An Automated System for Near-Real-Time Monitoring of Trace

A new gas chromatographic method developed to quantitatively determine important atmospheric halocarbons is described. Target compounds include ...
0 downloads 4 Views 122KB Size
Anal. Chem. 1998, 70, 958-965

An Automated System for Near-Real-Time Monitoring of Trace Atmospheric Halocarbons Matthew R. Bassford,* Peter G. Simmonds, and Graham Nickless

School of Chemistry, Cantocks Close, University of Bristol, Bristol BS8 1TS, U.K.

A new gas chromatographic method developed to quantitatively determine important atmospheric halocarbons is described. Target compounds include replacement CFCs, chlorinated solvents, and biosynthesized (naturally produced) organohalogens, all trace gases in the atmosphere at concentrations ranging from 0.1 to 600 pptv (where pptv ) 1 part in 10-12 by volume). A combination of ultralow concentrations and relatively small electron attachment cross sections renders these compounds very difficult to routinely measure in the background air typical of remote atmospheric monitoring stations. Detection is achieved by preconcentration of a 200-mL air sample using an adsorbent-filled microtrap and enhancement of electron capture detector response by oxygen doping one of two detectors connected in series. Oxygen doping specifically targets halocarbons with relatively poor electron attachment rate coefficients. The work described here details construction of a novel analytical system, laboratory trials, and optimization followed by an extended field campaign at a remote atmospheric monitoring station, Mace Head, Ireland. A calibration standard or ambient air sample was acquired every hour using a cyclic, automated procedure without employing cryogenic preconcentration or refocusing. Overall precision of the analytical method for the target compounds is between 0.3 and 1.5%.

ODP. Recent publications on atmospheric measurements reveal that the rate of accumulation of CF2Cl2 (CFC-12) has decreased and concentrations of the principal halocarbons, CFCl3 (CFC-11), CFCl2CF2Cl (CFC-113), CH3CCl3, and CCl4, and the halons, CF2BrCl and CF3Br, are now in decline.3-6 The detection and accelerating abundance of replacement compounds in the atmosphere has also been reported.7,8 Long-term monitoring of all these species is important to verify the expected decrease in atmospheric halogen burden as a consequence of the Montreal Protocol and to assess the environmental impact of the new substitutes, especially since many HCFCs and HFCs are effective greenhouse gases with high global warming potentials.9 In addition to anthropogenic sources, there are a variety of halocarbons known to be biogenically produced. Although CH3Cl is the most abundant biosynthesized organohalogen, there are many other natural halocarbons that contribute to the total halogen budget. These provide a significant contribution to the total atmospheric halogen load and are synthesized predominately by oceanic biota10-13 and by fungi14 or released during biomass burning.15-17 Detailed information about the sources, sinks, and characterized emission pattern on diurnal, seasonal, and annual cycles for many of these naturally occurring halocarbons is sparse and high-frequency, high-precision measurements are needed to quantify their global atmospheric budgets. CH3Br is now thought to play an important part in stratospheric ozone depletion, and consequently, there is immense interest in quantifying fluxes

Production of halocarbons by the chemical industry is now restricted under terms laid out in the Montreal Protocol and subsequent revisions.1 Manufacture and use of specific halocarbons, chlorofluorocarbons (CFCs), halons (bromine-containing compounds used in fire extinguishing), and certain chlorinated solvents is now prohibited (except for very small “special use” applications). CFCs have been replaced by a group of interim surrogate compounds, the hydrochlorofluorocarbons (HCFCs) which have a labile C-H bond along with lower chlorine and bromine content. Consequently, tropospheric destruction of HCFCs is more rapid than their CFC predecessors, and the ozone depletion potential (ODP) is smaller;2 however, HCFCs are in turn being replaced by hydrofluorocarbons (HFCs), which have zero * Corresponding author: (e-mail) [email protected]; (fax) +44 117 9251295. (1) World Meteorological Organisation, Scientific Assessment of Ozone Depletion: 1991. Global Ozone Research and Monitoring Project; WMO, Geneva, 1992; Report No. 25. (2) Solomon, S.; Albritton, D. L. Nature 1992, 357, 33-37.

(3) Cunnold, D.; Fraser, P.; Weiss, R. F.; Prinn, R.; Simmonds, P. G.; Miller, B. R.; Alyea, F. N.; Crawford, A. J. J. Geophys. Res. 1994, 99, 1107-1126. (4) Butler, J. H.; Elkins, J. W.; Hall, B. D.; Cummings, S. O.; Montzka, S. A. Nature 1992, 359, 403-405. (5) Simmonds, P. G.; Cunnold, D. M.; Dollard, G. J.; Davies, T. J.; McCulloch, A.; Derwent, R. G. Atmos. Environ. 1993, 27A, 1397-1407. (6) Montzka, S. A.; Butler, J. H.; Myers, J. C.; Thompson, T. M.; Swanson, T. H.; Clarke, A. D.; Lock, L. T.; Elkins, J. W. Science 1996, 272, 1318-1322. (7) Montzka, S. A.; Myers, R. C.; Butler, J. H.; Elkins, J. W. Geophys. Res. Lett. 1994, 21, 2483-2486. (8) Oram, D. E.; Reeves, C. E.; Sturges, W. T.; Penkett, S. A.; Fraser, P. J.; Langenfelds, R. L. Geophys. Res. Lett. 1996, 23, 1949-1952. (9) Daniel, J. S.; Solomon, S.; Albritton, D. L. J. Geophys. Res. 1995, 100, 12711285. (10) Nightingale, P. D.; Malin, G.; Liss, P. S. Limnol. Oceanogr. 1995, 40, 680689. (11) Tokarczyk, R.; Moore, R. M. Geophys. Res. Lett. 1994, 21, 285-288. (12) Scarratt, M. G.; Moore, R. M. Mar. Chem. 1996, 54, 263-272. (13) Klick, S.; Abrahamsson, K. J. Geophys. Res. 1992, 97, 12683-12687. (14) Harper, D. B. Nature 1985, 315, 55-57. (15) Blake, N. J.; Blake, D. R.; Sive, B. C.; Chen, T. Y.; Rowland, F. S.; Collins, J. E. J. Geophys. Res. 1996, 101, 24151-24164. (16) Rudolph, J.; Khedim, A.; Koppmann, R.; Bonsang, B. J. Atmos. Chem. 1995, 22, 67-80. (17) Mano, S.; Andrae, M. O. Science 1994, 263, 1255-1256.

958 Analytical Chemistry, Vol. 70, No. 5, March 1, 1998

S0003-2700(97)00861-5 CCC: $15.00

© 1998 American Chemical Society Published on Web 02/03/1998

especially because the current estimation of sinks exceeds the estimated sources.18-19 Proposed sources of CH3Br include biomass burning, fumigant use, leaded gasoline, and oceanic organisms. The electron capture detector (ECD), is a highly sensitive detector routinely used for direct determination of certain halocarbons without preconcentration; an air sample (3 mL) is taken and compounds are separated on a packed column.3 This method is suitable for those halocarbons with inherently large cross sections for thermal electron attachment and sufficiently high atmospheric abundance. Continuous monitoring of other target compounds presents a formidable analytical challenge arising from very low atmospheric concentrations, augmented in some cases by a weak ECD response. Direct chromatographic analysis is thus precluded, and sample preconcentration is necessary prior to analysis. Methods of preconcentrating ambient air samples normally involve trapping the halocarbons cryogenically (typically at -100 °C) followed by GC separation coupled with ECD or mass spectrometric detection.20,21 However, the supply of liquid cryogen is costly and presents logistic difficulties when monitoring in remote locations. Another disadvantage to cryogenic trapping is the large amount of water collected during sampling, which induces chromatographic difficulties. The method described here uses a microtrap to adsorb an air sample at ambient temperature, followed by thermal desorption directly onto a capillary column, and is based on methods our laboratory has developed.22-24 A judicious choice of adsorbents combined with near-capillary dimensions of the trap facilitate rapid, one-stage desorption and bypass the need for cryogenic refocusing. Compounds such as CH3Cl, CH2Cl2, and CHClF2 (HCFC-22), exhibit a weak ECD response that prevents ECD measurement even after sample enrichment. Grimsrud and Miller25,26 demonstrated that the response of certain compounds can be enhanced by the presence of oxygen in the detector. Since this discovery, “oxygen doping” of either the carrier gas or makeup gas to the detector has successfully been used in other studies to sensitize detection of CH3Cl, CH2Cl2, CH3Br, and CHClF2.20,24 Increased response is caused by oxygen catalyzing electron capture in the following reaction scheme:

e- + O2 T O2O2- + AB f O2 + A + B(compared with normal associative ECD reaction: e- + AB f AB- ). (18) Butler, J. H.; Rodriguez, J. M. The Methyl Bromide Issue; Wiley & Sons: New York, 1996; Chapter 2. (19) Anbar, A. D.; Yung, Y. L.; Chavez, F. P. Global Biogeochem. Cycles 1996, 10, 175-190. (20) Kerwin, R. A.; Crill, P. M.; Talbot, R. W.; Hines, M. E.; Shorter, J. H.; Kolb, C. E.; Harriss, R. C. Anal. Chem. 1996, 68, 899-903. (21) Chen, L.; Makide, Y.; Tominaga, T. Chem. Lett. 1994, 3, 571-574. (22) Simmonds, P. G.; O’Doherty, S. J.; Nickless, G.; Sturrock, G. A.; Swaby, R.; Knight, P.; Ricketts, J.; Woffendin, G.; Smith, R. Anal. Chem. 1995, 67, 717-723. (23) O’Doherty, S. J.; Simmonds, P. G.; Nickless, G. J. Chromatogr. 1993, 657, 123-129. (24) Sturrock, G. A.; Simmonds, P. G.; Nickless, G. J. Chromatogr. 1995, 707, 255-265. (25) Miller, D. A.; Grimsrud, E. P. Anal. Chem. 1979, 51, 851-858. (26) Grimsrud, E. P.; Miller, D. A. Anal. Chem. 1978, 50, 1141-1145.

EXPERIMENTAL SECTION Instrumentation. Instrumentation consisted of a gas chromatograph (GC) (model 6890, Hewlett-Packard Ltd.) equipped with two ECDs and modified to direct the effluent from the first ECD into the second detector. The two detectors were connected by a short length of deactivated fused silica (60 cm × 0.53 mm i.d., SGE Ltd.). Makeup gas for the first ECD was N2 (BOC Gases), and a mixture of oxygen (2% v/v) in nitrogen (BOC Gases) was used as makeup gas for the second ECD. This furnished the second detector with the low concentration of oxygen (0.2% v/v) required to increase ECD response to weakly electron capturing species; previous studies have shown 0.2% O2 to be the optimum concentration.24,25 This unique serial detection system was designed to be extremely sensitive for analysis of both strongly and weakly electron capturing species. The arrangement means that each detector can be independently optimized and provide a high degree of selectivity and sensitivity for a wide range of halocarbons. An automated front-end, adsorption-desorption system (ADS) was constructed based on that described by Simmonds22 to perform routine analysis of air samples and calibration standard runs in a continuous 2-h cycle. Figure 1 is a schematic illustration of the ADS. The microtrap design is similar to that described by Sturrock,24 with near-capillary dimensions and rapid resistive heating (from ambient temperature to 230 °C in under 3 s) maintaining good chromatographic peak shape. Three carbon-based adsorbents were packed into the trap in the following order: Carbotrap (5 mg), Carboxen 1003 (4 mg), and Carboxen 1000 (3 mg) (Supelco, Bellefonte, PA). The three-stage trap was required in order to trap the least volatile compounds on the weakest adsorbent, and most volatile compounds on the strongest adsorbent. This ensured quantitative trapping and desorption of the range of target components present in ambient air, and the halocarbons detected in a nominal 200-mL air sample are listed in Table 1. All valves were electronically actuated with purged housings (Valco Instruments Inc, Houston, TX) to minimize contamination. The air sample or standard mixture passed through a Nafion drier, counterpurged with dry nitrogen gas, to remove water prior to adsorption on the trap. Sample desorption was facilitated by backflushing the trap with helium (2 mL min-1, grade 6.0, Air Products Ltd.) during resistive heating. Operational details are contained in the Simmonds reference.22 ECD Optimization. Electron capture detector response is very temperature dependent, especially when the detector is oxygen doped. To optimize signal-to-noise (S/N) response, a gas standard was prepared containing a mixture of components (CF2Cl2, CFCl3, CF2BrCl, CF3Br, CHCl3, CH3Br, CH2Cl2, CHClF2, CH3Cl) present at approximately atmospheric concentrations in order to replicate real air analysis. These halocarbons display a wide range of thermal electron attachment coefficients and consequently large differences in relative ECD response. Preconcentration of the gas standard was performed using the ADS, and detector temperatures varied between 250 and 350 °C. At each temperature, the mean signal response of four runs was calculated and then S/N ratios were obtained by dividing peak height by noise magnitude. Analytical Chemistry, Vol. 70, No. 5, March 1, 1998

959

Figure 1. Schematic diagram of adsorption-desorption system.

Quantitative Atmospheric Monitoring. Once variable parameters had been optimized and retention times of the target compounds determined, the analytical system was deployed at the atmospheric monitoring station at Mace Head, Co. Galway, Ireland (54°N, 10°W) from June 28 to August 8, 1996. This was part of the Atmospheric Chemistry Studies in an Oceanic Environment (ACSOE) field campaign, funded by a NERC thematic program. Air was drawn into the ADS through a 1/4-in.-o.d. stainless steel tube at a sampling height of 7 m. The sample lines were purged for at least 10 min (at 200 mL min-1) prior to sample acquisition on the microtrap, when the sampling flow was reduced to 30 mL min-1. To quantify atmospheric measurements and allow for any sensitivity drift, alternate analyses of a calibration standard and an ambient air sample were repeated every 2 h. Each air measurement was calibrated by interpolating the average concentration of two bracketing standards. A gravimetrically prepared calibration standard was obtained containing 16 atmospheric halocarbons present at ppm concentrations with a stated accuracy of (1%, (Linde Gases). An accurately known aliquot (250 µL) of this primary standard was injected into an electropolished 36 liter stainless steel flask and diluted volumetrically using ultrahigh-purity zero air (Air Products Ltd.). This yielded a 16-component surrogate air standard with nearambient concentrations, which are given in Table 4. Prior to injection of primary halocarbon standard, the internal walls of the flask were further deactivated by introducing water (250 µL, double distilled, helium sparged) to the flask. The dilution system is ultraclean and based on a design by the Scripps Institute of Oceanography;3 all gas lines and components are stainless steel 960

Analytical Chemistry, Vol. 70, No. 5, March 1, 1998

and built to withstand the high pressures that are encountered in standard preparation. The final calibration standard was later compared with the absolute calibration scale that the Scripps Institute of Oceanography maintains. For those compounds known or suspected to be unstable in a gaseous mixture at very low concentrations, a liquid calibration standard was prepared by performing a volumetric dilution of pure components into methanol (HPLC grade). This standard was injected directly onto the chromatographic column and the ECD response for each compound determined. Assuming that the chromatographic peak area (A) is proportional to the concentration (C) of any compound in a sample, a relationship between compound x and C2Cl4 (perchloroethylene) can be expressed thus:

Ax ) kxCx

(1)

AC2Cl4 ) kC2Cl4CC2Cl4

(2)

k is the ECD response factor and the ratio of kx to kC2Cl4 yields the relative response factor K.

K)

kx kC2Cl4

)

AxCC2Cl4 AC2Cl4Cx

(3)

In this manner, atmospheric mixing ratios can be calculated retrospectively using C2Cl4 as a surrogate standard. RESULTS Effect of Oxygen Doping on ECD Response. The application of oxygen doping dramatically enhanced the ECD response

Table 1. List of Atmospheric Halocarbons elution no.a

formula

compound

dominant source(s)

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30

CF3Cl CF3Br CHClF2 CF2Cl2 CF2ClCH3 CH3Cl CF2ClCF2Cl CF2BrCl CH3Br C2H5Cl CF3CHCl2 CFCl3 CFCl2CH3 C2H5Br CH3I CH2Cl2 CCl2FCClF2 CH2BrCl CHCl3 CH2ClCH2Cl CH3CCl3 CCl4 CH2Br2 CHCl2Br CHCldCCl2 CH2ClI CHBr2Cl CH2BrCH2Br CCl2dCCl2 CHBr3

CFC-13 halon-1301 HCFC-22 CFC-12 HCFC-142b methyl chloride CFC-114 halon-1211 methyl bromide ethyl chloride HCFC-123 CFC-11 HCFC-141b bromoethane methyl iodide dichloromethane CFC-113 chlorobromomethane chloroform 1,2-dichloroethane methylchloroform carbon tetrachloride dibromomethane bromodichloromethane trichloroethylene (TCE) chloroiodomethane chlorodibromomethane 1,2-dibromoethane perchloroethylene (PCE) bromoform

byproduct of CFC manufacture fire extinguisher replacement CFC-11, -12 aerosols, foam blowing, refrigeration replacement CFC-11, -12 biomass burning, macroalgae aerosol propellant fire extinguisher biomass burning, macroalgae, fumigant unknown replacement for CFC-11 aerosols, foam blowing, refrigeration replacement for CFC-11 oceanic phytoplankton solvent, biomass burning solvent oceanic solvent and oceanic manufacture of PVC solvent solvent oceanic oceanic degreasing oceanic oceanic oceanic degreasing and dry cleaning oceanic

a The elution number corresponds to the order of elution using the optimized chromatographic conditions. It should be noted that this numbering is used in the chromatograms, Figures 2 and 4.

for certain weakly responding compounds, which are not possible to detect in air using conventional electron capture detection. A pair of chromatographic traces obtained from a typical calibration standard run are shown as Figure 2 (the top trace is from detector 1, and the bottom from detector 2) and illustrate the different responses of the serial ECDs for a range of compounds. Peak numbers correspond to those in Table 1. It was found that the addition of oxygen also caused an increase in baseline noise associated with nonspecific electron capture. Noise was determined by the GC software (Chemstation 4.01, Hewlett-Packard Ltd.) which employs the standard method defined by the American National Standards Society (ASTM E 19.09). For compounds that exhibited an increased response, relative enhancements and signal-to-noise ratios are presented in Table 2. Compared with normal electron capture detection, the dual ECD system reduced the air sample volume required very significantly. For example, using a nondoped ECD, air sample volumes of 12 and 1.4 L, respectively, would be necessary to obtain similar signals for CH3Cl and 1,2-C2H4Cl2. Accurate quantitation is dependent on very accurate integration of chromatographic peaks, and due to the sensitivity of dual detection, a 200-mL sample proved sufficient to monitor small changes in atmospheric mixing ratios of a wide range of halocarbons. Optimization of ECD Temperature. The response of each ECD in terms of S/N ratio was studied over the temperature range 250-350 °C. For the nondoped ECD, highly responsive compounds such as CFCl3 and CHCl3 have a decreasing S/N ratio with an increase in detector temperature. More weakly electron capturing species such as CF2Cl2 exhibit a small increase in response as the temperature is raised. Little temperature depen-

Figure 2. ECD 1 and ECD 2 chromatograms for calibration standard analysis (oxygen-doped ECD is the lower trace).

dence is exhibited by CF3Br or CF2BrCl. An ECD temperature of 300 °C was therefore selected for the first detector as the best compromise in overall sensitivity. Analytical Chemistry, Vol. 70, No. 5, March 1, 1998

961

Table 2. Signal Enhancement Using Oxygen Doping peak area (Hz s)

CHF2Cl CH3Cl CH3Br CH2Cl2 1,2-C2H4Cl2

response

ECD 1

ECD 2

ECD 2/ECD 1

S/N enhance.a

12.0 31.4 252.5 179.2 78.5

1264.3 6285.7 1131.5 2369.6 1862.3

105.3 200.2 4.5 13.2 23.7

31.0 58.9 1.3 3.9 7.0

a

The signal-to-noise values are calculated by dividing the increased peak area response by the increase in baseline noise.

Table 3. GC Conditions for Analysis of 200-mL Air Sample ECD 1

makeup gas

2.0 mL min-1 (6.0 grade He)a 30 °C (8 min), 5 °C min-1 ramp to 150 °C, hold 6 min 3.0 mL min-1 carrier gas, oven temperature 160 °C, hold 5 min 25.0 mL min-1 (6.0 grade N2)

detector temp

300 °C

carrier gas flow oven temp program postrun

ECD 2

3.0 mL min-1 (2% O2/N2) 280 °C

a The column head pressure was ramped to maintain a constant carrier gas flow.

As shown in Figure 3, the oxygen-doped ECD exhibited optimum S/N values for the compounds of interest in the range 270-290 °C, with a decrease in S/N values above 300 °C. Although temperature versus response profiles for each species are different, an ECD temperature of 280 °C provides a reasonable compromise and a high degree of sensitivity for the compounds in the second, oxygen-doped detector. Chromatography. A consequence of sample enrichment and detector sensitivity is the need for high resolving power and chromatographic separation of the complex mixture of gases present in a concentrated 200-mL air sample. This is necessary for the correct identification of individual peaks and to facilitate accurate, reproducible quantitation but is by no means a trivial task because of the large number of components and wide boiling point range (-81 to +150 °C). After three capillary columns were evaluated, the column that offered the best compromise between resolution and analysis time was a gas-liquid wall-coated open tubular (WCOT) CP Sil-5 methyl silicone column (0.32 mm × 50 m, 5µm film thickness, Chrompack International BV). This column has an essentially nonpolar stationary phase (100% dimethylsiloxane), and analytes are eluted primarily in terms of their boiling point. Separation was optimized by experimenting with different temperature programs and carrier gas flow rates

Figure 3. Signal-to-noise temperature dependence for ECD 2. 962 Analytical Chemistry, Vol. 70, No. 5, March 1, 1998

to attain the best possible separation for the target compounds, and the GC conditions used for routine analyses are listed in Table 3. Even with this highly efficient capillary column, certain compounds remain incompletely resolved, namely, CHF2Cl and CF2Cl2, CFCl3 and CFCl2CH3, and CH2ClI with C2Cl3H. Although overall resolution of the low-boiling components can be improved by subambient temperature programming, this involves the use of cryogens, and research is continuing to evaluate the performance of new capillary columns. One of the benefits of serial detection is that separation of closely eluting peaks can be improved in the second detector. Strongly electronegative compounds efficiently attach electrons during passage through the first detector and produce an attenuated response in the second oxygen-doped detector. This results in a decrease in peak width and, consequently, the potential for an increase in resolution. Desorption Studies. Efficient sample recovery from the trap is important in adsorption-desorption work. To investigate this, a 200-mL standard sample was trapped and desorbed in the normal manner and then successive thermal desorptions of the same trap volume were performed. A “desorption recovery” value can then

Table 4. Standard Composition and Recovery from Microtrapa % recoveryb calibrn std concn (pptv)

detector no.

first desorptn

three desorptns

61.00 126.26 510.70 573.12 62.42 58.17 28.37 246.36 28.65 56.75 57.29 28.08 111.34 111.72 28.65 28.37

1 2 1 2 1 2 1 1 1 2 1 2 1 1 1 1

97.6 100.0 98.5 90.4 99.8 95.2 97.7 99.3 100.0 100.0 98.5 100.0 100.0 100.0 100.0 100.0

100 100 100 100 100 100 100 100 100 100 100 100 100 100 100 100

CF3Br CHF2Cl CF2Cl2 CH3Cl CF2ClBr CH3Br CF3CHCl2 CFCl3 CFCl2CH3 CH2Cl2 CHCl3 1,2-C2H4Cl2 CH3CCl3 CCl4 C2HCl3 C2Cl4

a The results in this table are the mean of three trials. b Desorption recovery is defined as (amount desorbed in a trial × 100)/(∑amounts desorbed in three trials).

Table 5. Precision of Standard Components (n ) 300) % RSD compound

ECD 1

CF3Br CHF2Cl CF2Cl2 CH3Cl CF2BrCl CH3Bra CF3CHCl2 CFCl3 CFCl2CH3 CH2Cl2 CHCl3 1,2-C2H4Cl2 CH3CCl3 CCl4 C2HCl3 C2Cl4

0.58

ECD 2

Figure 4. ECD 1 and ECD 2 chromatograms for typical 200-mL air sample analysis (oxygen-doped ECD is the lower trace).

1.33 0.17 1.27 0.23 0.79 0.76 0.27 0.56

1.11

1.41 0.49 0.85 0.69 0.92 0.15 0.20

a The signal-to-noise ratio for CH Br is similar in both detectors, so 3 percent RSD is reported twice.

be calculated as a percentage, and defined as (amount desorbed in a trial × 100)/(∑amounts desorbed in three trials). Although the majority of compounds were recovered with a single desorption, in certain cases (notably methyl chloride and methyl bromide) recovery was less than 100%, probably due to the polar nature of these compounds and formation of especially strong bonds to the adsorbent. The percentage recovery of each component of interest from a single thermal desorption is summarized in Table 4. To ensure run-to-run consistency, each thermal desorption of the microtrap was followed by two thermal cleanup cycles where the microtrap was desorbed under a helium atmosphere and vented. Successive desorptions of a standard sample confirmed that there were no residual traces of any components following three desorptions, which demonstrated that there was no sample carry-over. In these experiments, the method reproducibility is

important and laboratory studies of synthetic air analyses (sample volume 200 mL) showed that when 50 standard runs were carried out, the precision (RSD) for each of the 16 components was below 2% and in most cases