14.01.2013 Views

Analytical Chemistry Chemical Cytometry Quantitates Superoxide

Analytical Chemistry Chemical Cytometry Quantitates Superoxide

Analytical Chemistry Chemical Cytometry Quantitates Superoxide

SHOW MORE
SHOW LESS

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

Figure 6. The average measured response time, τmeas, is independent<br />

of the volume fraction of absorbed analyte, φ, and thus the<br />

analyte activity. This response time, however, is strongly analyte<br />

dependent, enabling discrimination between analytes having nearly<br />

identical chemical affinities such as toluene and undecane. At low<br />

concentrations (where the slope of the response curve is low) and<br />

for long response times, sensor drift becomes an issue and response<br />

times are difficult to reliably measure; such is the case for low<br />

concentrations of undecane.<br />

variations, we normalized the response time of each sensor to<br />

agree with that of an arbitrarily chosen reference sensor, by<br />

exposing the sensors to a mesitylene activity of 0.058, which<br />

resulted in a swelling of 0.005. These renormalized time scales<br />

were then used to determine both the average response time<br />

to an analyte and the associated measurement error in terms<br />

of the standard deviation. The dimensionless correction factors<br />

ranged from 0.6 to 1.3.<br />

To determine if it is a reasonable assumption that the variation<br />

in the measured response time is caused by variations in sensor<br />

thickness, we made 10 chemiresistors on a glass slide using<br />

methods and materials identical to those for the originals. Sensor<br />

thickness measurements, made with a Nikon mm-800 measuring<br />

microscope with a quadra-chek 200 advanced digital readout<br />

system, give a sensor thickness of 216 ± 26 µm. The average<br />

response time, τ meas, of the original five sensors to mesitylene<br />

at an activity of 0.058 was 48 ± 13 s. Using this average response<br />

time and the average composite thickness of the newly made<br />

sensors (i.e., 216 µm), we calculate a measured diffusion<br />

coefficient of 4.5 × 10 -2 cm 2 /s. From this diffusion coefficient<br />

we computed the predicted response times for the 10 new<br />

sensors, yielding 48 ± 8 s. Therefore, the variation in response<br />

times that we measured for the original sensors is similar to<br />

the variation in response time that we predict from the newly<br />

made sensors. It should be noted that this measured diffusion<br />

coefficient is not a true diffusion coefficient because it is<br />

calculated from τmeas, which is a convolution of the actual<br />

diffusion time scale and the time scale from the flow cell fill<br />

time as we will discuss in the following section.<br />

An accurate determination of the sorption kinetics ideally<br />

requires that the flow cell fill time is fast compared to the sorption<br />

kinetics. In the following we will determine this fill time from the<br />

measured sorption kinetics and use this value to extract the true<br />

sorption kinetics from the measured values. The true sorption<br />

time can then be compared to the flow cell fill time to determine<br />

whether this fast fill time condition is met in our experiments.<br />

Flow Cell Fill Time. After the inlet stream starts to deliver<br />

an analyte of fixed activity to the flow cell, the analyte activity<br />

rises continuously, due to the finite volume of pure nitrogen that<br />

must be displaced. The transient analyte activity can be obtained<br />

by modeling the flow cell as a continuously stirred tank and is of<br />

the form<br />

a(t) ) a ∞ (1 - e -t/τf ) (8a)<br />

In terms of the volume fraction of the vapor this is<br />

φ vap (t) ) φ vap (∞)(1 - e -t/τf ) (8b)<br />

The characteristic time for the flow cell to reach a steady-state<br />

concentration is given by τf ) Vf/F, where Vf is the volume of<br />

the flow cell and F is the volumetric flow rate of nitrogen from<br />

the inlet stream.<br />

Convolution of Time Scales. The measured composite<br />

swelling in Figure 4 is a convolution of the increase in the analyte<br />

vapor activity in the flow cell and the diffusive kinetics, which for<br />

simplicity we take to be of the form φ(t) ) φ∞(1 - e -t/τ d) for a<br />

step increase in the analyte activity at t ) 0. Using eq 5, the<br />

expression is<br />

t dφvap (t)<br />

φ(t) ) K∫ 0<br />

t)s dt [1 - e-(t-s)/τd ]ds (9)<br />

Using eq 8a to evaluate the derivative gives<br />

φ(t) ) Kφvap (∞) t<br />

τ ∫ e<br />

0<br />

f<br />

-s/τf -(t-s)/τd [1 - e ]ds (10)<br />

A straightforward integration leads to the final expression for the<br />

sorption kinetics”<br />

φ(t) ) Aφvap (∞)[ 1 - τd e<br />

τd - τf -t/τd +<br />

τ f<br />

e<br />

τd - τf -t/τf]<br />

(11)<br />

The measured lifetime can then be computed from this equation,<br />

and the surprisingly simple result is<br />

τmeas ) 1 ∞<br />

φ ∫ [1 - φ(t)] dt ) τ<br />

0<br />

d + τf ∞<br />

(12)<br />

The true diffusion time can thus be obtained from the measured<br />

time by τd ) τmeas s τf. (The case where τd = τf leads to division<br />

by zero in eq 11. This division looks troublesome, but can be<br />

handled by defining τd ) τf(1 + ε) and carefully taking the limit<br />

as ε f 0.)<br />

Corrected Sorption Times. The correction of the measured<br />

sorption times for the flow cell fill time requires a determination<br />

of the fill time. The fill time can be extracted from the measured<br />

sorption times themselves, through a limiting process, as we will<br />

now describe. To obtain accurate measured sorption times, we<br />

first average these measured times over all analyte activities to<br />

obtain a mean response time we call τjmeas. This averaging is valid,<br />

because the measured sorption time is independent of the<br />

<strong>Analytical</strong> <strong>Chemistry</strong>, Vol. 82, No. 16, August 15, 2010<br />

6973

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!