower than the predicted<br />
FeCl interferences on 93 Nb <strong>of</strong> ~0.6 mg/kg (Apps. 1 <strong>and</strong> 2).<br />
The apparent poor accuracy <strong>of</strong> the JUB determination for Ti is considered to<br />
arise from reasons similar to those proposed for Ti determinations in FeR-4, i.e., a<br />
combination <strong>of</strong> low concentration (84 mg/kg) <strong>and</strong> analytical methodology for<br />
determination <strong>of</strong> the reference values (e.g., sample fusion <strong>and</strong> XRF analyses). The<br />
wide discrepancy observed for the Hf, Ta, Pb, <strong>and</strong> Th data <strong>of</strong> this study is primarily<br />
due to the inclusion <strong>of</strong> values from Govindaraju (1994) when calculating the<br />
reference average for these elements. Compared to more recent ICPMS analyses by<br />
Dulski (2001) <strong>and</strong> Bohlar et al. (2004), JUB data for these elements are quite<br />
38
consistent (App. 1), <strong>and</strong> similar to FeR-4, it is concluded that older data compilations<br />
frequently overestimate the abundance <strong>of</strong> many trace elements that are present at very<br />
low concentrations.<br />
7.3. Shales <strong>and</strong> clastic sediments<br />
Fine-grained clastic rocks <strong>and</strong> sediments commonly analyzed during<br />
geochemical research at JUB include shales <strong>and</strong> river sediments. Shales <strong>and</strong> clastic<br />
sediments may have Al 2 O 3 concentrations as high as ~15%, <strong>and</strong> the high<br />
aluminosilicate mineral content <strong>of</strong> these samples produces a relative enrichment in<br />
many trace metals. For example, the REY are incompatible trace metals that behave<br />
similarly to Al in crustal processes, <strong>and</strong> the REY are good proxies for the Al-rich<br />
continental crust that erodes to form river sediments <strong>and</strong> shales. In SCo-1, a shale<br />
CRM issued by the USGS, the summed REY concentrations are approximately 170<br />
mg/kg, whereas the IF-G iron-formation discussed above contains ~22 mg/kg <strong>of</strong><br />
REY.<br />
This trace metal enrichment facilitates geochemical analyses <strong>of</strong> shales <strong>and</strong><br />
clastic sediments in two ways, with the first being that reference values are better<br />
constrained, as older, less sensitive analytical methods can produce precise <strong>and</strong><br />
accurate data for many trace metals. The second advantage arises from the ability to<br />
significantly dilute shale <strong>and</strong> sediment samples for ICPMS analyses while remaining<br />
well above the IQLs for many trace metals, thereby reducing matrix effects <strong>and</strong> major<br />
element interferences. This latter effect is illustrated using SCo-1 <strong>and</strong> IF-G, which, as<br />
mentioned above, have very different REY concentrations. Assuming SCo-1 was<br />
diluted by a factor 7-8 times greater than the dilution factor used for IF-G, then<br />
similar ICPMS signal intensities would be expected for the REY when analyzing<br />
these two CRMs.<br />
SCo-1<br />
The SCo-1 shale CRM (Cody shale) contains 13.7% Al 2 O 3 , 5.14% Fe 2 O 3 ,<br />
approximately 2.6% each MgO <strong>and</strong> CaO, <strong>and</strong> only trace amounts <strong>of</strong> Mn (410 mg/kg).<br />
Figure 16 depicts the analytical accuracy observed for SCo-1. Of the 32 elements<br />
analyzed, 27 are within ±10% <strong>of</strong> the reference value, with lower masses displaying<br />
the greatest deviation, as the JUB data are significantly higher for these elements (Sc,<br />
Ti, Co, Ni). Some <strong>of</strong> the discrepancy observed for Sc, Ti, Co, <strong>and</strong> Ni can be ascribed<br />
39
1.50 SCo-1 (n=4)<br />
1.25<br />
JUB / reference average<br />
1.00<br />
0.75<br />
0.50<br />
0.25<br />
0.00<br />
Sc Ti Co Ni Rb Sr Y Zr Nb Mo Cs Ba La Ce Pr Nd Sm Eu Gd Tb Dy Ho Er Tm Yb Lu Hf Ta W Pb Th U<br />
Figure 16. Accuracy estimation for ICPMS analysis <strong>of</strong> shale SCo-1, where n equals the number <strong>of</strong><br />
separate HF-HClO 4 sample decompositions <strong>and</strong> ICPMS analyses in 0.5 M HCl. Grey bars represent<br />
variability in the calculated ratio that is due solely to uncertainty in the reference average. Vertical lines<br />
represent the variability in the calculated ratio due to the uncertainty <strong>of</strong> the JUB data, <strong>and</strong> all data are<br />
presented in App. 1. The triangles for Sc, Ti, Co, <strong>and</strong> Ni represent average JUB data that excludes a<br />
single analysis that determined anomalously high values for these four elements, <strong>and</strong> when the<br />
‘anomalous’ data are not considered the JUB averages more closely match the reference values for Sc,<br />
Ti, Co, <strong>and</strong> Ni.<br />
to a single JUB analysis <strong>of</strong> SCo-1 that determined much higher concentrations for<br />
these elements, <strong>and</strong> when these ‘anomalous’ data are not considered (see Fig. 16), the<br />
measured concentrations for 30 <strong>of</strong> 32 analyzed elements are between 90–110% <strong>of</strong> the<br />
reference value.<br />
Potential major element interferences on trace metal determinations in SCo-1<br />
would include 27 Al 35 Cl on 62 Ni, <strong>and</strong> FeCl on 93 Nb (Table 4). The impact <strong>of</strong> Alchloride<br />
interferences on 62 Ni is significant enough that it is recommended that 62 Ni<br />
not be used for Ni determinations in shales (or sediments) that contain more that a<br />
few percent Al 2 O 3 <strong>and</strong> that are diluted in HCl (see App. 2). Note that the Ni data<br />
presented in App. 1 <strong>and</strong> Fig. 16 reflect concentrations for SCo-1 that were determined<br />
solely by monitoring <strong>of</strong> 60 Ni. As for Nb, the presence <strong>of</strong> ~5% Fe 2 O 3 in SCo-1 would<br />
be predicted to increase Nb concentrations by roughly 0.1 mg/kg, which is<br />
insignificant when compared to the SCo-1 reference Nb value <strong>of</strong> 11.9 mg/kg. It is<br />
therefore concluded that Fe 2 O 3 concentrations in shales (or sediments) <strong>of</strong> 3-7% are<br />
40
1.50 SGR-1b (n=3)<br />
1.25<br />
JUB / reference average<br />
1.00<br />
0.75<br />
0.50<br />
0.25<br />
0.00<br />
Sc Ti Co Ni Rb Sr Y Zr Nb Mo Cs Ba La Ce Pr Nd Sm Eu Gd Tb Dy Ho Er Tm Yb Lu Hf Ta W Pb Th U<br />
Figure 17. Accuracy estimation for ICPMS analysis <strong>of</strong> shale SGR-1b, where n equals the number <strong>of</strong><br />
separate HF-HClO 4 sample decompositions <strong>and</strong> ICPMS analyses in 0.5 M HCl. Grey bars represent<br />
variability in the calculated ratio that is due solely to uncertainty in the reference average. Vertical lines<br />
represent the variability in the calculated ratio due to the uncertainty <strong>of</strong> the JUB data, <strong>and</strong> all data are<br />
presented in App. 1. The measured concentrations for Ni, Y, Tb, <strong>and</strong> Tm are not necessarily considered<br />
anomalous, <strong>and</strong> likely represent small interferences (Ni) <strong>and</strong> relatively large uncertainties in the<br />
reference data (Y, Tb, Tm; see text <strong>and</strong> App. 1).<br />
not likely to adversely affect Nb determinations, assuming Nb is present in the sample<br />
at concentrations <strong>of</strong> several mg/kg or higher.<br />
SGR-1b<br />
The SGR-1b (Green River) shale is a petroleum <strong>and</strong> carbonate rich CRM<br />
issued by the USGS. The SGR-1b shale is identical to the original SGR-1 shale (S.A.<br />
Wilson, USGS, personal communication) <strong>and</strong> contains 6.5% Al 2 O 3 , 4.4% MgO, 8.4%<br />
CaO, 3.0% Fe 2 O 3 , <strong>and</strong> minor amounts <strong>of</strong> Mn (267 mg/kg). While the Al content <strong>of</strong><br />
SGR-1b is low compared to the SCo-1 shale, it is still high enough to produce<br />
significant Al-chloride interferences on 62 Ni. As a result Ni concentration data<br />
reported for SGR-1b are determined solely from analyses <strong>of</strong> the 60 Ni isotope. For the<br />
remaining major elements, significantly higher MgO <strong>and</strong> CaO in SGR-1b compared<br />
to SCo-1 would be expected to erroneously increase measured signal intensities for<br />
60 Ni, due to MgCl <strong>and</strong> CaO(H) inteferences. The combined magnitude <strong>of</strong> these Mg<br />
<strong>and</strong> Ca interferences on 60 Ni is estimated to be ~2 mg/kg (App. 2), <strong>and</strong> this is<br />
41
consistent with the average Ni concentration for SGR-1b measured at JUB <strong>of</strong> 30.3<br />
mg/kg, which is slightly higher than the Ni reference average <strong>of</strong> 28 mg/kg (App. 1).<br />
Figure 17 compares measured JUB trace element data with reference values,<br />
<strong>and</strong> 28 <strong>of</strong> the 32 analyzed elements are within ±10% <strong>of</strong> the average reference value.<br />
Measured concentrations for Y, Tb, <strong>and</strong> Tm are 10-20% lower than the average<br />
reference value, whereas measured Nb concentrations are >20% higher than the<br />
literature Nb values. The discrepancies for Y, Tb, <strong>and</strong> Tm may arise from the wide<br />
range observed in reference values, <strong>and</strong>/or the aforementioned higher trace metal<br />
contents frequently reported by compilations <strong>of</strong> older data, as concentrations for these<br />
elements as reported by Dulski (2001) are indistinguishable from the JUB data (App.<br />
1). The higher Nb content in SGR-1b as determined at JUB is unlikely to result from<br />
FeCl interferences (see above discussion for SCo-1), <strong>and</strong> is not necessarily anomalous<br />
considering that literature Nb values solely reflect older data compilations.<br />
7.4. High Si content rocks (cherts)<br />
Occasionally, trace metal analyses are required for rocks which are almost<br />
exclusively composed <strong>of</strong> quartz (SiO 2 ). Examples <strong>of</strong> such rocks include microcrystalline<br />
SiO 2 chemical precipitates (cherts), <strong>and</strong> relatively pure quartz clastic<br />
sediments (e.g., s<strong>and</strong>stones). To date, little research has been performed at JUB<br />
concerning quartz-dominated clastic rocks, <strong>and</strong> therefore this section focuses on very<br />
pure SiO 2 chemical precipitates. These cherts are similar to the Si-rich, Fe-poor ironformations<br />
discussed earlier, though for this discussion the term chert refers to a SiO 2<br />
chemical precipitate that contains very little (≤0.5%) Al 2 O 3 , MgO, <strong>and</strong> CaO.<br />
Concentrations <strong>of</strong> Fe 2 O 3 may range between 1-5%, <strong>and</strong> as Fe contents increase<br />
beyond this range the previous discussion regarding iron-formations is considered<br />
more appropriate.<br />
The only chert CRM analyzed within the JUB Geochemistry Lab is JCh-1<br />
(Geological Survey <strong>of</strong> Japan), which is 98% SiO 2 , 0.73% Al 2 O 3 ,
1.50<br />
JCh-1 (n=3)<br />
1.25<br />
JUB / reference average<br />
1.00<br />
0.75<br />
0.50<br />
0.25<br />
0.00<br />
Sc Ti Co Ni Rb Sr Y Zr Nb Mo Cs Ba La Ce Pr Nd Sm Eu Gd Tb Dy Ho Er Tm Yb Lu Hf Ta W Pb Th U<br />
Figure 18. Accuracy estimation for ICPMS analysis <strong>of</strong> chert JCh-1, where n equals the number <strong>of</strong><br />
separate HF-HClO 4 sample decompositions <strong>and</strong> ICPMS analyses in 0.5 M HCl. Grey bars represent<br />
variability in the calculated ratio that is due solely to uncertainty in the reference average. Vertical lines<br />
represent the variability in the calculated ratio due to the uncertainty <strong>of</strong> the JUB data, <strong>and</strong> all data are<br />
presented in App. 1. The JUB data are quite consistent <strong>and</strong> reproducible, with the exception <strong>of</strong> Ni. The<br />
reference data for higher mass elements have a wide range <strong>of</strong> values (i.e., larger grey bars), with JUB<br />
data generally lower. Considering the uncertainty in the reference data, <strong>and</strong> the very low trace metal<br />
contents, the JUB data have additionally been normalized to the data <strong>of</strong> Dulski (2001), who has<br />
published the only complete REY dataset for JCh-1 (App. 1). The JUB/Dulski (2001) values are plotted<br />
as white triangles, <strong>and</strong> the majority <strong>of</strong> these ratios are close to 1.<br />
with measured concentrations for many metals significantly lower than reference data.<br />
However, the JUB data tend to be much more precise, particularly for higher mass<br />
elements. The exception is Ni, <strong>and</strong> the large variation in Ni measured at JUB results<br />
from one analysis, which determined a concentration <strong>of</strong> 14.0 mg/kg. The remaining<br />
two analyses at JUB determined Ni <strong>of</strong> 7.41 <strong>and</strong> 8.59 mg/kg, similar to the average<br />
reference value <strong>of</strong> 8.13 mg/kg (App. 1), <strong>and</strong> it is concluded that the single, 14.0 mg/kg<br />
Ni analysis is likely anomalous.<br />
The large uncertainty in the average reference values for many elements may<br />
be due to the difficulty in determining the very low trace metal contents in JCh-1, <strong>and</strong><br />
few reference data exist for some elements (e.g., the monoisotopic REE Pr, Tb, Ho,<br />
<strong>and</strong> Tm). One <strong>of</strong> the most complete trace element datasets yet published for JCh-1 is<br />
from Dulski (2001), whose data compare favorably with the concentrations<br />
determined at JUB, <strong>and</strong> these data are presented in Fig. 18.<br />
43
7.5. Carbonate rocks<br />
A common type <strong>of</strong> rock analyzed at JUB are carbonates, which are chemical<br />
precipitates generally containing low trace metal contents, similar to cherts. Carbonate<br />
rocks are dominated by (Ca,Mg)CO 3 minerals, <strong>and</strong> Ca <strong>and</strong> Mg may exist in solid<br />
solution within the carbonate mineral structure. As a result, carbonate rock samples<br />
typically analyzed may range from limestones (CaCO 3 ), through dolomites<br />
((Ca,Mg)CO 3 ), to pure magnesites (MgCO 3 ). All <strong>of</strong> these carbonate rock types are<br />
effectively decomposed using the HNO 3 carbonate decomposition method, though, as<br />
noted previously, any accessory aluminosilicate minerals <strong>and</strong> refractory organic<br />
carbon are relatively unaffected by dissolution with HNO 3 .<br />
This section discusses the application <strong>of</strong> both decomposition methods (HF-<br />
HClO 4 <strong>and</strong> HNO 3 ) to carbonate rocks, <strong>and</strong> describes the applicability <strong>and</strong> accuracy <strong>of</strong><br />
these methods with respect to specific elements <strong>of</strong> interest. As the literature reference<br />
values are generally determined by analytical methods capable <strong>of</strong> thoroughly<br />
characterizing the whole-rock abundance <strong>of</strong> elements, the inability <strong>of</strong> the carbonate<br />
decomposition method to dissolve refractory silicate minerals results in ‘poor’<br />
analytical results for elements typically hosted within these minerals (e.g., Zr, Nb, Hf,<br />
Ta, Th, among others). It is therefore necessary to first examine data obtained using<br />
the HF-HClO 4 whole-rock decomposition method, as these data are most comparable<br />
to the average reference value calculated from literature sources.<br />
The most commonly utilized carbonate CRM is the dolomite JDo-1, issued by<br />
the Geological Survey <strong>of</strong> Japan (GSJ). The JDo-1 dolomite is 34.0% CaO <strong>and</strong> 18.5%<br />
MgO, with only trace amounts <strong>of</strong> Al, Mn, or Fe. Therefore, major element<br />
interferences would be expected from the high Ca <strong>and</strong> Mg contents, <strong>and</strong> should<br />
primarily affect determinations <strong>of</strong> Co, Ni, Zr, <strong>and</strong> Nb. Figure 19 presents comparisons<br />
between concentrations determined at JUB using the HF-HClO 4 decomposition<br />
method with the average reference values. Determinations <strong>of</strong> Co <strong>and</strong> Ni are not shown<br />
in Fig. 19, as the elements are severely compromised in HCl acid matrices by<br />
interferences generated from Mg <strong>and</strong> Ca (Table 4). The JUB measured Co <strong>of</strong> 1.46<br />
mg/kg is several times higher than the average reference value <strong>of</strong> 0.234 mg/kg, <strong>and</strong><br />
measured Ni is 11.1 mg/kg, also several times higher than the reference value <strong>of</strong> 2.9<br />
mg/kg.<br />
Of the remaining elements Ti, Sr, Y, the REE, <strong>and</strong> W show good agreement<br />
with the average reference values, <strong>and</strong> the reference values themselves are consistent<br />
44
1.50 JDo-1 (n=5)<br />
1.25<br />
JUB / reference average<br />
1.00<br />
0.75<br />
0.50<br />
0.25<br />
0.00<br />
Sc Ti Co Ni Rb Sr Y Zr Nb Mo Cs Ba La Ce Pr Nd Sm Eu Gd Tb Dy Ho Er Tm Yb Lu Hf Ta W Pb Th U<br />
Figure 19. Accuracy estimation for ICPMS analysis <strong>of</strong> dolomite JDo-1, where n equals the number <strong>of</strong><br />
separate HF-HClO 4 sample decompositions <strong>and</strong> ICPMS analyses in 0.5 M HCl. Grey bars represent<br />
variability in the calculated ratio that is due solely to uncertainty in the reference average. Vertical lines<br />
represent the variability in the calculated ratio due to the uncertainty <strong>of</strong> the JUB data, <strong>and</strong> all data are<br />
presented in App. 1. Data for Co <strong>and</strong> Ni not shown due to large MgCl <strong>and</strong> CaO(H) interferences. The<br />
reference averages for several elements display very large uncertainties (Sc, Rb, Zr, Ba, <strong>and</strong> Pb, see<br />
App. 1). The JUB data for Sc is close to the IQL (Fig. 4), <strong>and</strong> Sc measurements at the concentration<br />
levels reported for JDo-1 are unlikely to be reliable.<br />
<strong>and</strong> calculated from a relatively large number <strong>of</strong> published studies (App. 1).<br />
Exceptions are Ti <strong>and</strong> W, as previously published data are only available from Imai et<br />
al. (1996), with Ti reported as 0.00133% TiO 2 . The observed discrepancies for other<br />
elements are variously attributed to either anomalously high reference data from<br />
individual studies that skew the average reference value (Sc, Rb, Zr, <strong>and</strong> Ba), or to a<br />
lack <strong>of</strong> published reference data (Nb, Cs, Ta). For Sc, concentrations in JDo-1 are<br />
similar to the IQL in 0.5 M HCl (Table 4), indicating that accurate Sc measurements<br />
in JDo-1 are likely to prove difficult. With respect to Zr <strong>and</strong> Ba the measured JUB<br />
concentrations agree well with at least two other published studies (see App. 1).<br />
The elements which display the greatest discrepancies, either as uncertainty in<br />
reference values or as deviation <strong>of</strong> the measured JUB concentration from the average<br />
<strong>of</strong> published data, are generally incompatible in carbonate minerals. As the JDo-1<br />
dolomite is a very pure marine chemical precipitate, these incompatible elements<br />
(e.g., Sc, Rb, Zr, Nb, Cs, Hf, Ta, Th) are likely to be concentrated in discrete,<br />
45
1.50<br />
HF-HClO 4<br />
carbonate (HNO 3<br />
)<br />
1.25<br />
JUB / reference average<br />
1.00<br />
0.75<br />
0.50<br />
0.25<br />
0.00<br />
Sc Ti Co Ni Rb Sr Y Zr Nb Mo Cs Ba La Ce Pr Nd Sm Eu Gd Tb Dy Ho Er Tm Yb Lu Hf Ta W Pb Th U<br />
Figure 20. Comparison <strong>of</strong> analytical accuracy relative to average reference values for both HF-HClO 4<br />
<strong>and</strong> carbonate (HNO 3 ) decompositions <strong>of</strong> the JDo-1 dolomite. Based upon similar results for the two<br />
decomposition methods, Sr, Y, the REE, W, Th, <strong>and</strong> U are considered to be quantifiable using the<br />
carbonate decomposition method. Additionally, note the excellent results for Ni determined from the<br />
carbonate decomposition method, which were calculated using the 62 Ni isotope. Elements likely to be<br />
hosted in refractory aluminosilicate phases that resist complete dissolution using the carbonate<br />
decomposition include Ti, Rb, Zr, Nb, Cs, Hf, <strong>and</strong> Ta.<br />
refractory mineral phases. If that is the case, then these refractory minerals may not be<br />
homogenously distributed within the JDo-1 powder, <strong>of</strong>fering a possible explanation<br />
for the wide variation in reported concentrations for incompatible elements.<br />
Compared to the HF-HClO 4 decomposition method the carbonate<br />
decomposition method is not expected to completely dissolve all mineral phases (e.g.,<br />
aluminosilicates), <strong>and</strong> data for JDo-1 obtained using the two different decomposition<br />
methods are presented in Figure 20. Regarding the carbonate decomposition <strong>of</strong> JDo-1,<br />
<strong>and</strong> many other carbonate samples as well, a black residue is commonly filtered out <strong>of</strong><br />
the sample solution, <strong>and</strong> this residue is presumed to be refractory organic carbon. The<br />
black solid retained during the filtering step does not appear to be <strong>of</strong> significance for<br />
the determination <strong>of</strong> Ni, Sr, Y, the REE, W, Th, <strong>and</strong> U, <strong>and</strong> it is concluded that these<br />
elements are hosted primarily by carbonate minerals, <strong>and</strong> therefore are quantifiable<br />
using the carbonate decomposition method. The accurate quantification <strong>of</strong> Ni using<br />
the carbonate method is notable, as this is only possible in an HNO 3 acid matrix <strong>and</strong><br />
by monitoring the 62 Ni isotope. Unlike 60 Ni, which is generally not suitable for Ni<br />
46
determinations in carbonate rocks regardless <strong>of</strong> the acid matrix, concentration<br />
determinations using 62 Ni provide excellent results for samples that are diluted in<br />
HNO 3 , primarily through a reduction in the 25 Mg 37 Cl interference.<br />
Elements that appear to be hosted primarily in refractory mineral phases<br />
resistant to decomposition using the HNO 3 carbonate method include Ti, Rb, Zr, Nb,<br />
Cs, Hf, <strong>and</strong> Ta, which would be expected for these incompatible elements. Two<br />
elements that may be concentrated in different mineral phases are Ba <strong>and</strong> Pb (C- <strong>and</strong><br />
S-rich minerals?), for which the carbonate decomposition recovers approximately<br />
90% <strong>and</strong> 80%, respectively, <strong>of</strong> the concentrations observed for the HF-HClO 4<br />
decomposition method.<br />
As stated above, it appears that the carbonate decomposition provides<br />
satisfactory results for W. However, W is one <strong>of</strong> three elements apparently ‘lost’<br />
during the analytical recovery test <strong>of</strong> the carbonate decomposition method, along with<br />
Ta, <strong>and</strong> to a lesser extent Nb (see Section 6, Fig. 10). The low analytical recovery for<br />
Nb, Ta, <strong>and</strong> W are attributed to the filtering step <strong>of</strong> the carbonate decomposition. For<br />
Nb <strong>and</strong> Ta, the poor analytical recoveries during the carbonate decomposition method<br />
are inconsequential, as regardless, sample dissolution in HNO 3 is not expected to<br />
provide accurate determinations <strong>of</strong> these elements. In the case <strong>of</strong> W, the observed loss<br />
during the filtering step is estimated at ~0.002 mg/kg (Fig. 10), which is also<br />
inconsequential when considering the reference concentration <strong>of</strong> W in the JDo-1<br />
dolomite <strong>of</strong> ~0.260 mg/kg. It is therefore unlikely that any loss <strong>of</strong> W during the<br />
filtering step for the carbonate decomposition would significantly affect W<br />
determinations, assuming W is present at concentrations ≥0.050 mg/kg in the sample<br />
powder. This is particularly true when considering the uncertainty in the JUB<br />
measured W data for JDo-1 (~15% RSD).<br />
7.6. Marine ferromanganese nodules <strong>and</strong> crusts<br />
Samples which comprise a large fraction <strong>of</strong> the geochemical research<br />
performed with the Geochemistry Lab at JUB are marine ferromanganese nodules <strong>and</strong><br />
crusts (referred to collectively as Fe-Mn crusts). These Fe-Mn crusts are typically 13-<br />
15% SiO 2 , 3-5% Al 2 O 3 , ~3% each CaO <strong>and</strong> MgO, 33-38% MnO, <strong>and</strong> 8-15% Fe 2 O 3 .<br />
Therefore, the high metal content <strong>and</strong> interferences from Mn <strong>and</strong> Fe are expected to<br />
provide the greatest obstacles to accurate trace metal determinations. However, unlike<br />
other metal-rich rocks such as iron-formations, Fe-Mn crusts are highly enriched in<br />
47
many trace metals, <strong>and</strong> therefore may be diluted significantly following<br />
decomposition. For example, the combined REY concentration within the Fe-Mn<br />
crust JMn-1 is >800 mg/kg, whereas the iron-formation IF-G contains only ~22 mg/kg<br />
REY. While IF samples typically may not be diluted by a factor greater than 1000, Fe-<br />
Mn crust analyses may benefit from higher dilution factors (≥5000).<br />
Several Fe-Mn crust CRMs are available for geochemical studies at JUB, <strong>and</strong><br />
include JMn-1 (Table 1), NOD-A-1 <strong>and</strong> NOD-P-1 (USGS), <strong>and</strong> GSPN-2 <strong>and</strong> GSPN-3<br />
(Institute <strong>of</strong> Rock <strong>and</strong> Mineral Analysis, People’s Republic <strong>of</strong> China). For this<br />
discussion only the JMn-1 ferromanganese crust issued by the Geological Survey <strong>of</strong><br />
Japan is considered, though additional information regarding analyses at JUB <strong>of</strong> other<br />
Fe-Mn crust CRMs is available from K. Schmidt.<br />
Figure 21 displays comparisons between measured JUB trace metal<br />
concentrations in JMn-1 with the average reference value. Measured concentrations<br />
are lower than the average reference value for all analyzed elements, with JUB data<br />
for 24 <strong>of</strong> 32 elements between 80-95% <strong>of</strong> the reference value. The precision <strong>of</strong> the<br />
JUB is generally quite good (2-3 % RSD), suggesting that the low measured values<br />
are reproducible <strong>and</strong> do not result from a single anomalous analysis. As discussed<br />
earlier, at high TDS contents samples may suffer from signal suppression effects<br />
(Section 4.2), which may result in low measured concentrations. However, this does<br />
not appear to be the case with JMn-1, as all <strong>of</strong> the JUB analyses were performed at<br />
dilution factors <strong>of</strong> 2500-5000, <strong>and</strong> the internal st<strong>and</strong>ard correction factors for these<br />
analyses were typically between 0.90–1.10.<br />
Rather, it is hypothesized that absorption <strong>of</strong> water vapor by JMn-1 sample<br />
powder is primarily responsible for the low measured concentrations. For this study,<br />
the JMn-1 powder was not dried prior to analysis, as for most CRMs drying the rock<br />
powder (e.g., overnight at 105-110°C) has not proven to significantly affect the<br />
measured concentrations <strong>of</strong> the elements <strong>of</strong> interest. However, Fe-Mn crusts are<br />
microcrystalline marine precipitates composed <strong>of</strong> poorly ordered mineral phases that,<br />
unlike other rock CRMs, have never undergone mineral recrystallization during<br />
diagenesis <strong>and</strong> lithification. As a result it is possible that powdered Fe-Mn crusts may<br />
more readily absorb water from ambient laboratory air (even though all CRMs are<br />
stored in dessicators until ready for use), <strong>and</strong> any increase in sample mass due to this<br />
effect would drive measured concentrations <strong>of</strong> trace metals towards lower values.<br />
This hypothesis is supported by detailed studies <strong>of</strong> Fe-Mn crusts by K. Schmidt<br />
48
1.50<br />
JMn-1 (n=4)<br />
1.25<br />
JUB / reference average<br />
1.00<br />
0.75<br />
0.50<br />
0.25<br />
0.00<br />
Sc Ti Co Ni Rb Sr Y Zr Nb Mo Cs Ba La Ce Pr Nd Sm Eu Gd Tb Dy Ho Er Tm Yb Lu Hf Ta W Pb Th U<br />
Figure 21. Accuracy estimation for ICPMS analysis <strong>of</strong> Fe-Mn nodule JMn-1, where n equals the<br />
number <strong>of</strong> separate HF-HClO 4 sample decompositions <strong>and</strong> ICPMS analyses in 0.5 M HCl. Grey bars<br />
represent variability in the calculated ratio that is due solely to uncertainty in the reference average.<br />
Vertical lines represent the variability in the calculated ratio due to the uncertainty <strong>of</strong> the JUB data, <strong>and</strong><br />
all data are presented in App. 1. The generally low concentrations for many elements as measured at<br />
JUB are hypothesized to result from the adsorption <strong>of</strong> water vapor by the JMn-1 powder, which would<br />
effectively dilute the sample <strong>and</strong> lower the measured concentrations (see text for details).<br />
(personal communication), who observed that concentrations measured in non-dried<br />
JMn-1 powders were typically 10-20% lower than concentrations determined for<br />
dried JMn-1 powders.<br />
7.7. Basalts<br />
The last rock type frequently used in geochemical studies at JUB is basalt,<br />
which is representative <strong>of</strong> oceanic crust. Basaltic rocks are relatively Al-poor, Mg<strong>and</strong><br />
Fe-rich igneous rocks that have formed throughout Earth’s history, <strong>and</strong> have been<br />
the focus <strong>of</strong> numerous geochemical studies related to crustal differentiation, hot spot<br />
volcanism, seawater-crust interaction, etc. As a result <strong>of</strong> this research attention<br />
basaltic CRMs are very well characterized, with numerous published studies reporting<br />
data for a large number <strong>of</strong> trace metals.<br />
One <strong>of</strong> the best characterized basalt CRMs is BHVO-2, a Hawaiian basalt<br />
issued by the USGS. The BHVO-2 basalt is 13.5% Al 2 O 3 , 7.2% MgO, 11.4% CaO,<br />
0.17% MnO, <strong>and</strong> 12.3% Fe 2 O 3 . Based on these major element abundances,<br />
49
1.50 BHVO-2 (n=7)<br />
1.25<br />
JUB / reference average<br />
1.00<br />
0.75<br />
0.50<br />
0.25<br />
0.00<br />
Sc Ti Co Ni Rb Sr Y Zr Nb Mo Cs Ba La Ce Pr Nd Sm Eu Gd Tb Dy Ho Er Tm Yb Lu Hf Ta W Pb Th U<br />
Figure 22. Accuracy estimation for ICPMS analysis <strong>of</strong> basalt BHVO-2, where n equals the number <strong>of</strong><br />
separate HF-HClO 4 sample decompositions <strong>and</strong> ICPMS analyses in 0.5 M HCl. Grey bars represent<br />
variability in the calculated ratio that is due solely to uncertainty in the reference average. Vertical<br />
lines represent the variability in the calculated ratio due to the uncertainty <strong>of</strong> the JUB data, <strong>and</strong> all data<br />
are presented in App. 1. Of all the CRMs analyzed at JUB, data for BHVO-2 basalt most closely<br />
matches the average reference values, <strong>and</strong> does so for the greatest number <strong>of</strong> elements.<br />
interferences would be expected on Co <strong>and</strong> Ni (from MgCl, AlCl, <strong>and</strong> CaO(H)), <strong>and</strong><br />
to a lesser extent on Nb from FeCl. However, concentrations <strong>of</strong> Co, Ni, <strong>and</strong> Nb in<br />
BHVO-2 are significant (46.0, 119, <strong>and</strong> 17.5 mg/kg, respectively), <strong>and</strong> major element<br />
interferences do not appear to affect analytical accuracy for Co, Ni, <strong>and</strong> Nb (Figure<br />
22).<br />
Of the various CRMs analyzed at JUB, data for BHVO-2 most closely<br />
matches the average reference values, <strong>and</strong> does so for the greatest number <strong>of</strong><br />
elements. Of the 32 elements measured, 30 are within ±10% <strong>of</strong> the reference value,<br />
<strong>and</strong> 26 are within ±5% (Fig. 22). The excellent agreement between measured <strong>and</strong><br />
reference data is attributed to the fact that BHVO-2 contains significantly higher<br />
concentrations <strong>of</strong> many elements that display ‘poor’ accuracy for a number <strong>of</strong> CRMs<br />
(e.g., Co, Ni, Zr, Nb, Hf, Ta, Th), <strong>and</strong> these higher concentrations facilitate<br />
geochemical analyses. Additionally, the numerous published studies for a large<br />
number <strong>of</strong> trace metals in BHVO-2 ensures a robust dataset for calculating an average<br />
50
eference value. The result is that for many elements the uncertainty in the average<br />
reference value is similar to that observed for the JUB data (2-5% RSD, App. 1).<br />
The close agreement illustrated for most elements in Fig. 22 is not observed<br />
for Ti in BHVO-2, which has consistently produced low measured values in the JUB<br />
data compared to the compiled literature value (App. 1). The reason for this behavior<br />
is not yet clear, yet it is hypothesized that it may result from a combination <strong>of</strong> the high<br />
Ti content in BHVO-2 <strong>and</strong> the relatively low 10 μg/kg Ti calibration st<strong>and</strong>ard used for<br />
determination <strong>of</strong> Ti concentrations. The ionization efficiency <strong>of</strong> Ti is quite low, <strong>and</strong><br />
the 10 μg/kg Ti calibration st<strong>and</strong>ard produces an ICPMS signal intensity <strong>of</strong> only<br />
8000-10,000 cps, relative to rather high blank intensities <strong>of</strong> 500-800 cps due to CCl<br />
<strong>and</strong> NO 2 interferences (Table 4). In comparison, the abundance <strong>of</strong> Ti in BHVO-2<br />
routinely produces 7-9 million cps at dilution factors <strong>of</strong> 1000. A benefit <strong>of</strong> ICPMS<br />
instruments is that they frequently <strong>of</strong>fer linear signal responses for many elements at<br />
concentration ranges that exceed six orders <strong>of</strong> magnitude. However, in the case <strong>of</strong> Ti,<br />
the combination <strong>of</strong> blank intensities roughly one order <strong>of</strong> magnitude less than the<br />
calibration st<strong>and</strong>ard intensities, which are themselves approximately three orders <strong>of</strong><br />
magnitude less than the sample intensity, may produce a non-linear instrument<br />
response. The fact that good results are observed in those CRMs that contain ~1000-<br />
1500 mg/kg Ti (FeR-2 <strong>and</strong> SGR-1b), with poorer results at higher Ti abundances<br />
(e.g., >6000 mg/kg as in JMn-1), <strong>of</strong>fers indirect support <strong>of</strong> this hypothesis. However,<br />
further investigation is warranted, <strong>and</strong> an approach <strong>of</strong> higher dilution factors for<br />
BHVO-2 (2500-5000), coupled with an increase in the Ti concentration <strong>of</strong> the<br />
calibration st<strong>and</strong>ard (50 μg/kg?) is recommended.<br />
7.8. Rare earth element ratios<br />
A significant emphasis <strong>of</strong> much <strong>of</strong> the research conducted within the <strong>Jacobs</strong><br />
<strong>University</strong> Geochemistry Lab regards the behavior <strong>of</strong> the rare earth elements <strong>and</strong><br />
yttrium in geological samples, including rocks <strong>and</strong> minerals, as well as natural water<br />
samples such as seawater, hydrothermal fluids, <strong>and</strong> ground <strong>and</strong> river waters.<br />
Analytical studies <strong>of</strong> the REY are generally more focused upon the relative<br />
distribution <strong>of</strong> these elements within any given sample, rather than their<br />
concentrations, <strong>and</strong> the REY plot in Fig. 14 is a typical manner <strong>of</strong> presenting REY<br />
data. Frequently, the shape <strong>of</strong> REY patterns for interpreting geological samples are<br />
51
Table 5. Select REY ratios <strong>and</strong> chondrite-normalized REE<br />
anomalies as calculated from reference data <strong>and</strong> JUB data for<br />
BHVO-2 basalt.<br />
REY ratio avg. ref. data 1 JUB data JUB/ref. avg.<br />
Pr/Sm 0.8885 0.8691 0.978<br />
Pr/Dy 1.017 1.000 0.983<br />
Pr/Yb 2.711 2.726 1.006<br />
Sm/Dy 1.145 1.151 1.005<br />
Sm/Yb 3.051 3.137 1.028<br />
Dy/Yb 2.665 2.726 1.023<br />
Y/Ho 26.44 25.31 0.957<br />
Ce/Ce* 2 1.00 1.01 0.99<br />
Ce/Ce* 3 0.933 0.964 0.97<br />
Eu/Eu* 4 1.00 1.01 0.99<br />
Eu/Eu* 5 1.01 1.02 0.99<br />
1 average reference data (see Appendix 1).<br />
2 calculated as Ce/Ce* = Ce/(0.5La+0.5Pr).<br />
3 calculated as Ce/Ce* = Ce/(2Pr-1Nd) after Bolhar et al. (2004).<br />
4 calculated as Eu/Eu* = Eu/(0.5Sm+0.5Gd).<br />
5 calculated as Eu/Eu* = Eu/(0.67Sm+0.33Tb).<br />
more useful than the absolute REY abundances, because fractionation between REY<br />
elements (<strong>and</strong> the specific REY patterns that result), may <strong>of</strong>fer clues to the processes<br />
responsible for the formation <strong>and</strong> subsequent history <strong>of</strong> many geological samples.<br />
Since the relative distribution <strong>of</strong> the REY is <strong>of</strong>ten more informative for<br />
interpreting geochemical data, it is useful to examine the ability <strong>of</strong> the described<br />
analytical methods in accurately determining the REY patterns in geological<br />
materials. As this study has focused upon the analyses <strong>of</strong> CRMs, <strong>and</strong> the BHVO-2<br />
basalt is considered the best characterized reference material with respect to the REY,<br />
the discussion is limited to relative REY ratios in BHVO-2.<br />
Table 5 lists REY ratios that characterize the relative distributions <strong>of</strong> the light<br />
rare earth elements (Pr/Sm), the middle rare earth elements (Sm/Dy), <strong>and</strong> the heavy<br />
rare earth elements (Dy/Yb) for the BHVO-2 basalt. Ratios that include La, Ce, Gd,<br />
Eu, <strong>and</strong> Lu are not used, as these elements may display anomalous behavior in some<br />
geological samples (e.g., La <strong>and</strong> Ce in seawater, <strong>and</strong> Eu in hydrothermal fluids <strong>and</strong><br />
plagioclase-rich rocks). Data used for calculating all REY ratios are tabulated in<br />
Appendix 1. No consideration <strong>of</strong> the relative errors in determining the concentrations<br />
<strong>of</strong> individual REY elements as reported in Appendix 1 is present in the data <strong>of</strong> Table<br />
5, as for many <strong>of</strong> the elements the %RSD is very similar between the average<br />
52
eference value <strong>and</strong> the measured JUB data (e.g., 3.5% <strong>and</strong> 2.7% for Yb,<br />
respectively). Relative to the REY ratios calculated from the average reference data,<br />
the JUB ratios are accurate to within 1% for Pr/Yb, 2% for Pr/Dy, 3% for Pr/Sm,<br />
Sm/Yb, <strong>and</strong> Dy/Yb, <strong>and</strong> to within 5% for Y/Ho. Considering the well-characterized<br />
nature <strong>of</strong> the BHVO-2 basalt <strong>and</strong> the large number <strong>of</strong> highly precise ICPMS analyses<br />
reported for this CRM, these values are considered the best estimates <strong>of</strong> the accuracy<br />
<strong>of</strong> the JUB analytical methods in determining REY ratios.<br />
Also presented in Table 5 are examples <strong>of</strong> calculated REE anomalies for<br />
chondrite-normalized data <strong>of</strong> Ce <strong>and</strong> Eu, two rare earth elements that may be<br />
fractionated from neighboring REE in natural systems (e.g., in seawater for Ce, <strong>and</strong> in<br />
high-temperature hydrothermal systems for Eu). Normalized REE data that is<br />
anomalous is commonly quantified by means <strong>of</strong> X/X* ratios, where X is the<br />
normalized measured concentration for a particular rare earth element, <strong>and</strong> X* is the<br />
predicted normalized concentration for that element. The predicted REE concentration<br />
X* may be calculated in various ways (e.g., Bolhar et al., 2004), <strong>and</strong> X* is commonly<br />
obtained by interpolating between adjacent REE, as in the calculation <strong>of</strong> Eu* by<br />
means <strong>of</strong> (0.5Sm + 0.5Gd). Alternatively, X* may be obtained by extrapolating from<br />
adjacent REE, as in the case <strong>of</strong> Ce* by means <strong>of</strong> (2Pr - 1Nd). Different methods for<br />
calculating chondrite-normalized Ce/Ce* <strong>and</strong> Eu/Eu* are presented in Table 5, <strong>and</strong><br />
demonstrate that REE anomalies calculated from measured JUB data accurately<br />
reproduce the anomalies as determined from the average reference data <strong>of</strong> the BHVO-<br />
2 basalt.<br />
8. Summary <strong>and</strong> conclusions<br />
The ICPMS analytical methods utilized within the JUB Geochemistry Lab are<br />
capable <strong>of</strong> reproducibly determining accurate concentrations <strong>of</strong> many trace metals in<br />
a variety <strong>of</strong> rock types. The relative precision <strong>of</strong> the ICPMS measurements is<br />
primarily controlled by the inherent ICPMS instrument precision, <strong>and</strong> not by sample<br />
preparation <strong>and</strong> decomposition techniques. On average, for all rock types<br />
decomposed using the HF-HClO 4 decomposition method, the ICPMS instrument<br />
precision (RSD) is better than 3% for eleven elements (Ti, Co, Sr, Y, Zr, Ba, La, Ce,<br />
Pr, Nd, Pb), between 3-5% for sixteen elements (Sc, Ni, Rb, Nb, Sm, Eu, Gd, Tb, Dy,<br />
Ho, Er, Tm, Yb, Lu, Th, U), <strong>and</strong> between 5-8% for only five elements (Mo, Cs, Hf,<br />
Ta, W). Certainly, the poorer precision observed for some elements is due to low<br />
53
concentrations <strong>of</strong> these elements in many rock types (e.g., Cs <strong>and</strong> Hf in JDo-1),<br />
<strong>and</strong>/or possible inhomogeneities in the CRM sample powders.<br />
The sample decomposition methods employed appear to conservatively retain<br />
all 32 elements analyzed throughout the dissolution procedures, with the only<br />
exceptions being Ni, Ta, <strong>and</strong> W during the carbonate decomposition. Greater than<br />
expected recoveries <strong>of</strong> Ni in spike solutions during the carbonate decomposition<br />
suggest a Ni contamination <strong>of</strong> 2-5 μg/kg occurs, likely during filtration <strong>of</strong> the<br />
dissolved sample solution. Analytical recoveries <strong>of</strong> Ta <strong>and</strong> W are low (24% <strong>and</strong> 80%,<br />
respectively), <strong>and</strong> observed to a lesser extent for Nb (92% recovery), <strong>and</strong> appear to<br />
reflect retention <strong>of</strong> these elements on the 0.2 μm cellulose acetate filters used during<br />
the carbonate decomposition.<br />
Some analyzed elements are significantly affected by interferences generated<br />
by high major element contents in certain rock types. The affected elements have<br />
atomic masses 10 mg/kg). Determination <strong>of</strong> low Nb contents (less than ~20 mg/kg) in Ferich<br />
samples is also likely to prove difficult, <strong>and</strong> Appendix 2 provides a thorough<br />
treatment <strong>of</strong> observed major element interferences <strong>and</strong> recommendations regarding<br />
analyses <strong>of</strong> specific rock types. The potential for reducing chloride-derived<br />
interferences that arise from the use <strong>of</strong> HCl during sample dilutions should be<br />
investigated more thoroughly. Diluting samples prior to ICPMS analyses in mixtures<br />
<strong>of</strong> HNO 3 <strong>and</strong> HCl may satisfactorily mitigate many major element interferences on<br />
low mass trace metals such as Co, Ni, Zr, Nb, <strong>and</strong> Mo.<br />
The analytical accuracy <strong>of</strong> the ICPMS measurements is considered excellent,<br />
though accuracy assessments are hampered by the lack <strong>of</strong> well constrained reference<br />
values for many trace metals in the CRMs discussed. It appears that older data<br />
compilations frequently overestimate the abundances <strong>of</strong> many trace metals, <strong>and</strong> for<br />
some elements very few published data exist (e.g., Nb, Mo, Ta, W). The best<br />
constrained CRM is the BHVO-2 basalt, for which the data produced at JUB is in<br />
excellent agreement. Based upon multiple analyses <strong>of</strong> BHVO-2 it is concluded that<br />
the JUB methods described here provide accurate element determinations for 31 <strong>of</strong><br />
54
the 32 analyzed elements, with Ti being the only exception. Element concentrations<br />
in CRMs measured at JUB also agree well with recent published studies that focused<br />
on similar trace metals, <strong>and</strong> that used similar analytical methods. While the task <strong>of</strong><br />
ensuring accuracy in analytical data is a continuous process, it is concluded that the<br />
ICPMS methods described here are suitable for reproducibly determining highly<br />
precise <strong>and</strong> accurate trace element data in a variety <strong>of</strong> geologic materials.<br />
55
References<br />
Abbey S., McLeod C.R., <strong>and</strong> Liang-Guo W. (1983) FeR-1, FeR-2, FeR-3, <strong>and</strong> FeR-4 four<br />
Canadian iron-formation samples prepared for use as reference materials. Geol. Sur.<br />
Canada paper 83-19.<br />
Baker J.A., Waight T., <strong>and</strong> Ulfbeck D. (2002) Rapid <strong>and</strong> highly reproducible analysis <strong>of</strong> rare<br />
earth elements by multiple collector inductively coupled plasma mass spectrometry.<br />
Geochim. Cosmochim. Acta, 66, 3635-3646.<br />
Barrat J.A., Boulègue J.J., Tiercelin J.-J., <strong>and</strong> Lesourd M. (2000) Strontium isotopes <strong>and</strong> rareearth<br />
element geochemistry <strong>of</strong> hydrothermal carbonate deposits from Lake<br />
Tanganyika, East Africa. Geochim. Cosmochim. Acta, 64, 287-298.<br />
Bédard L.P. <strong>and</strong> Barnes S.-J. (2002) A comparison <strong>of</strong> N-type semi-planar <strong>and</strong> coaxial INAA<br />
detectors for 33 geochemical reference samples. J. Radioanalytical Nuclear Chem.,<br />
254, 485-497.<br />
Bolhar R., Kamber B.S., Moorbath S., Fedo C.M., <strong>and</strong> Whitehouse M.J. (2004)<br />
Characterisation <strong>of</strong> early Archaean chemical sediments by trace element signatures.<br />
Earth Planet. Sci. Lett., 222, 43-60.<br />
Dai Kin F., Prudêncio M.I., Gouveia M.A., <strong>and</strong> Magnusson E. (1999) Determination <strong>of</strong> rare<br />
earth elements in geological reference materials: A comparative study by INAA <strong>and</strong><br />
ICP-MS. Geost<strong>and</strong>. Newsletter, 23, 47-58.<br />
Doherty W. (1989) An internal st<strong>and</strong>ardization procedure for the determination <strong>of</strong> yttrium <strong>and</strong><br />
the rare earth elements in geological materials by inductively coupled plasma-mass<br />
spectrometry. Spectrochim. Acta, 44B, 263-280.<br />
Dulski P. (1994) Interferences <strong>of</strong> oxide, hydroxide <strong>and</strong> chloride analyte species in the<br />
determination <strong>of</strong> rare earth elements in geological samples by inductively coupled<br />
plasma-mass spectrometry. Fresenius J. Anal. Chem., 350, 194-203.<br />
Dulski P. (2001) Reference materials for geochemical studies: new analytical data by ICP-MS<br />
<strong>and</strong> critical discussion <strong>of</strong> reference values. Geost<strong>and</strong>. Newsletter, 25, 87-125.<br />
Eggins S.M. (2003) Laser ablation ICP-MS analysis <strong>of</strong> geological materials prepared as<br />
lithium borate glasses. Geost<strong>and</strong>. Newsletter, 27, 147-162.<br />
Faure G. (1986) Principles <strong>of</strong> Isotope Geology. John Wiley <strong>and</strong> Sons, New York.<br />
Flem B. <strong>and</strong> Bédard L.P. (2002) Determination <strong>of</strong> trace elements in BCS CRM 313/1 (BAS)<br />
<strong>and</strong> NIST SRM 1830 by inductively coupled plasma-mass spectrometry <strong>and</strong><br />
instrumental neutron activation analysis. Geost<strong>and</strong>. Newsletter, 26, 287-300.<br />
Garbe-Schönberg C.-D. (1993) Simultaneous determination <strong>of</strong> thirty-seven trace elements in<br />
twenty-eight international rock st<strong>and</strong>ards by ICP-MS. Geost<strong>and</strong>. Newsletter, 17, 81-<br />
97.<br />
Govindaraju K. (1994) 1994 compilation <strong>of</strong> working values <strong>and</strong> sample descriptions for 383<br />
geost<strong>and</strong>ards. Geost<strong>and</strong>. Newsletter, 18, 1-158.<br />
56
Govindaraju K. (1995) 1995 working values with confidence limits for twenty-six CRPG,<br />
ANRT <strong>and</strong> IWG-GIT geost<strong>and</strong>ards. Geost<strong>and</strong>. Newsletter, 19 (special), 1-95.<br />
Huang S. <strong>and</strong> Frey F.A.(2003) Trace element abundances <strong>of</strong> Mauna Kea basalt from phase 2<br />
<strong>of</strong> the Hawaii Scientific Drilling Project: Petrogenetic implications <strong>of</strong> correlations<br />
with major element content <strong>and</strong> isotopic ratios. Geochem. Geophys. Geosys., 4, 1-43.<br />
Imai N., Terashima S., Itoh S., <strong>and</strong> Ando A. (1996) 1996 compilation <strong>of</strong> analytical data on<br />
nine GSJ geochemical reference samples: "Sedimentary Rock Series". Geost<strong>and</strong>.<br />
Newsletter, 20, 165-216.<br />
Imai N., Terashima S., Itoh S., <strong>and</strong> Ando A. (1999) 1998 compilation <strong>of</strong> analytical data for<br />
five GSJ geochemical reference samples: The "Instrumental Analysis Series".<br />
Geost<strong>and</strong>. Newsletter, 23, 223-250.<br />
Jochum K.P., Nohl U., Herwig K., Lammel E., Stoll B., <strong>and</strong> H<strong>of</strong>mann A.W. (2005) GeoReM:<br />
A New Geochemical Database for Reference Materials <strong>and</strong> Isotopic St<strong>and</strong>ards.<br />
Geost<strong>and</strong>. Geoanalytical. Res, 29, 333-338.<br />
Korotev R.L. (1996) A Self-Consistent compilation <strong>of</strong> elemental concentration data for 93<br />
geochemical reference samples. Geost<strong>and</strong>. Newsletter, 20, 217-245.<br />
MacDougall D., Crummett W.B., <strong>and</strong> et al. (1980) Guidelines for Data Acquisition <strong>and</strong> Data<br />
Quality Evaluation in Environmental Chemistry. Anal. Chem., 52, 2242–2249.<br />
McLennan S. (1989) Rare earth elements in sedimentary rocks: influence <strong>of</strong> provenance <strong>and</strong><br />
sedimentary processes. in Geochmemstry <strong>and</strong> Mineralogy <strong>of</strong> the Rare Earth Elements<br />
(eds. B.R. Lipin <strong>and</strong> G.A. McKay). Reviews in Mineralogy, Vol. 21.<br />
Meisel T., Schöner N., Paliulionyte V. <strong>and</strong> Kahr E. (2002) Determination <strong>of</strong> rare earth<br />
elements, Y, Th, Zr, Hf, Nb <strong>and</strong> Ta in geological reference materials G-2, G-3, SCo-1<br />
<strong>and</strong> WGB-1 by sodium peroxide sintering <strong>and</strong> inductively coupled plasma-mass<br />
spectrometry. Geost<strong>and</strong>. Newsletter, 26, 53-61.<br />
Münker C. (1998) Nb/Ta fractionation in a Cambrian arc/back arc system, New Zeal<strong>and</strong>:<br />
source constraints <strong>and</strong> application <strong>of</strong> refined ICPMS techniques. Chem. Geol., 144,<br />
23-45.<br />
Plumlee G. (1998) USGS Certificate <strong>of</strong> Analysis Basalt, Hawaiian Volcanic Observatory,<br />
BHVO-2, USGS, Boulder CO.<br />
Polat A., Li Jianghai, Fryer B.J., Kusky T., Gagnon J., <strong>and</strong> Zhang S. (2006) Geochemical<br />
characteristics <strong>of</strong> the Neoarchean (2800-2700 Ma) Taishan greenstone belt, North<br />
China Craton: Evidence for plume-craton interaction. Chem. Geol., 230, 60-87<br />
Ryder C.H., Gill J., Tepley III F., Ramos F., <strong>and</strong> Reagan M. K. (2006) Closed- to opensystem-differentiation<br />
at Arenal volcano (1968-2003). J. Volcano. Geotherm. Res.,<br />
157, 75-93.<br />
Schramm B., Devey C.W., Gillis K.M., <strong>and</strong> Lackschewitz K. (2005) Quantitative assessment<br />
<strong>of</strong> chemical <strong>and</strong> mineralogical changes due to progressive low-temperature alteration<br />
<strong>of</strong> East Pacific Rise basalts from 0 to 9 Ma. Chem. Geol., 218, 281-313.<br />
57
Shafer J.T., Neal C.R., <strong>and</strong> Regelous M. (2005) Petrogenesis <strong>of</strong> Hawaiian postshield lavas:<br />
Evidence from Nintoku Seamount, Emperor Seamount Chain. Geochem. Geophys.<br />
Geosys., 6, Q05L09, doi:10.1029/2004GC000875.<br />
Smith D.B. (1995) USGS Certificate <strong>of</strong> Analysis Cody Shale, SCo-1, United States Geologic<br />
Survey, Boulder, CO.<br />
Terashima S., Usui A., <strong>and</strong> Imai N. (1995) Two new GSJ geochemical reference samples:<br />
Syenite JSy-1 <strong>and</strong> manganese nodule JMn-1. Geost<strong>and</strong>. Newsletter, 19, 221-229.<br />
Thomas R. (2003) Practical Guide to ICPMS. Practical Spectroscopy Series, Taylor <strong>and</strong><br />
Francis, London.<br />
Thompson J.J <strong>and</strong> Houk R.S. (1987) A Study <strong>of</strong> Internal St<strong>and</strong>ardization in Inductively<br />
Coupled Plasma-Mass Spectrometry. Appl. Spectroscopy, 41, 801-806.<br />
Willbold M. <strong>and</strong> Jochum K.P. (2005) Multi-element isotope dilution sector field ICP-MS: A<br />
precise technique for the analysis <strong>of</strong> geological materials <strong>and</strong> is application to<br />
geological reference materials. Geost<strong>and</strong>. Geoanalytical. Res., 29, 63-82.<br />
Wilson S.A.(2001) USGS Certificate <strong>of</strong> Analysis Green River Shale, SGR-1. United States<br />
Geologic Survey, Boulder CO.<br />
Yamamoto Koshi, Itoh N., Matsumoto T., Tanaka Tsuyoshi, <strong>and</strong> Adachi M. (2004)<br />
Geochemistry <strong>of</strong> Precambrian carbonate intercalated in pillows <strong>and</strong> its host basalt:<br />
implications for the REE composition <strong>of</strong> circa 3.4Ga seawater. Precam. Res., 135,<br />
331-344.<br />
Yu Z., Robinson P., <strong>and</strong> McGoldrick P. (2001) An evaluation <strong>of</strong> methods for the chemical<br />
decomposition <strong>of</strong> geological materials for trace element determination using ICP-MS.<br />
Geost<strong>and</strong>. Newsletter, 25, 199-217.<br />
58
Appendix 1. Analytical data<br />
Appendix 1 contains the literature data used to calculate the average CRM<br />
reference values discussed in Section 7 regarding analytical accuracy. The average<br />
measured trace metal concentrations as determined within the JUB Geochemistry Lab<br />
for the various CRMs are also presented. As discussed within the main text <strong>of</strong> this<br />
study, some isotopes with atomic masses
for JMn-1 as described in Section 7.6 preclude definitive statements as to the reason<br />
for this discrepancy, <strong>and</strong> further studies <strong>of</strong> this effect in JMn-1 are recommended.<br />
Concentrations <strong>of</strong> Zr as reported for all CRMs in this study have been<br />
determined solely using 90 Zr. Interferences on the 91 Zr isotope primarily result from<br />
56 M 35 Cl interferences, where M can be Fe, CaO(H), or ArO, <strong>and</strong> the use <strong>of</strong> HNO 3 for<br />
diluting samples will reduce these interferences. Good results using 91 Zr in an HCl<br />
acid matrix appear to be limited to samples which contain more than ~40 mg/kg Zr,<br />
<strong>and</strong> that contain less than ~10% Fe 2 O 3 . The use <strong>of</strong> 91 Zr for quantifying Zr should be<br />
used with caution, <strong>and</strong> is likely to prove most useful only for samples diluted in<br />
HNO 3 , or for those samples diluted in HCl for which the Mn content is high enough<br />
to produce significant 55 Mn 35 interferences that preclude the use <strong>of</strong> 90 Zr for<br />
quantifying Zr (see App. 2).<br />
60
Appendix 1. Literature reference values <strong>and</strong> measured element concentration data for CRMs analyzed within JUB Geochemistry Lab. Data in mg/kg.<br />
Govindaraju<br />
(1994)<br />
Dulski<br />
(2001)<br />
Yu et al.<br />
(2001)<br />
FeR-2 FeR-4 IF-G 4<br />
Abbey et al.<br />
(1983)<br />
this<br />
study<br />
Govindaraju<br />
(1994)<br />
Dulski<br />
(2001)<br />
Yu et al.<br />
(2001)<br />
Abbey et al.<br />
(1983)<br />
method 1 compiled ICPMS ICPMS compiled ref. avg. 2 %RSD JUB avg. 3<br />
(n=3)<br />
%RSD compiled ICPMS ICPMS compiled ref. avg. %RSD JUB avg.<br />
(n=2)<br />
%RSD compiled ICPMS ICPMS ID MC-<br />
ICPMS<br />
ICPMS ref. avg. %RSD JUB avg.<br />
(n=2)<br />
Sc 6 5.31 6 5.77 5.6 5.34 3.2 1.5 1.09 1.5 1.36 14.2 0.875 0.3 0.2846 0.2923 2.6
Appendix 1 continued. Data in mg/kg.<br />
Smith<br />
(1995)<br />
Korotev<br />
(1996)<br />
Dai Kin<br />
(1999)<br />
SCo-1 SGR-1b JCh-1<br />
Dulski<br />
(2001)<br />
Bédard<br />
<strong>and</strong> Barnes<br />
(2002)<br />
Meisel<br />
et al.<br />
(2002)<br />
Eggins<br />
(2003)<br />
this<br />
study<br />
Govindaraju Korotev Dai Kin<br />
(1994) (1996) (1999)<br />
Wilson<br />
(2001)<br />
Dulski<br />
(2001)<br />
this<br />
study<br />
Govindaraju<br />
(1994)<br />
Imai<br />
et al.<br />
(1996)<br />
Dulski Flem <strong>and</strong><br />
(2001) Bédard<br />
(2002)<br />
method 1 compiled INAA ICPMS ICPMS INAA ICPMS LA- ref. avg. 2 %RSD JUB avg. 3 %RSD compiled INAA ICPMS compiled ICPMS ref. avg. %RSD JUB avg. %RSD compiled compiled ICPMS INAA ref. avg. %RSD JUB avg. %RSD<br />
ICPMS<br />
(n=4)<br />
(n=3)<br />
(n=3)<br />
Sc 11 11.57 11.53 12.7 11.70 5.3 12.8 13.7 4.6 4.94 4.6 4.71 3.4 5.01 2.6 0.85 0.979 1 0.943 7.0 0.931 6.3<br />
Ti 3776 3561 3669 2.9 4080 17.6 1580 1520 1550 1.9 1530 7.9 180 189 185 2.4 174 6.5<br />
Co 11 11.13 11.6 11.24 2.3 12.7 15.1 11.8 11.73 12 11.84 1.0 12.2 3.3 15 15.5 14.4 15.0 3.0 16.5 5.5<br />
Ni 27 24 26 5.9 30.3 22.6 29 26 29 28 5.1 30.3* 5.0 7.5 8.76 8.13 7.7 10.0 28.9<br />
Rb 110 111.5 113 106 111.7 110.4 2.2 113 8.4 83 79 80 81 2.1 82.0 4.5 8.5 8.61 8.6 8.3 8.50 1.5 8.90 0.1<br />
Sr 170 175 169 172.7 171.7 1.4 168 4.3 420 393 420 381 404 4.2 391 3.7 4.6 4.2 4.2 4.3 4.4 4.03 1.6<br />
Y 26 22.8 24.2 26.8 25.0 6.2 22.8 4.4 13 13 9.6 11.9 13.5 9.74 2.2 1.84 1.81 1.8 1.82 0.9 1.77 1.6<br />
Zr 160 166 153 116 165 188.6 158.1 13.8 157 6.6 53 45 53 42 48 10.1 43.7 3.5 11.7 11.5 6.2 9.8 26.0 6.60 3.2<br />
Nb 11 11.6 13.2 11.9 7.8 11.8 4.3 5.2 5.2 5.2 0.0 6.41 1.1 1.7 1.7 0.590 1.5<br />
Mo 1.4 1.4 1.22 5.7 35.1 35 35.1 0.1 34.1 9.2 0.245 10.5<br />
Cs 7.8 7.77 7.7 7.4 7.46 7.63 2.2 7.67 4.1 5.2 5.19 5.2 5.1 5.17 0.8 5.13 2.2 0.3 0.243 0.288 0.29 0.280 7.8 0.277 5.7<br />
Ba 570 559 578 567 592 573 2.0 562 4.3 290 270 290 265 279 4.1 278 1.6 302 302 291 298 1.7 276 4.5<br />
La 30 29.3 28.9 29 30.3 29.7 31.1 29.8 2.4 28.1 3.9 20.3 18.4 18.2 20 18 19.0 5.1 18.2 2.8 1.5 1.52 1.44 1.44 1.48 2.4 1.34 4.8<br />
Ce 62 56.7 57.3 57 60.3 57.5 58.8 58.5 3.1 56.1 3.7 36 34.4 34 36 33.8 34.8 2.8 34.0 3.3 4.72 5.21 4.7 4.66 4.82 4.7 4.42 4.5<br />
Pr 6.6 6.7 7.1 7 6.9 3.0 6.69 3.6 3.9 3.85 4 3.92 1.6 3.91 2.7 0.5 0.37 0.44 14.9 0.335 4.6<br />
Nd 26 25.4 28.4 26 24 27 27.3 26.3 5.0 25.7 3.0 15.5 13.8 13.7 16 13.8 14.6 6.8 14.2 2.4 1.7 2.05 1.41 1.4 1.64 16.2 1.32 4.4<br />
Sm 5.14 5.2 5.0 5.12 5.31 5.52 5.22 3.2 4.96 3.5 2.7 2.6 1.9 2.7 2.42 2.46 12.2 2.52 2.0 0.4 0.359 0.30 0.35 0.352 10.1 0.281 4.0<br />
Eu 1.088 1.3 1.13 1.11 1.17 1.18 1.163 5.9 1.10 3.4 0.56 0.466 0.48 0.56 0.47 0.507 8.5 0.466 2.1 0.08 0.0594 0.060 0.1 0.0749 22.3 0.0484 3.0<br />
Gd 5.3 4.6 4.72 4.8 4.86 5.5 4.53 6.0 2 1.9 2 2.05 1.99 2.7 2.04 1.5 0.304 0.304 0.0 0.286 2.4<br />
Tb 0.675 0.72 0.69 0.69 0.7 0.70 2.1 0.675 3.1 0.36 0.297 0.37 0.3 0.332 10.1 0.295 1.2 0.09 0.0385 0.046 0.04 0.0536 39.5 0.0438 2.5<br />
Dy 4.2 4.0 4.3 4.43 4.23 3.7 4.01 2.7 1.9 1.8 1.9 1.71 1.83 4.3 1.73 0.9 0.4 0.378 0.289 0.356 13.5 0.273 2.4<br />
Ho 0.76 0.8 0.86 0.81 5.1 0.803 2.5 0.38 0.38 0.4 0.34 0.38 5.8 0.347 0.7 0.1 0.060 0.080 25.0 0.0567 2.1<br />
Er 2.2 2.38 2.53 2.68 2.45 7.3 2.35 1.6 1.11 1.2 1.1 0.99 1.10 6.8 1.01 1.0 0.3 0.233 0.174 0.236 21.8 0.167 1.8<br />
Tm 0.4 0.35 0.36 0.37 5.8 0.334 4.8 0.17 0.18 0.17 0.147 0.167 7.3 0.144 1.6 0.025 0.025 0.0240 3.6<br />
Yb 2.27 1.7 2.34 2.37 2.57 2.25 13.0 2.23 1.4 0.94 0.966 1.1 0.94 0.96 0.981 6.1 0.949 1.0 0.1 0.182 0.171 0.16 0.153 20.7 0.162 2.8<br />
Lu 0.341 0.31 0.36 0.35 0.37 0.388 0.353 6.9 0.345 3.3 0.14 0.146 0.14 0.146 0.143 2.1 0.146 1.7 0.04 0.0344 0.026 0.023 0.0309 21.8 0.0269 6.6<br />
Hf 4.75 4.3 4.4 4.8 5.1 4.67 6.2 4.10 5.9 1.39 1.371 1.4 1.32 1.370 2.2 1.30 2.8 0.2 0.195 0.15 0.09 0.159 27.8 0.173 2.0<br />
Ta 0.804 0.9 0.84 0.921 0.866 5.4 0.803 0.5 0.42 0.402 0.411 2.2 0.404 1.6 0.182 0.11 0.146 24.7 0.135 5.8<br />
W 1.4 1.5 1.5 3.4 1.53 1.9 2.57 2.3 2.6 2.49 5.4 2.52 0.3 92.3 92.3 84.7 2.3<br />
Pb 31 30.0 33.7 31.6 5.0 29.7 2.4 38 38 42 39 4.8 40.5 0.3 2 2 2.0 2.0 0.0 1.63 0.9<br />
Th 9.7 9.01 9.5 8.72 9.3 10.23 9.41 5.2 8.92 5.7 4.78 4.48 4.8 4.64 4.68 2.7 4.46 1.6 0.735 0.56 0.63 0.642 11.2 0.555 0.9<br />
U 3 3.1 2.7 3.22 3.01 6.4 3.01 3.8 5.4 5.31 5.4 5.2 5.33 1.5 5.18 2.1 0.736 0.60 0.59 0.642 10.4 0.590 0.3<br />
1 data sources, whether compiled from studies or from single analytical method, <strong>and</strong> arranged chronologically: INAA = instrumental neutron activation analysis, LA-ICPMS = laser ablation-ICPMS.<br />
2 reference average reported to number <strong>of</strong> significant digits observed in most precisely measured data from literature sources.<br />
3 JUB average calculated from number <strong>of</strong> sample decompositions reported in Table 1 <strong>and</strong> reported to three significant digits only for illustrative purposes (see Section 5 regarding analytical precision).<br />
* values greater than reference average considered to include significant contributions from interfering molecular species, arising from high major element contents (Mg, Al, Ca, Mn, Fe).<br />
this<br />
study<br />
62
Appendix 1 continued. Data in mg/kg.<br />
JDo-1 JMn-1 BHVO-2<br />
Garbe- Govindaraju Imai Dulski Yamamoto<br />
this<br />
this<br />
Terashima Imai Dulski<br />
this<br />
Plumlee Huang Willbold Shafer Schramm Ryder Polat<br />
this<br />
Schönberg (1994) et al. (2001) (2004)<br />
study<br />
study<br />
et al. et al. (2001)<br />
study<br />
(1998) <strong>and</strong> Frey <strong>and</strong> et al. et al. (2006) et al.<br />
study<br />
(1993)<br />
(1996)<br />
HF-HClO4 carbonate 4<br />
(1995) (1999)<br />
(2003) Jochum (2005) (2005)<br />
(2006)<br />
(2005)<br />
method 1 ICPMS compiled compiled ICPMS ICPMS ref. avg. 2 %RSD JUB avg. 3 %RSD JUB avg. 3 %RSD ICPMS compiled ICPMS ref. avg. %RSD JUB avg. %RSD compiled ICPMS ICPMS ICPMS ICPMS ICPMS ICPMS ref. avg. %RSD JUB avg. %RSD<br />
HF-HClO4<br />
(n=5)<br />
(n=7)<br />
(n=4)<br />
(n=7)<br />
Sc 0.5 0.14 0.136 0.259 66.0 0.195 23.7 0.265 6.8 13 13 10.1 12.7 32 38 30.1 32 33.0 9.0 30.9 7.4<br />
Ti 7.97 7.97 8.80 10.1 2.10 8.9 6360 6360 5160 10.5 16300 16300 12700 15.3<br />
Co 0.3 0.168 0.234 28.2 1.46* 29.6 0.689* 16.9 1692 1732 1712 1.2 1080 5.9 45 56.2 41 45.7 42 46.0 11.8 44.3 6.0<br />
Ni 2.9 2.9 2.9 0.0 11.1* 2.8 3.04 5.9 12552 12632 12592 0.3 7770 12.1 119 137 107 115 115 119 8.4 124 19.3<br />
Rb 1.5 0.14 0.82 82.9 0.0940 9.1 0.0434 25.0 11.04 10.9 11.6 11.18 2.7 9.99 3.1 9.8 9.48 8.56 10.1 8.71 9.77 8.69 9.30 6.3 9.42 5.6<br />
Sr 108 119 116 117 115 3.6 117 5.7 113 2.6 737 792 800 776 3.6 653 5.5 389 399 394 407 389 396 368 392 2.9 379 4.3<br />
Y 8.9 11.2 10.3 10.4 10.2 8.1 10.6 3.3 11.0 3.0 106.49 111 114 110.50 2.8 95.1 2.5 26 28.3 29 25.8 23.2 28 23 26.2 8.5 24.3 2.6<br />
Zr 0.4 6.21 0.56 2.39 113.1 0.567 1.4 0.237 12.8 350 344 392 362 5.9 318 2.9 172 178 172 174 168 181 158 172 4.0 168 4.9<br />
Nb 0.2 0.2 0.253 40.6 0.0072 12.5 27.42 22.3 24.86 10.3 24.1 10.2 18 19 18.9 19.2 16.6 15.8 15 17.5 9.0 17.5 4.5<br />
Mo 0.4 0.4 0.4 0.0 0.291 7.5 0.282 34.3 316 318 317 0.3 268 2.0 4 3.94 3.97 0.8 4.58 14.9<br />
Cs 0.019 0.019 0.0072 8.8 0.0030 19.8 0.41 0.604 0.37 0.461 22.2 0.324 3.8 0.1 0.11 0.09 0.11 0.09 0.10 8.9 0.0954 4.6<br />
Ba 24 6.14 7 12.4 66.4 6.26 19.1 5.54 2.5 1702 1714 1511 1642 5.7 1450 2.9 130 135 131 142 129 133 125 132 3.8 128 3.4<br />
La 7.1 7.87 7.93 7.7 7.56 7.63 3.9 7.75 3.7 7.84 2.8 124.88 122 121 122.63 1.3 109 2.6 15 15.2 15.3 15.5 14.5 15.3 14.23 15.00 2.9 14.8 2.8<br />
Ce 1.85 2.54 2.49 2.02 2.06 2.19 12.5 2.07 3.4 2.12 2.3 286 277 271 278 2.2 274 19.4 38 38.4 37.6 38.4 35.3 37.5 35.73 37.28 3.1 36.8 3.1<br />
Pr 0.97 0.9 0.956 1.05 0.986 0.972 5.0 1.01 3.9 1.02 2.5 29.72 31.4 32 31.04 3.1 28.3 3.7 5.35 5.57 5.31 5.72 5.1 5.39 4.96 5.34 4.5 5.18 2.9<br />
Nd 4 5.33 5.25 4.09 4.21 4.58 12.8 4.18 3.6 4.25 2.4 127.88 137 123 129.29 4.5 118 3.1 25 24.9 24.5 24.8 23.4 24.6 22.8 24.3 3.2 24.0 3.6<br />
Sm 0.73 0.84 0.788 0.68 0.686 0.745 8.2 0.708 2.9 0.712 2.7 29.25 30.2 27 28.82 4.7 26.7 3.6 6.2 6.16 6.04 5.99 5.9 6.12 5.69 6.01 2.7 5.96 3.1<br />
Eu 0.15 0.19 0.176 0.162 0.154 0.166 8.9 0.159 4.0 0.161 3.0 7.26 7.58 7.53 7.46 1.9 6.60 2.0 2.07 2.03 2.05 2.07 1.96 2.07 1.96 2.03 2.3 2.03 3.2<br />
Gd 0.87 0.87 0.9 0.872 0.878 1.4 0.910 4.1 0.904 3.5 26.05 29.8 32.1 29.32 8.5 27.8 0.5 6.3 6.13 6.23 6.48 5.9 6.32 6.13 6.21 2.7 6.11 2.1<br />
Tb 0.12 0.12 0.116 0.12 0.113 0.118 2.4 0.121 3.9 0.121 3.3 4.42 4.81 4.92 4.72 4.5 4.27 1.6 0.9 0.963 0.933 0.9 0.9 0.959 0.89 0.921 3.1 0.908 2.9<br />
Dy 0.71 1 0.814 0.75 0.747 0.804 12.9 0.754 3.6 0.755 3.4 25.65 28.3 28.6 27.52 4.8 25.3 2.1 5.31 5.3 5.29 5.25 5.2 5.34 5.07 5.25 1.6 5.18 3.0<br />
Ho 0.17 0.2 0.164 0.167 0.164 0.173 7.9 0.168 3.9 0.168 3.6 5.49 5.76 5.58 5.61 2.0 4.93 2.4 1.04 1.01 0.964 1.03 0.95 1.01 0.93 0.991 4.0 0.960 2.5<br />
Er 0.44 0.44 0.46 0.462 0.451 2.3 0.466 3.0 0.465 3.2 13.26 14.6 15.6 14.49 6.6 13.8 1.8 2.54 2.5 2.49 2.49 2.4 2.53 2.39 2.48 2.2 2.49 2.6<br />
Tm 0.06 0.06 0.058 0.056 0.055 0.058 3.5 0.0550 3.8 0.0550 3.9 2.04 2.14 2.21 2.13 3.3 1.94 2.1 0.33 0.35 0.321 0.34 0.32 0.34 0.31 0.330 3.9 0.316 3.0<br />
Yb 0.29 0.36 0.323 0.305 0.303 0.316 7.7 0.301 3.3 0.297 3.5 12.86 13.8 14.4 13.69 4.6 12.5 2.4 2 2.05 1.95 2.01 1.91 2.03 1.84 1.97 3.5 1.90 2.7<br />
Lu 0.05 0.05 0.0494 0.043 0.042 0.0469 7.7 0.0426 4.2 0.0427 3.9 2.06 2.07 2.16 2.10 2.1 1.91 3.1 0.28 0.286 0.269 0.28 0.26 0.278 0.25 0.272 4.4 0.272 3.1<br />
Hf 0.12 0.1 0.0897 0.1032 12.2 0.0148 5.9 0.0053 16.7 6.1 6.23 6.4 6.24 2.0 5.59 4.3 4.1 4.42 4.2 4.58 4.4 4.38 4.35 3.6 4.27 4.7<br />
Ta 0.009 0.009 0.0137 75.1 0.0010 0.0 0.68 0.635 0.658 3.4 0.485 10.6 1.4 1.22 1.08 1.23 1.09 0.686 0.92 1.089 19.8 1.06 6.7<br />
W 0.26 0.26 0.276 13.5 0.248 17.5 37.49 45.3 41.40 9.4 34.7 7.5 0.21 0.29 0.25 16.0 0.228 6.7<br />
Pb 0.5 1 0.19 0.77 0.62 49.2 0.450 4.4 0.352 23.7 444 430 434 436 1.4 360 4.3 1.6 1.53 1.8 1.22 1.58 2.89 1.58 1.74 28.4 1.63 9.2<br />
Th 0.04 0.0429 0.0415 3.5 0.0556 3.3 0.0521 6.1 11.93 11.7 13 12.21 4.6 10.2 3.9 1.2 1.3 1.13 1.32 1.2 1.22 1.53 1.27 9.5 1.23 4.7<br />
U 0.8 0.858 0.88 0.846 4.0 1.03 3.9 0.987 6.1 4.81 5.01 5.3 5.04 4.0 4.32 3.2 0.403 0.446 0.403 0.45 0.42 0.425 0.4 0.421 4.6 0.430 3.4<br />
1 data sources, whether compiled from studies or from single analytical method, <strong>and</strong> arranged in chronologically.<br />
2 reference average reported to number <strong>of</strong> significant digits observed in most precisely measured data from literature sources.<br />
3 JUB average calculated from number <strong>of</strong> sample decompositions reported in Table 1 <strong>and</strong> reported to three significant digits only for illustrative purposes (see Section 5 regarding analytical precision), with exception <strong>of</strong> some elements for JDo-1 carbonate decomposition method.<br />
4 data for JDo-1 includes measured concentrations as determined using both the HF-HClO 4 <strong>and</strong> carbonate (HNO 3 ) sample decomposition methods (see Fig. 1). Note that Ni data for carbonate decomposition solely determined from the 62 Ni isotope.<br />
* values greater than reference average considered to include significant contributions from interfering molecular species, arising from high major element contents (Mg, Al, Ca, Mn, Fe).<br />
63
Appendix 2. Interferences due to major elements<br />
Appendix 2 contains figures illustrating significant molecular interferences for<br />
atomic masses
the concentration <strong>of</strong> the major element being considered, since as concentrations<br />
increase (e.g., to 500 mg/kg) the high TDS content <strong>of</strong> the solution begins to suppress<br />
analyte intensities, similar to that observed for rock samples that are only minimally<br />
diluted before analysis.<br />
The quantification <strong>of</strong> the observed interferences as concentrations in mg/kg<br />
allows predictions to be made regarding the potential impact these interferences might<br />
have upon trace metal determinations. For example, as the CaO content in a carbonate<br />
rock increases from 10% to 40%, then Ni concentrations determined in an HCl acid<br />
matrix would be expected to increase from ~2.5 mg/kg to ~6 mg/kg, due to Ca<br />
interferences on 60 Ni (Fig. A2.1). These calculated interferences in concentration<br />
units <strong>of</strong> m/kg are obtained from the long term average instrument response as<br />
determined from the 10 μg/kg calibration st<strong>and</strong>ard (described in Fig. 2, Section 4).<br />
For example, the 60 Ni isotope has an average instrument response <strong>of</strong> 2158 cps/μg·kg -1<br />
in 0.5 M HCl, <strong>and</strong> a sample containing 14% CaO <strong>and</strong> diluted 1000x would be<br />
expected to generate an CaO(H) interference <strong>of</strong> ~6700 cps on mass 60. Therefore, Ni<br />
quantified in this sample using measurement <strong>of</strong> the 60 Ni isotope would be erroneously<br />
overestimated by approximately 3.1 μg/kg (i.e., 2158/6700 = 3.1).<br />
It should be noted that, similar to estimates <strong>of</strong> interferences based upon raw<br />
data (cps), interference estimates reported as concentrations (mg/kg) will vary with<br />
instrument performance. As the long term average instrument response for the 10<br />
μg/kg calibration st<strong>and</strong>ard varies by as much as 30% (RSD), <strong>and</strong> this is the basis for<br />
the calculation <strong>of</strong> the interference magnitude in mg/kg, then these calculated<br />
concentrations will vary similarly.<br />
Data for interferences are presented for the following isotopes in the order;<br />
60 Ni, 62 Ni, 88 Sr, 89 Y, 90 Zr, 91 Zr, 93 Nb, <strong>and</strong> 95 Mo. Regardless <strong>of</strong> the element <strong>of</strong> interest,<br />
or the specific interferences that inhibit accurate concentration measurements <strong>of</strong> these<br />
elements, the key factor is always the relative abundance <strong>of</strong> the interfering species to<br />
the isotope <strong>of</strong> interest. In other words, the greatest analytical difficulties arise when<br />
trying to quantify small concentrations <strong>of</strong> an element in the presence <strong>of</strong> large<br />
concentrations <strong>of</strong> interfering elements. As a result, while some interferences may be<br />
large, e.g., several mg/kg in the case <strong>of</strong> Ca 2 <strong>and</strong> ArCa on Sr, the effect <strong>of</strong> these<br />
interferences is minor if the element being measured is abundant in the sample (e.g.,<br />
115 mg/kg Sr in the JDo-1 dolomite).<br />
65
The case <strong>of</strong> Ni determinations in carbonates has been mentioned previously,<br />
<strong>and</strong> the conclusion is that accurate Ni determinations in Ca- <strong>and</strong> Mg-rich rocks are<br />
only possible in HNO 3 acid matrices <strong>and</strong> by utilizing the 62 Ni isotope. Figures A2.1<br />
through A2.4 illustrate the numerous Ca <strong>and</strong> Mg interferences on 60 Ni <strong>and</strong> 62 Ni.<br />
Additionally, Figure A2.5 demonstrates that Al 2 O 3 abundances typical for shales (10-<br />
15%) that are analyzed in 0.5 M HCl are expected to contribute 2-3 mg/kg to the<br />
measured Ni concentration. The data suggest that 62 Ni, when measured in a HNO 3<br />
acid matrix, <strong>of</strong>fers the best isotope for Ni determinations in Mg-,Ca-, <strong>and</strong> Al-rich<br />
samples. However, 62 Ni is a low abundance isotope <strong>of</strong> Ni (3.63%), <strong>and</strong> will produce a<br />
relatively low signal response during ICPMS measurements, <strong>and</strong> this factor must be<br />
considered during attempts to quantify Ni.<br />
Figure A2.6 illustrates significant interferences on 88 Sr due to various Ca <strong>and</strong><br />
presumed Ar species ( 40 Ca 48 Ca, 40 Ar 48 Ca, <strong>and</strong> 44 Ca 44 Ca). These interferences exist<br />
regardless <strong>of</strong> the acid matrix used to prepare the samples for ICPMS analysis.<br />
Concentrations <strong>of</strong> Sr in many rocks are in the range <strong>of</strong> 50 to several hundred mg/kg<br />
(App. 1), <strong>and</strong> as Ca <strong>and</strong> Sr are both alkaline earth metals, their concentrations in<br />
many rocks vary proportionally. As a result, even though high Ca contents may<br />
produce an interference <strong>of</strong> 1-3 mg/kg on Sr, this frequently is <strong>of</strong> no great significance<br />
as Sr concentrations themselves may be high in the sample (e.g., ~115 mg/kg in the<br />
case <strong>of</strong> the JDo-1 dolomite).<br />
Interferences on Y due to high Fe contents in samples are presented in Figure<br />
A2.7 for informative purposes only. The impact <strong>of</strong> these interferences is likely to be<br />
small (~0.10 mg/kg) relative to typical Y contents in rocks (~10-50 mg/kg, see App.<br />
1), <strong>and</strong> will not be discussed further.<br />
Figures A2.8 <strong>and</strong> A2.9 demonstrate that significant interferences due to Mn<br />
<strong>and</strong> Fe may adversely impact determinations <strong>of</strong> Zr. The 90 Zr isotope suffers an<br />
interference <strong>of</strong> several mg/kg in 0.5 M HCl due to 55 Mn 35 Cl (Fig. A2.8). The effects<br />
<strong>of</strong> high Fe contents on measurements <strong>of</strong> 91 Zr are much greater due to 56 Fe 35 Cl, <strong>and</strong><br />
interferences <strong>of</strong> 10-20 mg/kg would be expected in Fe-rich samples such as IFs (Fig.<br />
A2.9). Smaller, but still significant effects are noted for 40 Ca 16 O 35 Cl interferences on<br />
91 Zr. These effects would be minimized by the use <strong>of</strong> HNO 3 as the diluting acid.<br />
Based upon these observations, the 91 Zr isotope is not recommended for most rock<br />
types, unless the use <strong>of</strong> HCl is avoided, or if Ca- <strong>and</strong> Fe-poor, Mn-rich samples<br />
require analysis.<br />
66
Potential interferences on monoisotopic 93 Nb are presented in Fig. A2.10, <strong>and</strong><br />
are very similar to those described for 91 Zr. This is due to the substitution <strong>of</strong> the 37 Cl<br />
isotope for 35 Cl in the Fe <strong>and</strong> Ca molecular species listed above (i.e., 56 Fe 37 Cl <strong>and</strong><br />
40 Ca 16 O 37 Cl). Unlike Zr, Nb <strong>of</strong>fers no additional isotope for analysis, <strong>and</strong> the low<br />
natural abundance <strong>of</strong> Nb exacerbates the problem <strong>of</strong> interfering species due to major<br />
elements. Accurate Nb analyses are unlikely in Fe- <strong>and</strong> Ca-rich rock types, unless the<br />
use <strong>of</strong> HCl is avoided. However, as discussed in Section 7.1 <strong>of</strong> the main text, Nb is<br />
unstable in pure HNO 3 acid matrices, <strong>and</strong> the presence <strong>of</strong> HCl stabilizes Nb, as well<br />
as other HFSE like Ta, in solution long enough for accurate ICPMS measurements.<br />
As mentioned previously, future work examining the optimum acid matrix for ICPMS<br />
determinations <strong>of</strong> Nb <strong>and</strong> Ta is recommended.<br />
The last element that appears to suffer from major element interferences is<br />
Mo, <strong>and</strong> in particular Mn interferences on the 95 Mo isotope (Fig. A2.11). Similarly to<br />
the Ca-Ar interferences on 88 Sr, Mn affects measured intensities for 95 Mo regardless<br />
<strong>of</strong> the acid matrix. This is apparently due to the 55 Mn 40 Ar molecular species, which<br />
forms independently <strong>of</strong> the bulk solution composition. Whereas the use <strong>of</strong> HCl<br />
appears to suppress 55 Mn 40 Ar formation (Fig. A2.11), the magnitude <strong>of</strong> the<br />
interference is significant regardless <strong>of</strong> the acid matrix (e.g., ~3-6 mg/kg at 20-30%<br />
MnO). It is therefore recommended that the 97 Mo isotope be used for Mo<br />
determinations in rock samples that contain more than a few percent MnO, unless Mo<br />
concentrations in the sample are the range <strong>of</strong> 50 to hundreds <strong>of</strong> mg/kg.<br />
67
20000<br />
0.5 M HCl<br />
10<br />
9<br />
interference on 60 Ni in sample solution (cps)<br />
15000<br />
10000<br />
5000<br />
CaO(H) +<br />
JDo-1 dolomite (2.9 mg/kg Ni)<br />
MgCl +<br />
8<br />
7<br />
6<br />
5<br />
4<br />
3<br />
2<br />
interference on 60 Ni in sample powder (mg/kg)<br />
1<br />
0<br />
0<br />
0 10 20 30 40 50 60 70 80 90 100<br />
MgO, CaO in sample powder (wt.%)<br />
Figure A2.1. Interferences observed at mass 60 (Ni) in 0.5 M HCl due to sample Mg <strong>and</strong> Ca content.<br />
JDo-1 contains 18.4% MgO <strong>and</strong> 34.0% CaO, effectively prohibiting in carbonate rocks<br />
determinations <strong>of</strong> Ni concentrations that are similar to JDo-1 (2.9 mg/kg Ni).<br />
15000<br />
18<br />
16<br />
interference on 60 Ni in sample solution (cps)<br />
10000<br />
5000<br />
0.5 M HNO 3<br />
JDo-1 dolomite (2.9 mg/kg Ni)<br />
CaO(H) +<br />
14<br />
12<br />
10<br />
8<br />
6<br />
4<br />
interference on 60 Ni in sample powder (mg/kg)<br />
N 2<br />
O 2 + (?)<br />
2<br />
0<br />
0<br />
0 10 20 30 40 50 60 70 80 90 100<br />
MgO, CaO in sample powder (wt.%)<br />
Figure A2.2. Interferences observed at mass 60 (Ni) in 0.5 M HNO 3 due to sample Mg <strong>and</strong> Ca<br />
content. Use <strong>of</strong> HNO 3 effectively removes MgCl interference, but does not affect the dominant<br />
CaO(H) interference.<br />
68
1500<br />
0.5 M HCl<br />
4.0<br />
interference on 62 Ni in sample solution (cps)<br />
1000<br />
500<br />
JDo-1 dolomite (2.9 mg/kg Ni)<br />
MgCl +<br />
CaO(H) + (Ni in acid blank?)<br />
3.5<br />
3.0<br />
2.5<br />
2.0<br />
1.5<br />
1.0<br />
0.5<br />
interference on 62 Ni in sample powder (mg/kg)<br />
0<br />
0.0<br />
0 10 20 30 40 50 60 70 80 90 100<br />
MgO, CaO in sample powder (wt.%)<br />
Figure A2.3. Interferences observed at mass 62 (Ni) in 0.5 M HCl due to sample Mg <strong>and</strong> Ca<br />
content. JDo-1 contains 18.4% MgO <strong>and</strong> 34.0% CaO, effectively prohibiting in carbonate rocks<br />
determinations <strong>of</strong> Ni concentrations that are similar to JDo-1 (2.9 mg/kg Ni).<br />
600<br />
4.5<br />
500<br />
4.0<br />
interference on 62 Ni in sample solution (cps)<br />
400<br />
300<br />
200<br />
100<br />
0.5 M HNO 3<br />
MgO, CaO in sample powder (wt.%)<br />
CaO(H) + (Ni in acid blank?)<br />
JDo-1 dolomite (2.9 mg/kg Ni)<br />
not MgCl + (Ni in Mg st<strong>and</strong>ard?)<br />
3.5<br />
3.0<br />
2.5<br />
2.0<br />
1.5<br />
1.0<br />
0.5<br />
interference on 62 Ni in sample powder (mg/kg)<br />
0<br />
0.0<br />
0 10 20 30 40 50 60 70 80 90 100<br />
Figure A2.4. Interferences observed at mass 62 (Ni) in 0.5 M HNO 3 due to sample Mg <strong>and</strong> Ca content.<br />
When monitored in HNO 3 matrices, the 62 Ni isotope is likely the most useful for trace Ni<br />
determinations in carbonate rocks, though the low abundance <strong>of</strong> 62 Ni (3.63%), which generates low<br />
ICPMS signal intensities, must be considered.<br />
69
1600<br />
1400<br />
0.5 M HCl<br />
4.5<br />
4.0<br />
interference on 62 Ni in sample solution (cps)<br />
1200<br />
1000<br />
800<br />
600<br />
400<br />
200<br />
0<br />
5.1% (FeR-2)<br />
AlCl +<br />
0.0<br />
0 5 10 15 20<br />
13.7% (SCo-1)<br />
3.5<br />
3.0<br />
2.5<br />
2.0<br />
1.5<br />
1.0<br />
0.5<br />
interference on 62 Ni in sample powder (mg/kg)<br />
Al 2<br />
O 3<br />
in sample powder (wt.%)<br />
Figure A2.5. Interferences observed at mass 62 (Ni) in 0.5 M HCl due to sample Al content. Shown are<br />
Al contents <strong>of</strong> two common CRMs that contain similar Ni (23 mg/kg for FeR-2, <strong>and</strong> 26 mg/kg for SCo-<br />
1). The lower Al/Ni <strong>of</strong> FeR-2 is expected to allow reasonable Ni determinations using 62 Ni, however the<br />
poor agreement between 60 Ni <strong>and</strong> 62 Ni argues against this (see App. 1). The greater Al/Ni <strong>of</strong> SCo-1<br />
significantly contributes to higher than expected Ni determinations in SCo-1 (see App. 1).<br />
interference on 88 Sr in sample solution (cps)<br />
60000<br />
50000<br />
40000<br />
30000<br />
20000<br />
10000<br />
0.5 M HNO 3<br />
0.5 M HCl<br />
5.2 mg/kg<br />
0<br />
0 10 20 30 40 50 60 70 80 90 100<br />
CaO in sample powder (wt.%)<br />
3.2 mg/kg<br />
Figure A2.6. Interferences observed at mass 88 (Sr) in both 0.5 M HCl <strong>and</strong> HNO 3 as a function <strong>of</strong> Ca<br />
content. Note right vertical axis denotes two concentration scales, dependent upon acid matrix.<br />
Significant interferences due to probable combinations <strong>of</strong> 40 Ca 48 Ca, 40 Ar 48 Ca, <strong>and</strong> 44 Ca 44 Ca exist<br />
regardless <strong>of</strong> acid matrix. Note that the discrepancy between acid matrices regarding the magnitude <strong>of</strong><br />
the interference arises from different ICPMS signal response (sensitivity) in HCl <strong>and</strong> HNO 3 . Caution<br />
should be exercised when interpreting Sr concentrations below ~75 mg/kg in samples with high Ca<br />
contents.<br />
0.5 M HCl<br />
3.5<br />
3.0<br />
2.5<br />
2.0<br />
1.5<br />
1.0<br />
0.5<br />
0.5 M HNO 3<br />
7<br />
6<br />
5<br />
4<br />
3<br />
2<br />
1<br />
interference on 88 Sr in sample powder (mg/kg)<br />
70
4000<br />
0.40<br />
0.20<br />
3500<br />
0.17 mg/kg<br />
0.18<br />
0.35<br />
interference on 89 Y in sample solution (cps)<br />
3000<br />
2500<br />
2000<br />
1500<br />
1000<br />
500<br />
0.5 M HCl<br />
0.5 M HNO 3<br />
0.13 mg/kg<br />
0.16<br />
0.14<br />
0.12<br />
0.10<br />
0.08<br />
0.06<br />
0.04<br />
0.02<br />
0<br />
0 10 20 30 40 50 60 70 80 90 100<br />
Fe 2<br />
O 3<br />
in sample powder (wt.%)<br />
0.5 M HCl<br />
0.5 M HNO 3<br />
0.30<br />
0.25<br />
0.20<br />
0.15<br />
0.10<br />
0.05<br />
interference on 89 Y in sample powder (mg/kg)<br />
Figure A2.7. Interferences observed at mass 89 (Y) in both 0.5 M HCl <strong>and</strong> HNO 3 as a function <strong>of</strong> Fe<br />
content. Note right vertical axis denotes two concentration scales, dependent upon acid matrix. A<br />
minor 54 Fe 35 Cl interference exists in 0.5 M HCl, <strong>and</strong> interferences are linear, though small, for both<br />
acid matrices, perhaps reflecting small Y contamination in artificial Fe solution. Concentrations <strong>of</strong> Y in<br />
all Fe-rich CRMs are 8-12 mg/kg, suggesting that Fe-interferences in these CRMs are not significant.<br />
12000<br />
1.3 mg/kg<br />
2.5<br />
10000<br />
1.2<br />
interference on 90 Zr in sample solution (cps)<br />
8000<br />
6000<br />
4000<br />
2000<br />
0.5 M HCl<br />
0.5 M HNO 3<br />
0.13 mg/kg<br />
0<br />
0 5 10 15 20<br />
MnO in sample powder (wt.%)<br />
0.5 M HCl<br />
1.0<br />
0.8<br />
0.6<br />
0.4<br />
0.2<br />
0.5 M HNO 3<br />
2.0<br />
1.5<br />
1.0<br />
0.5<br />
interference on 90 Zr in sample powder (mg/kg)<br />
Figure A2.8. Interferences observed at mass 90 (Zr) in both 0.5 M HCl <strong>and</strong> HNO 3 as a function <strong>of</strong> MnO<br />
content. Note right vertical axis denotes two concentration scales, dependent upon acid matrix. A strong<br />
55 Mn 35 Cl interference exists in 0.5 M HCl. Concentrations <strong>of</strong> MnO in Fe-Mn crusts (<strong>and</strong> Mn ores) may<br />
exceed 30%, <strong>and</strong> extrapolation <strong>of</strong> the data suggest possible interferences on 90 Zr <strong>of</strong> several mg/kg.<br />
71
40000<br />
35000<br />
0.5 M HCl<br />
20<br />
18<br />
interference on 91 Zr in sample solution (cps)<br />
30000<br />
25000<br />
20000<br />
15000<br />
10000<br />
5000<br />
FeCl +<br />
CaOCl +<br />
16<br />
14<br />
12<br />
10<br />
8<br />
6<br />
4<br />
2<br />
interference on 91 Zr in sample powder (mg/kg)<br />
0<br />
0<br />
0 10 20 30 40 50 60 70 80 90 100<br />
CaO, Fe 2<br />
O 3<br />
in sample powder (wt.%)<br />
Figure A2.9. Interferences observed at mass 91 (Zr) in 0.5 M HCl as a function <strong>of</strong> both Ca <strong>and</strong> Fe<br />
content. Interferences in HNO 3 acid not observed. Significant interferences produced at low<br />
concentrations (
interference on 95 Mo in sample solution (cps)<br />
5000<br />
4000<br />
3000<br />
2000<br />
1000<br />
0.5 M HNO 3<br />
0.5 M HCl<br />
0<br />
0 5 10 15 20<br />
MnO in sample powder (wt.%)<br />
4.3 mg/kg<br />
2.1 mg/kg<br />
0.5 M HCl<br />
2.0<br />
1.5<br />
1.0<br />
0.5<br />
0.5 M HNO 3<br />
4.5<br />
4.0<br />
3.5<br />
3.0<br />
2.5<br />
2.0<br />
1.5<br />
1.0<br />
0.5<br />
interference on 95 Mo in sample powder (mg/kg)<br />
Figure A2.11. Interferences observed at mass 95 (Mo) in both 0.5 M HCl <strong>and</strong> HNO 3 as a function <strong>of</strong><br />
MnO content. Note right vertical axis denotes two concentration scales, dependent upon acid matrix.<br />
Significant interferences produced for both HCl <strong>and</strong> HNO 3 acid matrices. For Mn-bearing (>5%),<br />
relatively Mo-poor (
Chapter 3. Continentally-derived solutes in shallow Archean seawater: Rare earth<br />
element <strong>and</strong> Nd isotope evidence in iron formation from the 2.9 Ga<br />
Pongola Supergroup, South Africa<br />
119
THIS PAGE INTENTIONALLY LEFT BLANK<br />
120
Available online at www.sciencedirect.com<br />
Geochimica et Cosmochimica Acta 72 (2008) 378–394<br />
www.elsevier.com/locate/gca<br />
Continentally-derived solutes in shallow Archean seawater:<br />
Rare earth element <strong>and</strong> Nd isotope evidence in iron<br />
formation from the 2.9 Ga Pongola Supergroup, South Africa<br />
Brian W. Alex<strong>and</strong>er a, *, Michael Bau a , Per Andersson b , Peter Dulski c<br />
a Earth <strong>and</strong> Space <strong>Science</strong>s, Campus Ring 1, Research III 102, <strong>Jacobs</strong> <strong>University</strong> Bremen, D-28759 Bremen, Germany<br />
b Laboratory for Isotope Geology, Swedish Museum <strong>of</strong> Natural History, SE-104 05 Stockholm, Sweden<br />
c GeoForschungsZentrum, D-14473 Potsdam, Germany<br />
Received 27 November 2006; accepted in revised form 25 October 2007; available online 17 November 2007<br />
Abstract<br />
The chemical composition <strong>of</strong> surface water in the photic zone <strong>of</strong> the Precambrian ocean is almost exclusively known from<br />
studies <strong>of</strong> stromatolitic carbonates, while b<strong>and</strong>ed iron formations (IFs) have provided information on the composition <strong>of</strong> deeper<br />
waters. Here we discuss the trace element <strong>and</strong> Nd isotope geochemistry <strong>of</strong> very shallow-water IF from the Pongola Supergroup,<br />
South Africa, to gain a better underst<strong>and</strong>ing <strong>of</strong> solute sources to Mesoarchean shallow coastal seawater. The Pongola<br />
Supergroup formed on the stable margin <strong>of</strong> the Kaapvaal craton 2.9 Ga ago <strong>and</strong> contains b<strong>and</strong>ed iron formations (IFs) that<br />
represent the oldest documented Superior-type iron formations. The IFs are near-shore, pure chemical sediments, <strong>and</strong> shalenormalized<br />
rare earth <strong>and</strong> yttrium distributions (REY SN ) exhibit positive La SN ,Gd SN , <strong>and</strong> Y SN anomalies, which are typical<br />
features <strong>of</strong> marine waters throughout the Archean <strong>and</strong> Proterozoic. The marine origin <strong>of</strong> these samples is further supported<br />
by super-chondritic Y/Ho ratios (average Y/Ho = 42). Relative to older Isua IFs (3.7 Ga) from Greenl<strong>and</strong>, <strong>and</strong> younger<br />
Kuruman IFs (2.5 Ga) also from South Africa, the Pongola IFs are depleted in heavy rare earth elements (HREE), <strong>and</strong><br />
appear to record variations in solute fluxes related to sea level rise <strong>and</strong> fall. Sm–Nd isotopes were used to identify potential<br />
sediment <strong>and</strong> solute sources within pongola shales <strong>and</strong> IFs. The Nd (t) for Pongola shales ranges from 2.7 to 4.2, <strong>and</strong> Nd (t)<br />
values for the coeval iron-formation samples (range 1.9 to 4.3) are generally indistinguishable from those <strong>of</strong> the shales,<br />
although two IF samples display Nd (t) as low as 8.1 <strong>and</strong> 10.9. The similarity in Nd isotope signatures between the shale<br />
<strong>and</strong> iron-formation suggests that mantle-derived REY were not a significant Nd source within the Pongola depositional environment,<br />
though the presence <strong>of</strong> positive Eu anomalies in the IF samples indicates that high-T hydrothermal input did contribute<br />
to their REY signature. Isotopic mass balance calculations indicate that most (P72%) <strong>of</strong> the Nd in these seawater<br />
precipitates was derived from continental sources. If previous models <strong>of</strong> Fe–Nd distributions in Archean IFs are applied, then<br />
the Pongola IFs suggest that continental fluxes <strong>of</strong> Fe to Archean seawater were significantly greater than are generally<br />
considered.<br />
Ó 2007 Elsevier Ltd. All rights reserved.<br />
1. INTRODUCTION<br />
* Corresponding author.<br />
E-mail address: b.alex<strong>and</strong>er@jacobs-university.de (B.W.<br />
Alex<strong>and</strong>er).<br />
Studies to determine the characteristics <strong>of</strong> seawater in<br />
the geologic past are dependent upon proxies which reliably<br />
reflect seawater composition. Such studies have<br />
examined sedimentary rocks <strong>and</strong> minerals which form directly<br />
from seawater solutions, including inorganic precipitates<br />
(e.g., evaporites) or biologically mediated<br />
precipitates (e.g., CaCO 3 ). Attempts to identify reliable<br />
proxies for paleo-seawater rely on chemical tracers, particularly<br />
elements with known <strong>and</strong> predictable behavior<br />
in modern seawater <strong>and</strong> modern seawater precipitates.<br />
0016-7037/$ - see front matter Ó 2007 Elsevier Ltd. All rights reserved.<br />
doi:10.1016/j.gca.2007.10.028
Nd isotopes in 2.9 Ga Archean surface seawater 379<br />
These elemental tracers must also be relatively unaffected<br />
by geologic processes which commonly affect marine precipitates,<br />
such as diagenesis <strong>and</strong> metamorphism. One<br />
such group <strong>of</strong> chemical tracers are the rare earth elements<br />
<strong>and</strong> yttrium (REY), which have been used as seawater<br />
proxies in a variety <strong>of</strong> sedimentary rocks <strong>of</strong> known<br />
marine origin, including limestones, phosphorites, <strong>and</strong><br />
b<strong>and</strong>ed iron formations. The usefulness <strong>of</strong> the REYs as<br />
seawater proxies has been discussed extensively by Bau<br />
<strong>and</strong> Dulski (1996), Webb <strong>and</strong> Kamber (2000), Shields<br />
<strong>and</strong> Stille (2001), Nothdurft et al. (2004), Shields <strong>and</strong><br />
Webb (2004), <strong>and</strong> Bolhar et al. (2004). These workers<br />
have concluded that REY distributions in marine chemical<br />
sediments can accurately reflect the REY distribution<br />
<strong>of</strong> the seawater from which they formed, <strong>and</strong> that this<br />
relationship can be reliably extended to sediments from<br />
throughout the geologic record.<br />
The information provided by the REYs regarding seawater<br />
composition is a function <strong>of</strong> the processes which control<br />
their distribution in seawater. In modern marine waters<br />
the variation in REY distributions is dominated by varying<br />
degrees <strong>of</strong> carbonate complexation, with REY principally<br />
sourced from weathering <strong>of</strong> continents, as suggested by<br />
the isotopic composition <strong>of</strong> Nd in modern seawater (Piepgras<br />
<strong>and</strong> Wasserburg, 1980). High-temperature hydrothermal<br />
inputs (e.g., black smoker fluids) are generally<br />
negligible REY sources to modern seawater, though this<br />
may not have been the case in Earth’s early oceans, a scenario<br />
assessed using Eu, which is enriched in high-temperature<br />
fluids emanating from basaltic, mid-ocean ridge<br />
sources (MORB, Michard, 1989; Bau <strong>and</strong> Dulski, 1999).<br />
This approach, coupled with Nd isotope systematics, has<br />
been extended to Archean b<strong>and</strong>ed iron formations (e.g.,<br />
Derry <strong>and</strong> <strong>Jacobs</strong>en, 1990; Danielson et al., 1992), in attempts<br />
to constrain continental- versus MORB-derived Fe<br />
inputs to Earth’s early oceans (Miller <strong>and</strong> O’Nions, 1985;<br />
<strong>Jacobs</strong>en <strong>and</strong> Pimentel-Klose, 1988a). Therefore, the<br />
REY are particularly well suited not only for identifying<br />
Archean seawater proxies, but also for <strong>of</strong>fering insight into<br />
the importance <strong>of</strong> mid-ocean ridge solute fluxes relative to<br />
solutes sourced from continental weathering.<br />
Numerous workers have used Precambrian iron formations<br />
to study ancient seawater indirectly (e.g., Fryer,<br />
1977; <strong>Jacobs</strong>en <strong>and</strong> Pimentel-Klose, 1988a; Derry <strong>and</strong> <strong>Jacobs</strong>en,<br />
1990; Shimizu et al., 1990; Towe, 1991; Bau <strong>and</strong><br />
Möller, 1993; among others). The similarity <strong>of</strong> REY patterns<br />
between modern seawater <strong>and</strong> both Archean iron formations<br />
<strong>and</strong> carbonates has been noted, <strong>and</strong> these chemical<br />
sediments have been used to constrain Archean seawater<br />
composition (e.g., Bau <strong>and</strong> Dulski, 1996; Kamber <strong>and</strong><br />
Webb, 2001; Bolhar et al., 2004). Though certain features<br />
are common for many Precambrian IFs (e.g., alternating<br />
b<strong>and</strong>s <strong>of</strong> Si- <strong>and</strong> Fe-rich oxide layers), variations are observed<br />
in chemical composition <strong>and</strong> mineralogy <strong>and</strong> are<br />
generally attributed to distinct differences in tectonic setting<br />
(e.g., Gross, 1983), or facies changes within a single depositional<br />
basin (Klein <strong>and</strong> Beukes, 1989). As such, Precambrian<br />
IFs formed in different marine environments may<br />
reflect potential variations in seawater composition between<br />
these environments.<br />
Two broad types <strong>of</strong> IF are generally recognized: the Algoma<br />
type which is <strong>of</strong>ten found in greenstone belts, <strong>and</strong> the<br />
Superior type associated with stable sedimentary basins <strong>and</strong><br />
cratonic margins (Gross, 1983). When considering the mass<br />
<strong>of</strong> sediment deposited, Superior-type IFs contain far more<br />
Fe than Algoma-type IFs (James <strong>and</strong> Trendall, 1982;<br />
Klemm, 2000), although this may represent a preservation<br />
bias due to their proposed tectonic settings. The association<br />
<strong>of</strong> Superior-type IFs with shelf deposits such as quartzites,<br />
carbonates, <strong>and</strong> shales is consistent with their depositional<br />
model, <strong>and</strong> Superior-type IFs are particularly useful for<br />
constraining trace element distributions in shallow Archean<br />
seawater.<br />
The Kaapvaal craton in South Africa is widely recognized<br />
as one <strong>of</strong> the earliest stable cratons formed (Tankard<br />
et al., 1982), <strong>and</strong> hosts the oldest known Superior-type IFs<br />
in the world within the 3.0 Ga Witwatersr<strong>and</strong> <strong>and</strong> Pongola<br />
Supergroups. Work by Beukes <strong>and</strong> Cairncross (1991)<br />
indicates that some units within both the Witwatersr<strong>and</strong><br />
<strong>and</strong> the Pongola are correlative, however, the Pongola differs<br />
most significantly from the Witwatersr<strong>and</strong> in that it<br />
experienced lower degrees <strong>of</strong> metamorphism <strong>and</strong> structural<br />
deformation. The age, depositional environment, <strong>and</strong> the<br />
relatively pristine preservation <strong>of</strong> the Pongola Supergroup<br />
provide unique opportunities for the investigation <strong>of</strong> Archean<br />
marine chemical sediments deposited in relatively<br />
shallow waters. Rare earth <strong>and</strong> yttrium distributions <strong>and</strong><br />
Sm–Nd isotope systematics in Pongola Supergroup IFs<br />
indicate significant continentally-derived solute fluxes to<br />
shallow Archean seawater.<br />
2. GEOLOGIC SETTING<br />
The 2.9–3.0 Ga Pongola Supergroup is located in eastern<br />
South Africa <strong>and</strong> southwestern Swazil<strong>and</strong> <strong>and</strong> crops<br />
out over a 270-km by 100-km area (Weilers, 1990; Fig. 1).<br />
The extent <strong>of</strong> the Pongola Supergroup is consistent with a<br />
minimum depositional area <strong>of</strong> 32,500 km 2 (Button et al.,<br />
1981), <strong>and</strong> the sequence as a whole consists <strong>of</strong> two stratigraphic<br />
units; the Nsuze Group <strong>and</strong> the overlying Mozaan<br />
Group. Sedimentary structures within Nsuze siliciclastic<br />
beds such as lenticular/flaser bedding <strong>and</strong> herringbone<br />
cross-lamination (von Brunn <strong>and</strong> Hobday, 1976) <strong>and</strong> stromatolite<br />
bearing Nsuze carbonates (Mason <strong>and</strong> von Brunn,<br />
1977; Beukes <strong>and</strong> Lowe, 1989) indicate a near-shore shallow<br />
water depositional environment (see also Matthews,<br />
1967; von Brunn <strong>and</strong> Mason, 1977; Tankard et al., 1982).<br />
This study focuses on samples from the Sinqeni Formation<br />
<strong>of</strong> the Mozaan Group, which outcrops within the Wit<br />
Mfolozi inlier (Fig. 1). The Sinqeni Formation is the most<br />
laterally extensive formation within the Mozaan Group<br />
(Nhleko, 2003) <strong>and</strong> is dominated by quartz arenite <strong>and</strong><br />
shale with minor conglomerate <strong>and</strong> b<strong>and</strong>ed iron formation<br />
(Matthews, 1967; Beukes <strong>and</strong> Cairncross, 1991). Specifically,<br />
the iron formation is hosted within the Ijzermijn<br />
Member, an approximately 15-m-thick unit which has shale<br />
at the base overlying with a gradational contact coarse<br />
s<strong>and</strong>stone <strong>of</strong> the older Dipka Member. The Ijzermijn shale<br />
grades upward into 3- to 5-m-thick iron formation intercalated<br />
with shale, followed again by shale which is capped
380 B.W. Alex<strong>and</strong>er et al. / Geochimica et Cosmochimica Acta 72 (2008) 378–394<br />
30 o 31 o 32 o 27 o<br />
25 o<br />
SWAZILAND<br />
MOZAMBIQUE<br />
REPUBLIC<br />
OF<br />
SOUTHAFRICA<br />
REPUBLIC<br />
OF<br />
SOUTHAFRICA<br />
Mozaan Group<br />
Nsuze Group<br />
0 50 km<br />
30 o 31 o<br />
WHITE MFOLOZI INLIER<br />
INDIAN<br />
OCEAN<br />
32 o<br />
28 o<br />
29 o<br />
LESOTHO<br />
INDIAN<br />
OCEAN<br />
30 o<br />
PONGOLA<br />
30 o SUPER GROUP<br />
LITHOLOGY AND SAMPLE POSITIONS<br />
shale<br />
conglomerate<br />
iron formation<br />
shale<br />
iron formation<br />
shale<br />
conglomerate<br />
19<br />
18<br />
13x<br />
11x<br />
12x<br />
10x<br />
8<br />
7x<br />
6x<br />
5x<br />
4x<br />
# = sample number (see caption for details)<br />
2<br />
Fig. 1. Map <strong>of</strong> study area showing extent <strong>of</strong> Pongola Supergroup, location <strong>of</strong> White Mfolozi inlier, <strong>and</strong> relative positions <strong>of</strong> sampling<br />
locations within a simplified sequence stratigraphy. The x following some sample numbers indicates that two samples were collected, <strong>and</strong><br />
numbered accordingly (e.g., 4x refers to samples 41 <strong>and</strong> 42). Thickness <strong>of</strong> the lithologic units is not to scale.<br />
with a sharp erosional contact formed by supermature<br />
orthoquartzite (Nhleko, 2003). Sedimentary structures<br />
found within the Mozaan Group are similar to those described<br />
in the Nsuze Group, <strong>and</strong> Beukes <strong>and</strong> Cairncross<br />
(1991) recognized only two major depositional environments<br />
for the Mozaan: fluvial braidplain <strong>and</strong> shallow marine<br />
shelf systems. Studies <strong>of</strong> the Mozaan Group (Watchorn,<br />
1980; Weilers, 1990) describe the IFs as being deposited<br />
within a distal shelf environment, conclusions that are consistent<br />
with the detailed analysis <strong>of</strong> Beukes <strong>and</strong> Cairncross<br />
(1991), who characterized the Mozaan IFs as forming on a<br />
shallow starved outer continental shelf during the peak <strong>of</strong> a<br />
marine transgression.<br />
Age constraints for the Mozaan are not directly available<br />
from units within the group, though it appears that<br />
deposition occurred between 2940 <strong>and</strong> 2870 Ma (Beukes<br />
<strong>and</strong> Cairncross, 1991). This is inferred from a U–Pb zircon<br />
age <strong>of</strong> 2940 ± 22 Ma for rhyolite within the Nsuze Group,<br />
<strong>and</strong> a 2871 ± 30 Ma Sm–Nd mineral isochron (Hegner<br />
et al., 1984) for pyroxenite from the post-Pongola intrusive<br />
Usushwana Complex. Following deposition, the Pongola<br />
Supergroup was intruded by differentiated gabbroic sills<br />
(the Usushwana Complex), primarily along the basal<br />
unconformity <strong>and</strong> as a series <strong>of</strong> sills in the Mozaan Group<br />
(Button et al., 1981). Granitic plutons emplaced at 2.5–<br />
2.7 Ga bound the Pongola Supergroup primarily along its<br />
eastern margins, resulting in narrow contact aureoles with<br />
relatively little deformation <strong>of</strong> Pongola strata (Button<br />
et al., 1981; Tankard et al., 1982). Mineral assemblages<br />
within the Pongola generally reflect greenschist facies regional<br />
metamorphism (Tankard et al., 1982; Hegner et al.,<br />
1984; Hunter <strong>and</strong> Wilson, 1988; Beukes <strong>and</strong> Cairncross,<br />
1991).<br />
3. SAMPLING AND ANALYTICAL METHODS<br />
Four shale samples <strong>and</strong> 16 iron formation samples were<br />
collected from the Sinqeni Formation in the White Mfolozi<br />
Inlier (Fig. 1). The shale <strong>and</strong> IF beds are intercalated, <strong>and</strong><br />
are bounded at the top <strong>and</strong> bottom by beds <strong>of</strong> small pebble<br />
conglomerate. Desiccation cracks are common in the clastic<br />
units immediately above <strong>and</strong> below the sequence studied,<br />
indicating that IF deposition was preceded <strong>and</strong> followed<br />
by relatively lower seawater levels. Field sampling avoided<br />
obvious fault zones <strong>and</strong> alteration features, <strong>and</strong> fresh<br />
unweathered samples were subsequently crushed <strong>and</strong> ana-
Nd isotopes in 2.9 Ga Archean surface seawater 381<br />
lysed at the GeoForschungsZentrum, Potsdam, Germany.<br />
Major element concentrations were obtained by X-ray fluorescence<br />
(XRF), <strong>and</strong> mineral phase determinations were<br />
obtained by X-ray diffraction (XRD). Trace element concentrations<br />
were determined with a Perkin-Elmer/Sciex<br />
Elan Model 5000 inductively coupled plasma mass spectrometer<br />
(ICP-MS) following the procedures <strong>of</strong> Dulski<br />
(2001). Samarium <strong>and</strong> Nd concentrations were also determined<br />
using thermal ionization mass spectrometry (TIMS).<br />
Agreement between TIMS <strong>and</strong> ICPMS data is better than<br />
2% for Nd (except shale WM2, 3.7%), <strong>and</strong> better than 4%<br />
for Sm (except shale WM2 <strong>and</strong> IF WM41, 6.1% <strong>and</strong><br />
6.0%, respectively). Iron-formation sample WM121 is atypical<br />
is displaying differences <strong>of</strong> 7.2% <strong>and</strong> 10.8% in respective<br />
Nd <strong>and</strong> Sm concentrations as determined by TIMS <strong>and</strong><br />
ICPMS, yet Sm/Nd for WM121 differs by only 2.8% between<br />
the two analytical methods, <strong>and</strong> Sm/Nd comparisons<br />
between TIMS <strong>and</strong> ICPMS data never exceeds 5.3%, indicating<br />
that consistent REE ratios were determined regardless<br />
<strong>of</strong> the method used. Accuracy for other ICPMS<br />
analyses is conservatively estimated to be ±10%, <strong>and</strong> rare<br />
earth element ratios are estimated to be accurate within<br />
±5%. Samples for Sm–Nd isotope determinations were<br />
spiked using a mixed 147 Sm/ 150 Nd spike, <strong>and</strong> processed at<br />
the Laboratory for Isotope Geology, Swedish Museum <strong>of</strong><br />
Natural History. Samarium <strong>and</strong> Nd were separated using<br />
ion exchange techniques, <strong>and</strong> isotopic ratios were determined<br />
with a five collector Finnigan MAT 261 TIMS. Total<br />
analytical blank for Nd was approximately 40 pg, <strong>and</strong> Nd<br />
was analysed as NdO + in multidynamic mode, with the<br />
data being reduced assuming exponential fractionation<br />
<strong>and</strong> normalized to 146 Nd/ 144 Nd = 0.7219. External precision<br />
was obtained by repeated analyses <strong>of</strong> the La Jolla st<strong>and</strong>ard<br />
with 143 Nd/ 144 Nd = 0.511853 ± 0.000025 (2r, n = 10).<br />
Samarium was analyzed as an oxide in static mode <strong>and</strong> normalized<br />
assuming 149 Sm/ 152 Sm = 0.51686. Uncertainties in<br />
TIMS determination <strong>of</strong> Sm <strong>and</strong> Nd concentrations are estimated<br />
to about 1% Al 2 O 3 <strong>and</strong> almost 15% MnO <strong>and</strong><br />
will be discussed in detail below. Unless otherwise noted,<br />
results refer only to the remaining 15 IF samples. Normative<br />
quartz <strong>and</strong> iron-oxides combined represent 96–99%<br />
(wt.) <strong>of</strong> the samples, with Fe ranging from 10% to 81%.<br />
Aluminium, Ca, Mg, Ti, <strong>and</strong> P are present only in trace<br />
amounts, <strong>and</strong> Na <strong>and</strong> K are absent or not measurable.<br />
Other than Si <strong>and</strong> Fe, only Mn is present in appreciable<br />
amounts <strong>of</strong> the major elements studied, <strong>and</strong> averages<br />
1%. The mineralogy <strong>of</strong> the iron oxide component for all<br />
samples varies between magnetite <strong>and</strong> hematite, with the<br />
relative proportion <strong>of</strong> magnetite increasing with increasing<br />
Fe content. The absence <strong>of</strong> chlorite or secondary minerals<br />
indicates that the samples have experienced little alteration<br />
or weathering. Immobile trace element concentrations in<br />
the IF samples are low (
Table 1<br />
Major element data for iron formation <strong>and</strong> shales, with mineral phase determinations for IF samples<br />
Iron-formation samples<br />
Shale samples a<br />
WM41 WM42 WM51 WM52 WM61 WM62 WM71 WM72 WM101 WM102 WM111 WM112 WM121 WM122 WM131 WM132 WM2 WM8 WM18 WM19 pelite b<br />
XRF (wt%)<br />
SiO 2 75.4 26.2 75.7 16.8 19.7 20.7 58.8 70.8 80.9 78.4 80.3 81.0 85.8 88.0 84.4 83.6 51.6 47.9 49.1 48.6 58<br />
Al 2 O 3 0.13 1.22 0.18 0.07 0.67 0.39 0.21 0.02 0.52 0.22 nd nd 0.06 0.06 0.34 0.10 23.3 11.9 24.7 26.8 18.8<br />
Fe 2 O 3 22.2 57.0 21.0 81.0 76.4 76.3 37.5 26.1 15.2 17.3 17.7 18.1 12.0 10.4 14.7 15.0 10.5 29.0 9.16 6.55 9.4<br />
MnO 0.92 14.5 1.85 0.34 1.54 2.50 1.93 2.39 2.11 2.57 1.59 1.27 1.14 1.30 0.35 0.63 0.12 0.43 0.13 0.11 0.27<br />
MgO 0.12 0.25 0.26 0.13 0.20 0.22 0.23 0.27 0.33 0.37 0.30 0.26 0.21 0.19 0.13 0.13 3.01 3.57 3.65 3.18 1.68<br />
CaO 0.15 0.18 0.18 0.18 0.11 0.17 0.24 0.17 0.14 0.21 0.16 0.18 0.14 0.12 0.11 0.11
Table 2<br />
Trace element concentrations in iron-formation <strong>and</strong> shale samples (mg/kg)<br />
Iron-formation samples<br />
Shale samples<br />
WM41 WM42 WM51 WM52 WM61 WM62 WM71 WM72 WM101 WM102 WM111 WM112 WM121 WM122 WM131 WM132 WM2 WM8 WM18 WM19 pelite a<br />
Rb
384 B.W. Alex<strong>and</strong>er et al. / Geochimica et Cosmochimica Acta 72 (2008) 378–394<br />
considerably lower (average 1.13). The REY patterns are<br />
not correlated with major element chemistry, nor with<br />
immobile trace element content, but the two groups <strong>of</strong> patterns<br />
can be distinguished on the basis <strong>of</strong> relative stratigraphic<br />
position; i.e., high (Dy/Yb) WMS corresponds to<br />
samples collected at the base <strong>and</strong> at the top <strong>of</strong> the sequence<br />
REY/WMS<br />
REY/WMS<br />
1<br />
0.1<br />
0.01<br />
0.001<br />
1<br />
0.1<br />
0.01<br />
top <strong>of</strong> sequence<br />
bottom <strong>of</strong> sequence<br />
La Ce Pr Nd Pm Sm Eu Gd Tb Dy Y Ho Er Tm Yb Lu<br />
middle <strong>of</strong> sequence<br />
WM42<br />
WM132<br />
WM131<br />
WM62<br />
WM61<br />
WM52<br />
WM51<br />
WM42<br />
WM41<br />
WM122<br />
WM121<br />
WM112<br />
WM111<br />
WM102<br />
WM101<br />
WM72<br />
WM71<br />
studied, <strong>and</strong> low (Dy/Yb) WMS is found in samples collected<br />
from the middle <strong>of</strong> the sequence.<br />
The REY patterns <strong>of</strong> all the IF samples exhibit WMSnormalized<br />
positive La, Eu, <strong>and</strong> Gd anomalies. Calculated<br />
(La/La * ) WMS <strong>and</strong> (Gd/Gd * ) WMS are above unity for all 16<br />
IF samples (Fig. 3), <strong>and</strong> (Eu/Eu * ) WMS ranges from 1.51 to<br />
2.09 (average 1.69). The magnitude <strong>of</strong> the La, Eu, <strong>and</strong> Gd<br />
anomalies are indistinguishable on the basis <strong>of</strong> stratigraphic<br />
position. Y/Ho averages 42 for all samples, <strong>and</strong> higher Y/<br />
Ho values are observed for samples collected from the middle<br />
<strong>of</strong> the sequence, though the difference between the two<br />
groups is small when samples WM121 <strong>and</strong> WM122 (Y/Ho<br />
<strong>of</strong> 61 <strong>and</strong> 66, respectively) are not considered.<br />
The REY pattern <strong>of</strong> sample WM42 is similar to samples<br />
obtained stratigraphically above <strong>and</strong> below it, possessing<br />
positive La, Gd, <strong>and</strong> Y anomalies. The RREY content is<br />
the highest <strong>of</strong> the IF samples, <strong>and</strong> though most major element<br />
concentrations for sample WM42 are comparable to<br />
the other IF samples, Al 2 O 3 is enriched at 1.2% <strong>and</strong> MnO<br />
is significantly higher (14.5%). WM42 also contains somewhat<br />
greater concentrations <strong>of</strong> Rb, <strong>and</strong> significantly more<br />
Zr, Hf <strong>and</strong> Th than any <strong>of</strong> the other 15 IF samples.<br />
Seven iron-formation samples <strong>and</strong> three shale samples<br />
were analyzed using TIMS, <strong>and</strong> Nd <strong>and</strong> Sm isotope data<br />
are presented in Table 3. Six <strong>of</strong> the samples (one shale<br />
<strong>and</strong> five IF samples) display similar 147 Sm/ 144 Nd between<br />
0.1164 <strong>and</strong> 0.1486, values common to other Archean shales<br />
<strong>and</strong> iron formation (e.g., Miller <strong>and</strong> O’Nions, 1985; Alibert<br />
<strong>and</strong> McCulloch, 1993; Jahn <strong>and</strong> Condie, 1995; Bau et al.,<br />
1997). Of the remaining four samples, two are iron formations<br />
displaying the highest 147 Sm/ 144 Nd observed (>0.17),<br />
whereas the lowest 147 Sm/ 144 Nd values are observed in the<br />
remaining two shale samples (
Nd isotopes in 2.9 Ga Archean surface seawater 385<br />
Table 3<br />
Nd <strong>and</strong> Sm concentrations (mg/kg) <strong>and</strong> isotopic data determined by TIMS for 2.9 Ga Mozaan shales <strong>and</strong> iron-formations<br />
Sample Nd (ppm) Sm (ppm)<br />
143 Nd/ 144 Nd ± 2r a 147 Sm/ 144 Nd ± 2r Nd (0) Nd (2.9 Ga)<br />
Shale<br />
WM2 30.7 6.34 0.511120 ± 25 0.1247 ± 6 29.61 2.7 ± 0.3<br />
WM8 8.19 1.33 0.510543 ± 25 0.0979 ± 5 40.87 4.0 ± 0.6<br />
WM18 64.7 9.70 0.510389 ± 25 0.0905 ± 5 43.87 4.2 ± 0.3<br />
Iron formation<br />
WM41 0.520 0.148 0.511754 ± 25 0.1721 ± 9 17.24 8.1 ± 0.7<br />
WM51 0.457 0.112 0.511587 ± 25 0.1486 ± 7 20.50 2.6 ± 0.8<br />
WM61 1.76 0.362 0.511029 ± 25 0.1242 ± 6 31.39 4.3 ± 0.6<br />
WM72 0.533 0.104 0.511203 ± 25 0.1302 ± 7 27.99 3.2 ± 0.6<br />
WM102 0.659 0.146 0.511322 ± 25 0.1330 ± 7 25.67 1.9 ± 0.7<br />
WM121 0.194 0.037 0.510921 ± 25 0.1164 ± 6 33.49 3.5 ± 0.5<br />
WM131 0.762 0.215 0.511698 ± 25 0.1766 ± 9 18.34 10.9 ± 0.6<br />
WM131r b 0.748 0.218 0.511734 ± 25 0.1763 ± 9 17.63 10.1 ± 0.8<br />
BCR-2 (3 mg) 28.8 6.6 0.512619 ± 25 0.1378 ± 7<br />
a The errors in the 143 Nd/ 144 Nd refer to the last digits <strong>and</strong> are estimated from repeated analysis <strong>of</strong> the La Jolla st<strong>and</strong>ard which gave<br />
143 Nd/ 144 Nd = 0.511853 ± 0.000025 (2r, n = 10).<br />
b Replicate digestion, column separation, <strong>and</strong> analysis <strong>of</strong> sample WM131.<br />
143 Nd/ 144 Nd <strong>of</strong> the sample, <strong>and</strong> 143 Nd/ 144 Nd within a chondritic<br />
uniform reservoir (CHUR) at any given time t. For<br />
the shale samples, Nd (t) falls within a narrow range from<br />
2.74 to 4.24, whereas the IF samples display a bimodal<br />
Nd (t) distribution, with most samples exhibiting Nd (t) similar<br />
to the shales, while the two IF samples with<br />
147 Sm/ 144 Nd >0.17 (WM41 <strong>and</strong> WM131) have significantly<br />
more negative Nd (t) values <strong>of</strong> 8.1 <strong>and</strong> 10.9, respectively.<br />
5. DISCUSSION<br />
5.1. Depositional controls on REYs in Mozaan IFs<br />
The Mozaan IFs REY distributions might have been affected<br />
by a number <strong>of</strong> processes including: (i) post- <strong>and</strong> syndepositional<br />
processes, (ii) the degree to which the REY<br />
were scavenged by particulate matter in the water column<br />
prior to deposition, <strong>and</strong> (iii) also by the various inputs to<br />
the seawater from which the IFs precipitated.<br />
Potential post-depositional mechanisms affecting trace<br />
metal <strong>and</strong> REY distribution in the Mozaan IFs include diagenetic<br />
<strong>and</strong> metamorphic processes. During diagenesis,<br />
mobility <strong>of</strong> the REY would tend to average out REY distributions<br />
in b<strong>and</strong>ed iron formations, yet Bau <strong>and</strong> Dulski<br />
(1992) observed highly variable REY ratios in individual<br />
Fe- <strong>and</strong> quartz-rich b<strong>and</strong>s within the 2.46 Ga Superior-type<br />
Kuruman IF. Such variation is also present in REY data<br />
for IF mesob<strong>and</strong>s from the Dales Gorge Member in the<br />
Hamersley Group <strong>of</strong> Australia (Morris, 1993). Bau (1993)<br />
studied the impact <strong>of</strong> post-burial processes on REY signatures<br />
in Precambrian IFs from the Hamersley Basin in Western<br />
Australia, the Kuruman <strong>and</strong> Penge IFs in South<br />
Africa, <strong>and</strong> the Broomstock IFs <strong>of</strong> Zimbabwe, <strong>and</strong> concluded<br />
that REYs are effectively immobile during diagenesis<br />
<strong>and</strong> lithification.<br />
The effects <strong>of</strong> metamorphism on REY mobility is a function<br />
<strong>of</strong> water/rock ratios, with LREE depletion <strong>and</strong> negative<br />
Eu anomalies expected in rocks that host significant<br />
amounts <strong>of</strong> metasomatic fluids during metamorphism<br />
(Grauch, 1989; Bau, 1993). Iron-formation samples from<br />
regions that have experienced high grade metamorphism<br />
(e.g., Isua, Greenl<strong>and</strong>) do not exhibit Eu or LREE depletion,<br />
<strong>and</strong> in general coeval detritus free IF samples display<br />
similar REY CN patterns regardless <strong>of</strong> metamorphic grade<br />
(Bau, 1991, 1993). Therefore, the relatively low-grade<br />
greenschist facies metamorphism which affected Mozaan<br />
IF samples would not have had a significant effect on their<br />
REY distribution.<br />
For syn-depositional processes, REY patterns in chemical<br />
precipitates may be fractionated with respect to contemporaneous<br />
seawater due to the presence <strong>of</strong> detrital<br />
aluminosilicates, or as a result <strong>of</strong> exchange effects between<br />
REY scavenging particulates <strong>and</strong> marine waters. The effect<br />
<strong>of</strong> clastic contamination can be assessed using trace elements<br />
that are essentially immobile in aqueous solutions,<br />
<strong>and</strong> hence occur only at ultra-trace levels in seawater. The<br />
low abundances in the Mozaan IFs <strong>of</strong> such elements as<br />
Al, Ti, Zr, Hf, Y, <strong>and</strong> Th demonstrates that these are very<br />
pure chemical sediments <strong>and</strong> that clastic contamination is<br />
negligible, which is consistent with a sediment-starved continental<br />
shelf depositional environment for the Mozaan IFs<br />
(e.g., Beukes <strong>and</strong> Cairncross, 1991).<br />
The influence <strong>of</strong> solution complexation, fractionation,<br />
<strong>and</strong> scavenging within the marine environment on REY<br />
distributions in IFs is more difficult to assess. The REY<br />
concentration in modern seawater is controlled primarily<br />
by scavenging particulate matter (e.g., Elderfield, 1988; Erel<br />
<strong>and</strong> Stolper, 1993), to the extent that seawater abundances<br />
<strong>of</strong> REYs are extremely low. The association between Ferich<br />
colloids <strong>and</strong> REYs has led many workers to conclude<br />
that Fe-oxyhydroxide particles dominated REY scavenging<br />
during the formation <strong>of</strong> ancient metalliferous sediments<br />
(e.g., Derry <strong>and</strong> <strong>Jacobs</strong>en, 1990). This scavenging should<br />
represent the balance between two competing effects; the<br />
solution complexation <strong>of</strong> the REYs <strong>and</strong> the Fe-oxyhydroxide<br />
surface complexation. This model assumes that the reac-
386 B.W. Alex<strong>and</strong>er et al. / Geochimica et Cosmochimica Acta 72 (2008) 378–394<br />
tion kinetics <strong>of</strong> adsorption/desorption are significantly faster<br />
than the particle residence time in an oxic water column.<br />
This assumption is supported by experimental evidence<br />
indicating that particle-surface/solution REY exchange<br />
equilibrium occurs within minutes (Bau, 1999), suggesting<br />
that most Fe-oxyhydroxide particles settling through an<br />
oxic water column (<strong>of</strong> relatively constant REY distribution)<br />
should be in equilibrium with the solution phase.<br />
Natural systems that provide insight into such equilibrium<br />
exchange processes are available in the form <strong>of</strong> modern<br />
marine hydrogenetic ferromanganese crusts <strong>and</strong><br />
terrestrial spring-water precipitates. Seafloor Fe–Mn crusts<br />
have trace metal distributions that reflect equilibrium<br />
adsorption–desorption processes between REYs <strong>and</strong> seawater,<br />
processes that strongly fractionate REYs. Very similar<br />
fractionation is observed between Fe-oxyhydroxides<br />
<strong>and</strong> terrestrial spring water (Bau et al., 1998). The most<br />
striking difference between these precipitates <strong>and</strong> the<br />
respective fluids which formed them is the lower Y/Ho observed<br />
in the precipitates, which display negative Y anomalies.<br />
This Y–Ho fractionation is due to the preferential<br />
sorption <strong>of</strong> Ho relative to Y on the scavenging Fe(–Mn)<br />
particles (e.g., Bau et al., 1996, 1998). Preferential adsorption<br />
<strong>of</strong> Ho over Y on Fe-oxyhydroxides has been demonstrated<br />
experimentally (Bau, 1999) <strong>and</strong> is completely<br />
consistent with the observed fractionation <strong>of</strong> Y <strong>and</strong> Ho<br />
during precipitation <strong>of</strong> hydrogenetic Fe–Mn crusts from<br />
seawater. We emphasize that such Y–Ho fractionation is<br />
not observed in Archean iron formations.<br />
If redox conditions in Archean seawater permitted the<br />
precipitation <strong>of</strong> ferric iron, <strong>and</strong> if scavenging Fe-oxyhydroxide<br />
particles achieved exchange equilibrium with surrounding<br />
seawater, the above observations predict subchondritic<br />
values <strong>of</strong> Y/Ho in the IFs. The significantly higher<br />
Y/Ho observed in IFs <strong>of</strong> different ages strongly suggests<br />
that Fe-oxyhydroxide particles that scavenged REYs could<br />
not be at or near exchange equilibria with respect to ambient<br />
seawater (Bau <strong>and</strong> Dulski, 1996). Although the actual<br />
reason why IFs display the REY seawater pattern has not<br />
been satisfactorily explained <strong>and</strong> requires further investigation,<br />
the striking similarity between REY patterns in Archean<br />
marine carbonates <strong>and</strong> IFs suggests that pure Archean<br />
IFs faithfully record the REY distributions <strong>of</strong> contemporaneous<br />
seawater.<br />
5.2. Archean seawater<br />
5.2.1. High-temperature hydrothermal input<br />
The geologic setting <strong>and</strong> low detritus content, coupled<br />
with the observed LREE depletions <strong>and</strong> positive La, Gd,<br />
<strong>and</strong> Y anomalies (Figs. 2 <strong>and</strong> 5), strongly suggests the Mozaan<br />
IFs represent 3.0 Ga seawater. Working under the<br />
assumption that these sediments possess trace element distributions<br />
reflective <strong>of</strong> contemporaneous seawater, some<br />
general constraints may be placed on the composition <strong>of</strong><br />
Archean seawater.<br />
The occurrence <strong>of</strong> the oxide-facies Mozaan IFs suggests<br />
that the redox level <strong>of</strong> even very shallow seawater permitted<br />
nearly continuous precipitation <strong>of</strong> significant amounts <strong>of</strong><br />
Fe(III)-bearing (hydr)oxides, while the lack <strong>of</strong> any Ce<br />
anomaly in normalized data indicates that no process operated<br />
during Fe-sedimentation that was capable <strong>of</strong> fractionating<br />
Ce. This scenario is clearly different from the oxic<br />
modern marine system, in which Ce is oxidized to immobile<br />
Ce(IV) on particle surfaces, resulting in fractionation from<br />
the other REY <strong>and</strong> producing the strong negative Ce anomaly<br />
observed in modern seawater. The lack <strong>of</strong> significant<br />
negative-Ce anomalies in many Archean chemical sediments<br />
has been discussed by numerous authors in terms<br />
<strong>of</strong> the oxidation history <strong>of</strong> the Earth’s surface (e.g., Fryer,<br />
1977; Derry <strong>and</strong> <strong>Jacobs</strong>en, 1990; Kato et al., 1998; among<br />
others). While the exact process that converted large<br />
amounts <strong>of</strong> Fe(II) to Fe(III) in seawater during sedimentation<br />
<strong>of</strong> the Mozaan IFs is unknown, the Ce distributions in<br />
these chemical precipitates indicate Archean seawater did<br />
not possess a negative Ce anomaly, <strong>and</strong> correspondingly,<br />
redox levels were lower than those observed in modern<br />
marine systems.<br />
Unlike Ce, the Mozaan IFs display pronounced positive<br />
Eu anomalies (Fig. 4), which in modern marine environments<br />
are only observed in high-temperature (>250 °C)<br />
hydrothermal fluids, such as those typically found at midocean<br />
ridges <strong>and</strong> back-arc spreading centers. It is reasonable<br />
that the REY CN patterns <strong>of</strong> these high-T fluids in<br />
the Early Precambrian were similar to those observed today<br />
<strong>and</strong> exhibited Eu anomalies >1 <strong>and</strong> enrichment in LREEs<br />
(Bau <strong>and</strong> Möller, 1993), features which are characteristic<br />
(Sm/Yb) CN<br />
10.0<br />
Pongola pelites<br />
PAAS<br />
1.0<br />
Pongola IF<br />
Penge IF<br />
Kuruman IF<br />
Isua IF<br />
Pacific seawater<br />
high-T hydrothermal<br />
fluids<br />
0.3<br />
0.25 0.50 0.75 1.00 1.25 1.50 1.75 2.00 2.25 5 10 15 20 25<br />
(Eu/Eu*) CN<br />
Fig. 4. Plot <strong>of</strong> chondrite-normalized Sm/Yb <strong>and</strong> Eu/Eu * for<br />
Mozaan IFs, <strong>and</strong> including data for 3.7 Ga Isua IFs (Bolhar et al.,<br />
2004), 2.5 Ga Kuruman <strong>and</strong> Penge IFs (Bau <strong>and</strong> Dulski, 1996),<br />
shallow (350 °C, Bau <strong>and</strong> Dulski, 1999). Note<br />
break in horizontal axis. Isua samples are those considered by<br />
Bolhar et al. (2004) to reflect contemporaneous seawater. The grey<br />
shaded area represents the range <strong>of</strong> values for 62 pelites from the<br />
Pongola Supergroup sampled by Wronkiewicz (1989), <strong>and</strong> the<br />
crossed-square represents Post-Archean Average Shale (PAAS,<br />
McLennan, 1989). The Mozaan IFs exhibit Eu/Eu * between that <strong>of</strong><br />
the Isua <strong>and</strong> Kuruman IFs, yet display Sm/Yb values similar to<br />
continental crust <strong>and</strong> significantly higher than any <strong>of</strong> the other IFs<br />
(except for two Isua <strong>and</strong> one Kuruman sample). All IF samples<br />
have significantly lower Sm/Yb <strong>and</strong> Eu/Eu * than high-T hydrothermal<br />
fluids.
Nd isotopes in 2.9 Ga Archean surface seawater 387<br />
<strong>of</strong> the Mozaan IF samples (Fig. 4). The presence <strong>of</strong> positive<br />
Eu anomalies is clear evidence that REY from high-T<br />
hydrothermal inputs influenced the continental shelf environment<br />
that hosted the Mozaan IFs.<br />
For constraints on ancient hydrothermal fluxes, modern<br />
black smoker fluids <strong>of</strong>fer the best analogs for the trace<br />
element compositions <strong>of</strong> high-T fluids which contributed<br />
to the REY distribution in Archean iron formations.<br />
For the Mozaan IFs, the magnitude <strong>of</strong> the positive Eu<br />
anomalies generally compare well with data for other<br />
IFs (Fig. 5). However, the general REY distributions <strong>of</strong><br />
the Mozaan IFs are significantly different when compared<br />
to well studied examples such as the 3.7 Ga Isua IF <strong>of</strong><br />
Greenl<strong>and</strong> (Bolhar et al., 2004) <strong>and</strong> the 2.5 Ga Kuruman<br />
IF from South Africa (Bau <strong>and</strong> Dulski, 1996), though<br />
regardless <strong>of</strong> age, these iron formations display the seawater<br />
characteristics <strong>of</strong> marine chemical sediments (Fig. 5).<br />
Conservative mixing calculations indicate that a high-T<br />
hydrothermal fluid contribution <strong>of</strong> less than 0.1% is adequate<br />
for producing the Eu/Sm ratios observed within<br />
the Pongola <strong>and</strong> Kuruman IFs, <strong>and</strong> reasonably accounts<br />
for most <strong>of</strong> the Eu/Sm ratios observed in the Isua IFs<br />
REY/WMS<br />
10 1<br />
10 0<br />
high-T hydrothermal fluid (x10 5 )<br />
10 -1<br />
10 -2<br />
10 -3<br />
Isua IF (x2)<br />
Kuruman IF<br />
top/bottom IFs<br />
middle IFs<br />
388 B.W. Alex<strong>and</strong>er et al. / Geochimica et Cosmochimica Acta 72 (2008) 378–394<br />
fluxes <strong>of</strong> trace elements would be expected to affect the<br />
REY distribution <strong>of</strong> the bottom water component, <strong>and</strong><br />
though such fluxes are at best difficult to characterize, it<br />
can be ruled out that such low-T input would have displayed<br />
positive Eu anomalies.<br />
Y/Ho<br />
Y/Ho<br />
Sm/Yb<br />
100<br />
80<br />
60<br />
40<br />
28<br />
20<br />
100<br />
80<br />
60<br />
40<br />
28<br />
20<br />
30<br />
10<br />
1<br />
seawater<br />
hydrogenetic<br />
Fe-Mn crusts<br />
0.1% hydrothermal<br />
fluid 1% hydrothermal<br />
fluid<br />
5% hydrothermal<br />
fluid<br />
high-T<br />
hydrothermal fluids<br />
0.1 1 10<br />
Eu/Sm<br />
seawater<br />
10<br />
0.2 1 10 20<br />
Sm/Yb<br />
hydrogenetic<br />
Fe-Mn crusts<br />
seawater<br />
hydrogenetic<br />
Fe-Mn crusts<br />
Pongola IF (2.9 Ga)<br />
Kuruman IF (2.5 Ga)<br />
Isua IF (~3.7 Ga)<br />
modern seawater<br />
modern hydrogenetic Fe-Mn crusts<br />
high-T hydrothermal fluids<br />
1% hydrothermal<br />
fluid<br />
0.1% hydrothermal<br />
fluid<br />
5% hydrothermal<br />
fluid<br />
high-T<br />
hydrothermal fluids<br />
high-T<br />
hydrothermal fluids<br />
5% hydrothermal<br />
fluid<br />
1% hydrothermal<br />
fluid<br />
0.3<br />
0.1 1<br />
10<br />
Eu/Sm<br />
5.2.3. Surface seawater input<br />
The major remaining REY input affecting the Mozaan<br />
IFs would be Archean surface seawater. Rare earth element<br />
distributions in modern seawater, while exhibiting<br />
local variability <strong>and</strong> distinct trends with depth, are generally<br />
consistent with the pattern shown in Fig. 5, <strong>and</strong> exhibit<br />
a (Sm/Yb) CN ratio <strong>of</strong> 0.8. A study by Elderfield<br />
et al. (1990) reported REY data for five coastal seas<br />
(salinity > 20‰) that exhibited (Sm/Yb) CN ranging from<br />
0.70 to 1.24, consistent with coastal seawater from the<br />
East Frisian isl<strong>and</strong>s (North Sea) that displays (Sm/Yb) CN<br />
between 0.61 <strong>and</strong> 0.73 (Kulaksiz <strong>and</strong> Bau, 2007). Elderfield<br />
et al., 1990 also presented REY data for waters<br />
from six estuaries <strong>and</strong> 15 rivers. The estuarine waters<br />
(salinity < 10‰) displayed (Sm/Yb) CN that ranged from<br />
0.63 to 4.74, while (Sm/Yb) CN <strong>of</strong> the river waters varied<br />
between 0.96 <strong>and</strong> 4.74. Much <strong>of</strong> the REY load transported<br />
by modern rivers is removed in estuaries by the<br />
settling <strong>of</strong> suspended detrital particles, <strong>and</strong> by the salinity-induced<br />
coagulation <strong>and</strong> settling <strong>of</strong> colloidal particles<br />
rich in the particle reactive REYs (e.g., Sholkovitz,<br />
1994), processes that homogenize the widely varying<br />
REY distributions observed in modern rivers to produce<br />
the REY pattern typical <strong>of</strong> modern seawater. Elderfield<br />
et al. (1990) deduced that the colloidal fraction carried<br />
by many rivers would be enriched in the MREEs<br />
(shale-normalized) relative to the light <strong>and</strong> heavy REYs,<br />
similar to patterns observed for the Mozaan IFs sampled<br />
from the bottom <strong>and</strong> top <strong>of</strong> the sequence (Fig. 2), which<br />
represent iron formation deposited under the shallowest<br />
sea-level conditions. Studies <strong>of</strong> the Kalix river in Sweden<br />
demonstrated that colloidal particles dominate the transport<br />
<strong>and</strong> export <strong>of</strong> rare earth elements (Ingri et al., 2000),<br />
<strong>and</strong> these colloids consist <strong>of</strong> two fractions, one Fe-rich<br />
<strong>and</strong> another C-rich (Andersson et al., 2006). General<br />
enrichments in MREEs for Fe-rich organic colloids have<br />
been observed in the Hudson <strong>and</strong> Connecticut rivers<br />
(Sholkovitz, 1994), <strong>and</strong> the presence <strong>of</strong> MREE-enriched<br />
river waters has also been documented by Sholkovitz<br />
<strong>and</strong> Szymczak (2000) <strong>and</strong> Hannigan <strong>and</strong> Sholkovitz<br />
(2001). The speculation that riverine-derived colloidal<br />
particles enriched in the MREEs could exert a strong<br />
influence on REY distributions in the Mozaan IFs relies<br />
on similar processes controlling REY distributions in<br />
b<br />
Fig. 6. Elemental ratio plots for data sets presented in Fig. 5, with<br />
two-component conservative mixing lines for Eu/Sm, Sm/Yb, <strong>and</strong><br />
Y/Ho (symbols for all plots defined in c): (a) Y/Ho versus Eu/Sm,<br />
showing that a 0.1% high-T hydrothermal (>350 °C, Bau <strong>and</strong><br />
Dulski, 1999) fluid contribution to waters with shallow (
Nd isotopes in 2.9 Ga Archean surface seawater 389<br />
both modern <strong>and</strong> Archean rivers <strong>and</strong> estuaries. However,<br />
such processes have not yet been supported by strong evidence<br />
from the geologic record, <strong>and</strong> thus the Mozaan IFs<br />
are a first hint at their presence in the Archean.<br />
10<br />
5<br />
depleted mantle<br />
5.3. Sources <strong>of</strong> Nd to Archean seawater<br />
5.3.1. Nd in Archean iron formations<br />
To further constrain solute sources supplying the margin<br />
<strong>of</strong> the Kaapvaal craton 2.9 Ga, three <strong>of</strong> the shale samples<br />
<strong>and</strong> seven <strong>of</strong> the iron-formation samples were selected for<br />
Nd isotopic analyses (Table 3). Neodymium isotopic variations<br />
in Archean IFs have been examined by previous<br />
workers in attempts to distinguish the source <strong>of</strong> REY <strong>and</strong><br />
Fe in these sediments. Miller <strong>and</strong> O’Nions (1985) concluded<br />
that continental Nd dominated the isotopic signature <strong>of</strong><br />
Precambrian IFs. <strong>Jacobs</strong>en <strong>and</strong> Pimentel-Klose (1988a) discussed<br />
the likelihood that Archean surface seawater displayed<br />
continental signatures derived from fluvial inputs,<br />
but concluded that these inputs were an insignificant source<br />
<strong>of</strong> Nd <strong>and</strong> Fe to Archean IFs <strong>and</strong> that both Nd <strong>and</strong> Fe were<br />
almost exclusively sourced from a hydrothermal component<br />
similar to the flux emanating from modern mid-ocean<br />
ridges. Subsequent studies also suggested REY in Archean<br />
IFs were derived from sources with depleted-mantle Nd isotopic<br />
ratios (Shimizu et al., 1990), <strong>and</strong> Alibert <strong>and</strong> McCulloch<br />
(1993) proposed that 50% <strong>of</strong> Nd in 2.5 Ga<br />
Hamersley IFs was derived from Archean mid-ocean ridge<br />
sources with Nd (t) = +4. A similar conclusion was drawn<br />
for the 2.5 Ga Kuruman <strong>and</strong> Penge IFs deposited on the<br />
Kaapvaal craton (Bau et al., 1997).<br />
Neodymium isotopic data for Precambrian IFs vary,<br />
with samples generally possessing Nd (t) between 5 <strong>and</strong><br />
+5 (Fig. 7). Therefore, it seems likely that the seawater<br />
from which these IFs precipitated possessed variable Nd<br />
values, <strong>and</strong> that Archean–Paleoproterozoic oceans commonly<br />
ranged within ±5 Nd (t)-units relative to CHUR.<br />
The IF samples from this study exhibit Nd (2.9 Ga) values<br />
that range over 9 -units, <strong>and</strong> the Nd isotope ratios observed<br />
in the Mozaan IFs are different from most mid- to<br />
late-Archean IFs, which tend towards chondritic or positive<br />
Nd (t) values. One highly negative Nd (t) value <strong>of</strong> 8.1 for<br />
3.45 Ga iron formation from the Fig Tree/Moodies sequence<br />
<strong>of</strong> the Barberton greenstone belt (BGB) was reported<br />
by Miller <strong>and</strong> O’Nions (1985). This sample (SL 28<br />
B) was a small fragment from an individual magnetite b<strong>and</strong><br />
within the whole-rock sample SL 28 A, which alternatively<br />
displayed Nd (t) = +1.1, similar to Nd (t) = +0.3 for a second<br />
whole-rock sample analysed by these authors. Two<br />
additional analyses <strong>of</strong> Fig Tree Group IF by <strong>Jacobs</strong>en<br />
<strong>and</strong> Pimentel-Klose (1988b) reported Nd (t) <strong>of</strong> +2.9 <strong>and</strong><br />
+4.1. If the highly negative value <strong>of</strong> 8.1 reported by Miller<br />
<strong>and</strong> O’Nions (1985) for sample SL 28 B is atypical <strong>of</strong><br />
Barberton IF, then these data suggest that IFs deposited<br />
prior to 3.0 Ga on or near the Kaapvaal craton would likely<br />
display chondritic or slightly positive Nd (t). This Nd isotopic<br />
behavior is also observed in younger South African IFs,<br />
as samples from the 2.5 Ga Kuruman <strong>and</strong> Penge IFs <strong>of</strong> the<br />
Transvaal Supergroup display a narrow range <strong>of</strong> Nd (2.5)<br />
between 0.2 <strong>and</strong> +1.9 (Bau et al., 1997).<br />
ε Nd (t)<br />
0<br />
-5<br />
-10<br />
-15<br />
continental crust<br />
Mozaan IFs (this study)<br />
Australia<br />
Southern Africa<br />
North America<br />
Greenl<strong>and</strong><br />
2.0 2.2 2.4 2.6 2.8 3.0 3.2 3.4 3.6 3.8 4.0<br />
stratigraphic age (Ga)<br />
Fig. 7. Initial Nd (t) values for Precambrian IFs from literature<br />
data. The grey area in the top portion <strong>of</strong> the figure represents Nd (t)<br />
values predicted for Nd isotopic evolution in a depleted mantle<br />
region with an intial Nd (4.56 Ga) = 0 <strong>and</strong> Nd (0) = +10, while the<br />
dashed line in the lower portion represents a simple continental<br />
crust trend for Nd isotopic evolution where Nd (4.56 Ga) = 0 <strong>and</strong><br />
Nd (0) = 17. In order to facilitate comparison with the present<br />
study, data have been screened for samples described as oxidefacies<br />
IF by authors, <strong>and</strong> are from: Miller <strong>and</strong> O’Nions (1985),<br />
<strong>Jacobs</strong>en <strong>and</strong> Pimentel-Klose (1988a,b; recalculated using<br />
146 Nd/ 144 Nd = 0.7219), Shimizu et al. (1990), Alibert <strong>and</strong> McCulloch<br />
(1993, 12 Marra Mamba IFs), <strong>and</strong> Bau et al. (1997). Three<br />
analyses <strong>of</strong> Isua IFs by Shimizu et al. (1990) which display<br />
Nd (3.7) > +10 are not shown. Most Archean oxide facies iron<br />
formations that appear to have retained a primary isotopic Nd<br />
signature cluster within ±5 -units <strong>of</strong> Nd (t) = 0 (chondritic<br />
evolution), with notable exceptions being samples from Isua, <strong>and</strong><br />
two samples from this study (see text). Nd isotopic data from Frei<br />
et al. (1999) for Isua are not included, as they represent additional<br />
analyses <strong>of</strong> sample 242573 which was originally described by Miller<br />
<strong>and</strong> O’Nions (1985). Sample 242573 provided all but one Nd<br />
isotope analyses <strong>of</strong> Isua IFs as reported by Miller <strong>and</strong> O’Nions<br />
(1985) <strong>and</strong> Frei et al. (1999), <strong>and</strong> displays Nd (t) values that range<br />
from 8.6 to +14.8, leading Frei et al. (1999) to suggest that this<br />
sample did not retain a primary Nd isotopic signature.<br />
5.3.2. Nd in Mozaan shales<br />
The shale samples are correlated with respect to their<br />
Sm–Nd isotope systematics (Fig. 8), <strong>and</strong> these data suggest<br />
that the shales observed relatively closed-system behavior<br />
with respect to their Nd isotopic evolution. The shales have<br />
model ages (T DM ; <strong>Jacobs</strong>en <strong>and</strong> Pimentel-Klose, 1988a) <strong>of</strong><br />
3.4 ± 0.1 Ga, that agree well with U–Pb SHRIMP dates for<br />
individual zircon grains from the siliciclastic Dipka member<br />
underlying the IFs, which are predominantly >3.2 Ga as reported<br />
by Nhleko (2003). However, previous Nd isotope<br />
studies <strong>of</strong> Mozaan Group fine-grained clastic sediments<br />
have revealed a wide range <strong>of</strong> T DM ages <strong>of</strong> 1.52–4.73 Ga<br />
(Dia et al., 1990) <strong>and</strong> 2.92–4.63 Ga (Jahn <strong>and</strong> Condie,<br />
1995), leading Jahn <strong>and</strong> Condie (1995) to surmise that Pongola<br />
pelites had very diverse provenances or that some samples<br />
were affected by open-system behavior during<br />
diagenesis or low-grade metamorphism. This large range<br />
in Sm–Nd isotopic data was not observed, however, by Stevenson<br />
<strong>and</strong> Patchett (1990), who sampled three Mozaan<br />
shales from a single locality <strong>and</strong> reported Nd (t) values be-
390 B.W. Alex<strong>and</strong>er et al. / Geochimica et Cosmochimica Acta 72 (2008) 378–394<br />
143<br />
Nd/<br />
144<br />
Nd<br />
0.5125<br />
0.5120<br />
0.5115<br />
0.5110<br />
0.5105<br />
iron-formation<br />
shale<br />
T = 2.9 Ga<br />
ε Nd = -1<br />
WM8<br />
WM18<br />
0.5100<br />
-50<br />
0.06 0.08 0.10 0.12 0.14 0.16 0.18 0.20<br />
147<br />
Sm/ 144 Nd<br />
T = 2.9 Ga<br />
ε Nd = -5<br />
WM41<br />
WM131<br />
Fig. 8. Analytical results for Nd isotope analyses <strong>of</strong> Mozaan shales<br />
<strong>and</strong> iron-formation samples, including Nd (0) values. Reference<br />
isochrons for Nd (t) <strong>of</strong> 1 <strong>and</strong> 5 bracket all samples from this<br />
study, except for IF samples WM41 <strong>and</strong> WM131. The similarity in<br />
the Nd isotopic evolution for the majority <strong>of</strong> the IFs <strong>and</strong> the three<br />
shale samples suggests that REY sources for the shallow seawater<br />
which precipitated the Mozaan IFs were predominately derived<br />
from continental crust low in radiogenic Nd.<br />
tween 1.9 to 0.9 (Fig. 8). This suggests that high-resolution<br />
sampling (as in this study) <strong>of</strong> Mozaan fine-grained clastic<br />
sediment provides consistent Nd isotopic data<br />
supporting relatively homogeneous provenances over short<br />
geologic time spans.<br />
Sample WM8 exhibits a similar Si concentration compared<br />
to the other three Mozaan shales, but has a much<br />
higher Fe content (29% compared to 6–10%, respectively),<br />
<strong>and</strong> a correspondingly lower Al content. Since this alumininous<br />
Fe-rich sample resides between two IF b<strong>and</strong>s, it is<br />
possible that its major element composition reflects dilution<br />
with IF-type precipitates, <strong>and</strong> the Fe increase occurred at<br />
the expense <strong>of</strong> aluminium-rich phases. Samples WM8 <strong>and</strong><br />
WM18 have Sm <strong>and</strong> Nd concentrations that differ by more<br />
than a factor <strong>of</strong> six, yet display very similar 147 Sm/ 144 Nd<br />
<strong>and</strong> 143 Nd/ 144 Nd.<br />
5.3.3. Nd in Mozaan iron formations<br />
The good agreement in Sm–Nd isotope systematics between<br />
the shales <strong>and</strong> the majority <strong>of</strong> the IF samples<br />
(Fig. 8) argues against a significant REY component derived<br />
from a depleted mantle source contributing to local<br />
seawater. However, the IF samples do display the positive<br />
shale-normalized Eu anomalies typical <strong>of</strong> Archean IFs,<br />
which requires some high-T hydrothermal REY input. It<br />
appears that such a high-T input, however, did not overwhelm<br />
the terrestrial flux in terms <strong>of</strong> the Nd isotope signature.<br />
The relationship between Eu anomalies <strong>and</strong> Sm–Nd<br />
isotopic signatures in Archean iron-formations was examined<br />
by Derry <strong>and</strong> <strong>Jacobs</strong>en (1990), who considered twocomponent<br />
conservative mixing between East Pacific Rise<br />
hydrothermal fluids <strong>and</strong> estimated river water. These<br />
authors concluded that Eu fractionation due to scavenging<br />
-10<br />
-20<br />
-30<br />
-40<br />
ε Nd (0)<br />
<strong>of</strong> Eu 2+ near vent sites renders these anomalies unsuitable<br />
for mass balance calculations in IFs <strong>and</strong> that more reliable<br />
mass fraction estimates are obtained using the Nd isotopic<br />
mass balance. However, the inability to reconcile Nd <strong>and</strong><br />
Eu mass balance calculations from the mixing <strong>of</strong> hydrothermal<br />
<strong>and</strong> river/seawater endmembers is expected, as only the<br />
Nd isotopic signature provides information regarding the<br />
relative mass fractions <strong>of</strong> the REY sources; Eu anomalies<br />
only provide information regarding the temperature <strong>of</strong><br />
REY sources. An extreme example <strong>of</strong> the lack <strong>of</strong> correlation<br />
between mantle-like Nd isotopic signatures <strong>and</strong> positive<br />
Eu anomalies in high-T marine hydrothermal fluids<br />
(<strong>and</strong> hence, IFs) may be found in the data <strong>of</strong> Piepgras<br />
<strong>and</strong> Wasserburg (1985). These authors reported Nd isotopic<br />
data for a hydrothermal system developed on a sediment<br />
covered ridge in the Guaymas Basin. This system<br />
vented hydrothermal fluid in excess <strong>of</strong> 300 °C, which migrated<br />
through a sedimentary package several hundred meters<br />
thick <strong>and</strong> possessed a positive Eu anomaly coupled<br />
with Nd (0) = 11.4, indicating that marine fluids hot enough<br />
to fractionate Eu from other REY will display Nd isotopic<br />
signatures similar to the host rock (in this case,<br />
continentally-derived sediments). This illustrates that positive<br />
Eu anomalies cannot be used to constrain MORB-derived<br />
hydrothermal inputs, as they provide little<br />
information regarding the source <strong>of</strong> REYs to high-T fluids,<br />
but rather reflect only the temperature <strong>of</strong> the hydrothermal<br />
system.<br />
Of the seven IF samples, five appear to have experienced<br />
an Nd isotopic evolution similar to that <strong>of</strong> the shales, consistent<br />
with these samples (both shales <strong>and</strong> IFs) having received<br />
Nd from an isotopically similar source. Two <strong>of</strong> the<br />
Pongola IF samples, WM41 <strong>and</strong> WM131, are not consistent<br />
with a single isotopic source or closed-system conditions<br />
following deposition. These samples differ primarily<br />
in their high 147 Sm/ 144 Nd (>0.17), compared to a<br />
147 Sm/ 144 Nd average <strong>of</strong> 0.13 ± 0.01 for the remaining five<br />
IFs. Considering that WM41 <strong>and</strong> WM131 represent the<br />
first <strong>and</strong> last IF beds deposited, respectively, it is unlikely<br />
that post-depositional processes (e.g., metamorphism)<br />
would selectively affect only these samples. The<br />
147 Sm/ 144 Nd average <strong>of</strong> 0.13 ± 0.01 for five <strong>of</strong> the seven<br />
Mozaan IFs is identical to the average obtained<br />
(0.13 ± 0.03) from a survey <strong>of</strong> 61 analyses <strong>of</strong> predominantly<br />
oxide facies Archean <strong>and</strong> Paleoproterozoic IFs (Miller <strong>and</strong><br />
O’Nions, 1985; <strong>Jacobs</strong>en <strong>and</strong> Pimentel-Klose, 1988a,b; Shimizu<br />
et al., 1990; Alibert <strong>and</strong> McCulloch, 1993; Bau et al.,<br />
1997). Of these literature data, only three samples (two<br />
from the Hamersley Basin <strong>and</strong> one Barberton greenstone<br />
belt sample) have higher<br />
147 Sm/ 144 Nd than samples<br />
WM31 <strong>and</strong> WM141. Alibert <strong>and</strong> McCulloch (1993) reported<br />
data for 26 samples from the Dales Gorge <strong>and</strong> J<strong>of</strong>fre<br />
IFs, concluding on the basis <strong>of</strong> Nd isotopes that these IFs<br />
had suffered a metamorphic resetting <strong>of</strong> their isotopic system<br />
some 500 Ma after deposition, a conclusion supported<br />
by Rb–Sr <strong>and</strong> Pb–Pb isotopic analyses <strong>of</strong> the underlying<br />
Fortescue lavas (Nelson et al., 1992). The 147 Sm/ 144 Nd <strong>of</strong><br />
these 26 metamorphosed IF samples averages 0.12 ± 0.02.<br />
It therefore seems that metamorphic events in Archean<br />
IFs capable <strong>of</strong> altering Nd isotope ratios have little effect
Nd isotopes in 2.9 Ga Archean surface seawater 391<br />
on overall 147 Sm/ 144 Nd values, an observation that supports<br />
the view <strong>of</strong> Bau (1993), who concluded that REY<br />
exhibited little mobility during all but the most severe<br />
grades <strong>of</strong> metamorphism. This suggests that the high<br />
147 Sm/ 144 Nd observed in samples WM41 <strong>and</strong> WM131 is<br />
primary in origin. If these samples represent closed-system<br />
behavior, as the other IF <strong>and</strong> shale samples appear to, then<br />
they require a source with a lower initial 143 Nd/ 144 Nd (characteristic<br />
<strong>of</strong> continental crust), coupled with a Sm/Nd ratio<br />
typical <strong>of</strong> more mafic sources, <strong>and</strong> as such, they remain<br />
ambiguous.<br />
The IF samples that closely follow the Nd systematics <strong>of</strong><br />
the shales display the two distinct types <strong>of</strong> REY distributions<br />
discussed earlier. WM51 <strong>and</strong> WM61 exhibit the<br />
MREE to HREE depleted patterns typical at the onset <strong>of</strong><br />
IF deposition, whereas the remaining three samples possess<br />
REY patterns that are relatively flat from the MREE to<br />
HREE. The isotopic similarity between the IFs <strong>and</strong> shales<br />
suggests that soluble REY delivered to the iron-formation<br />
originated from contemporaneous continental crust. The<br />
least radiogenic shale sample displays Nd (2.9 Ga) = 4.2,<br />
so it is reasonable that local continental crust possessed a<br />
similar Nd (2.9 Ga) value. Assuming the mid-ocean ridge<br />
hydrothermal flux displayed Nd (2.9 Ga) = +4, <strong>and</strong> assigning<br />
the continental flux an Nd (2.9 Ga) value <strong>of</strong> 4.2, then<br />
Nd isotopic mass balance calculations require that between<br />
72% <strong>and</strong> 100% <strong>of</strong> the Nd in these IF samples ( Nd (2.9 Ga)<br />
<strong>of</strong> 1.9 to 4.3) must have originated from a source(s) possessing<br />
an Nd isotopic signature similar to local continental<br />
crust. If this was the case, this source should possess a REY<br />
distribution similar to that <strong>of</strong> the Mozaan shale average<br />
WMS, <strong>and</strong> deviations from the WMS pattern within a pure<br />
chemical sediment would originate from processes occurring<br />
solely within the water column (i.e., chemical complexation<br />
<strong>and</strong> scavenging on Fe-oxyhydroxide particles, etc.).<br />
The geologic evidence for a transgressive–regressive cycle<br />
<strong>of</strong> deposition, coupled with the evolution <strong>of</strong> the REY patterns<br />
in the Mozaan IFs with stratigraphic position, suggests<br />
that varying proportions <strong>of</strong> a continentally-derived<br />
solute source dominated trace element evolution in shallow<br />
seawater near the Kaapvaal craton 3.0 Ga ago. Therefore,<br />
it is possible that the MREE-enriched patterns observed in<br />
modern low salinity coastal seas <strong>and</strong> estuaries (Elderfield<br />
et al., 1990) represent processes analogous to those that<br />
controlled the REY distributions observed in the Mozaan<br />
IFs. Unlike modern environments, the importance <strong>of</strong> organic<br />
compounds <strong>and</strong> organic-based colloidal particles on<br />
REY distributions in Archean marine settings is difficult<br />
to estimate, though the presence <strong>of</strong> stromatolitic carbonate<br />
in the underlying Nsuze Group (Mason <strong>and</strong> von Brunn,<br />
1977; Beukes <strong>and</strong> Lowe, 1989) indicates that the margin<br />
<strong>of</strong> the Kaapvaal craton was colonized by microbial communities<br />
prior to deposition <strong>of</strong> the Mozaan IFs. It is therefore<br />
not unreasonable to expect that organic constituents exerted<br />
some effect on local geochemical processes, <strong>and</strong> the<br />
above observations are consistent with non-redox sensitive<br />
processes similar to modern ones operating in shallow seas<br />
on Archean continental margins.<br />
The negative Nd (2.9 Ga) values for the Mozaan IFs are<br />
not entirely unusual, however, as supported by negative<br />
Nd (t) values for some younger IFs, as well as by recent<br />
investigations <strong>of</strong> Archean shallow water carbonates from<br />
southern Africa (Bolhar et al., 2002; Kamber et al., 2004).<br />
These studies concluded that near-shore seawater displayed<br />
REY <strong>and</strong> Nd isotope signatures indicative <strong>of</strong> local solute<br />
inputs derived from continental weathering, <strong>and</strong> <strong>of</strong>fer further<br />
support suggesting that shallow seawater during the<br />
mid- to late-Archean was strongly influenced by continentally-derived<br />
solute fluxes.<br />
A final comment regarding the correlation between Sm–<br />
Nd isotope ratios <strong>and</strong> Fe in Archean IFs is warranted. Previous<br />
workers have calculated the mass fraction <strong>of</strong> Fe derived<br />
from different sources by conservatively mixing a<br />
hydrothermal endmember possessing a depleted-mantle<br />
Nd isotope ratio with a river/seawater endmember which<br />
possesses a continental Nd isotope signature (e.g., Miller<br />
<strong>and</strong> O’Nions, 1985; <strong>Jacobs</strong>en <strong>and</strong> Pimentel-Klose, 1988a).<br />
While conclusions have varied regarding the dominant Fe<br />
source, these attempts have all relied on assumptions<br />
regarding the Fe/Nd ratio <strong>of</strong> the endmembers, <strong>and</strong> crucial<br />
to the models is the requirement that Fe/Nd did not change<br />
significantly in the Archean ocean. However, there existed<br />
processes in the Archean ocean, such as pyrite precipitation<br />
close to hydrothermal vent sites, <strong>and</strong> IF precipitation itself,<br />
for example, that fractionated Fe <strong>and</strong> Nd. Moreover, Fe<br />
<strong>and</strong> Nd behavior in the continentally-derived flux produces<br />
even more ambiguity due to uncertainties in various parameters<br />
such as the pH <strong>of</strong> river water, local redox conditions,<br />
<strong>and</strong> possible involvement <strong>of</strong> biological processes. As a result,<br />
the current state <strong>of</strong> knowledge precludes definitive<br />
statements regarding the Fe source to Archean IFs based<br />
upon Nd isotope systematics.<br />
6. CONCLUSIONS<br />
All <strong>of</strong> the Mozaan iron-formation samples exhibit major<br />
<strong>and</strong> trace element characteristics typical <strong>of</strong> very pure marine<br />
chemical sediments, including shale-normalized positive<br />
La, Gd, <strong>and</strong> Y anomalies, which are fully consistent with a<br />
marine depositional setting. Positive Eu anomalies in all IF<br />
samples indicate a high-T hydrothermal source supplied<br />
REYs to the seawater which precipitated the Mozaan IFs.<br />
The magnitude <strong>of</strong> this anomaly is consistent with the general<br />
trend <strong>of</strong> decreasing Eu/Eu * with decreasing age for Precambrian<br />
IFs <strong>and</strong> is within the range for other Archean IFs,<br />
though it cannot be used to quantify the magnitude <strong>of</strong> the<br />
high-T hydrothermal flux. The lack <strong>of</strong> Ce anomalies in all<br />
<strong>of</strong> the IF samples indicates that local, shallow Archean seawater<br />
was too reducing to permit oxidation <strong>of</strong> Ce(III) to<br />
Ce(IV), implying relatively low redox levels in the Archean<br />
atmosphere compared to present values.<br />
When compared to many other Archean IFs, the Pongola<br />
samples have higher Sm/Yb ratios <strong>and</strong> appear to record a systematic<br />
variation in water depth on the margin <strong>of</strong> the Kaapvaal<br />
craton during IF sedimentation. Dominant factors<br />
influencing the REY distribution <strong>of</strong> coastal seawater in the<br />
Archean would include atmospheric composition (e.g.,<br />
qCO 2 ) <strong>and</strong> fluxes <strong>of</strong> solutes from terrestrial sources. Isotope<br />
systematics <strong>of</strong> Sm–Nd reveal similarity between iron-formation<br />
<strong>and</strong> contemporaneous shale, arguing against an Arche-
392 B.W. Alex<strong>and</strong>er et al. / Geochimica et Cosmochimica Acta 72 (2008) 378–394<br />
an mid-ocean ridge hydrothermal system as the dominant<br />
REY source for these IFs. Rare earth elements in both shale<br />
<strong>and</strong> IFs appear to be derived primarily from a relatively<br />
homogenous cratonic source older than 3.2 Ga, <strong>and</strong> consistent<br />
Nd isotopic behavior can be observed in Pongola sediments<br />
that sample relatively short geologic time spans.<br />
These data imply that solutes within shallow seawater along<br />
Archean cratonic margins were sourced primarily from<br />
weathering <strong>of</strong> continental crust. The Nd isotopic data for<br />
the Mozaan shales (T DM <strong>of</strong> 3.4 ± 0.1 Ga) <strong>and</strong> the depositional<br />
age <strong>of</strong> these sediments suggests that local continental<br />
crust was stable on 100 Ma time-scales, <strong>and</strong> when coupled<br />
with the laterally extensive nature <strong>of</strong> the precipitated Mozaan<br />
IFs, this implies that by 2.9 Ga ago a significant amount <strong>of</strong><br />
continental crust was exposed to weathering.<br />
ACKNOWLEDGMENTS<br />
We appreciate the assistance <strong>of</strong> Marina Fischerström <strong>and</strong> Hans<br />
Schöberg with the Sm–Nd isotopic analyses performed at LIG.<br />
Furthermore, this manuscript significantly benefited from the comments<br />
<strong>of</strong> A. Bekker <strong>and</strong> one anonymous reviewer, <strong>and</strong> their efforts<br />
are gratefully acknowledged. We also thank T. Lyons for helpful<br />
suggestions <strong>and</strong> editorial guidance.<br />
REFERENCES<br />
Alibert C. <strong>and</strong> McCulloch M. T. (1993) Rare earth element <strong>and</strong> Nd<br />
isotopic compositions <strong>of</strong> the b<strong>and</strong>ed iron-formations <strong>and</strong><br />
associated shales from Hamersley, Western Australia. Geochim.<br />
Cosmochim. Acta 57, 187–204.<br />
Alibo D. S. <strong>and</strong> Nozaki Y. (1999) Rare earth elements in seawater:<br />
particle association, shale-normalization, <strong>and</strong> Ce oxidation.<br />
Geochim. Cosmochim. Acta 63, 363–372.<br />
Anders E. <strong>and</strong> Grevesse N. (1989) Abudances <strong>of</strong> the elements:<br />
meteoric <strong>and</strong> solar. Geochim. Cosmochim. Acta 53, 197–214.<br />
Andersson K., Dahlqvist R., Turner D., Stolpe B., Larsson T.,<br />
Ingri J. <strong>and</strong> Andersson P. (2006) Colloidal rare earth elements<br />
in a boreal river: changing sources <strong>and</strong> distributions during the<br />
spring flood. Geochim. Cosmochim. Acta 70, 3261–3274.<br />
Barrett T. J., Fralick P. W. <strong>and</strong> Jarvis I. (1988) Rare-earth-element<br />
geochemistry <strong>of</strong> some Archean iron formations north <strong>of</strong> Lake<br />
Superior, Ontario. Can. J. Earth Sci. 25, 570–580.<br />
Bau M. (1991) Rare earth element mobility during hydrothermal<br />
<strong>and</strong> metamorphic fluid–rock interaction <strong>and</strong> the significance <strong>of</strong><br />
the oxidation state <strong>of</strong> europium. Chem. Geol. 93, 219–230.<br />
Bau M. (1993) Effects <strong>of</strong> syn- <strong>and</strong> post-depositional processes on<br />
the rare-earth element distribution in Precambrian iron-formations.<br />
Eur. J. Mineral. 5, 257–267.<br />
Bau M. (1999) Scavenging <strong>of</strong> dissolved yttrium <strong>and</strong> rare earths by<br />
precipitating iron oxyhydroxide: experimental evidence for Ceoxidation,<br />
Y–Ho fractionation, <strong>and</strong> lanthanide tetrad effect.<br />
Geochim. Cosmochim. Acta 63, 67–77.<br />
Bau M. <strong>and</strong> Dulski P. (1992) Small-scale variations <strong>of</strong> the rareearth<br />
element distribution in Precambrian iron formations. Eur.<br />
J. Mineral. 4, 1429–1433.<br />
Bau M. <strong>and</strong> Dulski P. (1996) Distribution <strong>of</strong> yttrium <strong>and</strong> rare-earth<br />
elements in the Penge <strong>and</strong> Kuruman iron-formations, Transvaal<br />
Supergroup, South Africa. Precamb. Res. 79, 37–55.<br />
Bau M. <strong>and</strong> Dulski P. (1999) Comparing yttrium <strong>and</strong> rare earths in<br />
hydrothermal fluids from the Mid-Atlantic Ridge: implications<br />
for Y <strong>and</strong> REE behaviour during near-vent mixing <strong>and</strong> for the<br />
Y/Ho ratio <strong>of</strong> Proterozoic seawater. Chem. Geol. 155, 77–90.<br />
Bau M. <strong>and</strong> Möller P. (1993) Rare earth element systematics <strong>of</strong> the<br />
chemically precipitated component in Early Precambrian iron<br />
formations <strong>and</strong> the evolution <strong>of</strong> the terrestrial atmosphere–<br />
hydrosphere–lithosphere system. Geochim. Cosmochim. Acta<br />
57, 2239–2249.<br />
Bau M., Koschinsky A., Dulski P. <strong>and</strong> Hein J. R. (1996)<br />
Comparison <strong>of</strong> the partitioning behaviours <strong>of</strong> yttrium, rare<br />
earth elements, <strong>and</strong> titanium between hydrogenetic marine<br />
ferromanganese crusts <strong>and</strong> seawater. Geochim. Cosmochim.<br />
Acta 60, 1709–1725.<br />
Bau M., Höhndorf A., Dulski P. <strong>and</strong> Beukes N. J. (1997) Sources<br />
<strong>of</strong> rare-earth elements <strong>and</strong> iron in Paleoproterozoic ironformations<br />
from the Transvaal Supergroup, South Africa:<br />
Evidence from neodymium isotopes. J. Geol. 105, 121–129.<br />
Bau M., Usui A., Pracejus B., Mita N., Kanai Y., Irber W. <strong>and</strong><br />
Dulski P. (1998) Geochemistry <strong>of</strong> low-temperature water–rock<br />
interaction: evidence from natural waters, <strong>and</strong>esite, <strong>and</strong> ironoxyhydroxide<br />
precipitates at Nishiki-numa iron-spring, Hokkaido,<br />
Japan. Chem. Geol. 151, 293–307.<br />
Beukes N. J. <strong>and</strong> Cairncross B. (1991) A lithostratigraphicsedimentological<br />
reference pr<strong>of</strong>ile for the late Archaean Mozaan<br />
Group, Pongola Sequence: application to sequence stratigraphy<br />
<strong>and</strong> correlation with the Witwatersr<strong>and</strong> Supergroup.<br />
S. Afr. J. Geol. 94, 44–69.<br />
Beukes N. J. <strong>and</strong> Lowe D. R. (1989) Environmental control on<br />
diverse stromatolite morphologies in the 3000 Myr Pongola<br />
Supergroup, South Africa. Sedimentology 36, 383–397.<br />
Bolhar R., H<strong>of</strong>mann A., Woodhead J., Hergt J. <strong>and</strong> Dirks P.<br />
(2002) Pb- <strong>and</strong> Nd-isotope systematics <strong>of</strong> stromatolitic limestones<br />
from the 2.7 Ga Ngezi Group <strong>of</strong> the Belingwe Greenstone<br />
Belt: constraints on timing <strong>of</strong> deposition <strong>and</strong> provenance.<br />
Precamb. Res. 114, 277–294.<br />
Bolhar R., Kamber B. S., Moorbath S., Fedo C. M. <strong>and</strong><br />
Whitehouse M. J. (2004) Characterisation <strong>of</strong> early Archaean<br />
chemical sediments by trace element signatures. Earth Planet.<br />
Sci. Lett. 222, 43–60.<br />
Button A., Pretorius D. A., Jansen H., Stocklmayer V., Hunter D.<br />
R., Wilson J. F., Wilson A. H., Vermaak C. A., Lee C. A. <strong>and</strong><br />
Stagman J. G. (1981) The cratonic environment: the Pongola<br />
Supergroup. In Precambrian <strong>of</strong> the Southern Hemisphere (ed. D.<br />
R. Hunter). Elsevier Scientific Publications Co., Amsterdam,<br />
pp. 501–510.<br />
Danielson A., Möller P. <strong>and</strong> Dulski P. (1992) The europium<br />
anomalies in b<strong>and</strong>ed iron formations <strong>and</strong> the thermal history <strong>of</strong><br />
the oceanic crust. Chem. Geol. 97, 89–100.<br />
DePaolo D. J. <strong>and</strong> Wasserburg G. J. (1976) Nd isotopic variations<br />
<strong>and</strong> petrogenetic models. Geophys. Res. Lett. 3, 249–252.<br />
Derry L. A. <strong>and</strong> <strong>Jacobs</strong>en S. B. (1990) The chemical evolution <strong>of</strong><br />
Precambrian seawater: evidence from rare earth elements in<br />
b<strong>and</strong>ed iron formations. Geochim. Cosmochim. Acta 54, 2965–<br />
2977.<br />
Dia A., Allègre C. J. <strong>and</strong> Erlank A. J. (1990) The development <strong>of</strong><br />
continental crust through geological time: the South African<br />
case. Earth Planet. Sci. Lett. 98, 74–89.<br />
Dulski P. (2001) Reference materials for geochemical studies: new<br />
analytical data by ICP-MS <strong>and</strong> critical discussion <strong>of</strong> reference<br />
values. Geost<strong>and</strong>. Newslett. 25, 87–125.<br />
Elderfield H. (1988) The oceanic chemistry <strong>of</strong> the rare-earth<br />
elements. Phil. Trans. R. Soc. London Ser. A 325, 105–126.<br />
Elderfield H., Upstill-Goddard R. <strong>and</strong> Sholkovitz E. R. (1990) The<br />
rare earth elements in rivers, estuaries, <strong>and</strong> coastal seas <strong>and</strong><br />
their significance to the composition <strong>of</strong> ocean waters. Geochim.<br />
Cosmochim. Acta 54, 971–991.<br />
Erel Y. <strong>and</strong> Stolper E. M. (1993) Modeling <strong>of</strong> rare-earth element<br />
partitioning between particles <strong>and</strong> solution in aquatic environments.<br />
Geochim. Cosmochim. Acta 57, 513–518.
Nd isotopes in 2.9 Ga Archean surface seawater 393<br />
Frei R., Bridgewater D., Rosing M. <strong>and</strong> Stecher O. (1999)<br />
Controversial Pb–Pb <strong>and</strong> Sm–Nd isotope results in the early<br />
Archean Isua (West Greenl<strong>and</strong>) oxide iron formation: preservation<br />
<strong>of</strong> primary signatures versus secondary disturbances.<br />
Geochim. Cosmochim. Acta 63, 473–488.<br />
Fryer B. J. (1977) Rare earth evidence in iron formations for<br />
changing Precambrian oxidation states. Geochim. Cosmochim.<br />
Acta 41, 361–367.<br />
Grauch R. I. (1989) Rare earth elements in metamorphic rocks. In<br />
Geochemistry <strong>and</strong> Mineralogy <strong>of</strong> Rare Earth Elements (eds. B.<br />
R. Lipin <strong>and</strong> G. A. McKay); Rev. Mineral. 21, 147–167,<br />
Mineral. Soc. Amer.<br />
Gross G. A. (1983) Tectonic systems <strong>and</strong> the deposition <strong>of</strong> ironformation.<br />
Precamb. Res. 20, 171–187.<br />
Hannigan R. E. <strong>and</strong> Sholkovitz E. R. (2001) The development <strong>of</strong><br />
middle rare earth element enrichments in freshwaters: weathering<br />
<strong>of</strong> phosphate minerals. Chem. Geol. 175, 495–508.<br />
Hegner E., Kröner A. <strong>and</strong> H<strong>of</strong>mann A. W. (1984) Age <strong>and</strong> isotope<br />
geochemistry <strong>of</strong> the Archaean Pongola <strong>and</strong> Usushwana suites<br />
in Swazil<strong>and</strong>, southern Africa: a case for crustal contamination<br />
<strong>of</strong> mantle-derived magma. Earth Planet. Sci. Lett. 70, 267–279.<br />
Hunter D. R. <strong>and</strong> Wilson A. H. (1988) A continuous record <strong>of</strong><br />
Archaean evolution from 3.5 Ga to 2.6 Ga in Swazil<strong>and</strong> <strong>and</strong><br />
northern Natal. S. Afr. J. Geol. 91, 57–74.<br />
Ingri J., Widerlund A., L<strong>and</strong> M., Gustafsson O., Andersson P. <strong>and</strong><br />
Ohl<strong>and</strong>er B. (2000) Temporal variations in the fractionation <strong>of</strong><br />
the rare earth elements in a boreal river; the role <strong>of</strong> colloidal<br />
particles. Chem. Geol. 166, 23–45.<br />
<strong>Jacobs</strong>en S. B. <strong>and</strong> Pimentel-Klose M. R. (1988a) A Nd isotopic<br />
study <strong>of</strong> the Hammersley <strong>and</strong> Michipicoten b<strong>and</strong>ed iron<br />
formations: The source <strong>of</strong> REE <strong>and</strong> Fe in Archean oceans.<br />
Earth Planet. Sci. Lett. 87, 29–44.<br />
<strong>Jacobs</strong>en S. B. <strong>and</strong> Pimentel-Klose M. R. (1988b) Nd isotopic<br />
variations in Precambrian b<strong>and</strong>ed iron formations. Geophys.<br />
Res. Lett. 15, 393–396.<br />
Jahn B. <strong>and</strong> Condie K. C. (1995) Evolution <strong>of</strong> the Kaapvaal Craton<br />
as viewed from geochemical <strong>and</strong> Sm–Nd isotopic analyses <strong>of</strong><br />
intracratonic pelites. Geochim. Cosmochim. Acta 59, 2239–2258.<br />
James H. L. <strong>and</strong> Trendall A. F. (1982) B<strong>and</strong>ed iron formation:<br />
distribution in time <strong>and</strong> paleoenvironmental significance. In<br />
Mineral Deposits <strong>and</strong> Evolution <strong>of</strong> the Biosphere (eds. H. D.<br />
Holl<strong>and</strong> <strong>and</strong> M. Schidlowski). Springer, Berlin, pp. 199–217.<br />
Kamber B. S. <strong>and</strong> Webb G. E. (2001) The geochemistry <strong>of</strong> late<br />
Archaean microbial carbonate: Implications for ocean chemistry<br />
<strong>and</strong> continental erosion history. Geochim. Cosmochim. Acta<br />
65, 2509–2525.<br />
Kamber B. S., Bohlar R. <strong>and</strong> Webb G. E. (2004) Geochemistry <strong>of</strong><br />
late Archaean stromatolites from Zimbabwe: evidence for<br />
microbial life in restricted epicontinental seas. Precamb. Res.<br />
132, 379–399.<br />
Kato Y., Ohta I., Tsunematsu T., Watanabe Y., Isozaki Y.,<br />
Maruyama S. <strong>and</strong> Imai N. (1998) Rare earth element variations<br />
in mid-Archean b<strong>and</strong>ed iron formations: implications for the<br />
chemistry <strong>of</strong> ocean <strong>and</strong> continent <strong>and</strong> plate tectonics. Geochim.<br />
Cosmochim. Acta 62, 3475–3497.<br />
Khan R. M. K., Das Sharma S., Patil D. J. <strong>and</strong> Naqvi S. M. (1996)<br />
Trace, rare-earth element, <strong>and</strong> oxygen isotopic systematics for<br />
the genesis <strong>of</strong> b<strong>and</strong>ed iron-formations: evidence from the<br />
Kushtagi schist belt, Archaean Dharwar Craton, India. Geochim.<br />
Cosmochim. Acta 60, 3285–3294.<br />
Klein C. <strong>and</strong> Beukes N. J. (1989) Geochemistry <strong>and</strong> sedimentology<br />
<strong>of</strong> a facies transition from limestone to iron-formation deposition<br />
in the Early Proterozoic Transvaal Supergroup, South<br />
Africa. Econ. Geol. 84, 1733–1774.<br />
Klemm D. D. (2000) The formation <strong>of</strong> Palaeoproterozoic b<strong>and</strong>ed<br />
iron formations <strong>and</strong> their associated Fe <strong>and</strong> Mn deposits, with<br />
reference to the Griqual<strong>and</strong> West deposits, South Africa. J. Afr.<br />
Earth Sci. 30, 1–24.<br />
Kulaksiz S. <strong>and</strong> Bau M. (2007) Contrasting behaviour <strong>of</strong> anthropogenic<br />
gadolinium <strong>and</strong> natural rare earth elements in estuaries<br />
<strong>and</strong> the gadolinium input into the North Sea. Earth Planet. Sci.<br />
Lett. 260, 361–371.<br />
Mason T. R. <strong>and</strong> von Brunn V. (1977) 3-Gyr-old stromatolites<br />
from South Africa. Nature 266, 47–49.<br />
Matthews P. E. (1967) The pre-Karroo formations <strong>of</strong> the White<br />
Umfolozi inlier, northern Natal. Trans. Geol. Soc. S. Afr. 70,<br />
39–63.<br />
McLennan S. M. (1989) Rare earth elements in sedimentary rocks:<br />
influence <strong>of</strong> provenance <strong>and</strong> sedimentary processes. In Geochemistry<br />
<strong>and</strong> Mineralogy <strong>of</strong> Rare Earth Elements (eds. B.R.<br />
Lipin <strong>and</strong> G.A. McKay); Rev. Mineral. 21, 169–200, Mineral.<br />
Soc. Amer.<br />
McLennan S. M., Taylor S. R. <strong>and</strong> Kröner A. (1983) Geochemical<br />
evolution <strong>of</strong> Archean shales from South Africa. I. The<br />
Swazil<strong>and</strong> <strong>and</strong> Pongola Supergroups. Precamb. Res. 22, 93–<br />
124.<br />
Michard A. (1989) Rare earth element systematics in hydrothermal<br />
fluids. Geochim. Cosmochim. Acta 53, 745–750.<br />
Miller R. G. <strong>and</strong> O’Nions R. K. (1985) Sources <strong>of</strong> Precambrian<br />
chemical <strong>and</strong> clastic sediments. Nature 314, 325–330.<br />
Morris R. C. (1993) Genetic moelling for b<strong>and</strong>ed iron-formation <strong>of</strong><br />
the Hamersley Group, Pilbara Craton, Western Australia.<br />
Precamb. Res. 60, 243–286.<br />
Nelson D. R., Trendall A. F., de Laeter J. R., Grobler N. J.<br />
<strong>and</strong> Fletcher I. R. (1992) A comparative study <strong>of</strong> the<br />
geochemical <strong>and</strong> isotopic systematics <strong>of</strong> late Archean flood<br />
basalts from the Pilbara <strong>and</strong> Kaapvaal cratons. Precamb.<br />
Res. 54, 231–256.<br />
Nhleko N. (2003) The Pongola Supergroup in Swazil<strong>and</strong>, Ph.D.<br />
dissertation R<strong>and</strong> Afrikaans <strong>University</strong>.<br />
Nothdurft L. D., Webb G. E. <strong>and</strong> Kamber B. S. (2004) Rare earth<br />
element geochemistry <strong>of</strong> Late Devonian reefal carbonates,<br />
Canning Basin, Western Australia: confirmation <strong>of</strong> a seawater<br />
REE proxy in ancient limestones. Geochim. Cosmochim. Acta<br />
68, 263–283.<br />
Pichler T., Veizer J. <strong>and</strong> Hall G. E. M. (1999) The chemical<br />
composition <strong>of</strong> shallow-water hydrothermal fluids in Tutum<br />
Bay, Ambitle Isl<strong>and</strong>, Papua New Guinea <strong>and</strong> their effect on<br />
ambient seawater. Mar. Chem. 64, 229–252.<br />
Piepgras D. J. <strong>and</strong> Wasserburg G. J. (1980) Neodymium isotopic<br />
variations in seawater. Earth Planet. Sci. Lett. 50, 128–138.<br />
Piepgras D. J. <strong>and</strong> Wasserburg G. J. (1985) Strontium <strong>and</strong><br />
neodymium isotopes in hot springs on the East Pacific rise<br />
<strong>and</strong> Guaymas basin. Earth Planet. Sci. Lett. 72, 341–356.<br />
Shields G. <strong>and</strong> Stille P. (2001) Diagenetic constraints on the use <strong>of</strong><br />
cerium anomalies as palaeoseawater redox proxies: an isotopic<br />
<strong>and</strong> REE study <strong>of</strong> Cambrian phosphorites. Chem. Geol. 175,<br />
29–48.<br />
Shields G. A. <strong>and</strong> Webb G. E. (2004) Has the REE composition <strong>of</strong><br />
seawater changed over geological time? Chem. Geol. 204, 103–<br />
107.<br />
Shimizu H., Umemoto N., Masuda A. <strong>and</strong> Appel P. W. U. (1990)<br />
Sources <strong>of</strong> iron-formations in the Archean Isua <strong>and</strong> Malene<br />
supracrustals, West Greenl<strong>and</strong>: evidence from La–Ce <strong>and</strong> Sm–<br />
Nd isotopic data <strong>and</strong> REE. Geochim. Cosmochim. Acta 54,<br />
1147–1154.<br />
Sholkovitz E. R. (1994) The aquatic chemistry <strong>of</strong> rare earth<br />
elements in rivers <strong>and</strong> estuaries. Aqua. Geochem. 1, 3–34.<br />
Sholkovitz E. <strong>and</strong> Szymczak R. (2000) The estuarine chemistry <strong>of</strong><br />
rare earth elements: comparison <strong>of</strong> the Amazon, Fly, Sepik <strong>and</strong><br />
the Gulf <strong>of</strong> Papua systems. Earth Planet. Sci. Lett. 179, 299–<br />
309.
394 B.W. Alex<strong>and</strong>er et al. / Geochimica et Cosmochimica Acta 72 (2008) 378–394<br />
Stevenson R. K. <strong>and</strong> Patchett P. J. (1990) Implications for the<br />
evolution <strong>of</strong> continental crust from Hf isotope systematics <strong>of</strong><br />
Archean detrital zircons. Geochim. Cosmochim. Acta 54, 1683–<br />
1697.<br />
Tankard A. J., Jackson M. P. A., Eriksson K. A., Hobday D. K.,<br />
Hunter D. R. <strong>and</strong> Minter W. E. L. (1982) Crustal Evolution <strong>of</strong><br />
Southern Africa. Springer Verlag, New York.<br />
Taylor S. R. <strong>and</strong> McLennan S. M. (1985) The continental crust: its<br />
composition <strong>and</strong> evolution. Blackwell Scientific, Oxford.<br />
Towe K. M. (1991) Aerobic carbon cycling <strong>and</strong> cerium oxidation:<br />
significance for Archean oxygen levels <strong>and</strong> b<strong>and</strong>ed ironformation<br />
deposition. Palaeogeo. Palaeoclim. Palaeoecol. 97,<br />
113–123.<br />
von Brunn V. <strong>and</strong> Hobday D. K. (1976) Early Precambrian tidal<br />
sedimentation in the Pongola Supergroup <strong>of</strong> South Africa. J.<br />
Sediment. Petrol. 46, 670–679.<br />
von Brunn V. <strong>and</strong> Mason T. R. (1977) Siliciclastic-carbonate tidal<br />
deposits from the 3000 M.Y. Pongola Supergroup, South<br />
Africa. Sediment. Geol. 18, 245–255.<br />
Watchorn M. B. (1980) Fluvial <strong>and</strong> tidal sedimentation in the<br />
3000 Ma Mozaan basin, South Africa. Precamb. Res. 13, 27–42.<br />
Webb G. E. <strong>and</strong> Kamber B. S. (2000) Rare earth elements in<br />
Holocene reefal microbialites: a new shallow seawater proxy.<br />
Geochim. Cosmochim. Acta 64, 1557–1565.<br />
Weilers B. F. (1990) A review <strong>of</strong> the Pongola Supergroup <strong>and</strong> its<br />
setting on the Kaapvaal craton. Econ. Geol. Res. Unit, Info.<br />
Cir. No. 228, Univ. Witwatersr<strong>and</strong>.<br />
Wheat C. G., Mottl M. J. <strong>and</strong> Rudnicki M. (2002) Trace element <strong>and</strong><br />
REE composition <strong>of</strong> a low-Temperature ridge-flank hydrothermal<br />
spring. Geochim. Cosmochim. Acta 66, 3693–3705.<br />
Wronkiewicz, D. J. (1989) Geochemistry <strong>and</strong> mineralogy <strong>of</strong> mid-<br />
Archean to mid-Proterozoic sediments from the Kaapvaal<br />
craton, southern Africa: source area provenance, weathering<br />
tectonic setting <strong>and</strong> crustal evolution. Ph.D. dissertation New<br />
Mexico Inst. Mining Tech.<br />
Associate editor: Timothy W. Lyons
THIS PAGE INTENTIONALLY LEFT BLANK<br />
138
Chapter 4. Neodymium isotopes in Archean seawater <strong>and</strong> implications for the<br />
marine Nd cycle in Earth’s early oceans<br />
139
THIS PAGE INTENTIONALLY LEFT BLANK<br />
140
Author's personal copy<br />
Earth <strong>and</strong> Planetary <strong>Science</strong> Letters 283 (2009) 144–155<br />
Contents lists available at <strong>Science</strong>Direct<br />
Earth <strong>and</strong> Planetary <strong>Science</strong> Letters<br />
journal homepage: www.elsevier.com/locate/epsl<br />
Neodymium isotopes in Archean seawater <strong>and</strong> implications for the<br />
marine Nd cycle in Earth's early oceans<br />
Brian W. Alex<strong>and</strong>er a, ⁎, Michael Bau a , Per Andersson b<br />
a Earth <strong>and</strong> Space <strong>Science</strong>s, <strong>Jacobs</strong> <strong>University</strong> Bremen, Campus Ring 8, 28759 Bremen, Germany<br />
b Laboratory for Isotope Geology, Swedish Museum <strong>of</strong> Natural History, Box 50007, 104 05 Stockholm, Sweden<br />
article<br />
info<br />
abstract<br />
Article history:<br />
Received 31 August 2008<br />
Received in revised form 7 April 2009<br />
Accepted 7 April 2009<br />
Available online 9 May 2009<br />
Editor: M.L. Delaney<br />
Keywords:<br />
Pietersburg<br />
Archean<br />
seawater<br />
neodymium<br />
Isua<br />
Published neodymium (Nd) isotopic data for Archean iron-formations (IF) suggest that, overall, seawater<br />
throughout the Archean typically displayed 143 Nd/ 144 Nd close to bulk Earth values, with Є Nd (t) between<br />
−1.5 <strong>and</strong> +2.5. Neodymium isotopic ratios in seawater during deposition <strong>of</strong> the ~3.8 Isua (Greenl<strong>and</strong>) IF<br />
likely displayed positive Є Nd (3.8 Ga) <strong>of</strong> +2.5, as suggested by IF-G, an Isua reference IF that is considered the<br />
best archive for Early Archean seawater. Seawater 143 Nd/ 144 Nd ratios dominated by radiogenic Nd (positive<br />
Є Nd (t)) seem to have persisted for much <strong>of</strong> the Archean, as IF from the Pietersburg greenstone belt, South<br />
Africa, suggest seawater Є Nd (2.95 Ga)≥+1. However, similarly aged (~2.9 Ga) IFs from South Africa indicate<br />
that significant variations in seawater 143 Nd/ 144 Nd occurred, <strong>and</strong> clearly show influences from isotopically<br />
distinct crustal sources. These variations are apparently related to depositional environment, with cratonic<br />
margin, shallow-water IFs possessing a continental Є Nd (t) <strong>of</strong>−3, while IFs associated with sub-aqueous<br />
mafic volcanics display more radiogenic, positive Є Nd (t) values. Such variation in seawater 143 Nd/ 144 Nd is not<br />
possible in an isotopically well-mixed ocean, <strong>and</strong> similar to today, it appears that marine Nd cycling in the<br />
Archean produced water masses with distinct Nd isotopic ratios. Since the presence <strong>of</strong> b<strong>and</strong>ed ironformations<br />
requires a reducing Archean ocean capable <strong>of</strong> transporting Fe, metal-oxide precipitation <strong>and</strong><br />
scavenging processes near deep sea hydrothermal vent systems would not have scavenged mantle Nd, i.e., Nd<br />
sourced from alteration <strong>of</strong> oceanic crust. We propose that bulk anoxic seawater prior to 2.7 Ga possessed<br />
relatively constant positive Є Nd (t) <strong>of</strong> +1 to +2, whereas local shallow-water masses associated with<br />
exposed evolved crust could possess distinctly different, lower Є Nd (t).<br />
© 2009 Elsevier B.V. All rights reserved.<br />
1. Introduction<br />
Several lines <strong>of</strong> evidence suggest that the hydrothermal alteration<br />
<strong>of</strong> oceanic crust <strong>and</strong> its impact on ocean chemistry in the first half <strong>of</strong><br />
Earth's history was significantly greater than today. These include<br />
Archean (N2.5 Ga) marine carbonates with higher manganese (Mn)<br />
<strong>and</strong> iron (Fe) concentrations <strong>and</strong> low mantle-like strontium (Sr)<br />
isotopic ratios (e.g., Veizer et al., 1989; Eglington et al., 2003).<br />
Additionally, the presence <strong>of</strong> positive europium (Eu) anomalies in<br />
Archean marine precipitates (carbonates <strong>and</strong> iron-formations) similar<br />
to modern high-temperature ‘black-smoker’ fluids has been cited as<br />
evidence <strong>of</strong> an increased mantle-derived hydrothermal component<br />
within Archean seawater (e.g., Fryer et al., 1979; Appel, 1983; Derry<br />
<strong>and</strong> <strong>Jacobs</strong>en, 1990; Danielson et al., 1992; Kamber <strong>and</strong> Webb, 2001).<br />
These observations would be consistent with models <strong>of</strong> higher heat<br />
production (e.g., McKenzie <strong>and</strong> Weiss, 1975) <strong>and</strong> higher sea-floor<br />
spreading rates (e.g., Abbott <strong>and</strong> H<strong>of</strong>fman, 1984) during the Archean,<br />
which in turn implies increased seawater fluxes through oceanic crust.<br />
⁎ Corresponding author. Tel.: +49 421 200 3154; fax: +49 421 200 3229.<br />
E-mail addresses: b.alex<strong>and</strong>er@jacobs-university.de, m.bau@jacobs-university.de<br />
(B.W. Alex<strong>and</strong>er), per.<strong>and</strong>ersson@nrm.se (P. Andersson).<br />
Attempts to resolve the relative importance <strong>of</strong> Archean marine<br />
hydrothermal fluxes have been limited by the availability <strong>of</strong> suitable<br />
seawater archives. Archean carbonates are rare (Veizer <strong>and</strong> Mackenzie,<br />
2004), <strong>and</strong> thick well-preserved sequences only appear ca.<br />
~2.7 Ga ago (cf. Grotzinger, 1989). Therefore, studies regarding hydrothermal<br />
impacts upon Archean–Paleoproterozoic seawater chemistry<br />
have <strong>of</strong>ten used b<strong>and</strong>ed iron-formations (IFs) as seawater archives<br />
(e.g., Fryer et al., 1979; <strong>Jacobs</strong>en <strong>and</strong> Pimentel-Klose, 1988a; Derry <strong>and</strong><br />
<strong>Jacobs</strong>en, 1990; Alibert <strong>and</strong> McCulloch, 1993). Iron-formations are<br />
chemical precipitates whose ubiquitous presence throughout the<br />
Archean suggests an increased hydrothermal fluid flux, as modern<br />
vent fluids at oceanic spreading ridges are reducing <strong>and</strong> Fe-rich, which<br />
are necessary properties <strong>of</strong> the fluid which ultimately produced IFs.<br />
The conclusion that most IFs precipitated from seawater is generally<br />
accepted <strong>and</strong> indicated from different lines <strong>of</strong> evidence (cf. Simonson,<br />
2003). These include the frequent association <strong>of</strong> IF with marine<br />
carbonates <strong>and</strong> marine transgressions (Klein <strong>and</strong> Beukes, 1989;<br />
Simonson <strong>and</strong> Hassler, 1996), <strong>and</strong> geochemical evidence such as<br />
rare earth element (REE) distributions in many IF (Bau <strong>and</strong> Dulski,<br />
1996; Bolhar et al., 2004) that are remarkably similar to both<br />
Archean–Paleoproterozoic marine carbonates (Kamber <strong>and</strong> Webb,<br />
2001; Bau <strong>and</strong> Alex<strong>and</strong>er, 2006) <strong>and</strong> modern seawater.<br />
0012-821X/$ – see front matter © 2009 Elsevier B.V. All rights reserved.<br />
doi:10.1016/j.epsl.2009.04.004
Author's personal copy<br />
B.W. Alex<strong>and</strong>er et al. / Earth <strong>and</strong> Planetary <strong>Science</strong> Letters 283 (2009) 144–155<br />
145<br />
However, the exact process <strong>of</strong> IF deposition is unclear, as it implies<br />
contradictory mechanisms in which seawater is reducing enough to<br />
transport large amounts <strong>of</strong> soluble Fe(II), yet oxidizing enough to<br />
episodically precipitate sediments that may contain N90% Fe (measured<br />
as Fe 2 O 3 ). Therefore, at least with respect to the marine Fe cycle,<br />
Archean seawater during IF deposition was clearly distinct from<br />
modern seawater, <strong>and</strong> several studies have attempted to quantify the<br />
importance <strong>of</strong> an Fe-rich hydrothermal flux by examining samarium–<br />
neodymium (Sm–Nd) isotopes in IFs (e.g., Miller <strong>and</strong> O'Nions, 1985;<br />
<strong>Jacobs</strong>en <strong>and</strong> Pimentel-Klose; 1988a, b; Derry <strong>and</strong> <strong>Jacobs</strong>en, 1990;<br />
Alibert <strong>and</strong> McCulloch; 1993; Bau et al., 1997a; Alex<strong>and</strong>er et al., 2008).<br />
This application <strong>of</strong> Sm–Nd isotopes is possible because fractionation<br />
during crustal differentiation produces different Sm/Nd ratios in<br />
oceanic <strong>and</strong> continental crust, resulting in unique 143 Nd/ 144 Nd ratios<br />
in these reservoirs due to the subsequent decay <strong>of</strong> 147 Sm to 143 Nd.<br />
Measured 143 Nd/ 144 Nd ratios are typically expressed using the Є Nd (t)<br />
notation (DePaolo <strong>and</strong> Wasserburg, 1976), which describes the deviation<br />
<strong>of</strong> 143 Nd/ 144 Nd in parts per 10 4 from CHUR (chondritic uniform<br />
reservoir, i.e., bulk silicate Earth) at time t. Continental crust, with a<br />
low Sm/Nd ratio <strong>and</strong> little radiogenic 143 Nd, displays negative Є Nd (t),<br />
whereas oceanic crust derived from a depleted mantle has a higher<br />
Sm/Nd ratio, more radiogenic 143 Nd, <strong>and</strong> positive Є Nd (t) values.<br />
Therefore, Sm–Nd isotope ratios in a sample <strong>of</strong> known age can distinguish<br />
whether Nd has been primarily derived from continental or<br />
oceanic crust.<br />
The purpose <strong>of</strong> this study is to screen IFs for seawater provenance<br />
on the basis <strong>of</strong> REE patterns, <strong>and</strong> then re-evaluate the isotopic evolution<br />
<strong>of</strong> Nd in Archean seawater. Though data between 3.0 <strong>and</strong> 3.5 Ga<br />
are few, it seems that seawater possessed relatively constant Є Nd (t)<strong>of</strong><br />
approximately +1 to +2 from 3.8 Ga until ~2.6 Ga. New Nd isotopic<br />
results for 2.95 Ga IF from the Pietersburg greenstone belt (PGB) in<br />
South Africa support this conclusion, <strong>and</strong> the PGB samples show strong<br />
evidence for mixing between a detrital sediment source with slightly<br />
negative Є Nd (t) <strong>and</strong> ambient seawater with positive Є Nd (t) equal to or<br />
greater than +1. It is concluded that, unlike modern seawater, the<br />
great majority <strong>of</strong> Nd in bulk seawater prior to 2.7 Ga originated from<br />
mantle-derived mafic source rocks. However, comparison <strong>of</strong> the<br />
Pietersburg IF to similarly aged IF from South Africa indicates that<br />
where evolved local continental crust was present this mantle Nd<br />
signal was not discernible in shallow Archean seawater (Alex<strong>and</strong>er<br />
et al., 2008), implying that like modern oceans, the Archean ocean was<br />
not well-mixed with respect to its Nd isotopic composition.<br />
2. Nd in seawater <strong>and</strong> previous IF studies<br />
Seawater in modern oceans possesses negative Є Nd (0) between<br />
−1 <strong>and</strong> −20, though extreme values in this range are restricted to<br />
shallow waters (Frank, 2002; Goldstein <strong>and</strong> Hemming, 2003). There is<br />
no evidence that radiogenic Nd derived from hydrothermal alteration<br />
<strong>of</strong> mid-ocean ridge basalt contributes significantly to modern seawater,<br />
<strong>and</strong> in fact, efficient scavenging <strong>of</strong> REE by precipitating metal<br />
oxides near hydrothermal vent sites is considered to provide an<br />
ultimate sink for REE in seawater (German et al., 1990). As a result,<br />
Nd derived from continental crust dominates world seawater (e.g.,<br />
Piepgras <strong>and</strong> Wasserburg, 1980; Andersson et al., 2008, <strong>and</strong> references<br />
therein). The persistence <strong>of</strong> a ~20 Є Nd -unit variation between ocean<br />
basins indicates that the marine residence time <strong>of</strong> Nd (τ Nd ) is less than<br />
the mixing time <strong>of</strong> the oceans (~1500 yr, e.g., Broecker <strong>and</strong> Peng,<br />
1982), <strong>and</strong> estimates <strong>of</strong> τ Nd in modern oxic seawater based on isotopic<br />
studies typically range from ~500 yr (Tachikawa et al., 2003) to 1000–<br />
2000 yr (Je<strong>and</strong>el et al., 1995;). Temporal variability in seawater Є Nd (t)<br />
has also been noted, <strong>and</strong> Recent (b20 Ka) seawater Nd ratios appear to<br />
have fluctuated by 2–3 Є Nd -units on 1000–5000 yr time scales (e.g.,<br />
Piotrowski et al., 2005; Gutjahr et al., 2008).<br />
Early Nd isotopic studies <strong>of</strong> IFs attempted to constrain the source <strong>of</strong><br />
Fe to coeval seawater, based upon the premise that initial 143 Nd/ 144 Nd<br />
ratios in IFs would indicate if Fe was derived from hydrothermal fluids<br />
originating in ancient oceanic crust, or if the Fe was sourced from<br />
weathering <strong>of</strong> contemporaneous continental crust. However, this<br />
approach is problematic as recently discussed by Alex<strong>and</strong>er et al.<br />
(2008), as the different geochemical behaviors <strong>of</strong> Fe <strong>and</strong> Nd do not<br />
allow definitive statements to be made regarding the Fe source to IF<br />
based upon Nd isotopes.<br />
Regardless <strong>of</strong> the Fe source to IFs, the use <strong>of</strong> Sm–Nd isotopes for<br />
discerning the relative impact <strong>of</strong> oceanic crust-derived hydrothermal<br />
fluxes to Archean seawater should be possible. Neodymium isotopic<br />
data for IFs have varied, with Miller <strong>and</strong> O'Nions (1985) concluding<br />
that continental crust supplied much <strong>of</strong> the Nd to Archean oceans.<br />
However, later studies by <strong>Jacobs</strong>en <strong>and</strong> Pimentel-Klose (1988a, b),<br />
Alibert <strong>and</strong> McCulloch (1993), <strong>and</strong> Bau et al. (1997a) concluded that<br />
contemporaneous seawater Nd was primarily derived from hydrothermal<br />
alteration <strong>of</strong> mafic oceanic crust. A recent study <strong>of</strong> 2.9 Ga IF by<br />
Alex<strong>and</strong>er et al. (2008) indicates that shallow Archean seawater along<br />
cratonic margins was dominated by continentally-derived Nd, similar<br />
to coastal seawater in modern oceans. Whereas early studies have<br />
extrapolated data from relatively small numbers <strong>of</strong> samples <strong>and</strong> locations<br />
to be representative <strong>of</strong> the entire Archean ocean (e.g., Miller <strong>and</strong><br />
O'Nions, 1985; <strong>Jacobs</strong>en <strong>and</strong> Pimentel-Klose, 1988b), the conflicting<br />
interpretations could reflect spatial <strong>and</strong> temporal heterogeneity with<br />
respect to 143 Nd/ 144 Nd ratios in Earth's early oceans. This study<br />
addresses the possibility that Nd isotopic heterogeneity existed in the<br />
Archean ocean, similar to what is observed in modern oceans.<br />
It is necessary to examine the possibility that primary Nd isotopic<br />
signals recorded in IFs may be reset by post-depositional processes;<br />
however, the Sm–Nd isotopic system is resistant to metamorphism<br />
<strong>and</strong> consistent Nd isotopic results are obtained for IFs that have experienced<br />
a wide range <strong>of</strong> metamorphic overprint. Bau et al. (1997a)<br />
reported Sm–Nd data for the stratigraphically equivalent Kuruman<br />
<strong>and</strong> Penge IFs <strong>of</strong> the Transvaal Supergroup in South Africa (Beukes,<br />
1983). Whereas the Kuruman IFs were weakly metamorphosed<br />
(b200 °C) <strong>and</strong> the Penge IFs suffered intense metamorphic overprint<br />
(N500 °C), both IFs yield very similar Sm–Nd isotopic ratios, <strong>and</strong> REE<br />
patterns are indistinguishable between the two. Moreover, 2.5–3.7 Ga<br />
IFs from a number <strong>of</strong> locations <strong>and</strong> with very different metamorphic<br />
histories display remarkably similar, seawater-like REE distributions<br />
(e.g., Alibert <strong>and</strong> McCulloch, 1993; Bolhar et al., 2004), <strong>and</strong> in a<br />
detailed study <strong>of</strong> worldwide IFs Bau (1993) concluded that postdepositional<br />
REE mobility in IFs is negligible.<br />
3. Geologic setting<br />
The Pietersburg greenstone belt outcrops over a 150 km long by<br />
70 km wide area located in northeast South Africa (Fig. 1). The<br />
northeastern portion <strong>of</strong> the PGB consists <strong>of</strong> schists, is bounded by the<br />
Limpopo metamorphic province, <strong>and</strong> experienced the highest degrees<br />
<strong>of</strong> deformation <strong>and</strong> lower to upper amphibolite grade metamorphism<br />
(De Wit et al., 1992). The southwestern portion <strong>of</strong> the belt is<br />
surrounded by igneous intrusives <strong>and</strong> younger cover <strong>of</strong> the Traansvaal<br />
Supergroup, <strong>and</strong> experienced less tectonic deformation <strong>and</strong> lower<br />
degrees <strong>of</strong> metamorphism (De Wit et al., 1992). The PGB contains two<br />
major tectonic units separated by a distinct unconformity, above<br />
which lies the sedimentary Uitkyk Formation (De Wit et al., 1992).<br />
Below the unconformity, the South African Committee for Stratigraphy<br />
(SACS, 1980) recognized five formations, from oldest to youngest,<br />
the Mothiba, Ysterberg, Eersteling, Z<strong>and</strong>rivierspoort, <strong>and</strong> the Vrischgewaagd.<br />
Only the lowermost Mothiba <strong>and</strong> Ysterberg Formations<br />
contain IF. However, De Wit (1991) observed significant structural<br />
repetition in the southwestern portion <strong>of</strong> the PGB, where the IF<br />
samples for this study originated, <strong>and</strong> concluded that a strict<br />
SACS stratigraphy was not applicable. Therefore, we adopt the stratigraphy<br />
<strong>of</strong> De Wit (1991) for this area, which labels all units below<br />
the unconformity as simatic basement, due to the silica-magnesia
Author's personal copy<br />
146 B.W. Alex<strong>and</strong>er et al. / Earth <strong>and</strong> Planetary <strong>Science</strong> Letters 283 (2009) 144–155<br />
Fig. 1. Map showing the location <strong>of</strong> 2.9–3.5 Ga greenstone belts associated with the eastern margin <strong>of</strong> the Kaapvaal craton, including the Pietersburg greenstone belt (PGB), its<br />
distribution, <strong>and</strong> a geologic map <strong>of</strong> the southwestern region <strong>of</strong> the PGB (modified after De Wit et al., 1992). The oldest epicratonic sediments in South Africa are located in the<br />
~2.90 Ga Pongola Supergroup, which consists <strong>of</strong> the older Nsuze Group <strong>and</strong> the younger, IF-bearing Mozaan Group. Mozaan IF discussed in the text was sampled from the Wit Mfolozi<br />
inlier. Also shown on the detailed map are the type localities for the Pietersburg geologic formations.<br />
lithologies that dominate. The simatic basement primarily consists <strong>of</strong><br />
mafic volcanic rocks including metagabbros, serpentinized peridotites,<br />
<strong>and</strong> massive pillowed metatholeiites, with accessory interlayered<br />
IF, Fe-rich shales, <strong>and</strong> minor cherts <strong>and</strong> carbonates (De Wit,<br />
1991).<br />
Few radiometric age data exist for the PGB. The maximum age <strong>of</strong><br />
the Uitkyk Formation is interpreted as ~2.90 Ga, as determined<br />
by three detrital zircon uranium–lead (U–Pb) ages that range from<br />
2901±2 Ma to 2957±8 Ma (De Wit et al., 1993). A minimum age <strong>of</strong><br />
2687 ±2 Ma for the PGB is provided by a U–Pb single zircon analysis <strong>of</strong><br />
the Uitloop granite (Fig. 1), which cross-cuts all units <strong>of</strong> the PGB<br />
(De Wit et al., 1993). For the simatic basement below the unconformity<br />
a Pb–Pb date for mafic metavolcanics from the Eersteling area<br />
<strong>of</strong> 3455±128 Ma was reported by Byron <strong>and</strong> Barton (1990), but this<br />
age was interpreted as reflecting a secondary metamorphic overprint.<br />
The only other data for the simatic basement consists <strong>of</strong> two zircon<br />
analyses for a metaquartz porphyry associated with IF in the Ysterberg<br />
area. These analyses yielded a Pb–Pb evaporation age <strong>of</strong> 2939±26 Ma,<br />
<strong>and</strong> a precise U–Pb zircon age <strong>of</strong> 2949±0.2 Ma (Kröner et al., 2000). A<br />
granitoid located 15 km west <strong>of</strong> the Ysterberg area that intrudes into,<br />
<strong>and</strong> is deformed with, the lower simatic basement yielded a U–Pb<br />
zircon age <strong>of</strong> 2958±2 Ma (De Wit et al., 1993). These data indicate<br />
that the oldest sections <strong>of</strong> the PGB are greater than 2.96 Ga in age, <strong>and</strong><br />
the fact that iron-formation is described only in the oldest portions<br />
<strong>of</strong> the simatic basement (Mothiba <strong>and</strong> Ysterberg Formations) suggests<br />
that IF deposition occurred no later than 2.95 Ga. Therefore, considering<br />
the precision <strong>of</strong> the U–Pb date (2949±0.2 Ma) reported by<br />
Kröner et al. (2000) for the IF-bearing sequence in the Ysterberg<br />
Formation, we assign the IF in the southwest portion <strong>of</strong> the PGB a<br />
minimum age <strong>of</strong> 2.95 Ga for the purposes <strong>of</strong> interpreting Nd isotopic<br />
data.<br />
For comparison, the Pietersburg samples are discussed in relation<br />
to IF from the ~2.9 Ga Mozaan Group, which is part <strong>of</strong> the Pongola<br />
Supergroup located southeast <strong>of</strong> the PGB (Fig. 1). The Mozaan IF<br />
were part <strong>of</strong> the first sediments unconformably deposited on the older<br />
Nsuze Group, <strong>and</strong> are considered ideal archives for shallow Archean<br />
seawater (Alex<strong>and</strong>er et al., 2008). Few age constraints exist for<br />
the Mozaan Group. A rhyolite occurring in the upper third <strong>of</strong> the<br />
older Nsuze Group yielded a SHRIMP U–Pb date <strong>of</strong> 2977±5 Ma for<br />
single zircons (Nhleko, 2003), <strong>and</strong> is the best constraint for the upper<br />
age limit <strong>of</strong> the Mozaan Group (see also Hegner et al., 1994). The<br />
youngest depositional age for the Mozaan Group has been interpreted<br />
as 2837±5 Ma from a SHRIMP U–Pb zircon age obtained from a<br />
quartz-porphyry sill (Gutzmer et al., 1999). Based on the limited data,<br />
it appears that the oldest parts <strong>of</strong> the Mozaan Group (the IF) may be as<br />
old as ~2.95 Ga, <strong>and</strong> these sediments are unlikely to be younger than<br />
~2.84 Ga.<br />
4. Sampling <strong>and</strong> analytical methods<br />
The Pietersburg IF was sampled from an approximately 50 cm long<br />
continuous section <strong>of</strong> drill core that was subdivided into seven pieces
Author's personal copy<br />
B.W. Alex<strong>and</strong>er et al. / Earth <strong>and</strong> Planetary <strong>Science</strong> Letters 283 (2009) 144–155<br />
147<br />
varying between 5 <strong>and</strong> 10 cm in length <strong>and</strong> labelled as IF1 to IF7. The<br />
entire length <strong>of</strong> sampled drill core is highly magnetic. The IF is finely<br />
laminated on millimeter scales <strong>and</strong> displays thicker b<strong>and</strong>ing on<br />
centimeter scales. The longest individual pieces <strong>of</strong> drill core were<br />
subdivided two or three times <strong>and</strong> these samples are identified as<br />
such, e.g., IF6, IF6a, <strong>and</strong> IF6b are individual samples from the ~10 cm<br />
long IF6 piece <strong>of</strong> drill core. This resulted in a total <strong>of</strong> fourteen samples,<br />
each representing relatively homogenous ~1 cm thick b<strong>and</strong>s that<br />
contained varying amounts <strong>of</strong> Fe. Samples were crushed with a steel<br />
jaw crusher <strong>and</strong> powdered with an agate ball mill. Major element<br />
concentrations were determined by X-ray fluorescence (XRF) at the<br />
SpectRAU laboratory <strong>of</strong> the <strong>University</strong> <strong>of</strong> Johannesburg, South Africa.<br />
Trace element concentrations were determined at <strong>Jacobs</strong> <strong>University</strong><br />
Bremen using inductively-coupled plasma mass spectrometry<br />
(ICPMS) following the methods described by Dulski (2001). Briefly,<br />
for both trace metal <strong>and</strong> Sm–Nd isotopic analyses, powdered samples<br />
were digested under pressure at high temperature (TN160 °C) in<br />
Teflon vessels using concentrated ultrapure perchloric (HClO 4 ) <strong>and</strong><br />
hydr<strong>of</strong>luoric (HF) acids. Following HClO 4 –HF evaporation the samples<br />
were re-dissolved in hydrochloric acid prior to ICPMS analysis or REE<br />
separation. Analytical quality for major <strong>and</strong> trace element determinations<br />
was monitored using the iron-formation st<strong>and</strong>ard FeR-2<br />
(Geological Survey <strong>of</strong> Canada), analysis <strong>of</strong> which produced good<br />
agreement with reference values (Table 1).<br />
Samples for Sm–Nd isotope determinations were spiked prior<br />
to dissolution with a mixed 147 Sm/ 150 Nd spike, <strong>and</strong> analyzed at the<br />
Laboratory for Isotope Geology (LIG), Swedish Museum <strong>of</strong> Natural<br />
History. Samarium <strong>and</strong> Nd were separated using ion exchange chromatography<br />
techniques described in Andersson et al. (2008). The<br />
isotopic ratios were determined with a Thermo Scientific TritonTIMS<br />
(thermal ionization mass spectrometer) using multicollector static<br />
mode. Samarium <strong>and</strong> Nd were loaded on double rhenium filaments. The<br />
total Nd blank averaged 43 pg for three separate digestion batches <strong>and</strong> is<br />
an insignificant contribution to the sample isotopic composition. Measured<br />
isotope ratios were reduced assuming exponential fractionation.<br />
Samarium ratios were normalized to 149 Sm/ 152 Sm = 0.516747. Neodymium<br />
was run in static mode using rotating gain compensation, <strong>and</strong><br />
calculated ratios were normalized to 146 Nd/ 144 Nd=0.7219.<br />
The La Jolla st<strong>and</strong>ard yielded 143 Nd/ 144 Nd=0.511848 ±0.0000047<br />
(2σ, n=32). No corrections were applied to the measured ratios <strong>and</strong><br />
Table 1<br />
Major <strong>and</strong> trace element data for Pietersburg IF.<br />
Sample IF1 IF1a IF2 IF2a IF2b IF3 IF4 IF5 IF5a IF6 IF6a IF6b IF7 IF7a FeR-2 Ref a<br />
(wt.%)<br />
SiO 2 49.3 33.7 61.3 44.7 38.4 43.5 38.8 28.6 38.4 49.1 44.7 61.8 44.8 nd 50.4 49.21<br />
TiO 2 0.04 0.05 0.02 0.07 0.07 0.09 0.07 0.04 0.01 0.02 0.03 0.02 0.25 nd 0.19 0.18<br />
Al 2 O 3 0.73 0.72 0.32 1.06 0.91 1.50 1.53 0.33 0.13 0.26 0.54 0.30 5.30 nd 5.19 5.16<br />
Fe 2 O 3 45.6 60.5 36.8 43.7 56.4 48.3 52.5 68.7 58.8 48.7 49.8 37.0 42.1 nd 39.67 39.21<br />
MnO 0.20 0.28 0.05 0.13 0.11 0.21 0.53 0.79 0.64 0.08 0.08 0.09 0.12 nd 0.12 0.12<br />
MgO 3.23 0.07 1.65 4.25 3.55 3.90 3.19 2.67 1.42 1.84 2.17 2.22 5.15 nd 2.23 2.10<br />
CaO 1.75 1.74 0.89 3.45 1.28 2.13 2.77 1.18 1.43 1.07 1.64 2.79 1.33 nd 2.13 2.17<br />
Na 2 O bd bd bd bd bd bd bd bd bd bd bd bd bd nd 0.39 0.51<br />
K 2 O bd 0.02 bd 0.04 0.03 0.02 0.01 bd 0.01 bd 0.01 0.01 0.09 nd 1.34 1.33<br />
P 2 O 5 0.19 0.23 0.11 0.20 0.24 0.30 0.22 0.21 0.10 0.14 0.21 0.16 0.16 nd 0.27 0.27<br />
S bd bd 0.56 0.24 0.12 0.03 0.11 bd bd 0.57 0.17 0.22 0.54 nd 0.33 0.17<br />
Total 101.04 97.31 101.70 98.84 101.11 99.98 99.73 102.52 100.94 101.78 99.35 104.61 99.84 nd 102.27<br />
mg/kg<br />
Sc 1.81 2.06 b1.30 2.86 3.03 2.83 2.85 b1.30 b1.30 b1.30 b1.30 b1.30 7.20 b1.0 4.94 6<br />
Ti 195 306 101 438 488 384 352 172 47.4 80.7 161 91.9 1284 151 1045 1079<br />
Co 4.11 6.00 1.94 7.73 6.94 6.02 6.24 2.61 1.20 1.58 2.31 1.67 14.3 2.29 6.50 7<br />
Ni 26.2 22.3 12.3 57.6 47.2 33.3 31.0 13.9 5.38 8.94 15.1 11.2 128 15.2 22.0 21<br />
Rb 1.17 1.13 0.703 2.46 2.05 2.00 2.07 1.17 0.368 0.52 0.657 0.498 5.60 0.837 62.6 66<br />
Sr 29.5 28.5 11.0 51.2 15.8 33.8 41.0 24.2 29.8 14.8 25.0 38.4 17.9 21.7 60.8 58<br />
Y 10.3 12.6 4.98 10.5 8.96 15.0 18.3 10.8 5.91 6.01 9.68 7.54 12.7 8.74 12.4 15<br />
Zr 9.88 10.0 4.81 14.4 11.7 16.6 15.6 8.20 1.67 3.87 5.15 2.90 61.4 4.90 36.4 39<br />
Cs 0.366 0.417 0.160 0.633 0.60 0.681 0.661 0.467 0.166 0.144 0.265 0.155 1.45 0.273 4.55 5<br />
Ba 2.13 4.26 1.41 6.13 4.33 3.83 3.03 5.77 1.62 2.07 1.31 1.07 6.26 1.39 223 240<br />
La 4.62 4.96 1.96 4.45 5.28 7.69 9.06 4.47 1.43 2.22 3.42 2.44 8.49 2.60 11.8 14<br />
Ce 8.43 9.14 3.53 8.34 9.56 14.0 18.0 7.87 1.88 3.79 5.56 3.87 17.1 4.38 24.1 25<br />
Pr 1.01 1.12 0.425 1.02 1.11 1.65 2.12 0.942 0.204 0.439 0.633 0.454 2.01 0.505 2.83 3<br />
Nd 4.24 4.89 1.80 4.38 4.68 6.84 8.84 3.91 0.894 1.90 2.76 1.95 8.01 2.20 11.5 12<br />
Sm 0.983 1.16 0.393 0.990 1.08 1.55 2.02 0.890 0.183 0.410 0.592 0.431 1.68 0.504 2.44 2.6<br />
Eu 0.418 0.561 0.190 0.473 0.461 0.688 0.973 0.544 0.121 0.241 0.306 0.276 0.601 0.266 1.20 1.25<br />
Gd 1.27 1.55 0.539 1.31 1.39 1.96 2.55 1.24 0.326 0.594 0.879 0.676 1.87 0.743 2.25 2<br />
Tb 0.192 0.236 0.079 0.211 0.201 0.308 0.417 0.191 0.048 0.090 0.140 0.101 0.294 0.118 0.332 0.32<br />
Dy 1.30 1.59 0.560 1.36 1.25 2.07 2.82 1.31 0.344 0.644 0.960 0.708 1.91 0.809 2.10 2<br />
Ho 0.298 0.357 0.128 0.304 0.272 0.463 0.621 0.300 0.093 0.153 0.237 0.180 0.412 0.195 0.436 0.6<br />
Er 0.910 1.12 0.406 0.957 0.840 1.43 1.92 0.945 0.304 0.489 0.751 0.550 1.30 0.618 1.31 1.5<br />
Tm 0.132 0.158 0.061 0.135 0.118 0.199 0.271 0.136 0.045 0.070 0.107 0.081 0.193 0.089 0.187 0.2<br />
Yb 0.870 1.01 0.368 0.881 0.755 1.26 1.73 0.884 0.276 0.462 0.686 0.515 1.27 0.583 1.23 1.3<br />
Lu 0.137 0.166 0.062 0.143 0.123 0.203 0.268 0.141 0.047 0.076 0.114 0.084 0.200 0.096 0.191 0.2<br />
Hf 0.248 0.258 0.118 0.370 0.291 0.438 0.373 0.220 0.037 0.102 0.135 0.073 1.68 0.114 0.955 1<br />
Ta 0.040 0.044 0.017 0.052 0.046 0.074 0.064 0.036 0.007 0.015 0.022 0.013 0.264 0.020 0.147 0.2<br />
W 50.9 140 21.1 43.4 27.9 61.5 134 31.3 59.9 52.2 74.1 39.2 55.9 86.7 2.36<br />
Pb 1.80 2.09 1.29 2.94 1.94 2.25 2.71 2.34 1.41 1.41 1.37 1.69 2.21 1.27 8.43 11<br />
Th 0.330 0.380 0.155 0.473 0.430 0.523 0.506 0.280 0.066 0.174 0.208 0.124 1.46 0.174 2.44 3<br />
U 0.083 0.102 0.041 0.113 0.108 0.146 0.146 0.079 0.020 0.036 0.053 0.031 0.500 0.050 1.33 1.2<br />
bd=below determination limit.<br />
nd=not determined.<br />
a Ref =values for FeR-2 iron-formation st<strong>and</strong>ard as provided by issuing agency (Canada Centre for Mineral <strong>and</strong> Energy Technology-Mining <strong>and</strong> Mineral <strong>Science</strong>s Laboratories, Abbey<br />
et al., 1983), except for italicized data which are obtained from the GeoReM database (Jochum et al., 2005).
Author's personal copy<br />
148 B.W. Alex<strong>and</strong>er et al. / Earth <strong>and</strong> Planetary <strong>Science</strong> Letters 283 (2009) 144–155<br />
Table 2<br />
Sm–Nd concentrations (mg/kg) <strong>and</strong> isotope ratios for Pietersburg IF.<br />
Sample Sm Nd<br />
147 Sm/ 144 Nd<br />
143 Nd/ 144 Nd±2σ a Є Nd (0) Є Nd (t) Sm Nd<br />
TIMS TIMS t=2.95 Ga ICPMS %RSD ICPMS %RSD<br />
TIMS/ICPMS<br />
TIMS/ICPMS<br />
IF1 0.954 4.116 0.1401 0.511502±4 −22.2 −0.66 0.983 −3.0 4.24 −2.9<br />
IF2 0.426 1.915 0.1344 0.511466±5 −22.9 0.83 0.393 8.4 1.80 6.4<br />
IF3 1.525 6.694 0.1377 0.511464±4 −22.9 −0.48 1.55 −1.6 6.84 −2.1<br />
IF4 2.029 8.862 0.1384 0.511485±3 −22.5 −0.35 2.02 0.4 8.84 0.2<br />
IF4r b 2.035 8.874 0.1386 0.511486±3 −22.5 −0.39<br />
IF5 0.847 3.725 0.1375 0.511499±3 −22.2 0.29 0.890 −4.8 3.91 −4.7<br />
IF5r b 0.856 3.747 0.1381 0.511504±4 −22.1 0.14<br />
IF6 0.424 1.922 0.1334 0.511455±5 −23.1 0.99 0.410 3.4 1.90 1.2<br />
IF7 1.717 8.041 0.1291 0.511290 ±9 −26.3 −0.61 1.68 2.2 8.01 0.4<br />
t=3.8 Ga<br />
IF-G c 0.39 1.72 0.1371 0.511277 2.73<br />
IF-G 0.383 1.687 0.1373 0.511258±4 2.25<br />
BCR-2 d 6.574 28.64 0.1388 0.512628 ±2<br />
a<br />
b<br />
c<br />
2σ refers to last digit in the measured ratio.<br />
r in sample name denotes a replicate digestion <strong>and</strong> analysis <strong>of</strong> the same sample, <strong>and</strong> these data are not discussed in the text.<br />
Data from Stecher et al. (1986). IF-G distributed by GIT-IWG (Groupe International de Travail – International Working Group), Service d' Analyse des Roches et des Minéraux,<br />
CRPG-CNRS, V<strong>and</strong>oeuvre-lès-Nancy, France. Sample powder from the original IF-G processing run is no longer available, though a second batch <strong>of</strong> Isua IF-G iron-formation is<br />
currently available from GIT-IWG. Data for this study, <strong>and</strong> presumably from Stecher et al. (1986) as well, are for IF-G powder from the original processing run.<br />
d Average <strong>of</strong> 2 separate digestions <strong>of</strong> BCR-2 Columbia River basalt reference material (United States Geologic Survey).<br />
the overall reproducibility (external precision) is 9 ppm as estimated<br />
from the st<strong>and</strong>ards. Uncertainties in TIMS determination <strong>of</strong> Sm <strong>and</strong> Nd<br />
concentrations are estimated as b1% <strong>and</strong> 2%, respectively.<br />
Agreement between TIMS <strong>and</strong> ICPMS concentration data is better<br />
than 4.7% for both Sm <strong>and</strong> Nd, except for sample IF2 where TIMS<br />
values are respectively 8.4% <strong>and</strong> 6.4% higher relative to ICPMS data<br />
(Table 2). Rare earth element ratios calculated from TIMS <strong>and</strong> ICPMS<br />
data are significantly more consistent, with all samples displaying<br />
Sm/Nd that varies by no more than 2.2% between the two analytical<br />
methods. Therefore, we very conservatively estimate the accuracy for<br />
ICPMS analyses as better than ±10%, <strong>and</strong> accuracy for rare earth<br />
element ratios as better than ±5%.<br />
Fig. 2. Selected trace element concentrations plotted as a function <strong>of</strong> Ti. Trends are similar regardless <strong>of</strong> elements affinity for felsic (Rb, Th) or mafic crust (Ni, Co). The excellent<br />
correlation between Rb <strong>and</strong> more immobile metals such as Al <strong>and</strong> Ti argues against significant post-depositional mobility <strong>of</strong> trace metals following IF deposition.
Author's personal copy<br />
B.W. Alex<strong>and</strong>er et al. / Earth <strong>and</strong> Planetary <strong>Science</strong> Letters 283 (2009) 144–155<br />
149<br />
Fig. 3. REY distributions in IF samples normalized to Al-rich IF7 (5.3% Al 2 O 3 ). Inset<br />
displays REY distribution in Al-rich IF7 normalized to chondritic meteorite (Anders <strong>and</strong><br />
Grevesse, 1989) <strong>and</strong> continental crust as represented by Post Archean Australian Shale<br />
(PAAS, McLennan, 1989). PAAS possesses a negative Eu anomaly compared to bulk<br />
Earth, <strong>and</strong> PAAS-normalized data will correspondingly display positive Eu anomalies<br />
when no Eu fractionation exists. This is the case for Al-rich IF7, for which no anomalous<br />
behavior <strong>of</strong> Eu is observed. The remaining samples all display positive Eu anomalies<br />
relative to IF7 that significantly vary in magnitude, with greater Eu enrichments in<br />
samples <strong>of</strong> lower total REY abundance. Positive La, Gd, <strong>and</strong> Y anomalies are present in<br />
many <strong>of</strong> the samples <strong>and</strong>, similar to Eu, the magnitude <strong>of</strong> these anomalies increases<br />
with decreasing REY concentration. The effect <strong>of</strong> detrital contamination is observed in<br />
samples IF3 <strong>and</strong> IF4, which have the second highest Al 2 O 3 contents (1.5%) after IF7, <strong>and</strong><br />
relatively flat REY patterns. However, IF4 also possesses greater total REY concentrations<br />
that IF7, which coupled with the positive Eu anomaly in IF4 indicates that a significant<br />
fraction <strong>of</strong> its REY budget was supplied by the chemical precipitate which produced the<br />
IF. Sample IF7a illustrates the large range <strong>of</strong> REY concentrations (<strong>and</strong> REY fractionation)<br />
observed on cm-scales within the drill core, as it was obtained from a homogenous b<strong>and</strong><br />
adjacent to sample IF7. The range <strong>of</strong> concentrations <strong>and</strong> REY distributions observed<br />
throughout the core argues against the widespread mobility <strong>of</strong> REY at any point<br />
following deposition.<br />
nickel (Ni), <strong>and</strong> sc<strong>and</strong>ium (Sc)), including relatively mobile rubidium<br />
(Rb) (Fig. 2). These correlations remain even if the single sample with<br />
the highest Ti <strong>and</strong> Al content (IF7) is not considered (Fig. 2).<br />
The iron-formations possess REE <strong>and</strong> yttrium (REY) concentrations<br />
that vary by an order <strong>of</strong> magnitude between the samples, <strong>and</strong> the<br />
distribution patterns <strong>of</strong> these elements vary significantly as well. This<br />
may be illustrated by normalizing REY data to IF7, which possesses the<br />
highest Al 2 O 3 content (5.3%), a procedure that effectively removes the<br />
influence <strong>of</strong> any detrital material that may be present (Fig. 3). The Alrich<br />
IF7 itself is light rare earth element (LREE) enriched when<br />
normalized to chondrite (subscript CN , Anders <strong>and</strong> Grevesse, 1989)<br />
<strong>and</strong> shows gadolinium/ytterbium ratios ((Gd/Yb) CN ) <strong>of</strong> ~1 <strong>and</strong> no<br />
REY anomalies (Fig. 3). The two samples with the next highest Al 2 O 3<br />
concentrations (1.5%, IF3 <strong>and</strong> IF4) have higher REY abundances than<br />
the remaining samples, <strong>and</strong> REY concentrations change significantly<br />
on small scales (1–2 cm) throughout the core as a function <strong>of</strong> Al<br />
content. This is illustrated by samples IF5 <strong>and</strong> IF5a, which represent<br />
individual ≤1 cm thick b<strong>and</strong>s sampled within 2–3 cm <strong>of</strong> each other,<br />
<strong>and</strong> that contain Al, Ti, <strong>and</strong> REY contents that vary by a factor <strong>of</strong> ~3.<br />
However, a significant fraction <strong>of</strong> the REY in most <strong>of</strong> the samples<br />
must be derived from the chemical precipitate which produced the IF.<br />
Sample IF4 has one-third the Al <strong>of</strong> sample IF7, yet IF4 has higher REY<br />
concentrations (Fig. 3). In the remaining samples four <strong>of</strong> the REY<br />
display anomalous behavior to varying degrees, with strong positive<br />
europium (Eu) <strong>and</strong> Y anomalies <strong>and</strong> smaller positive lanthanum (La)<br />
<strong>and</strong> Gd anomalies. The IFs also possess slight positive lutetium (Lu)<br />
anomalies, <strong>and</strong> the magnitude <strong>of</strong> all anomalies generally increases as<br />
Sm–Nd analytical quality was monitored using the basalt reference<br />
st<strong>and</strong>ard BCR-2 (United States Geologic Survey), which yielded 143 Nd/<br />
144 Nd=0.512628±0.000002. Additionally, the Isua iron-formation<br />
st<strong>and</strong>ard IF-G, issued by IWG-GIT (International Working Group –<br />
Groupe International de Travail) was analyzed to provide reference<br />
data for subsequent Sm–Nd isotopic studies <strong>of</strong> iron-formations, <strong>and</strong><br />
these data match well with the single previously published Sm–Nd<br />
isotopic analysis <strong>of</strong> IF-G (Stecher et al., 1986).<br />
5. Results<br />
Analytical data are presented in Tables 1 <strong>and</strong> 2. Excluding samples<br />
IF2a <strong>and</strong> IF7, all samples are more than 91% silicon- <strong>and</strong> iron-oxides<br />
(SiO 2 <strong>and</strong> Fe 2 O 3 ), with SiO 2 between 29 <strong>and</strong> 62%, <strong>and</strong> Fe 2 O 3 between<br />
37 <strong>and</strong> 69%. Sample IF2a is enriched in calcium (Ca) <strong>and</strong> magnesium<br />
(Mg) compared to most samples, <strong>and</strong> IF7 possesses significantly more<br />
aluminum (Al) <strong>and</strong> Mg than any other sample. While Al contents are<br />
typically very low (b0.8%), consistent with very low amounts <strong>of</strong><br />
detrital aluminosilicates, IF3 <strong>and</strong> IF4 each contain 1.5% Al 2 O 3 , <strong>and</strong> IF7<br />
contains 5.3% Al 2 O 3 . Trace elements that are considered immobile,<br />
such as zirconium (Zr), hafnium (Hf), <strong>and</strong> thorium (Th), correlate well<br />
with Al <strong>and</strong> titanium (Ti), as do most trace metals (e.g., cobalt (Co),<br />
Fig. 4. Nd concentrations <strong>and</strong> Є Nd (t) <strong>of</strong> Pietersburg IF samples as a function <strong>of</strong> Al<br />
content. Top figure (a): Nd concentrations generally decrease with decreasing Al<br />
concentrations. However, significant Nd must be derived from the chemical precipitate<br />
that formed the IFs, as IF4 contains less than one-third the Al <strong>of</strong> IF7, but higher Nd<br />
concentrations. Bottom figure (b): Iron-formation samples with Al 2 O 3 N0.5% possess<br />
very consistent negative Є Nd (t), whereas positive Є Nd (t) values are only observed in<br />
samples with the lowest Al concentrations, indicating that contamination with minor<br />
detrital aluminosilicates drives Є Nd (t) to more negative values. This suggests that preexisting<br />
crust that provided aluminosilicates to the IF depositional environment possessed<br />
Є Nd (t) <strong>of</strong> approximately −0.5, <strong>and</strong> ambient Fe-rich seawater possessed Є Nd (t)<br />
equal to or greater than +1. Error bars for Є Nd (t) calculated by assuming average errors<br />
in 147 Sm/ 144 Nd ratios <strong>of</strong> 0.5%.
Author's personal copy<br />
150 B.W. Alex<strong>and</strong>er et al. / Earth <strong>and</strong> Planetary <strong>Science</strong> Letters 283 (2009) 144–155<br />
REY abundances decrease. However, when comparing samples <strong>of</strong><br />
similar REY content the relative size <strong>of</strong> the anomalies varies. For<br />
example, IF5 possesses a REY distribution <strong>and</strong> positive La <strong>and</strong> Y<br />
anomalies almost identical to samples <strong>of</strong> similar concentration, yet IF5<br />
displays a significantly larger Eu anomaly.<br />
The possibility <strong>of</strong> significant alteration <strong>of</strong> the primary REY distributions<br />
is considered unlikely. Mobility <strong>of</strong> the REY has not been observed<br />
in Archean IFs (Bau, 1993), <strong>and</strong> concentrations <strong>of</strong> REY may vary<br />
considerably between adjacent Fe- <strong>and</strong> Si-rich b<strong>and</strong>s in Archean–<br />
Paleoproterozoic IFs (Bau <strong>and</strong> Dulski, 1992). This behavior is also<br />
observed in samples IF5 <strong>and</strong> IF5a, which were obtained from adjacent<br />
layers in a ~4 cm section <strong>of</strong> drill core <strong>and</strong> yet possess distinct REY<br />
patterns <strong>and</strong> very different REY concentrations (Fig. 3). We emphasize<br />
that the lack <strong>of</strong> significant diagenetic or metamorphic alteration is<br />
further supported by the excellent correlation <strong>of</strong> relatively mobile Rb<br />
compared to highly immobile elements such as Ti <strong>and</strong> Th (Fig. 2).<br />
Similarly to immobile trace elements, Є Nd (2.95 Ga) <strong>of</strong> the IF<br />
samples varies as a function <strong>of</strong> Al content (Fig. 4), with samples containing<br />
more than 0.5% Al 2 O 3 displaying relatively uniform negative<br />
Є Nd (2.95 Ga) between −0.35 <strong>and</strong> −0.66. Samples containing less<br />
than 0.5% Al 2 O 3 are significantly more radiogenic, displaying positive<br />
Є Nd (2.95 Ga) between +0.29 <strong>and</strong> +0.99. The difference between the<br />
two groups <strong>of</strong> Є Nd (2.95 Ga) values exceeds the analytical error <strong>of</strong> the<br />
measurements. The correlation between increasing immobile element<br />
contents (e.g., Al, Th, Zr, Rb, <strong>and</strong> Hf) <strong>and</strong> decreasing Є Nd (t) suggests<br />
that crustal sources contributing detritus to the Pietersburg IF depositional<br />
basin possessed Є Nd (2.95 Ga) <strong>of</strong> approximately −0.5, <strong>and</strong><br />
seawater responsible for precipitating the IF was more radiogenic with<br />
Є Nd (2.95 Ga) similar to or greater than +1.<br />
6. Discussion<br />
6.1. Nature <strong>of</strong> the detrital aluminosilicate source<br />
The higher Al 2 O 3 content (≥1.5%) <strong>of</strong> some IF samples indicates<br />
contamination with detrital aluminosilicates, <strong>and</strong> c<strong>and</strong>idates for a<br />
detrital source(s) to the iron-formation include lithologies in the<br />
simatic basement. Published data from the Eersteling area are primarily<br />
for major elements <strong>and</strong> transition metals, <strong>and</strong> 80 analyses have<br />
characterized metagabbros, metabasalts, amphibolites, serpentinites,<br />
as well as quartz porphyrys (Saager <strong>and</strong> Meyer, 1982; Jones, 1990;<br />
Byron <strong>and</strong> Barton, 1990). Aluminum averaged 9.7% in these samples<br />
<strong>and</strong> only four contained more than 15% Al 2 O 3 (maximum 16.4%),<br />
suggesting that if the aluminosilicate fraction in the iron-formation<br />
was derived from pre-existing simatic basement then 15% Al 2 O 3 is the<br />
likely upper limit for this material, which agrees well with estimates<br />
for Archean mafic rocks (Condie, 1993). The strong correlations<br />
between immobile trace elements <strong>and</strong> Al allows reasonable estimates<br />
to be made regarding the trace metal distribution in the detrital source,<br />
<strong>and</strong> by comparing this estimated distribution with trace element data<br />
from the literature it may be possible to restrict the provenance <strong>of</strong> the<br />
detrital component. Estimated trace metal data (Fig. 5) for the most Alrich<br />
sample (IF7) at higher Al 2 O 3 contents <strong>of</strong> 10–15% match reasonably<br />
well with a mixture <strong>of</strong> Archean felsic volcanics <strong>and</strong> komatiites possessing<br />
compositions as proposed by Condie (1993), <strong>and</strong> komatiites in<br />
the PGB have been described previously (Saager <strong>and</strong> Mayer, 1982).<br />
However, few trace element data exist for PGB mafic rocks, making it<br />
difficult to characterize the nature <strong>of</strong> this aluminosilicate source,<br />
except to state that physical weathering <strong>of</strong> serpentinized material is an<br />
unlikely c<strong>and</strong>idate for any detritus present in the Pietersburg IF.<br />
The relatively high Al 2 O 3 content in some samples (e.g., IF3, IF4, <strong>and</strong><br />
IF7) renders them unsuitable as archives for the marine fluid that<br />
precipitated the Pietersburg IF, but these samples do provide Nd isotopic<br />
information regarding the clastic detrital source during deposition<br />
<strong>of</strong> the IF. Samples with Al 2 O 3 greater than ~0.5% possess very<br />
consistent Є Nd (t) close to −0.5 (Fig. 4), which is considered to reflect<br />
the Nd isotopic composition <strong>of</strong> the detrital aluminosilicate source. Close<br />
examination <strong>of</strong> the Al–Є Nd (t) relationships (Fig. 4) suggeststhattwocomponent<br />
conservative mixing between a detrital aluminosilicate<br />
source <strong>and</strong> the pure chemical precipitate that formed the IFs might<br />
adequately model the Nd isotopic data. However, two-component<br />
mixing models that attempt to account for detrital contamination<br />
Fig. 5. Trace element discrimination diagrams for simatic basement samples from the southwest region <strong>of</strong> the PGB. The grey area in both diagrams represents the range <strong>of</strong> predicted<br />
element concentrations in sample IF7, <strong>and</strong> was calculated from linear regressions <strong>of</strong> the metal contents as a function <strong>of</strong> Al. The lower border <strong>of</strong> the grey area represents 10% Al 2 O 3 <strong>and</strong><br />
the upper border represents 15% Al 2 O 3 , which is a reasonable range for the detrital source (see Section 6). Diagram a contains literature data for possible source lithologies (Jones,<br />
1990; Byron <strong>and</strong> Barton, 1990), <strong>of</strong> which pillow lavas <strong>and</strong> amphibolites are broadly consistent with the predicted distribution pattern. Diagram b shows the distribution patterns for<br />
model Archean komatiites <strong>and</strong> felsic volcanics <strong>of</strong> Condie (1993), a mixture <strong>of</strong> which could reasonably produce the pattern observed in the detrital fraction <strong>of</strong> the iron-formation<br />
samples, <strong>and</strong> komatiites are present within the oldest sections <strong>of</strong> the PGB (Saager <strong>and</strong> Meyer, 1982).
Author's personal copy<br />
B.W. Alex<strong>and</strong>er et al. / Earth <strong>and</strong> Planetary <strong>Science</strong> Letters 283 (2009) 144–155<br />
151<br />
generally fail to accurately describe Є Nd (t) in IFs, as pure, Al-free IFs are<br />
themselves best described as mixtures <strong>of</strong> two components; a siliceous<br />
Nd-poor component (chert), <strong>and</strong> a Nd-rich Fe-oxide component.<br />
Therefore, mixing between pure IF <strong>and</strong> aluminosilicates more realistically<br />
reflects three endmembers; an Nd-rich contaminant with a distinct<br />
143 Nd/ 144 Nd ratio, <strong>and</strong> two IF endmembers that presumably have<br />
similar initial 143 Nd/ 144 Nd ratios, yet very different Nd concentrations.<br />
The relatively constant trace metal ratios as a function <strong>of</strong> increasing<br />
aluminosilicate content (Fig. 2), <strong>and</strong> the strong correlation with<br />
decreasing Є Nd (t) suggests that the detrital component was wellmixed<br />
<strong>and</strong> that the sediment source did not change during the time <strong>of</strong><br />
IF deposition represented by our samples. The lack <strong>of</strong> any negative<br />
Eu CN anomaly in IF7 (Fig. 3) argues against significant contributions<br />
from a highly evolved felsic source that had experienced fractional<br />
crystallization <strong>of</strong> feldspars. The (Gd/Yb) CN ratios close to unity, i.e., the<br />
lack <strong>of</strong> HREE depletion, rules out a dominant tonalite–trondhjemite–<br />
granodiorite (TTG) provenance, or that mafic igneous rocks produced<br />
by small-scale partial melting <strong>of</strong> a garnet-bearing (mantle) source<br />
controlled the REY CN patterns <strong>of</strong> the detritus. Compared to data<br />
compiled by Wilson (1989), incompatible trace element ratios in IF7,<br />
such as Th/Yb (1.2), Ta/Yb (0.2), Zr/Ta (233), <strong>and</strong> Zr/Yb (48) as well<br />
as the REY CN patterns are similar to those <strong>of</strong> modern back-arc basalts.<br />
Assuming this trace element distribution represents the detrital<br />
source, the aluminosilicates in the IFs appear to have originated from<br />
the Archean equivalent <strong>of</strong> mafic back-arc basalts derived from a<br />
slightly depleted mantle source.<br />
6.2. IFs as seawater archives<br />
The majority <strong>of</strong> published Sm–Nd isotopic data for Archean IFs are<br />
for oxide-facies samples like the Pietersburg IF, <strong>and</strong> the following<br />
discussion is therefore restricted to oxide-facies IF. The range <strong>of</strong><br />
literature Є Nd (t) values is large, as illustrated by the ~3.8 Ga Isua IFs.<br />
Shimizu et al. (1990) reported Є Nd (3.8 Ga) for four separate Isua<br />
samples that ranged from −5.0 to +22.4. Another sample (242573<br />
<strong>of</strong> Miller <strong>and</strong> O'Nions, 1985) displays Є Nd (t) values from −15.2 to<br />
+14.8, <strong>and</strong> has provided ten Sm–Nd analyses <strong>of</strong> Isua IF (see also Frei<br />
et al., 1999). To our knowledge only eight individual samples <strong>of</strong> Isua IF<br />
have been analyzed for Sm–Nd isotopes, <strong>and</strong> two <strong>of</strong> these (242573 <strong>of</strong><br />
Miller <strong>and</strong> O'Nions, 1985, <strong>and</strong> 491243 <strong>of</strong> Frei <strong>and</strong> Polat, 2007) have<br />
produced 25 <strong>of</strong> the 30 individual Sm–Nd analyses reported in the<br />
literature.<br />
Considerable Є Nd (t) variability (between −4.7 to +1.6) is also<br />
observed in younger IFs (e.g., the 2.5 Ga Hamersley IF <strong>of</strong> Miller <strong>and</strong><br />
O'Nions, 1985). Whereas consistent Sm–Nd results may be obtained<br />
for some IFs (e.g., the Transvaal Supergroup, Bau et al., 1997a), any<br />
attempt at reconstructing the Nd isotopic signature <strong>of</strong> Archean seawater<br />
is hampered by the wide range <strong>of</strong> reported <strong>and</strong> <strong>of</strong>ten conflicting<br />
Є Nd (t) values. At least in the case <strong>of</strong> Isua this may reflect resetting <strong>of</strong><br />
the isotopic system (Miller <strong>and</strong> O'Nions, 1985; Shimizu et al., 1990)or<br />
possible contamination with clastic detritus. The last effect is difficult<br />
to judge, as frequently only Sm–Nd data exist for IFs <strong>and</strong> major element<br />
concentrations such as Al are not provided (e.g., Miller <strong>and</strong><br />
O'Nions, 1985; <strong>Jacobs</strong>en <strong>and</strong> Pimentel-Klose, 1988b).<br />
However, consistent Sm–Nd data for Isua IF has recently been<br />
reported by Frei <strong>and</strong> Polat (2007), who analyzed 15 b<strong>and</strong>s <strong>of</strong> a single<br />
sample <strong>and</strong> determined Є Nd (3.7 Ga) between +1.7 <strong>and</strong> +2.7. These<br />
data are very similar to the only other Sm–Nd data for Isua IF as<br />
reported in this study <strong>and</strong> Stecher et al. (1986) for the IF-G geost<strong>and</strong>ard<br />
(Table 2), which possesses Є Nd (3.8 Ga) <strong>of</strong> +2.3 to +2.7. Frei<br />
<strong>and</strong> Polat (2007) also noted the similarity between REY distributions<br />
in their Isua sample <strong>and</strong> modern seawater, consistent with similar<br />
observations for Isua IF as reported by Bolhar et al. (2004). Modern<br />
seawater, IF-G, <strong>and</strong> the Pietersburg IF all display very similar<br />
REY patterns (Fig. 6), suggesting that anomalously wide variations<br />
in Є Nd (t) may not be observed in IF samples that possess seawater-like<br />
Fig. 6. Comparison <strong>of</strong> REY in modern seawater with eleven Pietersburg iron-formation<br />
samples that contain b1.5% Al 2 O 3 . Samples are normalized to PAAS to facilitate<br />
comparison with literature data. The seawater pattern represents the average <strong>of</strong> 158<br />
worldwide seawater analyses <strong>of</strong> all depths (average depth 1560 m) for which complete<br />
REY datasets are available (Zhang <strong>and</strong> Nozaki, 1996; Alibo <strong>and</strong> Nozaki, 1999; Bau et al.,<br />
1997b; Nozaki et al., 1999; Nozaki <strong>and</strong> Alibo, 2003). Dashed lines are five Pietersburg IF<br />
samples containing N170 ppm Ti, <strong>and</strong> solid lines represent six samples containing<br />
b170 ppm Ti. Modern seawater displays pronounced positive Y <strong>and</strong> La anomalies, which<br />
are also observed in the Pietersburg IF <strong>and</strong> 3.8 Ga Isua IF-G data (Dulski, 2001), <strong>and</strong> the<br />
general shapes <strong>of</strong> the IF patterns are indistinguishable from seawater. Modern seawater<br />
also displays smaller positive Gd anomalies, <strong>and</strong> slight positive Lu anomalies, features<br />
which are present in the both the Pietersburg IF <strong>and</strong> IF-G. The only distinct differences<br />
between modern seawater <strong>and</strong> the Archean IFs are the strong negative Ce anomaly in<br />
seawater <strong>and</strong> the strong positive Eu anomaly in IF, which respectively reflect lower<br />
relative redox levels <strong>and</strong> high-T hydrothermal input.<br />
REY distributions, a conclusion supported by the 2.5 Ga Transvaal<br />
Supergroup data <strong>of</strong> Bau et al. (1997a).<br />
Therefore, to more clearly underst<strong>and</strong> Nd isotopic ratios in Archean<br />
seawater we have focused on those IFs which display clear seawaterlike<br />
REY patterns. The REY are particularly useful for identifying the<br />
marine origin <strong>of</strong> chemical precipitates (cf. Nothdurft et al., 2004, <strong>and</strong><br />
references therein) because seawater REY distributions are unique<br />
relative to REY sources to the oceans (i.e., oceanic or continental<br />
crust). This characteristic REY pattern in seawater is primarily due to<br />
the different complexation behavior <strong>of</strong> the light <strong>and</strong> heavy REY with<br />
regard to various aqueous carbonate species (Cantrell <strong>and</strong> Byrne,<br />
1987), resulting in preferential scavenging <strong>of</strong> the light rare earth<br />
elements (LREE) in the marine environment (Byrne <strong>and</strong> Kim, 1990).<br />
This produces the heavy rare earth element (HREE) enrichment characteristic<br />
<strong>of</strong> seawater (Fig. 6). Furthermore, pure marine precipitates<br />
such as modern corals (Sholkovitz <strong>and</strong> Shen, 1995), as well as<br />
Holocene microbialitic limestones (Webb <strong>and</strong> Kamber, 2000) accurately<br />
record REY distributions in coeval seawater. We therefore<br />
conclude that IF samples possessing seawater-like REY distributions<br />
represent the best archives <strong>of</strong> coeval Precambrian seawater 143 Nd/<br />
144 Nd, <strong>and</strong> using such an approach to screen IF samples facilitates<br />
comparisons between studies.<br />
The data available to date suggest that the overall shape <strong>of</strong> the REY<br />
pattern for seawater has not significantly varied for the past 3.8 Ga,<br />
based upon similar REY patterns observed in the Isua IF-G (see also<br />
Bolhar et al., 2004), as well as Archean–Paleoproterozoic marine<br />
carbonates (Kamber <strong>and</strong> Webb, 2001; Bau <strong>and</strong> Alex<strong>and</strong>er, 2006). The<br />
most significant differences between the REY distribution in modern<br />
seawater <strong>and</strong> Archean seawater are for cerium (Ce) <strong>and</strong> Eu, the only
Author's personal copy<br />
152 B.W. Alex<strong>and</strong>er et al. / Earth <strong>and</strong> Planetary <strong>Science</strong> Letters 283 (2009) 144–155<br />
REY that can have valence states other than +3. The oxic nature <strong>of</strong><br />
modern oceans allows Ce to be readily oxidized at particle surfaces to<br />
highly insoluble Ce(IV), preferentially removing Ce <strong>and</strong> producing the<br />
very negative Ce anomaly observed in modern seawater. The fact that<br />
negative Ce anomalies have not been observed in marine chemical<br />
sediments (IF or carbonates) older than ~2.3 Ga indicates relatively<br />
reducing oceans, <strong>and</strong> an atmosphere–hydrosphere redox state for the<br />
early Earth that was lower than today (cf. Holl<strong>and</strong>, 1984).<br />
Whereas Ce fractionation <strong>and</strong> removal occurs in the water column<br />
<strong>and</strong> at the sediment–water interface, Eu fractionation occurs during<br />
water–rock interaction at high temperatures via reduction <strong>of</strong> Eu 3+ to<br />
Eu 2+ . The alteration products <strong>of</strong> these water–rock reactions discriminate<br />
against the relatively large Eu 2+ ion <strong>and</strong> produce a subsequent<br />
enrichment <strong>of</strong> Eu in the fluid phase. This Eu enrichment is present<br />
only in fluids which have generally exceeded ~200 °C, <strong>and</strong> has been<br />
observed in numerous vent fluids from hydrothermal systems altering<br />
oceanic crust (e.g., Michard et al., 1983; Schmidt et al., 2007, <strong>and</strong><br />
references therein). The large positive Eu anomalies in Archean–<br />
Paleoproterozoic IFs <strong>and</strong> carbonates possessing seawater-like REY<br />
distributions are fully consistent with significant high-T hydrothermal<br />
fluid fluxes to Earth's early oceans. Unfortunately, Eu anomalies are<br />
not suitable for mass fraction calculations regarding the mixing <strong>of</strong><br />
high-T hydrothermal fluids <strong>and</strong> ambient seawater, <strong>and</strong> only reveal<br />
information concerning the temperature <strong>of</strong> the REY source (e.g., Bau<br />
<strong>and</strong> Möller, 1993; Alex<strong>and</strong>er et al., 2008).<br />
6.3. Nd isotope ratios in Archean seawater<br />
Generally, IFs older than 2.25 Ga that possess seawater-like REY<br />
patterns have Є Nd (t) between −5 <strong>and</strong> +5 (Fig. 7), with much <strong>of</strong> the<br />
data describing the ~2.5 Ga Hamersley IFs (Australia). The remaining<br />
data are primarily from South Africa, though Frei et al. (2007) have<br />
significantly exp<strong>and</strong>ed the Sm–Nd isotopic analyses <strong>of</strong> Archean IF<br />
from North America; unfortunately, possible ages for these IF are<br />
poorly constrained. Most Є Nd (t) values for IFs older than 2.25 Ga fall<br />
between −1.5 <strong>and</strong> +2.5, similar to that observed in the PGB samples.<br />
Data for IFs older than 3.0 Ga are dominated by studies <strong>of</strong> the<br />
relatively few samples from Isua discussed previously. Considering the<br />
wide range <strong>of</strong> Є Nd (t) values observed in the Isua IF <strong>and</strong> the few analyses<br />
for samples clearly possessing seawater-like REY distributions, we<br />
consider the IF-G Є Nd (3.8 Ga) value <strong>of</strong> approximately +2.5 as the best<br />
estimate for average ~3.8 Ga seawater, similar to the conclusions <strong>of</strong> Frei<br />
<strong>and</strong> Polat (2007). The Pietersburg data are similar to older IF, though<br />
tending to more CHUR-like values, <strong>and</strong> support the interpretation that<br />
Nd was significantly, if not dominantly, provided by ocean ridge<br />
hydrothermal systems possessing a combination <strong>of</strong> radiogenic Nd<br />
signatures <strong>and</strong> strong positive Eu anomalies. This trend seems consistent<br />
until ~2.6 Ga, when negative Є Nd (t) values <strong>and</strong> smaller Eu<br />
anomalies (Bau <strong>and</strong> Möller, 1993) become common in IFs, though in<br />
only a few instances do Archean–Paleoproterozoic IFs display Є Nd (t)<br />
lower than −1.5.<br />
Focusing on the period near 3.0 Ga, the Pietersburg IF Nd isotopic<br />
evolution (Fig. 8) is very similar to that observed for the 2.9 Ga<br />
Mozaan IFs from South Africa (Alex<strong>and</strong>er et al., 2008). The consistent<br />
Nd isotopic evolution observed in the two groups <strong>of</strong> IF is expected due<br />
to the similar Sm/Nd values observed in IFs with seawater-like REY<br />
distributions. The reasonable range <strong>of</strong> potential depositional ages for<br />
both the Pietersburg <strong>and</strong> the Mozaan IFs would be from ~2.84 to<br />
3.1 Ga, <strong>and</strong> within this time frame the Nd isotopic ratios remain<br />
distinctly different. As initial 143 Nd/ 144 Nd ratios in the Pietersburg IF<br />
are negatively correlated with increasing aluminosilicate content<br />
(Fig. 4), associated crustal rocks displayed Є Nd (2.95 Ga) <strong>of</strong> −0.5,<br />
whereas ambient seawater during deposition <strong>of</strong> these IFs displayed<br />
Є Nd (2.95 Ga) <strong>of</strong> approximately +1. This value is ~3–5 Є-units higher<br />
than seawater contemporaneous with deposition <strong>of</strong> the Mozaan<br />
samples IF (Alex<strong>and</strong>er et al., 2008). These different Є Nd (t) values<br />
indicated for seawater likely reflect different depositional environments,<br />
as the previously discussed trace element data suggests that<br />
the rocks associated with the Pietersburg IF were generated in an<br />
Fig. 7. Neodymium isotopic signatures in oxide <strong>and</strong> silicate facies IFs older than 2.4 Ga. World IFs generally possess Є Nd (t) between −5 <strong>and</strong> +5, though most <strong>of</strong> the data are more<br />
positive than −1.5. The dashed lines represent Є Nd (t) values predicted for simple Nd isotopic evolution in depleted mantle <strong>and</strong> continental crust with initial Є Nd (4.56 Ga)=0, <strong>and</strong><br />
possessing Є Nd (0) =+10 <strong>and</strong> −17, respectively. Also shown is the Nd evolution in a depleted mantle as modeled by Nägler <strong>and</strong> Kramers (1998). Data are screened to distinguish<br />
samples which display seawater-like REY patterns similar to those shown in Fig. 6, <strong>and</strong> are from Miller <strong>and</strong> O'Nions (1985), <strong>Jacobs</strong>en <strong>and</strong> Pimentel-Klose (1988a, 1988b), 12 samples<br />
considered pristine enough by Alibert <strong>and</strong> McCulloch (1993) for calculation <strong>of</strong> possible Archean seawater pH values, <strong>and</strong> Bau et al. (1997a). All IF samples <strong>of</strong> 3.7–3.8 Ga age are from<br />
Isua (Greenl<strong>and</strong>), with stratigraphic ages as reported in original datasets. All data have been calculated using a present day 143 Nd/ 144 Nd <strong>of</strong> 0.512638 in CHUR <strong>and</strong> normalized to<br />
146 Nd/ 144 Nd=0.7219. The majority <strong>of</strong> the data fall between −1.5 <strong>and</strong> +2.5, <strong>and</strong> the Є Nd (t) <strong>of</strong> Archean–Paleoproterozoic seawater as suggested by IFs was generally between +1 <strong>and</strong><br />
+2 until ~2.7 Ga. Only the 2.9 Ga Mozaan IFs possess distinctly different Є Nd (t) values, which are similar to igneous rocks (rhyolite, <strong>and</strong>esite, basalt, <strong>and</strong> gabbro) from the South<br />
Africa–Swazil<strong>and</strong> border area (Hegner et al., 1984, 1994).
Author's personal copy<br />
B.W. Alex<strong>and</strong>er et al. / Earth <strong>and</strong> Planetary <strong>Science</strong> Letters 283 (2009) 144–155<br />
153<br />
Fig. 8. Nd isotopic evolution lines for 2.9–3.0 Ga South African IFs. Mozaan IF data does<br />
not include the two samples shown in Fig. 7 with Є Nd (2.9 Ga) lower than −8.0. The<br />
left vertical axis represents the minimum age for the Mozaan Group <strong>of</strong> 2837±5 Ma<br />
(Gutzmer et al., 1999); the best estimate <strong>of</strong> the maximum age for the Mozaan IFs is<br />
2977±5 Ma (Nhleko, 2003). The lower age limit for the Pietersburg IFs is indicated as<br />
2958±2 Ma (De Wit et al., 1993). The geologic evidence <strong>and</strong> the limited radiometric<br />
age data do not preclude the possibility that deposition <strong>of</strong> the Pietersburg <strong>and</strong> Mozaan<br />
IFs overlapped in time. The isotopic evolution lines are generally subparallel for all IF<br />
samples, reflecting the similar 147 Sm/ 144 Nd ratios in these samples. The Pietersburg IFs<br />
are consistently more radiogenic than the Mozaan IFs, <strong>and</strong> are interpreted to represent<br />
Archean seawater masses with higher 143 Nd/ 144 Nd than the seawater that precipitated<br />
the Mozaan IFs.<br />
environment similar to a modern mafic back-arc setting, whereas<br />
the Mozaan IF were clearly deposited in a marine shelf setting along<br />
a stable cratonic margin (Beukes <strong>and</strong> Cairncross, 1991). The negative<br />
Є Nd (2.9 Ga) values observed in the Mozaan IF are not surprising<br />
considering the occurrence <strong>of</strong> ca. 2.9 Ga igneous rocks possessing<br />
negative Є Nd (t) near the South Africa–Swazil<strong>and</strong> border (Hegner<br />
et al., 1984, 1994). These felsic <strong>and</strong> mafic rocks (e.g., rhyolite, <strong>and</strong>esite,<br />
basalt, <strong>and</strong> gabbro) possess Є Nd (t) very similar to the Mozaan<br />
IFs (Fig. 7), <strong>and</strong> are consistent with published Mozaan Group<br />
shale Є Nd (t) values(Stevenson <strong>and</strong> Patchett, 1990; Alex<strong>and</strong>er et al.,<br />
2008).<br />
6.4. Implications <strong>of</strong> Nd isotopic variations within the Archean ocean<br />
The fact that many Archean IFs with seawater-like REY distributions<br />
possess positive Є Nd (t) implies a significant hydrothermal flux<br />
from mantle-derived source rocks to seawater until at least 2.5 Ga.<br />
Certainly more studies <strong>of</strong> Mesoarchean IF (3.0–3.5 Ga) are necessary,<br />
but existing Nd isotopic data are consistent with the increased<br />
mantle-derived hydrothermal flux implied from Sr isotopic studies <strong>of</strong><br />
Archean carbonates (Veizer <strong>and</strong> Mackenzie, 2004). With the exception<br />
<strong>of</strong> the Mozaan IF, the first clear evidence <strong>of</strong> seawater displaying<br />
negative Є Nd (t) is found in N. American IFs possessing Є Nd (2.65 Ga) <strong>of</strong><br />
−1.5 (Frei et al., 2007), which is considerably more radiogenic than<br />
the older Mozaan IF. However, it is possible that negative (<strong>and</strong> highly<br />
variable) Є Nd (t) values were more prevalent in ca. 2.9 Ga seawater<br />
than is suggested by Fig. 7, as it does not include data from Frei et al.<br />
(2007) for oxide-facies IF from the N. American Nemo iron-formation.<br />
This is because the Nemo IF is poorly constrained in age (2.56–2.9 Ga),<br />
<strong>and</strong> therefore could display Є Nd (2.9 Ga) between −2.76 <strong>and</strong> +3.71, or<br />
Є Nd (2.56 Ga) between −5.24 <strong>and</strong> 0.78.<br />
The clearly distinct Є Nd (t) values between the Pietersburg <strong>and</strong><br />
Mozaan IF are considered to arise from their different depositional<br />
environments, as the Mozaan IF are the oldest IF deposited on a stable<br />
cratonic margin, whereas older IF like the Pietersburg are associated<br />
with greenstone belts. If ca. 3.0 Ga seawater was homogenous with<br />
respect to 143 Nd/ 144 Nd, then world oceans would have become<br />
dramatically less radiogenic in the time span between deposition <strong>of</strong><br />
the Pietersburg <strong>and</strong> Mozaan IF. We consider this unlikely, as it would<br />
further require a reversion <strong>of</strong> world seawater Є Nd (t) back to chondritic<br />
or positive values for the remainder <strong>of</strong> the Archean. We therefore<br />
interpret the data as indicating that Archean seawater could be<br />
inhomogeneous with respect to 143 Nd/ 144 Nd ratios, primarily dependent<br />
upon the nature <strong>of</strong> local crust exposed for weathering, a situation<br />
that is similar to that observed in modern oceans.<br />
The degree <strong>of</strong> Nd isotopic inhomogeneity in Archean seawater is<br />
unclear; it may have been that only small parts <strong>of</strong> the world ocean<br />
deviated significantly from an ‘average’ 143 Nd/ 144 Nd ratio. Excluding<br />
the Mozaan IF, seawater appears to have displayed relatively constant<br />
Є Nd (t) between 0 <strong>and</strong> +2.5 from 2.7 to 3.8 Ga. This is remarkable, <strong>and</strong><br />
though detailed discussions <strong>of</strong> mantle evolution <strong>and</strong> the growth <strong>of</strong><br />
continental crust are beyond the scope <strong>of</strong> this study, crustal sources <strong>of</strong><br />
Nd were certainly evolving isotopically during this time period<br />
(Fig. 7). However, regardless <strong>of</strong> the model employed, there is little<br />
evidence that isotopic evolution within Nd source(s) to world oceans<br />
significantly altered the 143 Nd/ 144 Nd ratio <strong>of</strong> seawater throughout<br />
most <strong>of</strong> the Archean. This implies that prior to 2.7 Ga, the great bulk <strong>of</strong><br />
seawater at any time t was relatively homogeneous with respect to<br />
143 Nd/ 144 Nd, <strong>and</strong> water masses that deviated from a bulk seawater<br />
average Є Nd (t) <strong>of</strong> +1 to +2 (i.e., those that formed the Mozaan IF),<br />
were minor components <strong>of</strong> the world ocean.<br />
An isotopically homogenous world ocean is best explained by<br />
widespread anoxic conditions in Archean seawater (as suggested by<br />
the presence <strong>of</strong> IF themselves) that would increase the residence time<br />
<strong>of</strong> Nd in seawater. Evidence for deep water anoxia is provided by U <strong>and</strong><br />
Th behavior in altered Archean basalts (Nakamura <strong>and</strong> Kato, 2007),<br />
<strong>and</strong> this environment would prohibit the formation <strong>of</strong> important REY<br />
sinks, increasing the residence time <strong>of</strong> Nd in Archean bottom seawater.<br />
Shallow (coastal) seawater that could precipitate oxide-facies IF was<br />
most likely characterized by the same short marine residence time <strong>of</strong><br />
Nd as is observed in modern seawater, <strong>and</strong> was dominated by less<br />
radiogenic ‘local’ continental Nd. However, the relatively constant<br />
positive Є Nd (t) values observed in world IFs between 2.7 <strong>and</strong> 3.8 Ga in<br />
age that possess seawater-like REY distributions suggests hydrothermal<br />
alteration <strong>of</strong> oceanic crust was dominating the Nd (<strong>and</strong> hence Fe?)<br />
flux to bulk seawater throughout most <strong>of</strong> the Archean.<br />
7. Conclusions<br />
Prior to 2.7 Ga, iron-formations possessing seawater-like REY<br />
distributions are the best archives for reconstructing paleoseawater<br />
143 Nd/ 144 Nd ratios. Ambient seawater at the time <strong>of</strong> deposition <strong>of</strong> the<br />
Pietersburg IF possessed positive Є Nd (2.95 Ga) <strong>of</strong> at least +1, which is<br />
slightly lower than the best estimate for seawater Є Nd (3.8 Ga) as<br />
suggested by Isua IF. Local clastic aluminosilicate material, presumably<br />
derived from an Archean equivalent <strong>of</strong> a modern mafic back-arc<br />
setting, possessed lower Є Nd (t) <strong>of</strong> approximately −0.5, <strong>and</strong> relatively<br />
small contributions <strong>of</strong> this material (≥0.5% Al 2 O 3 ) were sufficient to<br />
dominate the Nd budget in the Pietersburg IF. Crustal sources <strong>of</strong> Nd to<br />
seawater near the Kaapvaal craton 2.9–3.0 Ga possessed distinctly<br />
different Є Nd (t), <strong>and</strong> could locally influence seawater 143 Nd/ 144 Nd<br />
ratios, as shown by the Mozaan IF, for example.<br />
The relatively consistent Nd isotopic ratios observed throughout<br />
most <strong>of</strong> the Archean for IFs with seawater-like REY distributions<br />
suggests that, on average, bulk Archean seawater displayed positive<br />
Є Nd (t) between +1 <strong>and</strong> +2 until at least 2.7 Ga. However, the range in<br />
Є Nd (t) values observed between the similarly aged Pietersburg <strong>and</strong><br />
Mozaan IFs implies that world oceans were not entirely well-mixed<br />
with respect to 143 Nd/ 144 Nd ratios, but that this variability was<br />
restricted to local water masses, <strong>and</strong> was likely the consequence <strong>of</strong>
Author's personal copy<br />
154 B.W. Alex<strong>and</strong>er et al. / Earth <strong>and</strong> Planetary <strong>Science</strong> Letters 283 (2009) 144–155<br />
different depositional environments. In conjunction with the ubiquitous<br />
positive Eu anomalies observed in IFs, the Nd isotopic data<br />
indicate high-temperature hydrothermal alteration <strong>of</strong> mantle-derived<br />
oceanic crust was a dominant control on bulk seawater composition<br />
for much <strong>of</strong> the Archean.<br />
Acknowledgements<br />
This research was funded by Exchange Grant 1327 within the<br />
European <strong>Science</strong> Foundation ArchEnviron Research Networking<br />
Programme. Mark Le Grange is thanked for his assistance with<br />
sampling. The authors would like to gratefully acknowledge the<br />
contributions <strong>of</strong> M. Fischerström <strong>and</strong> H. Schöberg regarding the Sm–<br />
Nd isotopic analyses performed at LIG. This manuscript significantly<br />
benefited from the comments <strong>of</strong> M. Gutjahr <strong>and</strong> one anonymous<br />
reviewer, as well as suggestions by M.L. Delaney.<br />
References<br />
Abbey, S., McLeod, C.R., Liang-Guo, W., 1983. FeR-1, FeR-2, FeR-3, <strong>and</strong> FeR-4 four<br />
Canadian iron-formation samples prepared for use as reference materials. Geol. Sur.<br />
Can. paper 83–119.<br />
Abbott, D.H., H<strong>of</strong>fman, S.E., 1984. Archaean plate tectonics revisited 1. Heat flow,<br />
spreading rate, <strong>and</strong> the age <strong>of</strong> subducting oceanic lithosphere <strong>and</strong> their effects on<br />
the origin <strong>and</strong> evolution <strong>of</strong> continents. Tectonics 3, 429–448.<br />
Alex<strong>and</strong>er, B.W., Bau, M., Andersson, P., Dulski, P., 2008. Continentally-derived solutes in<br />
shallow Archean seawater: rare earth element <strong>and</strong> Nd isotope evidence in iron<br />
formation from the 2.9 Ga Pongola Supergroup, South Africa. Geochim. Cosmochim.<br />
Acta 72, 378–394.<br />
Alibert, C., McCulloch, M.T., 1993. Rare earth element <strong>and</strong> Nd isotopic compositions <strong>of</strong><br />
the b<strong>and</strong>ed iron-formations <strong>and</strong> associated shales from Hamersley, Western<br />
Australia. Geochim. Cosmochim. Acta 57, 187–204.<br />
Alibo, D.S., Nozaki, Y., 1999. Rare earth elements in seawater: particle association, shale<br />
normalization, <strong>and</strong> Ce-oxidation. Geochim. Cosmochim. Acta 63, 363–372.<br />
Anders, E., Grevesse, N., 1989. Abundances <strong>of</strong> the elements: meteoric <strong>and</strong> solar.<br />
Geochim. Cosmochim. Acta 53, 197–214.<br />
Andersson, P.S., Porcelli, D., Frank, M., Björk, G., Dahlqvist, R., Gustafsson, Ö., 2008.<br />
Neodymium isotopes in seawater from the Barents Sea <strong>and</strong> Fram Strait Arctic–<br />
Atlantic gateways. Geochim. Cosmochim. Acta 72, 2854–2867.<br />
Appel, P.W.U., 1983. Rare earth elements in the early Archaean Isua iron-formation,<br />
West Greenl<strong>and</strong>. Precambrian. Res. 20, 243–258.<br />
Bau, M., 1993. Effects <strong>of</strong> syn- <strong>and</strong> post-depositional processes on the rare-earth element<br />
distribution in Precambrian iron-formations. Eur. J. Mineral. 5, 257–267.<br />
Bau, M., Dulski, P., 1992. Small-scale variations <strong>of</strong> the rare-earth element distribution in<br />
Precambrian iron formations. Eur. J. Mineral. 4, 1429–1433.<br />
Bau, M., Möller, P., 1993. Rare earth element systematics <strong>of</strong> the chemically precipitated<br />
component in Early Precambrian iron formations <strong>and</strong> the evolution <strong>of</strong> the<br />
terrestrial atmosphere–hydrosphere–lithosphere system. Geochim. Cosmochim.<br />
Acta 57, 2239–2249.<br />
Bau, M., Dulski, P., 1996. Distribution <strong>of</strong> yttrium <strong>and</strong> rare-earth elements in the Penge<br />
<strong>and</strong> Kuruman iron-formations, Transvaal Supergroup, South Africa. Precamb. Res.<br />
79, 37–55.<br />
Bau, M., Alex<strong>and</strong>er, B., 2006. Preservation <strong>of</strong> primary REE patterns without Ce anomaly<br />
during dolomitization <strong>of</strong> Mid-Paleoproterozoic limestone <strong>and</strong> the potential reestablishment<br />
<strong>of</strong> marine anoxia immediately after the “Great Oxidation Event”.<br />
S. Afr. J. Geol. 109, 81–86.<br />
Bau, M., Höhndorf, A., Dulski, P., Beukes, N.J., 1997a. Sources <strong>of</strong> rare-earth elements <strong>and</strong><br />
iron in Paleoproterozoic iron-formations from the Transvaal Supergroup, South<br />
Africa: evidence from neodymium isotopes. J. Geol. 105, 121–129.<br />
Bau, M., Möller, P., Dulski, P., 1997b. Yttrium <strong>and</strong> lanthanides in eastern Mediterranean<br />
seawater <strong>and</strong> their fractionation during redox-cycling. Mar. Chem. 56, 123–131.<br />
Beukes, N.J., 1983. Palaeoenvironmental setting <strong>of</strong> iron-formations in the depositional<br />
basin <strong>of</strong> the Transvaal Supergroup, South Africa, in: Trendall, A.F., Morris, R.C. (Eds.),<br />
Iron-formation Facts <strong>and</strong> Problems, 131–198, Vol. 6 Developments in Precambrian<br />
Geology (Ed. B.F. Windley), Elsevier, Amsterdam.<br />
Beukes, N.J., Cairncross, B., 1991. A lithostratigraphic sedimentological reference pr<strong>of</strong>ile<br />
for the late Archaean Mozaan Group, Pongola Sequence: application to sequence<br />
stratigraphy <strong>and</strong> correlation with the Witwatersr<strong>and</strong> Supergroup. S. Afr. J. Geol. 94,<br />
44–69.<br />
Bolhar, R., Kamber, B.S., Moorbath, S., Fedo, C.M., Whitehouse, M.J., 2004. Characterisation<br />
<strong>of</strong> early Archaean chemical sediments by trace element signatures. Earth<br />
Planet. Sci. Lett. 222, 43–60.<br />
Broecker, W.S., Peng, T.H., 1982. Tracers in the Sea. Eldigio Press, New York.<br />
Byrne, R.H., Kim, K., 1990. Rare earth element scavenging in seawater. Geochim.<br />
Cosmochim. Acta 54, 2645–2656.<br />
Byron, C.L., Barton, J.M., 1990. The setting <strong>of</strong> mineralization in a portion <strong>of</strong> the Eersteling<br />
goldfield, Pietersburg granite–greenstone terrane, South Africa. S. Afr. J. Geol. 93,<br />
463–472.<br />
Cantrell, K.J., Byrne, R.H., 1987. Rare earth element complexation by carbonate <strong>and</strong><br />
oxalate ions. Geochim. Cosmochim. Acta 51, 597–605.<br />
Condie, K.C., 1993. Chemical composition <strong>and</strong> evolution <strong>of</strong> the upper continental crust:<br />
contrasting results from surface samples <strong>and</strong> shales. Chem. Geol. 104, 1–37.<br />
Danielson, A., Möller, P., Dulski, P., 1992. The europium anomalies in b<strong>and</strong>ed iron<br />
formations <strong>and</strong> the thermal history <strong>of</strong> the oceanic crust. Chem. Geol. 97, 89–100.<br />
DePaolo, D.J., Wasserburg, G.J., 1976. Nd isotopic variations <strong>and</strong> petrogenetic models.<br />
Geophys. Res. Lett. 3, 249–252.<br />
Derry, L.A., <strong>Jacobs</strong>en, S.B., 1990. The chemical evolution <strong>of</strong> Precambrian seawater:<br />
Evidence from rare earth elements in b<strong>and</strong>ed iron formations. Geochim.<br />
Cosmochim. Acta 54, 2965–2977.<br />
De Wit, M.J., 1991. Archaean greenstone belt tectonism <strong>and</strong> basin development: some<br />
insights from the Barberton <strong>and</strong> Pietersburg greenstone belts, Kaapvaal Craton,<br />
South Africa. J. Afr. Earth Sci. 13, 45–63.<br />
De Wit, M.J., Jones, M.G., Buchanan, D.L., 1992. The geology <strong>and</strong> tectonic evolution <strong>of</strong> the<br />
Pietersburg Greenstone Belt, South Africa. Precambrian Res. 55, 123–153.<br />
De Wit, M.J., Armstrong, F.L., Kamo, S.L., Erlank, A.J., 1993. Gold-bearing sediments in the<br />
Pietersburg greenstone belt: age equivalents <strong>of</strong> the Witwatewrsr<strong>and</strong> Supergroup<br />
sediments. S. Afr. Econ. Geol. 88, 1242–1252.<br />
Dulski, P., 2001. Reference materials for geochemical studies: new analytical data by<br />
ICP-MS <strong>and</strong> critical discussion <strong>of</strong> reference values. Geost<strong>and</strong>. Newsl. 25, 87–125.<br />
Eglington, B.M., Talma, A.S., Marais, S., Matthews, P.E., Dixon, J.G.P., 2003. Isotopic<br />
composition <strong>of</strong> Pongola Supergroup limestones from the Buffalo River gorge, South<br />
Africa: constraints on their regional depositional setting. S. African J. Geol.106,1–10.<br />
Frank, M., 2002. Radiogenic isotopes: tracers <strong>of</strong> past ocean circulation <strong>and</strong> erosional<br />
input. Rev. Geophys. 40, 1–38.<br />
Frei, R., Polat, A., 2007. Source heterogeneity for the major components <strong>of</strong> ~3.7 Ga<br />
B<strong>and</strong>ed Iron Formations (Isua Greenstone Belt, Western Greenl<strong>and</strong>): tracing the<br />
nature <strong>of</strong> interacting water masses in IF formation. Earth Planet. Sci. Lett. 253,<br />
266–281.<br />
Frei, R., Bridgewater, D., Rosing, M., Stecher, O., 1999. Controversial Pb–Pb <strong>and</strong> Sm–Nd<br />
isotope results in the early Archean Isua (West Greenl<strong>and</strong>) oxide iron formation:<br />
preservation <strong>of</strong> primary signatures versus secondary disturbances. Geochim.<br />
Cosmochim. Acta 63, 473–488.<br />
Frei, R., Dahl, P.S., Duke, E.F., Frei, K.M., Hansen, T.R., Fr<strong>and</strong>sson, M.M., Jensen, L.A., 2007.<br />
Trace element <strong>and</strong> isotopic characterization <strong>of</strong> Neoarchean <strong>and</strong> Paleoproterozoic<br />
iron formations in the Black Hills (South Dakota, USA): assessment <strong>of</strong> chemical<br />
change during 2.9–1.9 Ga deposition bracketing the 2.4–2.2 Ga first rise <strong>of</strong><br />
atmospheric oxygen. Precam. Res. 162, 441–474.<br />
Fryer, B.J., Fyfe, W.S., Kerrich, R., 1979. Archaean volcanogenic oceans. Chem. Geol. 24,<br />
25–33.<br />
German, C.R., Klinkhammer, G.P., Edmond, J.M., Elderfield, H., Mitra, A., 1990.<br />
Hydrothermal scavenging <strong>of</strong> rare-earth elements in the ocean. Nature 345, 516–518.<br />
Goldstein, S.L., Hemming, S.R., 2003. In: Elderfield, H. (Ed.), Long-lived Isotopic Tracers<br />
in Oceanography, Paleoceanography <strong>and</strong> Ice Sheet Dynamics. Treatise on<br />
Geochemistry, Vol. 6. Elsevier-Pergamon, Oxford, pp. 453–489.<br />
Grotzinger, J.P., 1989. Facies <strong>and</strong> evolution <strong>of</strong> Precambrian carbonate depositional<br />
systems: emergence <strong>of</strong> the modern platform archetype. Controls on Carbonate<br />
Platform <strong>and</strong> Basin Development, SEPM (Soc. Econ. Paleont. Mineral). Spec. Pub.,<br />
vol. 44, pp. 79–106.<br />
Gutjahr, M., Frank, M., Stirling, C.H., Keigwin, L.D., Halliday, A.N., 2008. Tracing the Nd<br />
isotope evolution <strong>of</strong> North Atlantic Deep <strong>and</strong> Intermediate Waters in the western<br />
North Atlantic since the Last Glacial Maximum from Blake Ridge sediments. Earth<br />
Planet. Sci. Lett. 266, 61–77.<br />
Gutzmer, J., Nhleko, N., Beukes, N.J., Pickard, A., Barley, M.E., 1999. Geochemistry <strong>and</strong> ion<br />
microprobe (SHRIMP) age <strong>of</strong> a quartz porphyry sill in the Mozaan Group <strong>of</strong> the<br />
Pongola Supergoup: implications for the Pongola <strong>and</strong> Witwatersr<strong>and</strong> Supergroups.<br />
S. Afr. J. Geol. 102, 139–146.<br />
Hegner, E., Kröner, A., H<strong>of</strong>mann, A.W., 1984. Age <strong>and</strong> isotope geochemistry <strong>of</strong> the<br />
Archaean Pongola <strong>and</strong> Usushwana suites in Swazil<strong>and</strong>, southern Africa: a case for<br />
crustal contamination <strong>of</strong> mantle-derived magma. Earth Planet. Sci. Lett. 70,<br />
267–279.<br />
Hegner, E., Kröner, A., H<strong>of</strong>mann, A.W., 1994. A precise U–Pb zircon age for the Archaean<br />
Pongola Supergroup volcanics in Swazil<strong>and</strong>. J. Afr. Earth Sci. 18, 339–341.<br />
Holl<strong>and</strong>, H.H., 1984. The Chemical Evolution <strong>of</strong> the Atmosphere <strong>and</strong> the Oceans.<br />
Princeton Univ. Press.<br />
<strong>Jacobs</strong>en, S.B., Pimentel-Klose, M.R., 1988a. A Nd isotopic study <strong>of</strong> the Hamersley <strong>and</strong><br />
Michipicoten b<strong>and</strong>ed iron formations: the source <strong>of</strong> REE <strong>and</strong> Fe in Archean oceans.<br />
Earth Planet. Sci. Lett. 87, 29–44.<br />
<strong>Jacobs</strong>en, S.B., Pimentel-Klose, M.R., 1988b. Nd isotopic variations in Precambrian<br />
b<strong>and</strong>ed iron formations. Geophys. Res. Lett. 15, 393–396.<br />
Je<strong>and</strong>el, C., Bishop, K., Zindler, A., 1995. Exchange <strong>of</strong> neodymium <strong>and</strong> its isotopes<br />
between seawater <strong>and</strong> small <strong>and</strong> large particles in the Sargasso Sea. Geochim.<br />
Cosmochim. Acta 59, 535–547.<br />
Jochum, K.P., Nohl, U., Herwig, K., Lammel, E., Stoll, B., H<strong>of</strong>mann, A.W., 2005. GeoReM: A<br />
new geochemical database for reference materials <strong>and</strong> isotopic st<strong>and</strong>ards.<br />
Geost<strong>and</strong>. Geoanal. Res. 29, 333–338.<br />
Jones, M.G., 1990. The geology <strong>of</strong> the Mt Mare area, Pietersburg greenstone belt, South<br />
Africa. Ph.D. thesis, Imperial College, Univ. London.<br />
Kamber, B.S., Webb, G.E., 2001. The geochemistry <strong>of</strong> late Archaean microbial carbonate:<br />
Implications for ocean chemistry <strong>and</strong> continental erosion history. Geochim.<br />
Cosmochim. Acta 65, 2509–2525.<br />
Klein, C., Beukes, N.J., 1989. Geochemistry <strong>and</strong> sedimentology <strong>of</strong> a facies transition from<br />
limestone to iron-formation in the Early Proterozoic Transvaal Supergroup, South<br />
Africa. Econ. Geol. 84, 1733–1774.<br />
Kröner, A., Jaeckl, P., Br<strong>and</strong>l, G., 2000. Single zircon ages for felsic to intermediate rocks<br />
from the Pietersburg <strong>and</strong> Giyani greenstone belts <strong>and</strong> bordering granitoid<br />
orthogneisses, northern Kaapvaal Craton, South Africa. J. Afr. Earth Sci. 30, 773–793.
Author's personal copy<br />
B.W. Alex<strong>and</strong>er et al. / Earth <strong>and</strong> Planetary <strong>Science</strong> Letters 283 (2009) 144–155<br />
155<br />
McKenzie, D., Weiss, N., 1975. Speculations on the thermal <strong>and</strong> tectonic history <strong>of</strong> the<br />
Earth,. Geophys. J. R. Astr. Soc. 42, 131–174.<br />
McLennan, S.M., 1989. Rare earth elements in sedimentary rocks: influence <strong>of</strong><br />
provenance <strong>and</strong> sedimentary processes. In: Lipin, B.R., McKay, G.A. (Eds.),<br />
Geochemistry <strong>and</strong> Mineralogy <strong>of</strong> Rare Earth Elements. Rev. Mineral. 21, Mineral.<br />
Soc. Amer., pp. 169–200.<br />
Michard, A., Albarède, F., Michard, G., Minster, J.F., Charlou, J.L., 1983. Rare-earth<br />
elements <strong>and</strong> uranium in high-temperature solutions from East Pacific Rise<br />
hydrothermal vent field (13°N). Nature 303, 795–797.<br />
Miller, R.G., O'Nions, R.K., 1985. Sources <strong>of</strong> Precambrian chemical <strong>and</strong> clastic sediments.<br />
Nature 314, 325–330.<br />
Nägler, T.F., Kramers, J.D., 1998. Nd isotopic evolution <strong>of</strong> the upper mantle during the<br />
Precambrian: models, data <strong>and</strong> the uncertainty <strong>of</strong> both. Precambrian Res. 91,<br />
233–252.<br />
Nakamura, K., Kato, Y., 2007. A new geochemical approach for constraining a marine<br />
redox condition <strong>of</strong> Early Archean. Earth Planet. Sci. Lett. 261, 296–302.<br />
Nhleko, N., 2003. The Pongola Supergroup in Swazil<strong>and</strong>, Ph.D. thesis, Univ.<br />
Johannesburg.<br />
Nothdurft, L.D., Webb, G.E., Kamber, B.S., 2004. Rare earth element geochemistry <strong>of</strong> Late<br />
Devonian reefal carbonates, Canning Basin, Western Australia: confirmation <strong>of</strong> a<br />
seawater REE proxy in ancient limestones. Geochim. Cosmochim. Acta 68, 263–283.<br />
Nozaki, Y., Alibo, D.S., 2003. Importance <strong>of</strong> vertical geochemical processes in controlling<br />
the oceanic pr<strong>of</strong>iles <strong>of</strong> dissolved rare earth elements in the northeastern Indian<br />
Ocean. Earth Planet. Sci. Lett. 205, 155–172.<br />
Nozaki, Y., Alibo, D.S., Amakawa, H., Gamo, T., Hasumoto, H., 1999. Dissolved rare earth<br />
elements <strong>and</strong> hydrography in the Sulu Sea. Geochim. Cosmochim. Acta 63,<br />
2171–2181.<br />
Piepgras, D.J., Wasserburg, G.J., 1980. Neodymium isotopic variations in seawater. Earth<br />
Planet. Sci. Lett. 50, 128–138.<br />
Piotrowski, A.M., Goldstein, S.L., Hemming, S.R., Fairbanks, R.G., 2005. Temporal relationships<br />
<strong>of</strong> carbon cycling <strong>and</strong> ocean circulation at glacial boundaries. <strong>Science</strong> 307,<br />
1933–1938.<br />
Saager, R., Meyer, M., 1982. Gold distribution in supracrustal rocks from the Archean<br />
greenstone belts <strong>of</strong> Southern Africa <strong>and</strong> from Paleozoic ultramafic complexes <strong>of</strong> the<br />
European Alps: metallogenic <strong>and</strong> geochemical implications. Econ. Geol. 77, 1–24.<br />
SACS - S. Afr. Comm. Strat.,1980. Part 1: lithostratigraphy <strong>of</strong> the Republic <strong>of</strong> South Africa,<br />
South West Africa/Namibia <strong>and</strong> the republics <strong>of</strong> Bophuthatswana, Transkei <strong>and</strong><br />
Venda. In: Kent, L.E. (Ed.), Stratigraphy S. Afr. 8. Geol. Surv. S. Afr, pp. 71–80.<br />
Schmidt, K., Koschinsky, A., Garbe-Schönberg, D., de Carvalho, L.M., Seifert, R., 2007.<br />
Geochemistry <strong>of</strong> hydrothermal fluids from the ultramafic-hosted Logatchev<br />
hydrothermal field, 15°N on the Mid-Atlantic Ridge: temporal <strong>and</strong> spatial<br />
investigation. Chem. Geol. 242, 1–21.<br />
Shimizu, H., Umemoto, N., Masuda, A., Appel, P.W.U., 1990. Sources <strong>of</strong> iron-formations in<br />
the Archean Isua <strong>and</strong> Malene supracrustals, West Greenl<strong>and</strong>: evidence from La–Ce<br />
<strong>and</strong> Sm–Nd isotopic data <strong>and</strong> REE. Geochim. Cosmochim. Acta 54, 1147–1154.<br />
Simonson, B.M., 2003. Origin <strong>and</strong> evolution <strong>of</strong> large Precambrian iron formations. Geol.<br />
Soc. Amer. Spec. Paper 370, 231–243.<br />
Simonson, B.M., Hassler, S.W., 1996. Was the deposition <strong>of</strong> large Precambrian iron<br />
formations linked to major marine transgressions? J. Geol. 104, 665–676.<br />
Sholkovitz, E., Shen, G.T., 1995. The incorporation <strong>of</strong> rare earth elements in modern<br />
coral. Geochim. Cosmochim. Acta 59, 2749–2756.<br />
Stecher, O., Carlson, R.W., Shirey, S.B., Nielson, T., 1986. Nd isotope evidence for the<br />
evolution <strong>of</strong> metavolcanic rocks from the Archaean block <strong>of</strong> Greenl<strong>and</strong> <strong>and</strong><br />
Labrador. Terra Cognita 6, 236.<br />
Stevenson, R.K., Patchett, P.J., 1990. Implications for the evolution <strong>of</strong> continental crust<br />
from Hf isotope systematics <strong>of</strong> Archean detrital zircons. Geochim. Cosmochim. Acta<br />
54, 1683–1697.<br />
Tachikawa, K., Athias, V., Je<strong>and</strong>el, C., 2003. Neodymium budget in the modern ocean <strong>and</strong><br />
paleo-oceanographic implications. J. Geophys. Res. 108, 1–13.<br />
Veizer, J., Mackenzie, F.T., 2004. Evolution <strong>of</strong> sedimentary rocks, in: Mackenzie, F.T.<br />
(Ed.), Sediments, Diagenesis, <strong>and</strong> Sedimentary Rocks, 369–407, Vol. 7 Treatise on<br />
Geochemistry (Eds. H.D. Holl<strong>and</strong> <strong>and</strong> K.K. Turekian), Elsevier-Pergamon, Oxford.<br />
Veizer, J., Hoefs, J., Lowe, D.R., Thurston, P.C., 1989. Geochemistry <strong>of</strong> Precambrian<br />
carbonates: II. Archean greenstone belts <strong>and</strong> Archean sea water. Geochim.<br />
Cosmochim. Acta 53, 859–871.<br />
Webb, G.E., Kamber, B.S., 2000. Rare earth elements in Holocene reefal microbialites: a<br />
new shallow seawater proxy. Geochim. Cosmochim. Acta 64, 1557–1565.<br />
Wilson, M., 1989. Igneous Petrogenesis. Chapman <strong>and</strong> Hall, London.<br />
Zhang, J., Nozaki, Y., 1996. Rare earth elements <strong>and</strong> yttrium in seawater: ICP-MS<br />
determinations in the East Caroline, Coral Sea, <strong>and</strong> South Fiji basins <strong>of</strong> the western<br />
South Pacific Ocean. Geochim. Cosmochim. Acta 60, 4631–4644.
Chapter 5. Anoxygenic photoautotrophs <strong>and</strong> the origin <strong>of</strong> b<strong>and</strong>ed iron-formation<br />
153
THIS PAGE INTENTIONALLY LEFT BLANK<br />
154
Anoxygenic photoautotrophs <strong>and</strong> the origin <strong>of</strong> b<strong>and</strong>ed iron-formation<br />
Brian W. Alex<strong>and</strong>er * <strong>and</strong> Michael Bau<br />
<strong>Jacobs</strong> <strong>University</strong> Bremen, D28759 Bremen, Germany<br />
*corresponding author<br />
Oxide-facies b<strong>and</strong>ed iron-formation (BIF) comprised <strong>of</strong> alternating layers <strong>of</strong> Fe oxides<br />
<strong>and</strong> quartz formed as a chemical sediment at the Early Precambrian seafloor. Despite<br />
more than 100 years <strong>of</strong> research, however, the oxidative precipitation mechanism is still<br />
a matter <strong>of</strong> debate. Some propose an indirect biogenic origin via oxidation <strong>of</strong> Fe(II) by<br />
molecular oxygen produced photosynthetically by ancient cyanobacteria 1,2 , while others<br />
suggest an abiotic origin via photo-oxidation <strong>of</strong> Fe(II) 3,4 . Although hypothesized 35<br />
years ago 5 , it was the discovery <strong>of</strong> Fe(II) oxidizing photoautotrophs 6 that stipulated<br />
further investigation <strong>of</strong> the potential role <strong>of</strong> direct biogenic Fe(II) oxidation by<br />
anoxygenic photosynthesis (phot<strong>of</strong>errotrophy) in BIF genesis. These studies suggest that<br />
anoxygenic photoautotrophs could theoretically have oxidized enough Fe(II) to form<br />
oxide-facies BIF in the Early Precambrian ocean 7,8 before the onset <strong>of</strong> oxygenic<br />
photosynthesis. However, evidence from the geological record for Fe(II) oxidation in the<br />
absence <strong>of</strong> oxygen is missing. We present chemical data for ferruginous shales deposited<br />
with oxide-facies BIF from the 2.9 Ga old Pongola Supergroup, South Africa, that<br />
indicate that oxide-facies Precambrian BIF formed in the absence <strong>of</strong> photosynthetically<br />
produced oxygen. In contrast to non-ferruginous shales deposited below <strong>and</strong> above the<br />
Pongola BIF, the ferruginous shales are strongly depleted in K, Rb, Cs, Ba <strong>and</strong> Na, due<br />
to alkali metal exchange for ferrous iron, suggesting that oxide-facies BIF formed in<br />
anoxic, ferruginous seawater. While this is fully compatible with Fe(II) oxidation by<br />
anoxygenic photoautotrophs, it renders the presence <strong>of</strong> biogenic molecular oxygen<br />
unlikely <strong>and</strong> suggests that phot<strong>of</strong>errotrophy, not oxygenic photosynthesis, was a<br />
prerequisite for the formation <strong>of</strong> oxide-facies BIF on Early Earth.<br />
Although their abundance appears to have peaked at around 2.5 Ga ago 9 , Fe(III) oxide<br />
bearing chemical sediments such as BIF, jasper, jaspilite or ferruginous chert (here subsumed<br />
under oxide-facies iron-formation, IF) occur in the oldest marine sedimentary sequence on<br />
Earth at Isua, Greenl<strong>and</strong>, in all Archean greenstone belts, <strong>and</strong> in the oldest supracrustal<br />
successions in the Pongola <strong>and</strong> Witwatersr<strong>and</strong> Supergroups, South Africa. Hence, formation<br />
155
<strong>and</strong> preservation <strong>of</strong> oxide-facies IF was common throughout the first 2 billion years <strong>of</strong><br />
Earth’s geologic history. Oxide-facies IF in Archean marine basins is frequently associated<br />
with Fe-rich shales, with the latter commonly assumed to represent mixtures between detrital<br />
clay minerals <strong>and</strong> Fe(III) oxide minerals.<br />
The Archean (2.85-2.95 Ga) Pongola Supergroup in southern Africa consists <strong>of</strong> the<br />
predominantly volcanic Nsuze Group, which is unconformably overlain by the predominantly<br />
sedimentary Mozaan Group. Previous studies have concluded that the Pongola Supergroup<br />
was deposited in near-shore shallow marine waters on a stable cratonic margin 10,11 , <strong>and</strong><br />
subsequently experienced lower greenschist facies regional metamorphism 10 .<br />
This study discusses samples from the Sinqeni Formation <strong>of</strong> the Mozaan Group, that<br />
were obtained from the White Mfolozi inlier where a shallow-water sequence <strong>of</strong> BIF <strong>and</strong><br />
ferruginous shale is bounded top <strong>and</strong> bottom by non-ferruginous shales 12 . The sample set<br />
includes the non-ferruginous shales deposited immediately below <strong>and</strong> above the BIF <strong>and</strong> the<br />
ferruginous shales within the BIF. Contacts between the ferruginous shales <strong>and</strong> the pure<br />
oxide-facies BIF layers can be sharp, but some BIF beds are very impure chemical sediments<br />
due to the co-deposition <strong>of</strong> detrital aluminosilicates.<br />
Major element, carbon, nitrogen, <strong>and</strong> trace element data for the samples are provided in<br />
Table 1. Variation in major elements such as Al, Fe, <strong>and</strong> K is related to the relative position<br />
<strong>of</strong> the samples in the stratigraphic pr<strong>of</strong>ile. The non-ferruginous shales located below <strong>and</strong><br />
above the BIF display a major element composition typical for Archean shales (Fig. 1). They<br />
are indistinguishable from fine-grained clastic rocks described in a comprehensive study <strong>of</strong><br />
the Pongola Supergroup conducted by Wronkiewicz <strong>and</strong> Condie 13 (see Supplementary<br />
Table), <strong>and</strong> represent sediment weathered from the Kaapvaal craton ~2.85-3.0 Ga ago.<br />
In contrast, Fe in the ferruginous shales is inversely correlated with Al content (Fig. 1),<br />
a relationship that reflects mixing between Al-rich clastic sediment <strong>and</strong> the Fe-rich chemical<br />
precipitate that produced the BIF. On average, the ferruginous shales display half the Al 2 O 3<br />
content <strong>and</strong> a five-fold increase in Fe 2 O 3 compared to the non-ferruginous shales. Similar to<br />
Al <strong>and</strong> Fe, the alkali metals (with the exception <strong>of</strong> Na) display variable concentrations <strong>and</strong> a<br />
distinct bimodal distribution with strong depletions for K, Rb, <strong>and</strong> Cs in the ferruginous<br />
shales (Fig. 1). Potassium <strong>and</strong> Rb average 6.3% <strong>and</strong> 205 ppm, respectively, in the nonferruginous<br />
shales, whereas the ferruginous shales contain less than 0.05% K 2 O <strong>and</strong> typically<br />
less than 1 ppm Rb. Barium concentrations, unlike the other alkaline earth elements, are also<br />
distinct between the non-ferruginous <strong>and</strong> ferruginous shales, averaging 567 ppm <strong>and</strong> less than<br />
22 ppm, respectively (Fig. 1).<br />
156
Table 1. Major <strong>and</strong> trace element concentration data for Pongola Supergroup shales.<br />
wt.% 1<br />
bottom<br />
_________________________________________________<br />
sample position in studied stratigraphic section ______________________________________________________<br />
WM1 WM-A WM-B WM-C WM2 WM3 WM8 WM9 WM10 WM11 WM12 WM14 WM15 WM16 WM171 WM172 WM18 WM19 WM20<br />
Fe-poor<br />
shale<br />
Fe-poor<br />
shale<br />
Fe-poor<br />
shale<br />
Fe-poor<br />
shale<br />
Fe-poor<br />
shale Fe-shale Fe-shale Fe-shale Fe-shale Fe-shale Fe-shale Fe-shale Fe-shale Fe-shale<br />
impure<br />
BIF<br />
impure<br />
BIF<br />
Fe-poor<br />
shale<br />
Fe-poor<br />
shale<br />
top<br />
Fe-poor<br />
shale<br />
AVG<br />
(n = 9)<br />
Fe-shale<br />
AVG<br />
(n = 8)<br />
AVG<br />
(n = 50)<br />
Fe-poor<br />
shale pelite 4<br />
SiO 2 56.6 56.1 47.4 65.3 51.6 52.4 47.9 54.2 44.9 49.7 49.2 44.5 46.9 36.3 38.0 32.1 49.1 48.6 71.8 47.3 55.8 59.5<br />
TiO 2 0.94 1.11 0.95 0.73 0.90 0.76 0.72 0.82 0.81 0.58 0.47 0.59 0.48 0.83 0.16 0.08 1.35 1.21 0.60 0.67 0.97 0.90<br />
Al 2 O 3 29.2 29.8 30.1 22.2 23.3 14.2 11.9 11.7 14.9 13.3 10.4 11.5 11.0 17.1 4.46 1.86 24.7 26.8 16.7 12.9 25.4 21.1<br />
Fe 2 O 3 0.65 0.88 7.13 3.87 10.5 21.0 29.0 24.2 31.0 29.8 29.0 28.3 27.4 32.2 53.8 63.5 9.16 6.55 2.52 28.0 5.15 6.07<br />
MnO
TOP<br />
relative stratigraphic position (~10m)<br />
SiO 2<br />
Fe 2<br />
O 3<br />
BOTTOM<br />
0 10 20 30 40 50 60 70 80 0 5 10 15 20 25 30 35 40<br />
wt.%<br />
wt.%<br />
TOP<br />
Al 2<br />
O 3<br />
0 5 10 15 20 25 30 35 40<br />
wt.%<br />
Na 2<br />
O<br />
0.0 0.2 0.4 0.6 0.8 1.0 1.2<br />
wt.%<br />
relative stratigraphic position (~10m)<br />
predicted<br />
concentrations<br />
K 2<br />
O<br />
BOTTOM<br />
0 1 2 3 4 5 6 7 8 9 10<br />
wt.%<br />
Rb<br />
0 50 100 150 200 250 300<br />
mg/kg<br />
Cs<br />
0 1 2 3 4 5 6 7 8 9 10<br />
mg/kg<br />
Ba<br />
0 200 400 600 800 1000<br />
mg/kg<br />
Figure 1. Pongola shale major <strong>and</strong> trace element concentrations plotted in relative stratigraphic order.<br />
The studied sequence is ~10 m thick, <strong>and</strong> represents a cycle <strong>of</strong> typical Pongola shale deposition (bottom),<br />
through ferruginous shale <strong>and</strong> BIF deposition (middle), before returning to non-ferruginous shale deposition<br />
(top <strong>of</strong> sequence). The grey bar shows the onset <strong>and</strong> cessation <strong>of</strong> pure BIF deposition, <strong>and</strong> the depositional cycle<br />
is clearly evident in the increase in Fe content <strong>of</strong> the samples, coupled with a corresponding decrease in Al.<br />
Determinations <strong>of</strong> Na <strong>and</strong> K for high Fe samples were frequently below X-ray fluorescence quantification<br />
limits, <strong>and</strong> for these samples Na <strong>and</strong> K concentrations are plotted as equal to the analytical quantification limits<br />
(0.11% <strong>and</strong> 0.05%, respectively). Open circles represent predicted concentrations <strong>of</strong> K, Rb, Cs, <strong>and</strong> Ba as<br />
calculated from the average K/Al, Rb/Al, Cs/Al, <strong>and</strong> Ba/Al ratios <strong>of</strong> the eight ‘normal’ non-ferruginous shale<br />
samples that reside at the top (three samples) <strong>and</strong> bottom (five samples) <strong>of</strong> the studied sequence. Extreme<br />
depletion <strong>of</strong> alkali metals <strong>and</strong> Ba is inversely correlated with increasing Fe content <strong>of</strong> the samples <strong>and</strong> the<br />
occurrence <strong>of</strong> BIF deposition.<br />
Incompatible elements (Al, Ti, Zr, Th, Hf) have lower concentrations in the ferruginous<br />
shales compared to the non-ferruginous shales, but ratios <strong>of</strong> these elements are comparable to<br />
published data for Pongola shales 11,13 . For example, Al 2 O 3 /TiO 2 <strong>and</strong> Zr/Th ratios in Pongola<br />
shales are similar regardless <strong>of</strong> Fe content, indicating that this component in the ferruginous<br />
shales is geochemically similar to the local Archean continental crust that produced the nonferruginous<br />
shales. We also emphasize that Th/U ratios in both shale types are similar.<br />
158
The Pongola shales are mineralogically <strong>and</strong> chemically well characterized 11,13-16 <strong>and</strong> the<br />
observed depletion in K, Rb, Cs, Ba <strong>and</strong> Na is common in the Mozaan Group. It is, however,<br />
confined to shales with more than 20 wt.% Fe 2 O 3 , a relationship that is also observed in Ferich<br />
shales interbedded with IF from the ~3.0 Ga Buhwa greenstone belt, Zimbabwe 17 . The<br />
major minerals in non-ferruginous shales from the Mozaan Group are muscovite, chlorite,<br />
<strong>and</strong> illite, with accessory Na- <strong>and</strong> K-feldspar 11,13 . In seven ferruginous (≥19.8% Fe 2 O 3 )<br />
Mozaan Group shales, Nhleko 11 described quartz, muscovite, biotite, chlorite, albite,<br />
magnetite, siderite, <strong>and</strong> stipnomelane as abundant minerals in one or more <strong>of</strong> the samples.<br />
Therefore, the depletion in K, Rb, Cs, Ba <strong>and</strong> Na in the ferruginous shales is unlikely to be<br />
related to differences in the mineralogical composition <strong>of</strong> the source material.<br />
The magnitude <strong>of</strong> the K, Rb, Cs, Ba <strong>and</strong> Na depletion in the ferruginous shales is<br />
significant, as typical Archean shales, similar to modern shales, commonly contain several<br />
weight percent K 2 O <strong>and</strong> hundreds <strong>of</strong> ppm <strong>of</strong> Rb <strong>and</strong> Ba. Whereas the ferruginous shales have<br />
roughly half the Al content <strong>of</strong> typical Pongola shales, the alkali metals <strong>and</strong> Ba are depleted by<br />
a factor <strong>of</strong> approximately 100. Predicted concentrations <strong>of</strong> alkali metals <strong>and</strong> Ba in the<br />
ferruginous shales may be calculated based upon immobile Al <strong>and</strong> average K/Al, Cs/Al,<br />
Rb/Al, <strong>and</strong> Ba/Al ratios in the non-ferruginous shales. Predicted K, Rb, Cs, <strong>and</strong> Ba<br />
concentrations (Fig. 1) suggest that K <strong>and</strong> Rb were depleted by more than 97%, Cs by more<br />
than 94%, <strong>and</strong> Ba between 82-98% during periods <strong>of</strong> high Fe sedimentation. The lack <strong>of</strong> any<br />
unusual enrichment <strong>of</strong> these elements above or below the ferruginous shales suggests that K,<br />
Cs, Rb <strong>and</strong> Ba loss occurred either within the water column or near the sediment-water<br />
interface, as the mobilized metals apparently escaped into the water column.<br />
The mineralogy <strong>of</strong> Mozaan Group shales indicates that sheet silicates (e.g., muscovite,<br />
illite) are the primary hosts for K, Rb, Cs <strong>and</strong> Ba. For these minerals the alkali metals <strong>and</strong> Ba<br />
are electrostatically bound between silicate layers 18 , <strong>and</strong> are susceptible to ion-exchange with<br />
dissolved cations. Adsorption/desorption reactions with clay minerals significantly affect<br />
alkali metal distributions between sediments <strong>and</strong> pore fluids in modern marine systems, <strong>and</strong><br />
Cs <strong>and</strong> Rb, for example, are removed from sediment pore waters by adsorption processes.<br />
Such reactions are reversible, <strong>and</strong> alkali metals adsorbed to clay minerals may be returned to<br />
solution in the presence <strong>of</strong> NH +(19,20) 4 . The use <strong>of</strong> NH + 4 (or Ba 2+ ) to desorb alkali metals from<br />
interlayer atomic sites in sheet silicates is long-established experimental practice in the<br />
determination <strong>of</strong> clay mineral cation-exchange capacities 21,22 . However, N concentrations in<br />
the ferruginous shales are consistently below 0.01 %, <strong>and</strong> are similar to non-ferruginous<br />
shales 16 , suggesting that displacement <strong>of</strong> alkali metals by ammonium did not occur.<br />
159
The mechanism by which the alkali metals <strong>and</strong> Ba were mobilized from the clastic<br />
sediments is rather a consequence <strong>of</strong> high dissolved Fe 2+ concentrations. Ferrous iron may<br />
readily exchange with interlayer Na + in smectite 23 , for example, <strong>and</strong> such Fe 2+ -alkali metal<br />
exchange is also observed for montmorillonite at high ionic strength <strong>and</strong> high chloride<br />
activity in experiments intended to model seawater-clay particle interactions 24 . An Fe 2+ -alkali<br />
metal exchange is further supported by the fact that extreme K, Rb, Cs, Ba <strong>and</strong> Na depletion<br />
is confined to the Fe-rich clastic sediments.<br />
The ferruginous shales are closely associated with oxide-facies BIF. Although the<br />
contacts between ferruginous shale beds <strong>and</strong> BIF beds are <strong>of</strong>ten sharp, there also exist impure<br />
BIF samples that are transitional between ferruginous shales <strong>and</strong> pure BIF. Samples WM171<br />
<strong>and</strong> WM172 have Al 2 O 3 concentrations <strong>of</strong> 4.46% <strong>and</strong> 1.86%, respectively, <strong>and</strong> elevated<br />
concentrations <strong>of</strong> TiO 2 , Zr, Hf, <strong>and</strong> Th (Supplementary Table) compared to associated pure<br />
BIF. Ratios between these immobile elements are similar to those <strong>of</strong> the ferruginous shales,<br />
indicating that the deposition <strong>of</strong> ferruginous shale continued during the deposition <strong>of</strong> oxidefacies<br />
BIF. We emphasize that the detrital aluminosilicate component in WM171 <strong>and</strong><br />
WM172 is characterized by the same lack <strong>of</strong> K, Rb, Cs <strong>and</strong> Ba (<strong>and</strong> Na) as the ferruginous<br />
shales, demonstrating the abundance <strong>of</strong> dissolved ferrous iron in an environment which<br />
formed chemical sediments that are now oxide-facies BIF. This suggests that Fe(III) oxides<br />
precipitated <strong>and</strong> were preserved in a macro-environment that was reducing with respect to the<br />
Fe 2+ /Fe 3+ redox couple, i.e., in a macro-environment that was anoxic. This is supported by the<br />
close similarity <strong>of</strong> the Th/U ratios <strong>of</strong> the ferruginous <strong>and</strong> the non-ferruginous shales, as<br />
oxidation <strong>of</strong> U(IV) in the clay minerals during the exchange reaction would have resulted in<br />
the preferential mobilization <strong>of</strong> U(VI) over Th(IV) 25 , which is not observed.<br />
The oxidation <strong>of</strong> part <strong>of</strong> the available dissolved Fe 2+ <strong>and</strong> the production <strong>of</strong> ferric-ironrich<br />
sediment could result from a number <strong>of</strong> processes. These mechanisms include indirect or<br />
direct biological processes, i.e., oxidation by reaction with photosynthetically produced<br />
molecular oxygen 1,2 , or by anoxygenic photoautotrophs 7,8 , respectively. Abiotic<br />
photochemical oxidation <strong>of</strong> Fe 2+ has also been suggested 3,4 , but is unlikely in the light <strong>of</strong><br />
recent experimental results 26 .<br />
The ferruginous shales associated with the BIF formed when fine-grained clastic<br />
sedimentary material was deposited in ferrous iron rich seawater. Ferrous iron was<br />
incorporated into the clay minerals at the expense <strong>of</strong> K, Rb, Cs, Ba <strong>and</strong> Na which escaped<br />
into the water column. The contemporaneous precipitation <strong>and</strong> preservation <strong>of</strong> oxide-facies<br />
BIF demonstrates that Fe(II) oxidation occurred in an anoxic macro-environment, which is<br />
160
supported by the lack <strong>of</strong> Th-U fractionation. This renders it unlikely that molecular oxygen<br />
was responsible, as either molecular oxygen would have been abundant enough to<br />
quantitatively oxidize the available dissolved ferrous iron, preventing the formation <strong>of</strong> alkali<br />
element depleted ferruginous shales, or the Fe(III) oxides would have been reductively<br />
dissolved (by Fe 2+ , for example) as soon as all the molecular oxygen was consumed. Only the<br />
stabilization <strong>and</strong> preservation <strong>of</strong> Fe(III) oxides by photoautotrophs would have allowed for<br />
the simultaneous deposition <strong>of</strong> oxide-facies BIF <strong>and</strong> ferruginous shales depleted in K, Rb, Cs,<br />
Ba <strong>and</strong> Na.<br />
Thus, oxide-facies IF could form via phot<strong>of</strong>errotrophy throughout the Archean prior to<br />
the onset <strong>of</strong> oxygenic photosynthesis. The formation <strong>of</strong> Fe(III) (hydr)oxides, i.e., oxide-facies<br />
IF, would have been common in a ferrous-iron-rich Archean ocean capable <strong>of</strong> sustaining<br />
postulated anaerobic ecosystems dominated by anoxygenic photoautotrophs 27 , in full<br />
agreement with the ubiquitous presence <strong>of</strong> oxide-facies BIF, jasper, jaspilite <strong>and</strong> ferruginous<br />
chert in the Archean geological record.<br />
Acknowledgements This research was supported by the European <strong>Science</strong> Foundation<br />
research program Archean Environmental Studies: the Habitat <strong>of</strong> Early Life.<br />
References<br />
1. Cloud, P. Significance <strong>of</strong> the Gunflint (Precambrian) micr<strong>of</strong>lora. <strong>Science</strong> 148, 27-35<br />
(1965).<br />
2. Cloud, P. Paleoecological significance <strong>of</strong> the b<strong>and</strong>ed iron-formation. Econ. Geol. 68,<br />
1135-1143 (1973).<br />
3. Cairns-Smith, A.G. Precambrian solution photochemistry, inverse segregation, <strong>and</strong><br />
b<strong>and</strong>ed iron formations. Nature 276, 807-808 (1978).<br />
4. Braterman, P.S., Cairns-Smith, A.G., & Sloper, R.W. Photo-oxidation <strong>of</strong> hydrated Fe2+ -<br />
significance for b<strong>and</strong>ed iron formations. Nature 303, 163-164 (1983).<br />
5. Garrels, R.M., Perry Jr., E.A., & Mackenzie, F.T. Genesis <strong>of</strong> Precambrian ironformations<br />
<strong>and</strong> the development <strong>of</strong> atmospheric oxygen. Econ. Geol. 68, 1173-1179<br />
(1973).<br />
161
6. Widdel, F. et al. Ferrous iron oxidation by anoxygenic phototrophic bacteria. Nature 362,<br />
834-836 (1993).<br />
7. Konhauser, K.O. et al. Could bacteria have formed the Precambrian b<strong>and</strong>ed iron<br />
formations? Geology 30, 1079-1082 (2002).<br />
8. Kappler, A., Pasquero, C., Konhauser, K.O., & Newman, D.K. Deposition <strong>of</strong> b<strong>and</strong>ed<br />
iron formations by anoxygenic phototrophic Fe(II)-oxidizing bacteria. Geology 33,<br />
865-868 (2005).<br />
9. Klein, C. Some Precambrian b<strong>and</strong>ed iron-formations (BIFs) from around the world:<br />
Their age, geologic setting, mineralogy, metamorphism, geochemistry, <strong>and</strong> origin.<br />
Amer. Mineral. 90, 1473-1499 (2005).<br />
10. Beukes, N.J. & Cairncross, B. A lithostratigraphic-sedimentological reference pr<strong>of</strong>ile for<br />
the late Archaean Mozaan Group, Pongola Sequence; application to sequence<br />
stratigraphy <strong>and</strong> correlation with the Witwatersr<strong>and</strong> Supergroup. S. Afr. J. Geol. 94,<br />
44-69 (1991)<br />
11. Nhleko, N. The Pongola Supergroup in Swazil<strong>and</strong>, Ph.D. thesis, R<strong>and</strong> Afrikaans Univ.<br />
(2003).<br />
12. Alex<strong>and</strong>er, B.W., Bau, M., Andersson, P., & Dulski, P. Continentally-derived solutes in<br />
shallow Archean seawater: Rare earth element <strong>and</strong> Nd isotope evidence in iron<br />
formation from the 2.9 Ga Pongola Supergroup, South Africa. Geochim. Cosmochim.<br />
Acta 72, 378-394 (2008).<br />
13. Wronkiewicz, D.J. & Condie, K.C. Geochemistry <strong>and</strong> provenance <strong>of</strong> sediments from the<br />
Pongola Supergroup, South Africa: Evidence for a 3.0-Ga-old continental craton<br />
Geochim. Cosmochim. Acta 53, 1537-1549 (1989).<br />
14. McLennan, S. M., Taylor, S. R. & Kröner, A. Geochemical evolution <strong>of</strong> Archean shales<br />
from South Africa. I. The Swazil<strong>and</strong> <strong>and</strong> Pongola Supergroups. Precam. Res. 22, 93-<br />
124 (1983).<br />
15. Jahn, B. & Condie, K.C. Evolution <strong>of</strong> the Kaapvaal Craton as viewed from geochemical<br />
<strong>and</strong> Sm-Nd isotopic analyses <strong>of</strong> intracratonic pelites. Geochim. Cosmochim. Acta 59,<br />
2239-2258 (1995).<br />
162
16. Watanabe Y., Naraoka, H., Wronkiewicz, D.J., Condie, K.C., & Ohmoto, H. Carbon,<br />
nitrogen, <strong>and</strong> sulfur geochemistry <strong>of</strong> Archean <strong>and</strong> Proterozoic shales from the<br />
Kaapvaal Craton, South Africa. Geochim. Cosmochim. Acta 61, 3441-3459 (1997).<br />
17. Fedo, C.M., Eriksson, K.A., & Krogstad, E.J. Geochemistry <strong>of</strong> shales from the Archean<br />
(~3.0 Ga) Buhwa Greenstone Belt, Zimbabwe: Implications for provenance <strong>and</strong><br />
source-area weathering. Geochim. Cosmochim. Acta 60, 1751-1763 (1996).<br />
18. Deer, W.A., Howie, R.A., & Zussman, J. An Introduction to the Rock-Forming Minerals.<br />
(Longman Sci. & Tech., Essex, 1992).<br />
19. Comans, R.N.J. et al. Mobilization <strong>of</strong> radiocaesium in pore water <strong>of</strong> lake sediments.<br />
Nature 339, 367-369 (1989).<br />
20. James, R.H. & Palmer, M.R. Marine geochemical cycles <strong>of</strong> the alkali elements <strong>and</strong><br />
boron: The role <strong>of</strong> sediments. Geochim. Cosmochim. Acta 64, 3111-3122 (2000).<br />
21. Chapman, H. D. in Methods <strong>of</strong> Soil Analysis Part 2 (ed. Black, C.A.) 891-901 (Am. Inst.<br />
Agron., Madison, Wisconsin, 1965).<br />
22. Hendershot, W.H. & Duquette, M. A simple barium chloride method for determining<br />
cation exchange capacity <strong>and</strong> exchangeable cations. Soil Sci Soc Am J. 50, 605-608<br />
(1986).<br />
23. Kamei, G., Oda, C., Mitsui, S., Shibata, M., & Shinozaki, T. Fe(II )-Na ion exchange at<br />
interlayers <strong>of</strong> smectite: adsorption-desorption experiments <strong>and</strong> a natural analogue.<br />
Eng. Geol. 54, 15-20 (1999).<br />
24. Charlet, L. & Tournassat, C. Fe(II)–Na(I)–Ca(II) cation exchange on montmorillonite in<br />
chloride medium: evidence for preferential clay adsorption <strong>of</strong> chloride–metal ion pairs<br />
in seawater. Aqua. Geochem. 11, 115-137 (2005).<br />
25. Rosing, M.T. & Frei, R. U-rich Archaean sea-floor sediments from Greenl<strong>and</strong> -<br />
indications <strong>of</strong> >3700 Ma oxygenic photosynthesis. Earth Planet. Sci. Lett. 217, 237-<br />
244 (2004).<br />
26. Konhauser, K.O. et al. Decoupling photochemical Fe(II) oxidation from shallow-water<br />
IF deposition. Earth Planet. Sci. Lett. 258, 87-100 (2007).<br />
27. Canfield, D.E., Rosing, M.T., & Bjerrum, C. Early anaerobic metabolisms. Phil. Trans.<br />
R. Soc. B 361, 1819-1836 (2006).<br />
163
THIS PAGE INTENTIONALLY LEFT BLANK<br />
164
Chapter 6. Preservation <strong>of</strong> primary REE patterns without Ce anomaly during<br />
dolomitization <strong>of</strong> Mid-Paleoproterozoic limestone <strong>and</strong> the potential reestablishment<br />
<strong>of</strong> marine anoxia immediately after the“Great Oxidation<br />
Event”<br />
165
THIS PAGE INTENTIONALLY LEFT BLANK<br />
166
MICHAEL BAU AND BRIAN ALEXANDER<br />
81<br />
Preservation <strong>of</strong> primary REE patterns without Ce anomaly<br />
during dolomitization <strong>of</strong> Mid-Paleoproterozoic limestone <strong>and</strong> the<br />
potential re-establishment <strong>of</strong> marine anoxia immediately<br />
after the“Great Oxidation Event”<br />
Michael Bau <strong>and</strong> Brian Alex<strong>and</strong>er<br />
Geosciences <strong>and</strong> Astrophysics Program, <strong>School</strong> <strong>of</strong> <strong>Engineering</strong> <strong>and</strong> <strong>Science</strong>s,<br />
International <strong>University</strong> Bremen, P.O. Box 750561, D-28725 Bremen, Germany<br />
email: m.bau@iu-bremen.de; b.alex<strong>and</strong>er@iu-bremen.de<br />
© 2006 March Geological Society <strong>of</strong> South Africa<br />
ABSTRACT<br />
Comparison <strong>of</strong> shallow-water limestone <strong>and</strong> silicified dolomite from the Mid-Paleoproterozoic Mooidraai Formation, Transvaal<br />
Supergroup, South Africa, shows that the primary rare earth element (REE) distribution <strong>of</strong> these pure marine sedimentary carbonates<br />
has been preserved during dolomitization <strong>and</strong> silicification. Both lithologies display REE (<strong>and</strong> Y) patterns closely resembling those<br />
<strong>of</strong> present-day seawater, i.e. they show enrichment <strong>of</strong> the heavy relative to the light REE, positive anomalies <strong>of</strong> La, Gd <strong>and</strong> Lu, <strong>and</strong><br />
super-chondritic Y/Ho ratios. However, these shallow-water carbonates lack any Ce anomalies, indicating that in the Mid-<br />
Paleoproterozoic the redox-level <strong>of</strong> surface water in the Griqual<strong>and</strong>-West sub-basin <strong>of</strong> the Kaapvaal Craton did not allow for<br />
oxidation <strong>of</strong> Ce(III). The absence <strong>of</strong> Ce anomalies from shallow water Mooidraai carbonates indicates a return to marine anoxia<br />
immediately after large amounts <strong>of</strong> marine sedimentary Mn oxides had been deposited in a highly oxygenated marine environment<br />
in the underlying Hotazel Formation. This suggests that the “Great Oxidation Event” in the Paleoproterozoic was a transition period<br />
characterized by strong fluctuations <strong>of</strong> the redox level <strong>of</strong> the Earth’s surface ocean.<br />
Introduction<br />
Interpretation <strong>of</strong> trace element distribution <strong>and</strong> isotope<br />
ratios in Precambrian sedimentary carbonates suffers<br />
from the fact that these rocks may occur as (sometimes<br />
silicified) dolomites. Field evidence, such as<br />
dolomitization fronts, suggests that the dolomite is not a<br />
primary seawater precipitate <strong>and</strong> that dolomitization<br />
occurred during diagenesis, but in most cases such<br />
evidence is missing. Hence, conclusions based on the<br />
trace element <strong>and</strong> isotope compositions <strong>of</strong> dolomites are<br />
<strong>of</strong>ten questioned. This is unfortunate, since Neoarchean<br />
<strong>and</strong> Paleoproterozoic carbonates such as those from the<br />
Transvaal Supergroup in South Africa <strong>and</strong> from<br />
the Hamersley Group in Australia, for example, should<br />
have recorded potential changes in the seawater<br />
composition before <strong>and</strong> after the “Great Oxygenation<br />
Event” <strong>and</strong> during episodes <strong>of</strong> low-latitude glaciation in<br />
the Paleoproterozoic (for recent summaries <strong>of</strong> early<br />
Precambrian atmosphere-hydrosphere evolution see,<br />
e.g. Holl<strong>and</strong>, 2004, Canfield, 2005 <strong>and</strong> Catling <strong>and</strong> Claire,<br />
2005).<br />
The rare earths <strong>and</strong> yttrium (REY) hosted in marine<br />
sedimentary carbonates can be used as proxies for the<br />
REY distribution in ambient seawater, since the partition<br />
coefficients between carbonate <strong>and</strong> seawater do not<br />
show major differences within the REY series (Terakado<br />
<strong>and</strong> Masuda, 1988; Zhong <strong>and</strong> Mucchi, 1995; Webb <strong>and</strong><br />
Kamber, 2000; Tanaka et al., 2004). A positive Eu<br />
anomaly, for example, may reveal the presence <strong>of</strong> a<br />
high-temperature hydrothermal REY component even in<br />
shallow seawater <strong>and</strong> the presence or absence <strong>of</strong><br />
a Ce anomaly may indicate an oxygenated or<br />
anoxic atmosphere-hydrosphere system, respectively.<br />
Underst<strong>and</strong>ing the impact <strong>of</strong> dolomitization on the REY<br />
distribution in marine sedimentary carbonates, therefore,<br />
might enable us to significantly improve our knowledge<br />
<strong>of</strong> the chemical evolution <strong>of</strong> the atmospherehydrosphere<br />
system across the Archean-Proterozoic<br />
boundary <strong>and</strong> in the early Proterozoic. The topic <strong>of</strong> REY<br />
mobility during dolomitization had been addressed,<br />
amongst others, by Banner et al. (1988) <strong>and</strong> Nothdurft<br />
et al. (2003), for example, who studied sedimentary<br />
carbonates <strong>of</strong> Phanerozoic age. However, their<br />
conflicting conclusions do not prove or disprove the<br />
suitability <strong>of</strong> REY in Precambrian dolomites as proxies<br />
for REY in ambient seawater, but highlight the need for<br />
further study.<br />
Here, we compare the REY distribution in marine<br />
limestone <strong>and</strong> in silicified dolomite from the Paleoproterozoic<br />
Mooidraai Formation in the Postmasburg<br />
Group <strong>of</strong> the Transvaal Supergroup, South Africa.<br />
We show that dolomitization did not modify the primary<br />
REY distribution <strong>of</strong> the marine sedimentary carbonates,<br />
suggesting that these dolomites still provide information<br />
on the REY distribution in Precambrian seawater.<br />
We also show that neither the limestones nor the<br />
dolomites from the Mooidraai Formation display<br />
significant Ce anomalies <strong>and</strong>, hence, do not support the<br />
assumption <strong>of</strong> the existence <strong>of</strong> strongly oxygenated<br />
surface seawater during Mid-Paleoproterozoic<br />
“Mooidraai times”.<br />
Geology<br />
The Mooidraai Formation occurs within the uppermost<br />
part <strong>of</strong> the Neoarchean to Paleoproterozoic Transvaal<br />
Supergroup, Northern Cape Province, South Africa.<br />
SOUTH AFRICAN JOURNAL OF GEOLOGY, 2006,VOLUME 109 PAGE 81-86
82<br />
PRESERVATION OF PRIMARY REE PATTERNS WITHOUT CE ANOMALY<br />
Figure 1. Simplified stratigraphic position <strong>of</strong> the Mooidraai <strong>and</strong><br />
Hotazel formations <strong>of</strong> the Transvaal Supergroup, South Africa.<br />
Its stratigraphic position is indicated in Figure 1.<br />
Deposition <strong>of</strong> iron formations <strong>and</strong> intercalated<br />
manganiferous oxides <strong>and</strong> carbonates <strong>of</strong> the Hotazel<br />
Formation followed the extrusion <strong>of</strong> the Ongeluk<br />
basaltic <strong>and</strong>esites that define the base <strong>of</strong> the Voëlwater<br />
Subgroup. The Hotazel Formation is best known as host<br />
to the enormous resource <strong>of</strong> manganese ores that<br />
constitute the Kalahari Manganese Field. Conformably<br />
above the Hotazel Formation follow shallow marine<br />
limestones <strong>and</strong> dolomites <strong>of</strong> the Mooidraai Formation.<br />
Carbonate slump breccias <strong>and</strong> sideritic carbonate units<br />
are restricted to the lower part <strong>of</strong> the Mooidraai<br />
Formation, whereas microbiolaminated <strong>and</strong> stromatolitic<br />
carbonates predominate the upper part. The contact <strong>of</strong><br />
the Mooidraai Formation with the discordantly overlying<br />
Mapedi shales <strong>and</strong> quartzites <strong>of</strong> the Olifantshoek<br />
Formation is erosional. The Mooidraai carbonates did<br />
neither experience significant metamorphism nor<br />
deformation. A detailed description <strong>of</strong> the geological<br />
setting <strong>of</strong> the Mooidraai Formation can be found in<br />
Beukes (1983) <strong>and</strong> Swart (1999).<br />
The petrography <strong>of</strong> the limestones <strong>and</strong> <strong>of</strong> the<br />
silicified dolomites has been described by Tsikos et al.<br />
(2001) <strong>and</strong> Bau et al. (1999), respectively. Note that the<br />
dolomite samples examined here originate from the<br />
stromatolitic upper part <strong>of</strong> the Moodraai Formation (Bau<br />
et al., 1999); the limestone samples, on the other h<strong>and</strong>,<br />
are thought to represent the complete succession (Tsikos<br />
et al., 2001). The geology <strong>and</strong> the C-O-Sr isotope<br />
geochemistry <strong>of</strong> the Hotazel <strong>and</strong> Mooidraai formations<br />
are discussed by Schneiderhan et al. (2006).<br />
Unfortunately, the respective absolute ages <strong>of</strong> the<br />
Hotazel <strong>and</strong> the Mooidraai Formation are not very well<br />
constrained. No radiometric age is available for the<br />
Hotazel Formation <strong>and</strong> the Pb-Pb carbonate “age” for<br />
the Mooidraai dolomites discussed here is 2394 +/- 26<br />
Ma (Bau et al., 1999). Since the reliability <strong>of</strong> Pb-Pb<br />
carbonate ages is a matter <strong>of</strong> concern, the correct<br />
depositional age <strong>of</strong> the Hotazel <strong>and</strong> Mooidraai sediments<br />
is still unclear. The currently most widely accepted age<br />
for the Voëlwater Subgroup is bracketed by an illconstrained<br />
2222 Ma age for the underlying Ongeluk<br />
lava (the problems associated with the data on which<br />
the Ongeluk age is based have been addressed by Bau<br />
et al., 1999) <strong>and</strong> by the 2060 Ma ages <strong>of</strong> the Bushveld<br />
<strong>and</strong> Molopo Farms igneous complexes (e.g., Walraven<br />
et al., 1990, Walraven <strong>and</strong> Hattingh, 1993; Reichardt,<br />
1994; Buick et al., 2001). If the stratigraphic correlation<br />
between the Ongeluk <strong>and</strong> the Hekpoort extrusives is<br />
correct, a recent Re-Os age <strong>of</strong> 2316 +/- 7 Ma for pyrite<br />
in a carbonaceous shale from the base <strong>of</strong> the Timeball<br />
Hill Formation (Hannah et al., 2004) below the<br />
Hekpoort basalt narrows the depositional age <strong>of</strong> the<br />
Hotazel <strong>and</strong> Mooidraai formations down to between<br />
~2.32 <strong>and</strong> ~2.06 Ga. However, the “normal” 13 C values<br />
(Figure 2) <strong>of</strong> the Mooidraai carbonates (Bau et al., 1999;<br />
Tsikos et al., 2001) demonstrate that they formed<br />
before the Lomagundi Event, i.e. before the<br />
worldwide positive excursion <strong>of</strong> 13 C values <strong>of</strong> marine<br />
carbonate deposited between ~2.25 Ga <strong>and</strong> ~2.07 Ga<br />
ago (Aharon, 2005, <strong>and</strong> references therein). Hence, the<br />
current best estimate suggests that the Hotazel<br />
iron <strong>and</strong> manganese formations <strong>and</strong> the Mooidraai<br />
carbonates formed between ~2.32 <strong>and</strong> ~2.25 Ga ago.<br />
The Hotazel Formation <strong>and</strong> the Mooidraai Formation,<br />
therefore, represent a well-preserved Mid-<br />
Paleoproterozoic succession <strong>of</strong> shallow marine chemical<br />
sediments.<br />
Results<br />
The chemical compositions <strong>of</strong> limestone <strong>and</strong> silicified<br />
dolomite from the Mooidraai Formation have been<br />
presented by Tsikos et al. (2001) <strong>and</strong> Bau et al. (1999),<br />
respectively, <strong>and</strong> will only be summarized here (see<br />
these publications for data). Additional data can be<br />
found in Swart (1999) <strong>and</strong> Schneiderhan et al. (2006),<br />
but since REY data are incomplete or not available, we<br />
will not consider these in the present contribution which<br />
focusses on the REY distribution.<br />
The Mooidraai limestone samples (Figure 2) are only<br />
slightly silicified (CaO: 34.07 to 52.56%; MgO: 0.60 to<br />
1.94%; SiO 2 : 1.39 to 13.50%) <strong>and</strong> show low MnO (0.14<br />
to 1.30%) but high Fe 2 O 3 (2.42 to 23.92%). The most Feenriched<br />
carbonates occur in the bottom <strong>and</strong> top parts<br />
<strong>of</strong> the Mooidraai Formation (Tsikos et al., 2001). Low<br />
Al 2 O 3 contents (0.10 to 0.53%) indicate that the amount<br />
<strong>of</strong> detrital aluminosilicates in the chemical sediment is<br />
negligible. Ba <strong>and</strong> Sr concentrations range from 9 ppm<br />
to 105 ppm, <strong>and</strong> from 789 ppm to 1524 ppm,<br />
respectively. Concentrations <strong>of</strong> individual REY vary<br />
within a factor <strong>of</strong> six between samples; Nd<br />
concentrations, for example, range from 0.26 ppm to<br />
1.69 ppm. REY SN patterns (‘ SN ’: indicates normalization<br />
to post-Archean Australian Shale, PAAS, from McLennan,<br />
1989; note that Tsikos et al. (2001) do not report Y data)<br />
are strongly HREE-enriched (Figure 3).<br />
The Mooidraai dolomite (Figure 2) discussed here is<br />
strongly silicified (CaO: 15.68 to 17.74%; MgO: 10.04 to<br />
11.59%; SiO 2 : 39.3 to 46.5%). MnO <strong>and</strong> Fe 2 O 3 range from<br />
0.25 to 0.30% <strong>and</strong> from 1.26 to 1.44%, respectively,<br />
SOUTH AFRICAN JOURNAL OF GEOLOGY
MICHAEL BAU AND BRIAN ALEXANDER<br />
83<br />
parallel to those <strong>of</strong> the Mooidraai limestones (Figure 3).<br />
Both limestones <strong>and</strong> dolomites show similar HREE<br />
enrichment, positive La SN <strong>and</strong> Gd SN anomalies, <strong>and</strong> even<br />
small positive Lu SN anomalies (Figure 3), although the<br />
size <strong>of</strong> the positive La SN anomalies is somewhat larger in<br />
the limestones than it is in the dolomites. The presence<br />
<strong>of</strong> Eu SN anomalies is hard to verify because <strong>of</strong> the<br />
anomalous Gd content, but the Eu SN /Eu SN * ratios<br />
calculated as Eu SN /(0.67Sm SN +0.33Tb SN ) overlap, <strong>and</strong><br />
range from 1.11 to 1.51 with an average <strong>of</strong> 1.28 in the<br />
limestones compared to a range from 0.91 to 1.29 with<br />
an average <strong>of</strong> 1.11 in the dolomites.<br />
This preservation <strong>of</strong> primary REYSN patterns during<br />
dolomitization <strong>of</strong> marine sedimentary carbonates<br />
demonstrates that Precambrian dolomites can still<br />
provide valuable information on the REY distribution in<br />
Precambrian seawater.<br />
Figure 2. Major element <strong>and</strong> stable isotope characteristics <strong>of</strong><br />
limestone <strong>and</strong> dolomite <strong>of</strong> the Mooidraai Formation, Transvaal<br />
Supergroup, South Africa (data from Tsikos et al., 2001, <strong>and</strong> Bau et<br />
al., 1999). Note that the Mooidraai carbonates do not show<br />
anomalously positive 13 C values, suggesting that they formed<br />
before the Lomagundi Event, i.e. prior to ~2.25 Ga ago.<br />
Implications for Earth’s Mid-Paleoproterozoic<br />
atmosphere <strong>and</strong> oceans<br />
The REY distribution in the Mooidraai sedimentary<br />
carbonates indicates that these are marine chemical<br />
sediments, as the combination <strong>of</strong> HREE enrichment with<br />
positive La SN , Gd SN , Lu SN , <strong>and</strong> Y SN anomalies (or superchondritic<br />
Y/Ho ratios) cannot be observed in lakes <strong>and</strong><br />
rivers but is confined to seawater. These anomalies for<br />
which after compensating for the effects <strong>of</strong> silification is<br />
similar to what is seen in the Fe-poor samples <strong>of</strong><br />
Mooidraai limestone. Concentrations <strong>of</strong> elements<br />
typically associated with detrital aluminosilicates are<br />
very low (Rb: 0.05 to 0.17 ppm; Cs: 0.014 to 0.023 ppm;<br />
Al 2 O 3 , Zr, Hf, <strong>and</strong> Th are below the respective lower<br />
limit <strong>of</strong> determination), attesting to the purity <strong>of</strong> the<br />
chemical sediment. Ba <strong>and</strong> Sr concentrations<br />
are significantly lower than those <strong>of</strong> the limestone<br />
samples <strong>and</strong> range from 1.1 to 2.5 ppm <strong>and</strong> from<br />
12.8 to 15.3 ppm, respectively. REY concentrations are<br />
also low (e.g., Nd: 0.102 – 0.125 ppm) <strong>and</strong> REY SN<br />
patterns show pronounced enrichment <strong>of</strong> the HREE<br />
(Figure 3).<br />
Discussion<br />
Comparing REY in limestones <strong>and</strong> dolomites<br />
Compared to the limestones, the Mooidraai dolomites<br />
show significantly lower Sr <strong>and</strong> Ba concentrations <strong>and</strong><br />
reflect the loss <strong>of</strong> Sr, for example, typically observed<br />
during dolomitization (e.g., Veizer, 1983a; b). However,<br />
dolomitization did not produce elevated Fe <strong>and</strong> Mn<br />
concentrations as is observed in Phanerozoic carbonates<br />
(e.g., Veizer, 1983a; b), suggesting only minor<br />
remobilization <strong>of</strong> Mn <strong>and</strong> Fe. This may have implications<br />
for the interpretation <strong>of</strong> Fe isotope data <strong>of</strong> Precambrian<br />
sedimentary carbonates, since primary Fe isotope<br />
signatures may be preserved during dolomitization.<br />
Despite dolomitization <strong>and</strong> strong silicification, the<br />
Mooidraai dolomites show REY SN patterns that are<br />
Figure 3. Shale-normalized REY patterns <strong>of</strong> limestones <strong>and</strong><br />
silicified dolomites from the Mooidraai Formation <strong>and</strong> <strong>of</strong> a sample<br />
<strong>of</strong> manganese formation (braunite lutite; M. Bau, unpubl. data)<br />
from the underlying Hotazel Formation. The parallel patterns <strong>of</strong><br />
the carbonates demonstrate that the primary REY distribution <strong>of</strong><br />
the limestone was preserved during dolomitization <strong>and</strong><br />
silicification. Heavy REE enrichment <strong>and</strong> positive anomalies <strong>of</strong> La,<br />
Gd, Y, <strong>and</strong> Lu indicate that these carbonates are pure marine<br />
chemical sediments. Except for the lack <strong>of</strong> a negative Ce anomaly,<br />
these REY patterns are very similar to those <strong>of</strong> modern seawater<br />
<strong>and</strong> modern marine sedimentary carbonates. Note that in contrast<br />
to the Mooidraai carbonates, manganese formation from the<br />
Hotazel Formation does display a negative Ce SN anomaly.<br />
SOUTH AFRICAN JOURNAL OF GEOLOGY
84<br />
PRESERVATION OF PRIMARY REE PATTERNS WITHOUT CE ANOMALY<br />
Figure 4. Plot <strong>of</strong> Ce SN /Ce SN * vs. Pr SN /Pr SN * that allows to quantify<br />
Ce <strong>and</strong> La anomalies. The Mooidraai carbonates show Pr SN /Pr SN *<br />
ratios close to unity, i.e. similar to those <strong>of</strong> shales, which<br />
demonstrates the absence <strong>of</strong> negative Ce anomalies. Ce SN /Ce SN *<br />
ratios below unity indicate the presence <strong>of</strong> positive La anomalies<br />
in the Mooidraai carbonates, as commonly found in seawater <strong>and</strong><br />
marine chemical sediments. The absence <strong>of</strong> significant negative Ce<br />
anomalies suggests that surface seawater during “Mooidraai times”<br />
was too reducing to allow for oxidation <strong>of</strong> Ce(III). For further<br />
explanation see text. Data from Tsikos et al., 2001 (1); Bau et al.,<br />
1999 (2); Bau et al., 2003, <strong>and</strong> M. Bau, unpublished data (3);<br />
Taylor <strong>and</strong> McLennan, 1985 (4, 5).<br />
the non-redox-sensitive REY together with the lack <strong>of</strong><br />
siliciclastic detritus also preclude deposition <strong>of</strong> the<br />
Mooidraai Formation in coastal waters in close proximity<br />
to a l<strong>and</strong> mass, but rather support the drowned-platform<br />
environment suggested by Beukes (1983).<br />
The lack <strong>of</strong> clear positive Eu SN anomalies in PAASnormalized<br />
REY distribution patterns demonstrates the<br />
absence <strong>of</strong> a high-temperature hydrothermal REY<br />
component from shallow seawater during “Mooidraai<br />
times”. This is different from what is seen in Neoarchean<br />
carbonates, such as those from the Lower Transvaal<br />
Supergroup, South Africa (Kamber <strong>and</strong> Webb, 2001), or<br />
from the Hamersley Supergroup, Australia (M. Bau,<br />
unpublished data), <strong>and</strong> in marked contrast to what is<br />
observed in iron formation from the lower Transvaal<br />
Supergroup, that give strong evidence for a blacksmoker-type<br />
hydrothermal REY source (Bau <strong>and</strong> Dulski,<br />
1996; Bau et al., 1997).<br />
We emphasize that there exist no significant Ce SN<br />
anomalies in the Mooidraai limestones or dolomites.<br />
As discussed previously (Bau <strong>and</strong> Dulski, 1996;<br />
Kamber <strong>and</strong> Webb, 2001), the commonly used<br />
equation to quantifiy Ce SN anomalies (Ce SN /Ce SN * =<br />
Ce SN /(0.5La SN +0.5Pr SN )) must not be applied in studies<br />
<strong>of</strong> marine chemical sediments due to the presence <strong>of</strong><br />
positive La SN anomalies that suggest Ce SN * values that are<br />
much too high. A plot <strong>of</strong> Ce SN /Ce SN * vs. Pr SN /Pr SN * (Figure<br />
4; Pr SN * = 0.5Ce SN +0.5Nd SN ; for details on this approach<br />
see, for example, Bau <strong>and</strong> Dulski, 1996; Kamber <strong>and</strong><br />
Webb, 2001) clearly demonstrates that neither the<br />
limestones nor the dolomites from the Mooidraai<br />
Formation are characterized by distinct negative Ce SN<br />
anomalies (Figure 4). The Mooidraai carbonates,<br />
although free from detrital aluminosilicates, show<br />
Pr SN /Pr SN * ratios that are considerably smaller<br />
than those <strong>of</strong> Lower Carboniferous limestones from<br />
Irel<strong>and</strong> <strong>and</strong> Engl<strong>and</strong>, for example, that display<br />
negative Ce SN anomalies (Bau et al., 2003; M. Bau,<br />
unpublished data). Their Pr SN /Pr SN * ratios are, however,<br />
very similar to those typical <strong>of</strong> shales (Figure 4). The<br />
absence <strong>of</strong> Ce SN anomalies from the dolomites does not<br />
result from the dolomitization process but is a primary<br />
feature <strong>of</strong> the chemical sediment <strong>and</strong>, therefore, reflects<br />
the absence <strong>of</strong> any Ce SN anomaly from Mooidraai<br />
seawater.<br />
We emphasize that the Mooidraai REY data are yet<br />
another example <strong>of</strong> how neglecting the existence <strong>of</strong><br />
positive La SN anomalies in seawater <strong>and</strong> in some<br />
marine chemical sediments can lead to false claims <strong>of</strong><br />
the presence <strong>of</strong> negative Ce SN anomalies. Using a<br />
Pr SN /Pr SN * ratio <strong>of</strong> >1.1 as the positive criterion for<br />
a negative Ce SN anomaly, it appears that except for a few<br />
individual analyses <strong>of</strong> samples from iron-formations <strong>and</strong><br />
cherts in India (e.g., Kato et al., 2002, <strong>and</strong> references<br />
therein), there exist no published data from Archean or<br />
Early Paleoproterozoic chemical sediments that show<br />
primary Ce SN anomalies.<br />
The absence <strong>of</strong> distinct negative Ce SN anomalies from<br />
the Mooidraai carbonates, however, is surprising<br />
considering that the Mooidraai Formation was deposited<br />
immediately above the Hotazel Formation. The Hotazel<br />
Formation contains the first <strong>and</strong> hence oldest significant<br />
deposits <strong>of</strong> marine sedimentary Mn oxides in the<br />
geological record. The precipitation <strong>of</strong> these<br />
sedimentary Mn oxides represents a major event in<br />
Earth’s history as the Mn oxide deposits are widespread<br />
<strong>and</strong> several meters thick (for a recent discussion, see<br />
Schneiderhan et al., 2006). These Mn oxides are the<br />
prot-ore for the Mn ore in the Kalahari Manganese Field,<br />
the largest deposit <strong>of</strong> sedimentary Mn oxides on Earth.<br />
The precipitation <strong>and</strong>, in particular, the preservation <strong>of</strong><br />
such a huge amount <strong>of</strong> marine sedimentary Mn oxides<br />
requires a highly oxygenated depositional environment<br />
(e.g., Kirschvink et al., 2000; Kirschvink <strong>and</strong> Weiss, 2002;<br />
Kopp et al., 2005). The redox level necessary to stabilize<br />
significant amounts <strong>of</strong> Mn(IV) oxides, however, is<br />
significantly higher than that needed to oxidize Ce(III)<br />
<strong>and</strong> to form Ce anomalies. Hence, it comes as no<br />
surprise that, indeed, negative Ce SN anomalies occur in<br />
manganese formation from the Hotazel Formation<br />
(Figure 3).<br />
Their stratigraphic position above the Mn-oxidesbearing<br />
Hotazel Formation <strong>and</strong>, in particular, above the<br />
SOUTH AFRICAN JOURNAL OF GEOLOGY
MICHAEL BAU AND BRIAN ALEXANDER<br />
85<br />
~2.32 Ga-old Timeball Hill Formation that does not<br />
show mass-independent sulphur isotope fractionation<br />
(Bekker et al., 2004), indicates that the deposition <strong>of</strong> the<br />
Mooidraai carbonates post-dates the “Great Oxygenation<br />
Event”. The lack <strong>of</strong> specific Ce depletion in Mooidraai<br />
seawater, therefore, suggests that after the deposition <strong>of</strong><br />
the Hotazel Formation the redox-level <strong>of</strong> seawater in the<br />
Griqual<strong>and</strong>-West sub-basin had returned to a lower level<br />
insufficient to oxidize Ce(III). Apparently, Mn(II)<br />
oxidation had consumed the emerging oxygen reservoir<br />
<strong>and</strong> biogenic oxygen production was not yet able to<br />
balance the additional oxygen dem<strong>and</strong>. Thus, local<br />
marine anoxia was re-established during Mid-<br />
Paleoproterozoic Mooidraai times.<br />
It remains to be seen whether this retreat to marine<br />
anoxia in the Mid-Paleoproterozoic was a worldwide<br />
phenomenon or whether it was confined to the<br />
Griqual<strong>and</strong>-West sub-basin <strong>of</strong> the Kaapvaal Craton.<br />
Nevertheless, the REY data from the Mooidraai<br />
limestones <strong>and</strong> dolomites suggest that even the<br />
oxygenation <strong>of</strong> the surface water <strong>of</strong> the Earth’s oceans<br />
did not follow a unidirectional trend from generally<br />
anoxic to generally oxic conditions. Rather, oxygenated<br />
surface environments became progressively more<br />
abundant until the previously exotic “oxygen oases” had<br />
grown <strong>and</strong> become more persistent before they<br />
eventually represented the normal state <strong>of</strong> the system.<br />
Hence, it might be more apt to refer to the oxygenation<br />
<strong>of</strong> the Earth’s surface system in the Paleoproterozoic as<br />
the “Great Oxygenation Period”. Such a s<strong>of</strong>t rather than<br />
sharp transition is also more in line with, for example,<br />
the contemporaneous formation <strong>of</strong> oxic <strong>and</strong> anoxic<br />
paleosols (Rye <strong>and</strong> Holl<strong>and</strong>, 1998, Yang <strong>and</strong> Holl<strong>and</strong>,<br />
2003), the observation <strong>of</strong> abundant hydrothermal mantle<br />
Os in pyrites that do not show mass-independent<br />
sulphur isotope fractionation (Bekker et al., 2004;<br />
Hannah et al., 2004), <strong>and</strong> low Mo concentration <strong>and</strong><br />
unfractionated Mo isotope ratios in the Transvaal<br />
Supergroup (Siebert et al., 2004).<br />
Conclusions<br />
Comparison <strong>of</strong> marine shallow-water limestone <strong>and</strong><br />
silicified dolomite from the Mid-Paleoproterozoic<br />
Mooidraai Formation, Transvaal Supergroup, South<br />
Africa, demonstrates that the REY distribution <strong>of</strong> the<br />
marine sedimentary carbonate was preserved during<br />
dolomitization <strong>and</strong> silicification. With one exception,<br />
both lithologies display all the details <strong>of</strong> the REY<br />
distibution in present-day seawater, such as positive<br />
anomalies for La, Gd <strong>and</strong> Lu, <strong>and</strong> a super-chondritic<br />
Y/Ho ratio. However, the Mooidraai carbonates lack the<br />
negative Ce anomaly that indicates oxidation <strong>of</strong> Ce(III)<br />
in the Earth’s surface system. It appears that after the<br />
deposition <strong>of</strong> Mn oxides in the Hotazel Formation,<br />
which is indicative <strong>of</strong> a highly oxygenated supergene<br />
environment, conditions again became sigificantly less<br />
oxic. This suggests that during the transition period from<br />
a rather reducing to an oxygenated atmospherehydrosphere<br />
system in the Paleoproterozoic (the “Great<br />
Oxygenation Period”) the redox-level <strong>of</strong> the Earth’s<br />
surface ocean fluctuated between reducing <strong>and</strong> oxic.<br />
Acknowledgements<br />
We gratefully acknowledge many fruitful discussions<br />
<strong>of</strong> Precambrian geochemistry <strong>and</strong> geology with<br />
H.D. Holl<strong>and</strong>, J. Gutzmer, <strong>and</strong> J. Kasting. The paper also<br />
benefited from constructive reviews by A. Lepl<strong>and</strong>,<br />
G. Shields, <strong>and</strong> J. Gutzmer. In particular, however, we<br />
want to thank Nic Beukes for his advice, help, <strong>and</strong> kind<br />
hospitality during many years <strong>of</strong> collaborative work.<br />
His efforts have without doubt stimulated worldwide<br />
interest in <strong>and</strong> promoted South African chemical<br />
sediments as a valuable source <strong>of</strong> information on the<br />
chemical evolution <strong>of</strong> the Earth’s atmospherehydrosphere<br />
system.<br />
References<br />
Aharon, P. (2005) Redox stratification <strong>and</strong> anoxia <strong>of</strong> the early Precambrian<br />
oceans: Implications for carbon isotope excursions <strong>and</strong> oxidation events.<br />
Precambrian Research, 137, 207-222.<br />
Banner, J.K., Hanson, G.L. <strong>and</strong> Meyers, W.J. (1988) Rare earth element <strong>and</strong><br />
Nd isotopic variations in regionally extensive dolomites from the<br />
Burlington-Keokuk Formation (Mississippian): Implications for REE<br />
mobility during carbonate diagenesis. Journal <strong>of</strong> Sedimentary Petrology,<br />
58, 415-432.<br />
Bau, M. <strong>and</strong> Dulski, P. (1996) Distribution <strong>of</strong> yttrium <strong>and</strong> rare-earth elements<br />
in the Penge <strong>and</strong> Kuruman Iron-Formations, Transvaal Supergroup, South<br />
Africa. Precambrian Research, 79, 37-55.<br />
Bau, M., Höhndorf, A., Dulski, P. <strong>and</strong> Beukes, N.J. (1997) Sources <strong>of</strong> rareearth<br />
elements <strong>and</strong> iron in Paleoproterozoic iron-formations from the<br />
Transvaal Supergroup, South Africa: Evidence from neodymium isotopes.<br />
Journal <strong>of</strong> Geology, 105, 121-129.<br />
Bau, M., Romer, R., Lüders, V. <strong>and</strong> Beukes, N.J. (1999) Pb, O, <strong>and</strong> C isotopes<br />
in silicified Mooidraai dolomite (Transvaal Supergroup, South Africa):<br />
Implications for the composition <strong>of</strong> Paleoproterozoic seawater <strong>and</strong> “dating”<br />
the increase <strong>of</strong> oxygen in the Precambrian atmosphere. Earth <strong>and</strong><br />
Planetary <strong>Science</strong> Letters, 174, 43-57.<br />
Bau, M., Romer, R., Lüders, V. <strong>and</strong> Dulski, P. (2003) Tracing element sources<br />
<strong>of</strong> hydrothermal mineral deposits: REE <strong>and</strong> Y distribution <strong>and</strong> Sr-Nd-Pb<br />
isotopes in fluorite from MVT deposits in the Pennine Orefield, Engl<strong>and</strong>.<br />
Mineralium Deposita, 38, 992-1008.<br />
Bekker, A., Holl<strong>and</strong>, H.D., Wang, P.-L., Rumble, D. III, Stein, H.J., Hannah ,<br />
J.L., Coetzee, L.L. <strong>and</strong> Beukes, N.J. (2004) Dating the rise <strong>of</strong> atmospheric<br />
oxygen. Nature, 427, 117-120.<br />
Beukes, N. J. (1983) Palaeoenvironmental setting <strong>of</strong> iron-formations in the<br />
depositional basin <strong>of</strong> the Transvaal Supergroup, South Africa. In: A.F.<br />
Trendall <strong>and</strong> R.C. Morris (Editors), Iron-formation: Facts <strong>and</strong> Problems.<br />
Developments in Precambrian Geology, 6, 131-209.<br />
Buick, I.S., Maas, R. <strong>and</strong> Gibson, R. (2001) Precise U-Pb titanite age<br />
constraints on the emplacement <strong>of</strong> the Bushveld Complex, South Africa.<br />
Journal <strong>of</strong> the Geological Society London, 158, 3-6.<br />
Canfield, D.E. (2005) The early history <strong>of</strong> atmospheric oxygen: Homage to<br />
Robert M. Garrels. Annual Reviews Earth <strong>and</strong> Planetary <strong>Science</strong>s, 33, 1-36.<br />
Catling, D.C. <strong>and</strong> Claire, M.W. (2005) How Earth’s atmosphere evolved to an<br />
oxic state: A status report. Earth <strong>and</strong> Planetary <strong>Science</strong> Letters, 237, 1-20.<br />
Hannah, J.L., Bekker, A., Stein, H.J., Markey, R.J. <strong>and</strong> Holl<strong>and</strong>, H.D. (2004)<br />
Primitive Os <strong>and</strong> 2316 Ma age for marine shale: implications for<br />
Paleoproterozoic glacial events <strong>and</strong> the rise <strong>of</strong> atmospheric oxygen. Earth<br />
<strong>and</strong> Planetary <strong>Science</strong> Letters, 225, 43-52.<br />
Holl<strong>and</strong>, H.D. (2004) The geological history <strong>of</strong> seawater. In: H.D. Holl<strong>and</strong><br />
<strong>and</strong> K.K. Turekian (Editors), Treatise <strong>of</strong> Geochemistry, Elsevier,<br />
Amsterdam, The Netherl<strong>and</strong>s, 6, 583-625.<br />
Kamber, B.S. <strong>and</strong> Webb, G.E. (2001) The geochemistry <strong>of</strong> late Archaean<br />
microbial carbonate: Implications for ocean chemistry <strong>and</strong> continental<br />
erosion history. Geochimica et Cosmochimica Acta, 65, 2509-2525.<br />
Kato Y., Kano T. <strong>and</strong> Kunugiza, K. (2002) Negative Ce Anomaly in the Indian<br />
B<strong>and</strong>ed Iron Formations: Evidence for the Emergence <strong>of</strong> Oxygenated<br />
SOUTH AFRICAN JOURNAL OF GEOLOGY
86<br />
PRESERVATION OF PRIMARY REE PATTERNS WITHOUT CE ANOMALY<br />
Deep-Sea at 2.9~2.7 Ga. Resource Geology, 52, 101-110.<br />
Kirschvink, J.L. <strong>and</strong> Weiss, B.P. (2002) Mars, panspermia, <strong>and</strong> the origin <strong>of</strong><br />
life: Where did it all begin? Palaeontologia Electronica, 4, 8-15.<br />
Kirschvink, J.L., Gaidos, E.J., Bertani, L.E., Beukes, N.J., Gutzmer, J., Maepa,<br />
L.N. <strong>and</strong> Steinberger, R.E. (2000) Paleoproterozoic Snowball Earth: Extreme<br />
climatic <strong>and</strong> geochemical global change <strong>and</strong> its biological consequences .<br />
Proceedings <strong>of</strong> the National Acadamy <strong>of</strong> <strong>Science</strong>s, 97, 1400-1405.<br />
Kopp, R.E., Kirschvink, J.L., Hilburn, I.A. <strong>and</strong> Nash, C.Z. (2005) The<br />
Paleoproterozoic snowball Earth: A climate disaster triggered by the<br />
evolution <strong>of</strong> oxygenic photosynthesis. Proceedings <strong>of</strong> the National<br />
Acadamy <strong>of</strong> <strong>Science</strong>s, 102, 11131-11136.<br />
McLennan, S.M. (1989) Rare earth elements in sedimentary rocks: Influence<br />
<strong>of</strong> provenance <strong>and</strong> sedimentary processes. In: B.R. Lipin <strong>and</strong> G.A. McKay<br />
(Editors), Geochemistry <strong>and</strong> mineralogy <strong>of</strong> rare earth elements. Reviews in<br />
Mineralogy, 21, 169-200.<br />
Nothdurft, L.D., Webb, G.E. <strong>and</strong> Kamber, B.S. (2003) Rare earth element<br />
geochemistry <strong>of</strong> Late Devonian reefal carbonates, Canning basin, Western<br />
Australia: Confirmation <strong>of</strong> a seawater REE proxy in ancient limestones.<br />
Geochimica et Cosmochimica Acta, 68, 263-283.<br />
Reichardt, F.J. (1994) The Molopo Farms complex, botswana: history,<br />
stratigraphy, petrography, petrochemistry <strong>and</strong> Ni-Cu-PGE mineralization.<br />
Exploration <strong>and</strong> Mining Geology, 3, 264-284.<br />
Rye , R. <strong>and</strong> Holl<strong>and</strong> , H.D. (1998) Paleosols <strong>and</strong> the evolution <strong>of</strong><br />
atmospheric oxygen: A critical review. American Journal <strong>of</strong> <strong>Science</strong>,<br />
298, 621–672.<br />
Schneiderhan, E.A., Gutzmer, J., Strauss, H., Mezger, K. <strong>and</strong> Beukes, N.J.<br />
(2006) The chemostratigraphy <strong>of</strong> a Paleoproterozoic MnF-BIF-dolostone<br />
sucession – the Voëlwater Subgroup <strong>of</strong> the Transvaal Supergroup in<br />
Griqual<strong>and</strong> West, South Africa. South African Journal <strong>of</strong> Geology,<br />
109, 63-80.<br />
Siebert, C., Kramer, J.D., Meisel, Th., Morel, Ph. <strong>and</strong> Nägler, Th.F. (2005)<br />
PGE, Re-Os, <strong>and</strong> Mo isotope systematics in Archean <strong>and</strong> early Proterozoic<br />
sedimentary systems as proxies for redox conditions on the early Earth.<br />
Geochimica et Cosmochimica Acta, 69, 1787-1801.<br />
Swart, Q.D. (1999) Carbonate rocks <strong>of</strong> the Paleoproterozoic Pretoria <strong>and</strong><br />
Postmasburg Groups, Transvaal Supergroup. Unpublished M.Sc thesis,<br />
R<strong>and</strong> Afrikaans <strong>University</strong>, Johannesburg, South Africa, 126pp.<br />
Tanaka, K., Ohta, A. <strong>and</strong> Kawabe, I. (2004) Experimental REE partitioning<br />
between calcite <strong>and</strong> aqueous solution at 25°C <strong>and</strong> 1 atm: Constraints on<br />
the incorporation <strong>of</strong> seawater REE into seamount-type limestones.<br />
Geochemical Journal, 38, 19-32.<br />
Taylor, S.R. <strong>and</strong> McLennan, S.M. (1985) The continental crust: its composition<br />
<strong>and</strong> evolution. Oxford, Blackwell, United Kingdom, 312pp.<br />
Terakado, Y. <strong>and</strong> Masuda, A. (1988). The coprecipitation <strong>of</strong> rare-earth<br />
elements with calcite <strong>and</strong> aragonite. Chemical Geology, 69, 103-110.<br />
Tsikos, H., Moore, J.M. <strong>and</strong> Harris, C. (2001) Geochemistry <strong>of</strong> the<br />
Palaeoproterozoic Mooidraai Formation: Fe-rich limestone as end-member<br />
<strong>of</strong> iron-formation deposition, Transvaal Supergroup, South Africa. Journal<br />
<strong>of</strong> African Earth <strong>Science</strong>s, 32, 19-27.<br />
Veizer, J. (1983a) Chemical diagenesis <strong>of</strong> carbonate rocks: Theory <strong>and</strong><br />
application <strong>of</strong> trace element technique. In: M.A. Arthur et al. (Editors),<br />
Stable Isotopes in Sedimentary Geology, SEPM Short Course Notes,<br />
10, III/1 – III/100.<br />
Veizer, J. (1983b) Trace elements <strong>and</strong> isotopes in sedimentary carbonates.<br />
In: E. Roedder (Editor), Carbonates: Mineralogy <strong>and</strong> Chemistry, Reviews in<br />
Mineralogy, 11, 265-300.<br />
Walraven, F., Armstrong, R.A. <strong>and</strong> Kruger, F.J. (1990) A chronostratigraphic<br />
framework for the north-central Kaapvaal craton, the Bushveld Complex<br />
<strong>and</strong> the Vredefort structure. Tectonophysics, 171, 23-48.<br />
Walraven, F. <strong>and</strong> Hattingh, E. (1993) Geochronology <strong>of</strong> the Nebo Granite,<br />
Bushveld Complex. South African Journal <strong>of</strong> Geology, 96, 31-41.<br />
Webb, G. E. <strong>and</strong> Kamber, B. S. (2000) Rare earth elements in Holocene reefal<br />
microbialites: a new shallow seawater proxy. Geochimica et Cosmochimica<br />
Acta, 64, 1557-1565.<br />
Yang, W. <strong>and</strong> Holl<strong>and</strong>, H.D. (2003). The Hekpoort paleosol pr<strong>of</strong>ile in Strata<br />
1 at Gaborone, Botswana: Soil formation during the Great Oxidation Event.<br />
American Journal <strong>of</strong> <strong>Science</strong>, 303, 187-220.<br />
Zhong, S.Y. <strong>and</strong> Mucci, A. (1995) Partitioning <strong>of</strong> rare earth elements (REEs)<br />
between calcite <strong>and</strong> seawater solutions at 25ºC <strong>and</strong> 1 atm, <strong>and</strong> high<br />
dissolved REE concentrations. Geochimica et Cosmochimica Acta,<br />
59, 443-453.<br />
Editorial h<strong>and</strong>ling: J. Gutzmer<br />
SOUTH AFRICAN JOURNAL OF GEOLOGY
Chapter 7. Concluding remarks<br />
The data presented in this thesis presumably constrain certain geochemical<br />
features <strong>of</strong> seawater for the rather narrow period <strong>of</strong> time between 2.9-3.0 Ga. Studies <strong>of</strong><br />
seawater chemistry that focus on younger rocks benefit from both the increased<br />
preservation <strong>of</strong> potential seawater archives as well as the greater occurrences <strong>of</strong> marine<br />
carbonates. However, in many regards, the geochemistry <strong>of</strong> Archean seawater is poorly<br />
constrained, particularly for the period <strong>of</strong> time prior to ~3.0 Ga. Though rocks older<br />
than 3.0 Ga are not particularly rare, seawater archives from this time period are limited.<br />
Further work is necessary to identify seawater archives older than 3.0 Ga, <strong>and</strong> data not<br />
included in this thesis suggest that suitable material from southern Africa is available<br />
for the time period between 3.2-3.6 Ga. Neodymium isotopic studies <strong>of</strong> these samples<br />
would help refute or support the conclusion that bulk seawater Є Nd values were generally<br />
positive with values between 0 <strong>and</strong> +2 from 3.0-3.8 Ga.<br />
For younger rocks, unpublished Nd isotopic data indicates that shallow water<br />
marine carbonates from the Transvaal Supergroup in South Africa possess Є Nd (2.5 Ga)<br />
between -0.3 <strong>and</strong> -2.2, consistent with the conclusions <strong>of</strong> the Pongola IF study that<br />
weathering <strong>of</strong> continental crust significantly influenced seawater chemistry near the<br />
Kaapvaal craton after 2.9 Ga. Furthermore, Mn sediments from the ~2.25 Ga Hotazel<br />
Formation in South Africa display seawater-like REY patterns, <strong>and</strong> possess Є Nd values<br />
between -2.2 <strong>and</strong> -4.2. These data are consistent with an increasingly important flux <strong>of</strong><br />
solutes derived from weathering <strong>of</strong> continental crust in younger marine chemical<br />
precipitates. A possible test <strong>of</strong> this hypothesis would be Nd isotopic studies on late<br />
Archean (2.5-2.9 Ga) seawater archives <strong>of</strong> well defined age <strong>and</strong> representing a variety<br />
<strong>of</strong> depositional environments. Such studies may resolve the question <strong>of</strong> whether or not<br />
173
ulk, open ocean seawater tended to more positive Є Nd values as suggested by the<br />
Pietersburg IF data.<br />
Finally, the suggestion that iron-formations may have precipitated as Fe(II)<br />
minerals is worthy <strong>of</strong> future study. Many unanswered questions remain regarding this<br />
hypothesis, ranging from the primary mineralogy <strong>of</strong> such a precipitate to its conversion<br />
to the Fe(III) mineral phases that dominate oxide facies IFs. One intriguing aspect <strong>of</strong><br />
such a model regards the REY patterns <strong>of</strong> many Archean IFs. As discussed in the<br />
Introduction to this thesis, Fe oxy-hydroxides that precipitate from, <strong>and</strong> are in<br />
equilibrium with, modern seawater, display REY patterns distinctly different from<br />
ambient seawater. The seawater-like REY patterns observed in many Archean-<br />
Paleoproterozoic IFs have been attributed to quantitative scavenging <strong>of</strong> the REY from<br />
ancient seawater during the process(es) that produced IFs. If, however, the primary IF<br />
precipitate was predominantly composed <strong>of</strong> Fe(II) mineral phases, then different<br />
mechanisms for incorporating the REY into IFs may have operated, <strong>and</strong> further<br />
examination <strong>of</strong> this potential process is warranted.<br />
174