Analysis and Monitoring of Volatile Analytes from Aqueous Solutions

Microdialysis membranes (3 mm length × 200 μm i.d.) have been used to extract volatile analytes from aqueous samples into the gas phase and interfac...
0 downloads 0 Views 271KB Size
Anal. Chem. 2008, 80, 123-128

Analysis and Monitoring of Volatile Analytes from Aqueous Solutions by Extractions into the Gas Phase Using Microdialysis Membranes and Coupling to Fast GC Melissa A. Jones,† Ashley Kramer,† Matthew Humbert,† Tyler Vanadurongvan,† Jonathan Maurer,† Michael T. Bowser,‡ and Anthony J. Borgerding*,†

Department of Chemistry, University of St. Thomas, St. Paul, Minnesota 55105, and Department of Chemistry, University of Minnesota, Minneapolis, Minnesota 55455

Microdialysis membranes (3 mm length × 200 µm i.d.) have been used to extract volatile analytes from aqueous samples into the gas phase and interfaced with fast gas chromatography. Gas-phase extracts generated from aqueous samples reach steady-state concentrations and are transported to the detector in 5 s. The recovery of the system ranges from 1.28% for toluene to less than 0.1% for ethanol. The lowest detectable concentration without preconcentration was 5 mM for most compounds using a flame ionization detector, and as low as 0.01 mM for more volatile hydrophobic analytes. When interfaced with a fast GC system, changes in aqueous phase concentrations were monitored with a temporal resolution of 10 s. In this paper, we demonstrate for the first time the use of microdialysis membrane probes for the extraction of volatile analytes to the gas phase from aqueous solutions. The technique combines attributes of other gas-phase extractions using semipermeable membranes with attributes of microdialysis extractions into liquid-phase dialysates. Gas-phase extractions using membranes1-5 have been very popular for the analysis of environmental samples. They are simple and rugged and give strong signal for very low concentrations of volatile, nonpolar analytes in air and water. However, they have some limitations. Membrane materials that are readily available for construction of extractors are generally at least 2 mm in diameter and have thicknesses of at least 100 µm. The large size limits the applicability of probes made from these materials for use in very small volumes such as cells or other biological environments. To some extent, coating very small rods with nonpolar membrane materials such as poly(dimethylsiloxane) (PDMS) in solid-phase microextraction (SPME) experiments has * Corresponding author. E-mail: [email protected]. † University of St. Thomas. ‡ University of Minnesota. (1) Mendes, M. A.; Pimpim, R. S.; Kotiaho, T.; Eberlin, M. N. Anal. Chem. 1996, 68, 3502-3506. (2) Wong, P.; Cooks, R. G.; Cisper, M.; Hemberger, P. Environ. Sci. Technol. 1995, 29, A215-A218. (3) Guo, X.; Mitra, S. J. Chromatogr., A 2000, 904, 189-196. (4) Guo, X.; Mitra, S. J. Chromatogr., A 1998, 826, 39-47. (5) Mitra, S.; Zhang, L.; Zhu, N.; Guo, X. J. Microcolumn Sep. 1996, 8, 21-27. 10.1021/ac071530h CCC: $40.75 Published on Web 11/22/2007

© 2008 American Chemical Society

alleviated this problem. However, the thicknesses of the membrane materials, even in the SPME probes, requires several minutes for analytes to achieve equilibrium between the aqueous sample and the membrane material.6-8 This limits the speed with which samples can be analyzed. SPME is also hard to use for monitoring experiments, since it does not extract continuously. Furthermore, almost all of the membranes used for these experiments are nonpolar, and analyzing polar analytes in this way has been quite difficult. Allen et al. addressed this by sputtercoating porous polymer membranes with allyl alcohol in a membrane-inlet mass spectrometry (MIMS) experiment.9 In addition, SPME experiments have been performed using Nafion membranes.10 While these examples are noteworthy, they still were rather slow in their response, and they delivered limits of detection in the low part-per-million range (∼0.1 mM) for ethanol and methanol, indicating only marginal levels of preconcentration. Kou and Mitra have also used porous polymer membranes for gas-phase extraction but had slow response times and required further concentration on a microtrap before detection.11 Microdialysis sampling, and liquid-phase extraction using porous polymer membranes, overcomes some of these limitations. Porous polypropylene membranes have been used to extract a wide variety of analytes into polar, nonpolar, and ionic liquids.12-14 Microdialysis sampling has been used for aqueous extractions into aqueous buffers for analysis of peptides, amino acids, and other molecules of biological significance.15-19 Very small probes can (6) Saraullo, A.; Martos, P. A.; Pawliszyn, J. Anal. Chem. 1997, 69, 1992-1998. (7) Vaes, W. H. J.; Hamwijk, C.; Ramos, E. U.; Verhaar, H. J. M.; Hermens, J. L. M. Anal. Chem. 1996, 68, 4458-4462. (8) Louch, D.; Motlagh, S.; Pawliszyn, J. Anal. Chem. 1992, 64, 1187-1199. (9) Allen, T. M.; Falconer, T. M.; Cisper, M. E.; Borgerding, A. J.; Wilkerson, C. W., Jr. Anal. Chem. 2001, 73, 4830-4835. (10) Gorecki, T.; Martos, P.; Pawliszyn, J. Anal. Chem. 1998, 70, 19-27. (11) Kou, D.; Mitra, S. Anal. Chem. 2003, 75, 6355-6360. (12) Hauser, B.; Schellin, M.; Popp, P. Anal. Chem. 2004, 76, 6029-6038. (13) He, T.; Versteeg, L. A. M.; Mulder, M. H. V.; Wessling, M. J. Membr. Sci. 2004, 234, 1-10. (14) Peng, J.; Liu, J.; Hu, X.; Jiang, G. J. Chromatogr., A 2007, 1139, 165-170. (15) Ciriacks, C. M.; Bowser, M. T. Anal. Chem. 2004, 76, 6582-6587. (16) Huynh, B. H.; Fogarty, B. A.; Martin, R. S.; Lunte, S. M. Anal. Chem. 2004, 76, 6440-6447. (17) Miro, M.; Frenzel, W. TrAC, Trends Anal. Chem. 2005, 24, 324-333. (18) Bowser, M. T.; Kennedy, R. T. Electrophoresis 2001, 22, 3668-3676. (19) Brunner, M.; Muller, M. Recent Res. Dev. Cancer 2001, 3, 425-435.

Analytical Chemistry, Vol. 80, No. 1, January 1, 2008 123

be made using these membranes, which allows in vivo analysis. In addition, since the partitioning is not largely dependent on the chemistry of the membrane material, nonpolar analytes are commonly extracted. However, pumping liquid dialysates through the small i.d. capillaries in these tiny probes limits flows to ∼1 µL/min, which can lead to transport times of 1 min or more between the extraction probe and the detection system.15 In addition, while the pores of dialysis membranes can remove interferences from large molecules, salts will pass through the pores with analytes molecules which can depress signals in some detectors. The use of microdialysis membranes to sample volatile analytes into the gas phase offers many potential advantages. The small size and nonselective nature of these probes allow for very fast measurement of polar analytes as well as nonpolar analytes. Gases can flow through the probes at much higher flow rates, reducing transport time. In addition, extracting into the gas phase affords some level of selectivity in that only analytes with sufficient volatility will be extracted. Finally, the extraction technique can be easily interfaced with gas chromatography and the suite of gasphase detectors. Extracting analytes into the gas phase can expand the in vivo use of microdialysis probes to include applications involving volatile analytes commonly found in biological systems, including alcohols20-22 aldehydes, ketones,23,24 and other compounds of diagnostic interest.25-27 EXPERIMENTAL SECTION Microdialysis Probe Construction. “Side by side” probes were constructed in-house using fused-silica capillaries (Polymicro Technologies, Phoenix, AZ). Following the technique of Ciriacks and Bowser,10,15 two capillaries (40 µm i.d./105 µm o.d.), staggered by 3 mm, were placed in a regenerated cellulose microdialysis membrane fiber (200 µm i.d./216 µm o.d., 18 000 MWCO, Spectrum Laboratories, Inc., Rancho Dominguez, CA). The membrane was sealed on both ends with polyimide resin (see Figure 1). The probe was secured with superglue in a 21 gauge 2 in. luer lock needle with the luer hub removed (Sigma-Aldrich, St. Louis, MO). This needle had previously been inserted through the septum of an open top vial cap, and the probe was positioned with the membrane exposed from the needle tip at a depth that submerged it in the aqueous samples contained in the vials (see Figure 1). The figure also shows how the extract from the probe was analyzed, as described below. Instrumentation. Configuration 1: Microdialysis Probe Directly Interfaced to FID. With the use of 2 cm segments of 1/16 in. Teflon tubing (Alltech, Deerfield, IL), the probe was connected to the (20) Zhang, X.; Dong, F.; Li, Q.; Borgerding, A. J.; Klein, A. L.; Ren, J. J. Appl. Physiol. 2005, 99, 2246-2254. (21) Duan, J.; Esberg, L. B.; Ye, G.; Borgerding, A. J.; Ren, B. H.; Aberle, N. S.; Epstein, P. N.; Ren, J. Comp. Biochem. Physiol., Part A: Mol. Integr. Physiol. 2003, 134A, 607-614. (22) Robinson, D. L.; Lara, J. A.; Brunner, L. J.; Gonzales, R. A. J. Neurochem. 2000, 75, 1685-1693. (23) Kalapos, M. P. Biochim. Biophys. Acta 2003, 1621, 122-139. (24) Bailey, D. N. J. Toxicol. Clin. Toxicol. 1990, 28, 459-466. (25) St. Germain, F.; Vachon, B.; Montgomery, J.; Des Rosiers, C. Free Radical Biol. Med. 1997, 23, 166-172. (26) St. Germain, F.; Mamer, O.; Brunet, J.; Vachon, B.; Tardif, R.; Abribat, T.; Des Rosiers, C.; Montgomery, J. Anal. Chem. 1995, 67, 4536-4541. (27) Miekisch, W.; Schubert, J. K.; Vagts, D. A.; Geiger, K. Clin. Chem. 2001, 47, 1053-1060.

124 Analytical Chemistry, Vol. 80, No. 1, January 1, 2008

Figure 1. Schematic showing probe construction and the position of the probe in the sampling apparatus and the instrumental configuration. The GC was bypassed in some experiments.

flame ionization detector (FID) of a Hewlett-Packard 5890 GC with a 0.8 m length of 375/40 µm (o.d./i.d.) fused-silica capillary. The input to the probe was similarly connected to a tank of helium (Oxygen Service, St. Paul, MN) with a 0.15 m length of the 375/ 40 µm (o.d./i.d.) fused-silica capillary. Helium gas flow (flow rate 0.4 mL/min) swept compounds that permeated the membrane to the detector. The pressure at the helium tank was 80 psi, and the pressure inside the probes was estimated to be approximately 70 psi, based on lengths of restrictive capillary going into and out of the probes. As our results show, this pressure does not prevent gas-phase transport through the dialysis membrane, but it does prevent liquids from crossing the membrane and impacting gas flow. The membrane of the probe was periodically dipped into an aqueous solution of ethanol, isobutyl alcohol, or toluene and was held in solution for 20, 15, 10, and 5 s with 45 s in between immersions. Configuration 2: Microdialysis Probe Interfaced Directly to a Fast GC System. Sample flow from the probe was plumbed into a sixport diaphragm valve with 1/16 in. fittings (Valco Instruments Co. Inc., Houston, TX). The valve was configured for loop injection. The diaphragm valve was switched to the load position for a 50 ms time interval, resulting in partially filling the sample loop for this time and then immediately injecting the contents into the GC column. The flow of gas-phase extracts into the injector was 0.4 mL/min, resulting in an injection volume of 0.33 µL. Approximately 3 m of a polar DB-1701 column (0.25 mm i.d., 0.25 µm film; J&W Scientific, Folsom, CA) was utilized for separations, and the FID was used as the detector. Separations were performed isothermally at either 40 or 75 °C with a carrier gas flow rate of 3 mL/min. Control of the valve and data acquisition was performed by a program designed in-house using National Instruments LabVIEW 7.1. Reagents/Sample Preparation. Helium to control the valve, helium to flow through the probe, and the GC-FID gases (air, hydrogen, and high-purity helium) were obtained from Oxygen Service Company. Reagent grade hexanol, ethanol, isobutyl alcohol, toluene, and m-xylene were obtained from Aldrich Chemical Co. (Milwaukee, WI).

Figure 2. Plots generated by exposing the microdialysis probe to aqueous sample solutions for different amounts of time (time indicated above peaks). The gas-phase dialysates were sent directly to the FID. Solid line ) 200 mM ethanol, dashed line ) 100 mM toluene, dotted line ) 400 mM isobutyl alcohol.

Aqueous samples were prepared in 20 mL vials with open top screw caps containing a Teflon-lined septum. With the use of a 1.0 or 100.0 µL syringe, analytes were added to 10.0 mL of deionized water or other aqueous matrix that was pipetted into the vial. Gas-phase samples for recovery experiments were prepared by injecting appropriate volumes of neat analytes into a 1.0 L Tedlar gas sample bag (Alltech, Deerfield, IL). RESULTS AND DISCUSSION Response Time. The performance of microdialysis membranes for extractions of volatile analytes from aqueous samples into the gas phase has two main advantages when compared to extractions performed using traditional silicone and other nonpolar membranes. First, the transport time through the membrane is much faster, giving faster response. Second, polar analytes are transported to a much greater extent than has been shown previously using a nonpolar membrane such as PDMS.2,10 Figure 2 illustrates these performance features. The figure shows profiles generated by interfacing the microdialysis probe directly to the FID. As in many membrane extraction experiments1,15 the membrane is exposed to the sample for a period of time and then removed. Twelve such experiments, representing triplicate analysis of four different exposure times run consecutively, are shown in the figure. Separate experiments were performed for ethanol, isobutyl alcohol, and toluene, and the results are overlaid. The fact that similar maximum levels are reached, regardless of how long the microdialysis probe was exposed, indicates that steady-state concentrations at the detector are obtained within 5 s of the probe being exposed to the sample. Measuring the rise time from 10% to 90% peak height shows values of less than 1 s, indicating an extremely short equilibration time. This is much faster than typical membrane extractions into the gas phase, where several minutes are usually required to reach equilibrium.2,5,6,8,9 This fast response occurs for two reasons. The microdialysis membranes are much thinner (∼20 µm thickness) than silicone membranes typically used. In addition, transport occurs through

pores in the membrane material, which should be faster than diffusion though nonporous membrane materials. Not only is transport through the membrane rapid, but transport to the detector (lag time) is also very fast. In microdialysis extractions into a liquid dialysate, transport to the separation column or the detector is often limited by extremely low flows (∼1 µL/min) through the 40 µm i.d. tubing through which the dialysate flows. The use of helium as our extraction phase allows for a flow rate as high as 0.5 mL/min, which translates to transport times of 1 s or less between the microdialysis probe and the detector (or injection port for experiments using fast GC). This rapid response is potentially very valuable for monitoring of systems with rapidly changing analyte concentrations, as will be discussed in more detail below. Note also that the time required to reach steady-state concentrations in the gas phase is largely independent of the analyte. This also supports the idea that transport occurs though pores in the microdialysis membrane, so transport time should be less dependent on analyte polarity. One of the limitations of extractions into the gas phase using semipermeable nonpolar membranes is that diffusion of analytes that are more polar or larger in size can be exceedingly slow to the point where the analysis becomes impractical.2,9 Use of microdialysis membranes allows analysis of polar analytes, although it is somewhat limited in its sensitivity. Figures of Merit. Figure 2 also shows that the quantity of signal differs between analytes. We studied this further by interfacing the microdialysis extractor with a fast GC system. An example of the difference in analyte signal response is illustrated in Figure 3, which shows 10 consecutive separations of gas-phase extracts from an equimolar mixture containing five analytes. The top part of the figure shows the reproducibility of this system. There is less than 5% relative standard deviation (RSD) for all of the analytes and less than 2% RSD for all but the smallest peaks. This reproducibility likely is owed to the automated diaphragm valve injection in the fast GC. The bottom section of the figure Analytical Chemistry, Vol. 80, No. 1, January 1, 2008

125

Figure 4. Fast GC peak areas for analytes extracted from water samples at different temperatures using microdialysis probes. Error bars represent (1 standard deviation and in some cases are obscured by the data marker due to very small size.

more clearly shows the difference in signal for the five analytes. Note that, while nonpolar analytes like toluene have the largest signal, polar analytes such as isobutyl alcohol also have very strong signals. The signal levels depend largely on the extent to which analytes partition from the aqueous sample into the gas phase. This should be distinguished from the partitioning between the aqueous phase and a nonpolar membrane in previous membrane extraction studies.1-5,28,29 While partitioning between water and the gas phase obviously still depends on the polarity of the analytes, this property is not as limiting as is the case when using a nonpolar membrane.2,10 Consequently microdialysis extraction offers substantial improvements over nonpolar membranes for alcohols, aldehydes, and other volatile polar analytes. We calculated the extraction efficiency (recovery) for toluene and ethanol at room temperature by comparing the signal from our extraction experiments with calibration curves generated from gas-phase standards. Gas-phase standards were prepared and used to make a calibration curve by flowing them directly into the injection port of the fast GC system. This calibration curve was used to calculate the concentration of analyte in the gas-phase extracts of aqueous standards sampled with the microdialysis probe. Recovery is calculated by dividing the gas-phase concentration of the extracts by the aqueous phase concentration of the standard. In this way, we are defining recovery in a manner similar to other microdialysis experiments by comparing concentrations on either side of the membrane. The recovery for toluene was 1.28% and that for ethanol was 0.054%. As expected, these values

mirror the signal levels observed for extraction and separations of the equimolar mixture of these analytes in that nonpolar analytes have higher recoveries. Overall, these values are significantly lower than what has been reported for microdialysis experiments with liquid dialysates, which is usually 50% or higher.30,31 However, it is important to recognize that, unlike experiments with liquid-phase dialysates, we do not expect the concentration of the gas-phase extracts to equal those of the aqueous sample. In our experiments, concentrations of extracts are limited by the equilibrium of the analyte between the aqueous and gas phases. For example, using the Henry’s constant value of 2.0 × 102 M/atm for ethanol, the equilibrium concentration of this analyte in the gas phase above a 0.1 M aqueous solution is calculated to be 2.0 × 10-5 M. This indicates an expected recovery of only 0.02%. While this is an oversimplification that does not take into account the continuous nature of the extraction, it indicates that our recovery values are in the range that could be reasonably expected. As would be expected, more analytes are extracted into the gas phase as the temperature of the solution being sampled increases. Figure 4 shows signals for toluene, ethanol, and isobutyl alcohol over a temperature range from 0 to 37 °C. The data show that signal for ethanol and isobutyl alcohol approximately double as the temperature rises from room temperature to 37 °C (body temperature). While this indicates that detection limits are somewhat lower at higher temperatures, it also shows that sample temperature must be carefully controlled when sampling with these probes. We also performed experiments to determine the dynamic range and overall detection limits of the microdialysis extraction/ fast GC system. The response was linear for toluene, ethanol, acetone, and isobutyl alcohol over at least 3 orders of magnitude, and over at least 2 orders of magnitude for dimethyl sulfide (limited by solubility). Limits of detection were best for very volatile nonpolar species such as toluene, which could be detected with signal-to-noise levels of 3:1 at concentrations as low as 0.01 mM. Isobutyl alcohol, which leaves the water less readily, had a higher detection limit of 2 mM. Ethanol had the highest limit of

(28) Vaes, W. H. J.; Ramos, E. R.; Verhaar, H. J. M.; Seinen, W.; Hermens, J. L. M. Anal. Chem. 1996, 68, 4463-4467. (29) Zhang, Z.; Pawliszyn, J. J. High Resolut. Chromatogr. 1996, 19, 155-160.

(30) Sauernheimer, C.; Williams, K. M.; Brune, K.; Geisslinger, G. J. Pharmacol. Toxicol. Methods 1994, 32, 149-154. (31) Sun, L.; Stenken, J. A. J. Pharm. Biomed. Anal. 2003, 33, 1059-1071.

Figure 3. Series of chromatograms from an aqueous solution containing 100 mM each of ethanol, toluene, isobutyl alcohol, o-xylene, and hexanol sampled and analyzed using microdialysis sampling directly interfaced to a fast GC system. One GC injection/ min. Top: 10 consecutive injections. Bottom: first injection. See the Experimental Section for chromatographic details.

126 Analytical Chemistry, Vol. 80, No. 1, January 1, 2008

detection at 10 mM (all detection limits calculated at room temperature). These concentrations are at or below the concentrations at which these and other volatile analytes are measured in physiological samples,20-24,32 so the technique has applications at these detection limits. For nonpolar analytes, these limits of detection are somewhat lower than those measured for gas-phase extractions using nonpolar semipermeable membranes.2,33 While the microdialysis membranes offer very fast permeation of analytes, they do not concentrate analytes like the semipermeable membranes do. Recoveries may be enhanced by increasing the temperature of the sample, reducing the pressure of the gas inside the probe, or perhaps with chemical enhancement of the extractant phase. Overall detection limits may be achieved by interfacing the microdialysis extractors with more sensitive gas-phase detection systems34 or by using trapping techniques to concentrate the extracted analytes.1,3,4,35 Such experiments are underway in our laboratory. The performance of microdialysis probes show little deterioration or change in analyte transport time when exposed to most aqueous sample matrixes. Aqueous matrixes containing salts or proteins (10% solutions of NaCl or bovine albumin) and 1% H2SO4 were compared to samples in pure water to determine impact on signals. The salt and acid matrixes resulted in signals for alcohols that were elevated by ∼2× compared to a pure water matrix, possibly due to a decreased solubility. The protein matrix gave similar signals for alcohols compared to water, but toluene signals decreased by nearly a factor of 10 (see the Supporting Information for data). This may be a result of proteins filling the pores of the membranes, which could reduce the transport efficiency of nonpolar analytes. After prolonged exposure to high (10%) concentrations of protein, a film could be observed on the probe, and probe response in subsequent analyses of cleaner matrixes was also 10-30% lower. This was not observed when the probes were used in 1% concentrations of protein. Overall, the results indicate that the probes should be viable for extraction in physiological systems as long as the matrix stays consistent. In systems where the matrix may change over the course of an experiment, an internal standard may be necessary. While the probes are somewhat susceptible to leaks if the input and output tubes into the microdialysis membrane are not supported, overall they are quite rugged and can be expected to hold up to difficult sample environments, as has been previously shown for microdialysis probes with liquid dialysates.15,16,18 We have used the same probe for several weeks in aqueous solutions containing up to 1 M concentrations of salts, proteins, and organic analytes. The cellulose membranes will disintegrate after prolonged exposure to strong acid, but we have successfully used them for measurements in 10% H2SO4 for as long as 30 min. When not in use, the probes should be stored out of solution to prevent liquid from passing through the membrane. If liquid gets into the probes it clogs the transfer lines to the GC or FID and extracted analytes cannot flow. (32) Best, C. A.; Sarkola, T.; Eriksson, C. J. P.; Cluette-Brown, J. E.; Laposata, M. Alcohol. Clin. Exp. Res. 2006, 30, 1126-1131. (33) Mitra, S.; Zhang, L.; Zhu, N. Adv. Instrum. Control 1995, 50, 1-7. (34) Nyarady, S. A.; Barkley, R. M.; Sievers, R. E. Anal. Chem. 1985, 57, 11, 2074-2079. (35) Current, R. W.; Kozliak, E. I.; Borgerding, A. J. Environ. Sci. Technol. 2001, 35, 1452-1457.

Figure 5. Microdialysis extraction/fast GC analysis of a reaction mixture converting ethanol (from a 100 mM solution in 10% H2SO4) to acetaldehyde and acetic acid with toluene present as an internal standard. An injection of the gas-phase dialysate into the GC occurs every 10 s. Oxidizing agent (2 mL of saturated K2Cr2O7) was added 40 s into the experiment. The arrow indicates the time of addition. (a) Raw data for the entire experiment. (b) A magnified area from 45 to 115 s showing the appearance of acetaldehyde and, eventually, acetic acid.

Use in Monitoring Experiments. The rapid response of the probe, coupled with 10 s separations in a fast GC system, allow faithful monitoring of systems whose concentrations change rapidly. Figure 5 shows microdialysis probe/fast GC measurements of the oxidation of ethanol to acetaldehyde and acetic acid using dichromate ion in aqueous solution. Toluene was also added to the solution as a nonreactive standard. GC injections are made every 10 s. Note that each “peak” in Figure 5a actually corresponds to a series of peaks making up the 10 s chromatogram (see Figure 5b). The data are also represented in Figure 6, which shows the change in concentration of the reactant and products in graphical form. For the first 40 s, the data consists of repetitive measurements of ethanol and toluene. At 40 s, the dichromate oxidizing agent was added and the changes in the reaction mixture are immediately noted. Figure 5b shows the series of chromatograms recorded right after this addition. The ethanol peak decreases in size as it is oxidized to acetaldehyde, which appears and increases. Both of these changes show up immediately, indicating the very fast response of the system (lag time of less than 5 s). The last chromatogram in Figure 5b shows a tiny peak for acetic acid, an Analytical Chemistry, Vol. 80, No. 1, January 1, 2008

127

Figure 6. Plot showing the change in peak height for ethanol, its oxidation products acetaldehyde and acetic acid, and toluene (internal standard) in a reaction monitored using the microdialysis extraction system interfaced directly with a fast GC. Figure 5 shows the raw data.

oxidation product of acetaldehyde that appears at 100 s. Throughout the reaction, the peak for toluene remains essentially constant, even after the addition of the dichromate salt. This shows that changes in the intensity of signals for the reactants and products is due to actual changes in concentration in solution. These concentration changes for all of the analytes in this region are more easily seen in Figure 6, which clearly illustrates the decrease of ethanol and corresponding increases in acetaldehyde and eventually acetic acid. The reaction was temperature controlled at 30 °C in a water bath. Since the extraction is controlled mainly

128

Analytical Chemistry, Vol. 80, No. 1, January 1, 2008

by the extent to which the analyte partitions between the aqueous and gas phases, temperature is a very important parameter to control. This example demonstrates the power of the microdialysis to monitor rapid changes in the concentration of volatile analytes in a complex aqueous matrix. In this case, changes in salt concentration with the addition of dichromate did not impact the extent to which analytes were extracted by the probe, but an internal standard could correct any such changes that occurred if more drastic matrix changes occurred in the system during monitoring. As previously discussed, the response time of the system to changes in concentration is on the order of 5 s, which is remarkably fast. This evidence, together with the very small size of the probes, indicates a great potential for a system like this to be used for monitoring volatiles in biological systems. ACKNOWLEDGMENT This research was supported in part by the University of Minnesota Research Site for Educators in Chemistry (RSEC), which is funded in part through NSF-CHE-0113894. For ARB. SUPPORTING INFORMATION AVAILABLE Additional information as noted in text. This material is available free of charge via the Internet at http://pubs.acs.org. Received for review July 19, 2007. Accepted October 9, 2007. AC071530H