Anal. Chem. 1981, 53, 929-931 929

H2 + 2H+ t, 2e- (anodic reaction)

2AgC1 + 2e- + 2Ag + 2C1- (cathodic reaction)

The chloridation seems to occur during the conditioning procedures. The readings obtained are linear with respect to the amount of hydrogein dissolved in solution. The response is essentially complete within 90 s. The lower detectable limit can be estimated as follows: Assuming Henry’s law holds, a solution saturated with. Hz contains 1.5 ppm of H2 and gives a reading of 5.5 divisions on the panel. The lowest level of hydrogen attempted to be read was 20% Hz which gave a reading of 1.2.

1.5 ppm 5.5 X 1.2 = 0.33 ppm of H2

LITERATURE CITED (1) Sweet, W. J.; Houchins, J. P.; Rosen, P. R.; Arp, D. J. Anal. Blochem.

1980, 107, 337-340. (2) Clark, L. C., Jr.; Bargeron, L. M., Jr. Sclence 1959, 130, 709-710. (3) Clark, L. C., Jr. U.S. Patent 3380905, April 30, 1968. (4) Olsen, R. R.; Srlnivasan, V. S. Anal. Chem. 1977, 49, 853-857.

RECEIVED for review November 24, 1980. Accepted March 3,1981. This research was supported by the Faculty Research Committee, Bowling Green State University, Bowling Green, OH 43403.

Automatic Liquid Injector for Headspace Gas Chromatography

Rein Otson

Bureau of Chemical Hazards, Environmental Health Directorate, Health and Welfare Canada, Tunney‘s Pasture, Ottawa, Ontario K1A OL2, Canada

General aspects and practical applicatilons of headspace gas analysis have been discussed in detail (1, 2). Static headspace chromatographic techniques have been successfully applied in areas such as the analysis of food and beverages (3) and the analysis of water (4 ,5 ) . When a large number of samples must be analyzed, the technique requires time-consuming manual injection of headspace aliquots into the gas chroma- tograph unless an autornated headspace sample injector, such as the system of the Moldel F45 (Perkin-Elmer Corp., Norwalk, CT) is available. Since such an injector was not available when a survey of volatile compounds present in a large number of consumer products was initiated, adaptation of available equipment, a Model 8020 liquid autosampler (Varian Asso- ciates, Inc., Palo Alto, CA), to perform automated headspace analyses was investigated.

EXPERIMENTAL SEC’TION Gas Chromatography. All analyses were performed on a

Model 5840 gas chromatograph (Hewlett-Packard Co., Palo Alto, CA) equipped with an OIV-17, SCOT stainless steel column (50 ft X 0.020 in. i.d., Perkin Elmer Corp., Norwalk, CT) attached to a Hewlett-Packard capillary column inlet system operated in the splitless mode and maintained at 180 “C. For hydrocarbon analyses, the column was1 attached to a flame ionization detector (FID) and for trihalomethane (THM) analyses it was attached to a 63Ni, electron capture detector (ECD). Both detectors were held at 300 “C, and after each injection the column oven tem- perature was maintained at 60 “C for 3 min and then raised at a rate of 8 OC/min to 150 “C where it was held for 6 min. Nitrogen gas, passed through Oxiclear (Labclear, Oakland, CA) and mo- lecular sieve traps, was uiaed for column carrier gas, at 4 mL/min flow rate, and detector makeup gas, at 25 mL/min flow rate.

Autosampler. A Varian Model 8020 autosampler was inter- faced with the gas chromeitograph and was operated in the manner prescribed in the autosampler instruction manual. Zero grade nitrogen gas was used to operate the autosampler pneumatic system. Autosampler vials (2-mL capacity) sealed with Teflon- coated silicone disks and screw caps were used for preparation of samples for headspace analysis. A rinse volume of about 200 p L was used for both sample and rinse vials.

Reagents. The purity of all organic reagents was determined by gas chromatography. Composite stock solutions of four hy- drocarbons (Chem Service, Inc., West Chester, PA) in nitro- benzene (Fisher Scientific Co., Pittsburgh, PA) were prepared to contain hexane, benzene, toluene, and o-xylene in the con- centrations shown for sainples 1-5 in Table I. Three additional stock solutions, respectively containing lo9 of the hydrocarbon

concentrations for samples 1, 3, and 5, were also prepared for quantitation of hydrocarbons in headspace aliquots. Aliquots of methanolic (Caledon Laboratories Ltd., Georgetown, Ontario, Canada) solutions containing 0.20 mg of four trihalomethanes (Chem Service, Inc.) per milliliter of solution were injected into trihalomethane (THM) free water (2) to obtain aqueous, composite stock solutions. The aqueous solutions contained chloroform, bromodichloromethane, chlorodibromomethane, and bromoform at the concentrations corresponding to samples A-E in Table 11. Three composite solutions containing 5,10, and 50 ng/mL of each of the four trihalomethanes in hexane (Caledon Laboratories LM.) were also prepared for determination of THMs in headspace aliquots. A variety of consumer products, such as paints, paint removers, automotive and household cleaners, wood preservatives and sealants, and contact cements, were obtained.

Procedure. Sample, blank, and rinse vials for headspace analyses were prepared by injecting 200-pL aliquots of composite hydrocarbon solutions, consumer products, or nitrobenzene and 250-pL aliquots of THM solutions, THM free water (containing 0.1% of methanol by volume), methanol, or hexane into sealed autosampler vials. The storage time between aliquot injection and headspace analysis and the ambient temperature at the autosampler were recorded. Sealed autosampler vials containing ambient laboratory air only were used for air blank analyses and air rinses. Headspace aliquots of 5 pL volume were used both for the autosampler injections and for the manual injections performed by means of a 1 0 - ~ L Hamilton syringe.

Results from gas chromatographic analysis of aliquots of the solutions were used to construct calibration curves. These plots of peak area against the calculated amount of compound injected allowed estimation of the concentration of organics in headspace aliquots.

RESULTS AND DISCUSSION The Model 8020 autosampler principles of operation sug-

gested that it could be used to sample and inject headspace above samples in the autosampler vials. It was shown that headspace rather then liquid was sampled and injected if less than 300 p L of consumer product was added to a vial. Volatile organic compounds in the consumer products were thus readily detected by the automated headspace chromatography technique. For some products, contamination of the auto- sampler transfer and injection system by compounds in the headspace caused a “memory effect”. The “memory” for volatile compounds was reduced when the contaminated system was rinsed with air from an empty, sealed vial. The analysis of some consumer products caused such high

0003-2700/81/0353-0929$01.25/0 Published 1981 by the Amerlcan Chemical Society

930 ANALYTICAL CHEMISTRY, VOL. 53, NO. 6, MAY 1981

Table I. Typical Headspace Analyses Results for Composite Hydrocarbon Solutions -__.

hydrocarbon concentration (mg/mL), mean peak areaa (pV s), and RSD (%)

hexane benzene toluene o-xylene -___ fiv s P V s P V s PV S

air blank 9.5 5.2 1.8 22 0.5 32 <0.1

sample type mg/mL x l o 3 % mg/mL x l o 3 % mg/& x l o 3 % mg/mL x l o 3 %

Autosampler

0.2 90 nitrobenzene blank 10.1 28 1.6 29 0.7 33 1 0.33 0.9 170 0.44 2.2 14 0.43 <0.1 0.44 <0.1 2 3.30 39.0 5.9 4.39 9.0 5.0 4.33 1.7 11 4.41 0.2 89 3 16.5 180 4.7 22.0 35.9 6.4 21.7 9.1 17 22.0 1.3 58

2.9 30 4 33.0 336 3.0 43.9 81.6 4.2 43.3 20.0 5.3 44.1

2-IC 3.30 46.9 4.0 4.39 10.6 4.0 4.33 2.5 4.6 4.41 <0.1 3.30 28.4 1.2 4.39 8.6 1.8 4.33 1.7 11 4.41 <0.1

5 65.9 495 3.0 87.8 136 3.8 86.6 29.8 3.4 88.1 3.5 37

2-11 d

air blank <0.1 <0.1 <0.1 < O . l nitrobenzene blank <0.1 <0.1 <0.1 < 0.1

Manual

2-Ie 3.30 49.5 3.1 4.39 11.7 4.7 4.33 2.3 12 4.41 0.8 21

six vials stored for 3-8 h. one vial analyzed by manual injection before autosampler analyses.

Table 11. Typical Results for Headspace Analyses of Aqueous Trihalomethane Solutions

a For three sample vials stored for 13-25 h, except where noted otherwise. For five vials analyzed at random. For Reanalysis of headspace in sample 2-1 vials at 28-33 h after vial preparation. e Contents of

trihalomethane concentration (ng/mL), mean peak areaa (pV s), and RSD (%)

CHCl, CHBrCl, CHBr,CI CHBr, fiv s PV s fiv s PV s

sample type ng/mL x l o 3 % ng/mL x l o 3 % ng/mL x lo3 % ng/mL x l o3 %

Autosampler water blank 22 3.0 < 1 <1 < 1 A 50 67 1.3 10 14 4.3 5 4.3 14 5 <1 B 100 106 5.3 20 32 8.7 10 9.5 8.1 10 < l b C 150 140 3.5 50 86 12 50 51 14 50 > I b D 200 183 11 150 308 1.6 150 162 1.4 150 40 5.7 E 300 264 14 300 613 8.5 300 348 0.9 300 93 21

Manual water blank <1 <1 < 1 <1 CC 150 142 4.8 50 86 6.2 50 24 29 50 15 26

a For three sample vials stored for 2-7 h. Estimated, since improper integration setting. Contents of one vial analyzed by manual injection before autosampler analyses.

“memory” values that several rinses were required to prevent interference with subsequent sample analyses. Multiple rinses were readily incorporated during automated analyses.

Equipment design, sample type, and factors, such as the storage time and temperature of sample vials in autosampler trays, might affect the reliability of analytical results. The feasibility of using the equipment to perform reliable, auto- mated, qualitative, and quantitative headspace analyses was therefore evaluated. Since our interests included the analysis of potable water as well as the analysis of consumer products, two types of solutions (sample) were used for the investiga- tions. Aqueous solutions (samples) containing four commonly found trihalomethanes were used to test application of the automated headspace gas chromatography technique to po- table water analyses. Solutions (samples) containing four hydrocarbons were used to test the technique for the deter- mination of hydrocarbons. Nitrobenzene was used as the solvent since it has low vapor pressure and did not interfere with gas chromatographic analyses.

Hydrocarbon Analyses. The hydrocarbon concentrations in the sample vial headspace reached maximum values after about 2 h of sample storage a t 25 f 1 OC. Comparison of results for samples stored for 3-8 h (2-1, Table I) with those for samples stored for 13-25 h (2, Table I) shows that some decrease in headspace concentration occurred for three of the

hydrocarbons during storage over about 20 h. A decrease in headspace hydrocarbon concentration was also observed after reanalysis of samples as shown by comparison of results for samples 2 and 2-11 (28-33 h storage). For sample 2-11, the loss of hydrocarbons from the vial was likely due both to leakage from the sealed vial and to material lost during the initial analysis of sample 2.

The precision of headspace analyses of triplicate samples which had been stored for 10-25 h was usually better than 10% RSD (Table I). Precision in most instances improved to better than 5% RSD if the storage times of triplicate samples were within 5 h of each other. Precision suffered at low concentrations, due to inaccurate area integration of small peaks. Hexane, benzene, toluene, and o-xylene, respectively, could be detected in headspace aliquots when their concen- trations in the nitrobenzene solution were about 0.1,0.1,0.5, and 5 mg/mL. A plot of hydrocarbon concentration in so- lution against hydrocarbon peak area for the headspace aliquot gave a straight line for hexane concentration up to about 30 mg/mL and up to about 80 mg/mL concentration for the other three hydrocarbons.

Hydrocarbon “memory” values were reduced to, and maintained at, the levels shown for blank values in Table I, when the contaminated system was rinsed with air or nitro- benzene headspace after each composite solution sample

ANALYTICAL CHEMISTRY, VOL. 53, NO. 6, MAY 1981 931

parison of headspace analysis results for aqueous THM so- lutions with analytical results for hexane solutions of THMs allowed estimation of THM concentration in the headspace of aqueous solutions. THM concentrations ranging from about 1 to 40 ng/mL of headspace were detected and quantitated for the samples in Table 11.

Remarks. The commercially available autosampler, which is commonly used for injection of liquid samples, was readily applied to perform automated headspace injections. Antic- ipated mechanical problems, due to the lack of syring barrel lubrication, did not occur with more than 1000 headspace injections. Since the sample vials cannot be warmed, the technique is limited to the determination of liquid sample components which, a t ambient temperature, exert sufficient vapor pressure to allow their detection in the headspace gas. This limitation was demonstrated by the hydrocarbon analyses results where relatively poor sensitivity was obtained for o- xylene as compared to the more volatile hexane. Small changes in sample temperature did not have a significant effect on the concentration of volatile components in the headspace, in view of the analytical precision. Since ambient temperature control in most modern laboratories is comparable to that observed during this study (fl “C), the automated headspace analysis technique should be generally applicable with similar analytical precision. The technique allowed efficient screening of a large number of samples, such as consumer products, for the occurrence of volatile organic components. Only the preparation of sample and rinse vials required personnel time commitment.

The precision, linearity, and detection limits obtained for analyses of solutions containing hydrocarbons and trihalo- methanes suggest that the automated technique may be generally useful in qualitative and quantitative analytical studies. Results obtained elsewhere (4) for a manual head- space injection gas chromatography technique were similar in precision and detection limit values obtained here for the determination of trihalomethanes in water. Evaluation of the automated technique for the determination of a wider range of volatile organics in water is therefore suggested. Analytical conditions, such as, sample and headspace volumes and com- ponent concentrations, can be adjusted to suit the technique to a particular need. In fact, the autosampler design allows injection of the total purge volume, i.e., more than 250 pL. Practical and theoretical considerations for determining headspace analysis conditions have been discussed elsewhere (1,2).

ACKNOWLEDGMENT Helpful comments by David T. Williams are gratefully

acknowledged. The manuscript was reviewed in-house by David T. Williams and Ih Chu and was typed by Jean Ireland.

LITERATURE CITED (1) Hachenberg, H.; Schmidt, A. P. “Gas Chromatographic Head Space

Analysis”; Heyden: New York, 1977. (2) Drozd, Y.; Novak, Y. J. Chromtogr. 1979, 765, 141-165. (3) Charalambous, G. “Anaiysls of Food and Beverages: Head Space

Techniques”; Academic Press: New York, 1978. (4) Otson, R.; Williams, D. T.: Bothwell, P. D. Environ. Scl. Techno/.

analysis. No detectable hydrocarbon peaks were observed for blank analyses obtained by manual syringe injection (Table I), Results from manual injection of headspace in a vial containing sample 2-1 were similar to those obtained by au- tosampler injection. A calibration curve plotted from the results of analysis of standard solutions containing hydro- carbons in the microgram per milliliter concentration range allowed estimation of hydrocarbon content in the headspace aliquots for samples. It, was estimated that for samples 1-5 (Table I) roughly 1-1000 ng of hexane, 1-100 ng of benzene, <1-40 ng of toluene, and <1-4 ng of o-xylene were present in 5-pL headspace aliquots.

Trihalomethane Analyses. The THM concentrations in the sample vial headspace reached maximum values after about 3 h of storage at 25 f 1 “C. During the following 20 h of storage, the concentration did not usually change by more than i 5 % . Less than 50% of the original THM values were obtained upon reanalysis of the headspace in a sample vial which had been stored for 3 h after the original automated analysis.

The precision of results from headspace analyses of trip- licate samples which had been stored for 2-7 h (Table 11) was generally better than 1.0% RSD for TEIM concentrations ranging from 10 to 300 ng/mL. Results in Table I1 were obtained in one experiment when the samples were analyzed in random order. Chloroform determinations in the 10-50 ng/mL concentration range (not shown) showed an RSD of less than 10% when the sample transfer and injector system had been rinsed several times to decrease the chloroform “memory” well below the value shown for the water blank in Table 11. For all THMs, the precision of analytical results was improved when replicate samples bad been stored for identical time periods. Ihaccurate peak area integration, due to inadequate microprocessor commands, was a suspected cause of some high RSlD values when peak heights showed good similarity.

A plot of THM concentration in solution against THM peak area for the 5-pL headspace aliquot gave a straight line for THM concentrations between 5 and 300 ng of THM/mL of solution. Trihalometlhanes could be detected in 5-pL headspace aliquots when their concentration in 250-pL aqueous samples was about 5 ng/mL for CHC1, CHBrC12, and CHBr2Cl and about 10 ng/mL for CHBrS. The detection limit could be improved by optimization of chromatographic con- ditions and microprocetisor integration commands or by the analysis of larger headspace aliquots.

A trihalomethane “memory effect” was also observed during automated headspace analyses of aqueous THM solutions. Inconsistent reduction of “memory” THM values occurred after the system was rinsed with air, methanol vapor, or hexane vapor after each sample analysis. However, cross contamination was effectively reduced for three of the THMs when THM free water headspace rinses were incorporated between sample analyses (Table 11). The chloroform “memory” remained significant (22 X lo3 pV s), but consistent, throughout more than 60 headspace analyses of aqueous so- lutions containing THMk in the concentration range, 25-300 ng/mL. No detectable ‘THM peaks were observed for blank analyses performed by manual injection (Table 11). Results from manual injection of headspace in a vial containing sample C were similar to those aiubsequently obtained within 2 h for the same vial by the autosampler injection technique. Com-

1979, 13, 936-939. (5) Bush, B.; Narang, R. S.; Syrotynski, S. Bull. Envlron. Contam. Toxi-

col. 1977, 78, 436-441.

RECEIVED for review November 26,1980. Accepted January 29, 1981.