characterization of volatile organic compounds in smoke at

Journal of Toxicology and Environmental Health, Part A, 63:191–206, 2001 Copyright© 2001 Taylor & Francis 1528-7394 /01 $12.00 + .00

CHARACTERIZATION OF VOLATILE ORGANIC COMPOUNDS IN SMOKE AT EXPERIMENTAL FIRES C. C. Austin Department of Epidemiology, Biostatistics and Occupational Health, Faculty of Medicine, McGill University, Montreal, Quebec, Canada D. Wang Air Toxics Section, Analysis and Air Quality Division, Environment Canada, Ottawa, Ontario, Canada D. J. Ecobichon Department of Epidemiology, Biostatistics and Occupational Health, Faculty of Medicine, McGill University, Montreal, Quebec, Canada G. Dussault Division of Health, Safety, and Tactical Strategies, City of Montreal Fire Department, Montreal, Quebec, Canada Significant associations between firefighting and cancer have been reported; however, studies finding toxic products of combustion at municipal fires have been limited by ( 1) technical difficulties encountered at the scene of working fires, ( 2) the lack of a coherent sampling strategy, and ( 3) the absence of verified sampling methods. The objective of the present study was to characterize the presence of volatile organic compound ( VOC) combustion products in fire smoke. Air samples from experimental fires burning various materials commonly found at structural fires were collected into evacuated Summa canisters and analyzed for 144 target VOCs using cryogenic preconcentration and gas chromatography/ mass spectroscopy ( GC/ MSD) methodology. The resulting chromatograms were characterized by a small number of predominant peaks, with 14 substances ( propene, benzene, xylenes, 1-butene/ 2-methylpropene, toluene, propane, 1,2-butadiene, 2-methylbutane, ethylbenzene, naphthalene, styrene, cyclopentene, 1methylcyclopentene, isopropylbenzene) being found in proportionately higher concentrations in all experimental fires and accounting for 65% ( SD = ±12% ) by mass of total measured VOCs. Benzene, toluene, 1,3-butadiene, naphthalene, and styrene were found at higher concentrations than most other VOCs and increased with the time of combustion together with increasing levels of carbon monoxide. Benzene was found in the highReceived 7 September 2000; sent for revision 27 September 2000; accepted 8 December 2000. The authors thank the National Health Research and Development Program (NHRDP), Canada (grant 6605409858) and the Institut de recherche en santé et sécurité, du travail du Québec (IRSST) for their support during this study. Address correspondence to C. C. Austin, PhD, CIH, School of Engineering at Three Rivers, University of Quebec, PO Box 571, Station A, Montreal, Quebec, Canada H3C 2T6. E-mail: caustin@ sarec.ca 191

192

C. C. AUSTIN ET AL.

est concentrations, with peak levels ranging from 0.6 ppm to 65 ppm, while the levels of 1,3-butadiene, styrene, and naphthalene peaked at 0.1, 0.4, and 3 ppm, respectively. This study revealed that there were no new or novel, toxic nonpolar VOCs resulting from the burning of common building materials. This is important in view of the studies that have found associations between firefighting and various forms of cancer.

Guidotti (1995) and Guidotti and Clough (1992) discussed the occupational health concerns of firefighting, and the results of studies reporting chemical analysis of fire atmospheres have been summarized by McDiarmid et al. (1991), Golden et al. (1995), and Lees (1995). Chemical contaminants measured at fires include acrolein, benzene, methylene chloride, polyaromatic hydrocarbons (PAHs), perchloroethylene, toluene, trichloroethylene, trichlorophenol, xylene, formaldehyde, and a short list of gases that are of immediate concern due to their acute, toxic effects (Burgess et al., 1977, 1979; Gold et al., 1978, Treitman et al., 1980; Turkington, 1984; Brandt-Rauf et al., 1988, 1989; Jankovic et al., 1992). All of the studies conducted at actual working fires have been limited by (1) technical difficulties encountered at the scene of working fires, (2) the lack of a coherent sampling strategy, and (3) the absence of verified sampling methods. Those few studies reporting levels of a limited number of hydrocarbons at fires have used Draeger tubes, Tenax, or activated charcoal, none of which have been verified for use under the severe conditions (temperature, humidity, reactive atmospheres, multiple contaminants) of a fire. Little attention has been paid to volatile organic compounds (VOCs) present at fire sites, even though they are known to have chronic effects including kidney cancer and liver disease (Klaassen et al., 1996). The objectives of the present study were to (1) track the pattern and evolution of combustion products over time at experimental fires burning various materials found at municipal structural fires; (2) identify characteristic chromatographic “fingerprints” of combustion products generated at fires; and, (3) identify useful marker compounds. METHODS The data for this study were obtained from 15 experimental fires burned in one corner of an enclosed, concrete basement (9 × 9 × 2.2 m) of an abandoned, two-story brick house using different combustible materials including spruce wood, cardboard, plywood, a bed mattress, a foam sofa, gasoline, varsol, and solid white foam insulation. A firefighter’s service lamp was located on the ceiling in the corner opposite the fire and 1-L and 3-L Summa electropolished stainless-steel canisters fitted with a stainless-steel 2-µm Nupro prefilter that had been previously cleaned and evacuated were placed below this light, approximately 1 m above the floor. The collection of such small volumes of air minimized soot loading and possible desorption of benzene from particulates trapped by the stainless-stee l prefilter or adsorption onto the particulates of higher boiling point com-

CHARACTERIZATION OF VOCs IN FIRE SMOKE

193

pounds (Wilshire et al., 1994). The area was ventilated for approximately 15 min prior to experiments, at which point a field air blank was collected. “Grab” air samples (canisters filling to atmospheric pressure in approximately 20 s) were collected from each of the 15 experimental fires at the 2-, 5-, 10-, and 15-min time marks following ignition, simultaneously with photographs of both the fire and the light. Grab samples of biogas from a landfill site and 24-h continuous samples of outdoor ambient air were also collected for comparison in 3-L Summa canisters. The U.S. Environmental Protection Agency (EPA) TO-14A method has been modified and verified for the sampling and analysis of VOCs found in smoke at fires (U.S. EPA, 1997a, 1997b; Austin, 1997). Volatile organic compounds present in air and smoke samples were identified and quantified for 144 target compounds using cryogenic preconcentration at –183°C and a Hewlett-Packard model 5890 series II gas chromatograph (GC), capable of subatmospheric temperature programming, equipped with an HP model 5971 mass selective detector operated in selected ion monitoring (SIM) mode. The volatiles were separated on a Hewlett-Packa rd 50-m, 0.32-mm-ID fused silica capillary column with a 1.0-µm-thick film of HP1 bonded liquid phase. The revolatilized sample from the pre-concentrator was cryofocused on the column using liquid nitrogen. The initial temperature of the column, which was –60°C, was held for 3 min, then raised to 250°C at a rate of 8°C/min. The rate was then increased to 20°C/min and the temperature was held at 280°C for 8 min, after which it was lowered to 150°C. The total chromatographic separation time was 57.75 min, including a 5-min solvent delay time. Overnight and between runs, the GC was left on standby at 150°C. Samples were also analyzed in SCAN mode and examined for the appearance of substances that were not present in the instrument standard calibration mixture. The probability that a correct match was found with spectra in the NIST (1992) mass spectral library was required to be greater than 90%. The ratio of benzene to selected VOCs was determined by linear correlation, with the critical value for a statistically significant correlation coefficient being .95 (p £ .05, n = 4). The paired Student’s t-test (p £ .05) was used to compare the mean ratios of benzene to selected VOCs between fires burning liquids and fires burning solids for 5, 10, and 15 min (n = 4). RESULTS The chromatograms obtained from burning wood were distinctivel y different from those obtained from other sources such as outdoor ambient air or biogas produced at landfill sites (Figure 1). Burning spruce wood produced 108 different VOCs, but the chromatograms were characterized by a small number of prominent peaks (2-methylpentane, benzene, toluene, and naphthalene) and the presence of four peaks not found in ambient air

194

C. C. AUSTIN ET AL.

FIGURE 1. Derived chromatograms (CorelDraw v4.0), showing distinctive patterns or “fingerprints” of VOCs: (A) 24-h sample of outdoor, ambient urban air; (B) spruce wood fire burning for 10 min; and (C) grab sample of landfill gas (LFG).


195

samples (propene, 1,3-butadiene, styrene, and naphthalene). SCAN mode analysis of samples from the 15 experimental fires did not find any new substances not already present in the 144 target compound standard calibration mixture. Figure 2 shows derived chromatograms obtained from samples collected at the 10-min mark for 5 different experimental fires ignited with different materials commonly found in municipal structural fires (spruce wood, a mattress, gasoline, varsol, and white foam insulation). Although the pattern of peaks seen (chromatographic “fingerprint”) varied from one material to another, similarity in the nature of the combustion products generated by different materials was demonstrated by the characteristic prevalence of the 2-methylpentane, benzene, toluene, and naphthalene peaks. Derived chromatograms for plywood, cardboard, and sofa foam are shown in Figure 3 for analysis of samples taken at the 15-min mark. While these materials were observed to burn more slowly than other combustible materials, the combustion products were primarily propene, 2-methylpentane, benzene, and toluene. Naphthalene was essentially absent, replaced by decane, undecane, and dodecane. Total VOCs found in experimental fires, at a height of 1 m from the floor, ranged from 0.1 to 107 ppm depending on the material burned and on the duration of the fire. Of the 144 possible VOCs measured, 14 substances were found in proportionately higher concentrations in all experimental fires, accounting for 65% (SD = ±12%) by mass of total measured VOCs, with no statistically significant difference being found between fires burning liquids and fires burning solid combustible materials. The “fingerprints,” or relative amounts of these 14 VOCs, were also similar from fire to fire in the case of different solid combustible materials (Figure 4). In the case of liquids such as gasoline, varsol, and foam insulation (which melted prior to burning), higher relative levels of benzene and naphthalene were found compared to the VOCs of solid combustible materials (Figure 5). The fingerprints seen for the liquid combustible materials were distinct from each other. This may be explained by different combustion reactions occurring in hotter and more rapidly burning fires and by the contribution of solvent evaporation. Statistically significant positive correlations were found between increasing levels of benzene and levels of coeluents propene, 1-butene/2methylpropene, 1,3-butadiene, toluene, naphthalene, ethylbenzene, and isopropylbenzene in a number of experimental fires burning solid and liquid combustible materials (Table 1). The paired Student’s t-test revealed significantly different ratios of benzene to selected VOCs between fires burning liquids and fires burning solids (Table 2). These ratios, however, remained relatively constant during the course of fires that burned for 15 min, varying by 15% on average. The mean ratios of benzene to propene, 1,3-butadiene and toluene were 18-, 8-, and 1.6-fold higher, respectively, for fires burning liquids than for fires burning solids.

196

C. C. AUSTIN ET AL.

FIGURE 2. Derived chromatograms (CorelDraw v4.0) of the major components of the 10-min samples taken at the experimental fires of different combustible materials. The internal standards have been deleted to simplify comparisons. Propene, 1,3-butadiene, 2-methylpentane, benzene, toluene, styrene, 1,4-diethylbenzene, and naphthalene were common to all of these fire samples.


197

FIGURE 3. Derived chromatograms (CorelDraw v4.0) for fires from sofa foam, plywood, and cardboard, with the samples being taken 15 min after ignition The combustion products were primarily 2methylpentane, benzene, and toluene. Naphthalene was absent, replaced by decane, undecane, and dodecane.

198

FIGURE 4. Characteristic “fingerprints” of the dominant VOCs from fires of single combustible solid materials, showing the similarity in composition of the VOCs produced at the 5- and 10-min sampling times.

199

FIGURE 5. Characteristic “fingerprints” of the dominant VOCs from fires of combustible liquids (the foam melted before burning), showing the similarity in composition of the VOCs produced at the 5- and 10-min sampling time.

200

C. C. AUSTIN ET AL.

TABLE 1. Linear Correlation Coefficients for the Ratio of Benzene to Selected Volatile Organic Compounds at Experimental Fires Solids (n = 4) __________________

Liquids (n = 4) __________________

Ratio

5 min

10 min

15 min

5 min

10 min

15 min

Benzene/propene Benzene/(1-butene + 2-methylpropene) Benzene/1,3-butadiene Benzene/xylenes Benzene/toluene Benzene/naphthalene Benzene/ethylbenzene Benzene/isopropylbenzene

0.99a 0.98a 0.98a 0.55 0.99a –0.03 1.00a 0.22

0.91 0.85 0.95a 0.79 0.93 0.38 0.97a 0.96a

0.98a 0.87 1.00a 0.77 0.94 0.70 0.98a 0.90

0.97a –0.07 0.99a –0.23 0.76 1.00a 0.98a 1.00a

0.58 0.70 0.71 –0.09 0.98a 0.92 0.99a 0.99a

0.81 0.31 0.78 0.09 1.00a 0.95a 0.99a 0.99a

a

Statistically significant correlation (p £ .05).

Total levels of the measured VOCs increased with time for all fires burning solid combustible materials (Figure 6). However, total VOCs decreased with time in the case of gasoline (91 ppm at the 5-min mark to 5 ppm at the 15-min mark), as did the benzene and naphthalene levels for the other two liquids, varsol and foam insulation. Fires burning liquid combustibles such as gasoline and foam insulation produced much higher levels of smoke particulates (obscuration) than those burning solid combustible materials. Decreasing levels of VOCs with time in combination with increasing levels of smoke in the case of burning liquids suggest that a significant fraction of VOCs is adsorbed onto smoke particles, and that these escape detection when only the vapor phase is measured. Benzene, toluene, 1,3-butadiene, naphthalene, and styrene were found at higher concentrations than most other measured VOCs, and they were found to increase with time in all experimental fires together with increasing levels of carbon monoxide (Figures 6 and 7). Benzene was found in the highest concentrations, with peak levels ranging from 0.6 ppm to 65 ppm, while the levels of 1,3-butadiene, styrene, and naphthalene peaked at 0.1, 0.4, and 3.0 ppm, respectively. Carbon monoxide and carbon dioxide levels ranged from 25 to 465 ppm and from 830 to 14,549 ppm, respectively, increasing steadily with time during the course of 15-min experimental fires (n = 15). DISCUSSION Combustion toxicology to date has focused primarily on the acute toxicity of a limited number of gases and irritants, and little or no work has been done on the chronic effects of exposure to fire smoke. Reports have suggested that new toxicants arise from the combustion of new building materials;


201

TABLE 2. Ratios of Benzene to Selected VOCs at Experimental Fires Burning Solids and Liquids

VOC ratios Benzene/propane

Fire duration (min)

Solids (n = 4)

Liquids (n = 4)

Liquid/ solid

5 10 15

0.5 0.5 0.4 0.5 0.1 15.7

7.3 10.2 7.6 8.4 1.6 18.8

13.9 20.2 19.5 17.7 3.5 19.7

5 10 15

1.8 1.3 1.2 1.4 0.3 19.8

n.s.

n.s.

22.6 25.3 11.2 19.7 7.5 38.0

9.4 8.9 5.6 8.1 2.1 25.3

n.s.

n.s.

Average a SD %rsd Benzene/(1-butene + 2-methylpropene)

Average SD %rsd Benzene/1,3-butadiene

5 10 15

2.4 2.8 2.0 2.4 0.4 17.2

5 10 15

12.9 12.6 14.4 13.3 1.0 7.3

Average a SD %rsd Benzene/xylenes

Average SD %rsd Benzene/toluene

5 10 15

Average a SD %rsd Benzene/naphthalene

Average SD %rsd

Mean (ppm) ______________

3.3 3.2 3.0 3.2 0.2 5.7

5 10 15 n.s.

4.7 5.1 5.1 4.9 0.2 4.9 5.2 5.6 5.1 5.3 0.3 5.3

1.4 1.6 1.7 1.6 0.2 10.0

n.s.

Note. %rsd, Percent relative standard deviation = (standard deviation ¸ average) × 100; n.s., not significant. a The paired Student’s t-test revealed a statistically significant difference between means of fires burning solids and fires burning liquids (p £ .05).

202

C. C. AUSTIN ET AL.

FIGURE 6. Benzene, 1,3-butadiene, naphthalene, and styrene produced at fires burning solid combustible materials were found at higher levels than most other VOCs, and their concentrations were observed to increase with time together with increasing levels of carbon monoxide: (A) bed mattress, (B) cardboard boxes, and (C) spruce wood.


203

FIGURE 7. Benzene, 1,3-butadiene, naphthalene, and styrene produced at fires burning combustible liquids were found at higher levels than most other VOCs: (A) gasoline, (B) varsol, and (C) foam insulation which melted prior to burning. Their concentrations were observed to initially increase with time, together with increasing levels of carbon monoxide, and then to decrease, possibly as a result of adsorption onto particulates produced in large quantities at these fires.

204

C. C. AUSTIN ET AL.

however, it has also been claimed that reported increases in smoke casualties due to new, highly toxic combustion products formed from modern materials are unlikely to be correct (Purser, 1988; Guidotti & Clough, 1992; Hartzell, 1996). The present study revealed that there were no new or novel, toxic nonpolar VOCs resulting from the burning of common building materials. The analysis of air samples obtained at experimental fires during the present study revealed the predominant presence of contaminants, such as benzene, styrene, and 1,3-butadiene, whose toxic and carcinogenic affects have been established (IARC, 1985, 1987, 1992, 1994; ACGIH, 2000). The similarity in the nature of the combustion products from a variety of sources, demonstrated by the characteristic prevalence of benzene, toluene, and naphthalene, suggests that similar patterns might be found in working fires attended by firefighters. A common spectrum of chemicals (propene, benzene, xylenes, 1-butene/2-methylpropene, toluene, propane, 1,2-butadiene, 2-methylbutane, ethylbenzene, naphthalene, styrene, cyclopentene, 1-methylcyclopentene, isopropylbenzene) at high concentrations relative to other combustion products emerged from the analysis of samples obtained from different experimental fires. Along with naphthalene, the degradation products of polymeric materials were also the principal combustion products of wood, the predominant construction material in the past. Only in the case of solid sofa foam was there an absence of naphthalene, with peaks of decane, undecane, and dodecane appearing instead. The present study suggests that exposures to important carcinogens have occurred for decades before the introduction of polymers. The presence and concentration of contaminants such as benzene, styrene, and 1,3butadiene may not have been less in times when wood, cotton, and wool were dominant compared to present times when there is a prevalence of polymers. This is an important consideration in view of the epidemiological studies that have found associations between firefighting and various forms of cancer (IDSP, 1994; Guidotti, 1995). Burgess et al. (1979) found that benzene was highest at fires involving wood structures, with levels ranging from 0.3 to 180 mg/m 3. Many of the same contaminants have been found during overhaul (cleanup) at fire sites as were detected during the initial knockdown phase but at much lower levels (Jankovic et al., 1992). The constantly changing pattern of contaminant concentrations with time during a fire was observed in the present study. The present study also demonstrated that lower combustion temperatures characteristic of the later stages of a fire did not result in either higher levels or a change in the pattern of combustion products. Analysis of chromatograms obtained from experimental fire samples suggests that it is possible to identify certain types of combustible materials by the distinctive GC/MS “fingerprints” of their combustion products. The chromatographic “fingerprint” may be used to distinguish accidental fires burning primarily solid material from criminal fires where accelerants such as gasoline have been used. In particular, benzene showed an interesting


205

and unexpected correlation directly with the levels of a number of other important combustion products: propene, toluene, 1,3-butadiene, and naphthalene. While it was surprising that these ratios were found to be similar for different substances, the ratio concept might be of value in identifying cases of arson if appropriate air samples were obtained early enough during the fire. This aspect would require further study with different petrochemical products used as fire accelerants. Good correlations were also found between carbon monoxide and other products of combustion. This is consistent with the findings of Tewerson (1996), who found positive correlations between the equivalence ratio and the mass of carbon monoxide, hydrocarbons (methane), and smoke generated in fires. The equivalence ratio, describing the effect of ventilation on combustion, is determined by the ratio of the fuel generation rate to the flow rate of air multiplied by the air-to-fuel ratio, with this ratio increasing with reduced ventilation that favors incomplete combustion (Tewerson, 1996). The correlations found between benzene or carbon monoxide and other organic toxicants could be used as markers for exposure to fire smoke, simplifying considerably the development of a multicomponent exposure model and the ensuing calculations leading to an estimate of the potential chronic effects from exposure to fire atmospheres. Smoke rises and is therefore stratified in confined spaces. It is therefore expected that the concentration of combustion products would be higher if sampled nearer the ceiling rather than 1 m from floor level as in the present study. Nonetheless, those VOCs likely to be responsible for toxic effects of smoke appear to relatively few in number but at concentrations considerably higher than the remaining plethora of combustion products. They appear in the combustion of wood products, a traditional building material, as well as in newer, synthetic (polymeric) materials. Given the toxicity/carcinogenicity of those VOCs that were found in the highest concentrations, further investigatio n of VOC exposures of firefighters is warranted. REFERENCES American Conference of Government Industrial Hygienists. 2000. Threshold limit values for chemical substances and physical agents and biological exposure indices. Cincinnati, OH: ACGIH. Austin, C. C. 1997. Firefighter exposure to toxic gases and vapours. Doctoral thesis, McGill University. Brandt-Rauf, P. W., Fallon, L. F., Tarantini, T., Idema, C., and Andrews, L. 1988. Health hazards of firefighters: exposure assessment. Br. J. Ind. Med. 45:606–612. Brandt-Rauf, P. W., Cosman, B., Fallon, L. F., Tarantini, T., and Idema, C. 1989. Health hazards of firefighters: Exposure assessment. Br. J. Ind. Med. 46:209–211. Burgess, W. A., Sidor, R., Lynch, J. J., Buchanan, P., and Clougherty, E. 1977. Minimum respiratory protection factors for respiratory protective devices for firefighters. Am. Ind. Hyg. Assoc. J. 38: 18–23. Burgess, W. A., Treitman, R. D., and Gold, A. 1979. Air Contaminants in Structural Firefighting. Harvard School of Public Health, prepared for the National Fire Prevention and Control Administration, Springfield, VA, NTIS order number PB 299017.

206

C. C. AUSTIN ET AL.

Gold, A., Burgess, W. A., and Clougherty, E. V. 1978. Exposure of firefighters to toxic air contaminants Am. Ind. Hyg. Assoc. J. 39:534–539. Golden, A. L., Markowitz, S. B., and Landrigan, P. J. 1995. The risk of cancer in firefighters. Occup. Med. State Art Rev. 10:803–819. Guidotti, T. L. 1995. Occupational mortality among firefighters: Assessing the association. J. Occup. Environ. Med. 37:1348–1356. Guidotti, T. L., and Clough, V. M. 1992. Occupational health concerns of firefighting. Annu. Rev. Public Health 13:151–171. Harztell, G. E. 1996. Overview of combustion toxicology. Toxicology 115:7–23. IARC. 1985. 1,3-Butadiene. IARC Monogr. Eval. Carcinogen. Risks Hum. 39. IARC. 1987. Benzene; ethylene dibromide; vinyl chloride. In Overall evaluations of carcinogenicity: An updating of IARC Monogr. Vol. 1 to 42, Suppl. 7. IARC. 1992. 1,3-Butadiene. IARC Monogr. Eval. Carcinogen. Risks Hum. 54. IARC. 1994. Isoprene; propene; propylene oxide; styrene. IARC Monogr. Eval. Carcinogen. Risks Hum. 60. Industrial Disease Standards Panel. 1994. Report to the Workers’ Compensation Board on Cardiovascular Disease and Cancer Among Firefighters. Report 13, September. Toronto, Canada: IDSP. Jankovic, J., Jones, W., Burkhart, J., and Noonan, G. 1992. Environmental study of firefighters. Ann. Occup. Hyg. 35:581–602. Klaassen, C. D., Amdur, M. O., and Doull, J. eds. 1996. Casarett and Doull’s toxicology: The basic science of poisons, New York: McGraw-Hill. 5th ed. Lees, P. S. J. 1995. Combustion products and other firefighter exposures. Occup. Med. State Art Rev. 10:691–705. McDiarmid, M. A., Lees, P., Agnew, J., Midzenski, M., and Duffy, R. 1991. Reproductive hazards of firefighting. II. Chemical hazards. Am. J. Ind. Med. 19:447–472. NIST. 1992. NBS75K Mass Spectral Library. National Institute of Science and Technology, National Bureau of Standards, Gaithersburg, MD. Purser, D. A. 1988. Toxicity assessment of combustion products. In The SFPE handbook of fire protection engineering, ed. P. J. DiNeno, Chap. 14. National Fire Protection Association, Quincy, MA. Tewerson, A. 1996. Ventilation effects on combustion products. Toxicology 115:145–146. Treitman, R. D., Burgess, W. A., and Gold, A. 1980. Air contaminants encountered by firefighters. Am. Ind. Hyg. Assoc. J. 41:796–802. Turkington, R. P. 1984. An inexpensive, portable multi-sampler. Am. Ind. Hyg. Assoc. J. 45:BE-B22. U.S. Environmental Protection Agency. 1997a. Compendium of Methods for the Determination of Toxic Organic Compounds in Ambient Air—Compendium Method TO-14A. Determination of Volatile Organic Compounds (VOCs) in Ambient Air Using Specially Prepared Canisters With Subsequent Analysis by Gas Chromatography. Research Triangle Park, NC: U.S. EPA, Quality Assurance Division, Environmental Monitoring Systems Laboratory. U.S. Environmental Protection Agency. 1997b. Method TO-15: The Determination of Volatile Organic Compounds (VOCs) in Air Collected in SUMMA Canisters and Analyzed by Gas Chromatography/ Mass Spectrometry (GC/MSD). Research Triangle Park, NC: U.S. EPA, Quality Assurance Division, Environmental Monitoring Systems Laboratory. Wilshire, F. W., Johnson, L. D., and Hinshaw, G. D. 1994. Effect of soot build-up while sampling with the volatile organic sampling train (VOST). Haz. Waste Haz. Mater. 11:277–287.