A high-yield sampler for toxicological characterization of complex mixtures in combustion effluents.

Combustion sampling for toxicological assessment often requires that large (greater than 100 mg) lots of complex organic mixtures of wide volatility range be rapidly recovered from high temperature gases without contamination. A new sampler, meeting these criteria for studies of public health interest, has been developed and demonstrated. The device provides high sampling rates and intimate contacting of the samples stream with large volumes of a well-cooled, liquid solvent, dichloromethane (DCM). This promotes rapid organics dissolution from carrier gas and particulates and prompt dilution and quenching of the resulting solution, resulting in high organics collection efficiencies with minimal DCM losses. Solvent separation then remits large quantities of concentrated organics for chemical analysis and toxicological testing. One- to seven-hour interrogations of in-flame, post-flame, and flue gas regions gave 50- to 250-mg yields of complex organic mixtures. In side-by-side sampling of combustion exhaust, the DCM sampler provided higher yields of DCM solubles (identified with complex organic mixtures) and of S. typhimuirim mutagens (active without exogenous metabolizing agents) than did a filter/polymeric sorbent bed sampling train. The new sampler also collects polar and high volatile hydrocarbons such as benzaheyde, pentadiyne, m- and p-diethynyl-benzene, and 1-hexen-3,5-diyne. Nitration of naphthalene and pyrene in DCM solution (1 mg/mL each) was less than 1 part in 10(7) after a 345-min exposure to a bubbling flow of moist N2/air mixture (1:1 v/v) containing 107 ppm NO and 1.5 ppm NO2, indicating that for these condition a DCM sampler should resist artifactual nitration of aromatics. However, because of the very high bacterial mutagenicity of some nitroaromatics and the wide range of sampling conditions of environmental interest, nitration and all artifacts must still be scrutinized when using the DCM sampler. The DCM sampler is expected to contribute to public health impact assessments by facilitating detailed determinations of the identities, compositions, concentrations, sources, formation mechanisms, and biological activity of environmental toxicants in gaseous atmospheres.


Introduction
Combustion systems emit complex organic mixtures showing toxicological activity in bacterial cells (1)(2)(3), human cells (4,5), and rodents (W. F. Busby, Jr., personal communication). Progress in emissions monitoring and control has been encouraging. However, combustion will remain of interest as an important potential source of environmental toxicants because ofcontinued need for fossil and biomass-fueled stationary and mobile com-bustors and for incineration of municipal refuse and hazardous wastes. In-flame and exhaust region sampling are important for quantifying ambient concentrations, chemical composition, and bioactivity ofcombustion emissions, contributing critical data for determining toxicants emissions factors, sources, and formation mechanisms, and for devising control strategies for combustion-derived toxicants. Most characterization measurements are performed offline. Reliable toxicological assessments can easily require that > 100 mg samples of complex organic mixtures ofwide volatility range be expeditiously recovered from high temperature gases without contamination. For instance, assay for lung adenoma induction in the CD-1 strain ofnewborn mice requires up to 100 mg of test material for dose-response measurements in appropriate numbers of animals (6). Human cell mutagenicity assays require testing at doses of up to 0.1 mg/mL and up to 100 mL oftest solution, corresponding to a total sample requirement of 10 mg. Detailed chemical analyses to identify specific chemicals responsible for toxicological activity can consume a further 100 to 200 mg of mixture.
State-of-the-art methods for combustion sampling ofcomplex KRZEL  organic mixtures typically divert a known fraction of effluent through an invasive probe and collect polycyclics and other organic compounds on filters and within one or more postfilter traps, typically consisting ofpacked beds of polymeric sorbent particles. Organics are recovered for analysis by extracting each collection station with a strong, volatile organic solvent such as dichloromethane (DCM). Filter/sorbent bed sampling trains have exhibited several drawbacks in collecting complex organic mixtures for toxicological characterization. Sample distribution between filter and sorbent can be quite arbitrary, reflecting apparatus-specific effects ofsampling rate, sample composition, and collection station temperature. Filters contribute to mutagenic artifacts formation (7-11) by concentrating higher molecular weight organics and promoting their reaction with NOX and other effluent gases. Potential problems with some sorbents include high organics backgrounds; loss of collection efficiency with extended sampling; sorbent atrophy and fiagility; diminished performance under elevated temperatures; laborintensive preparation and sample workup; and inadequate quantitative understanding or organics collection efficiencies under actual combustion sampling conditions. This paper reports on the design, operation, and validation of an improved combustion sampler that furnishes high yields of complex organic effluents in reasonable collection times, while eliminating or significantly remediating the above deficiencies.

Design Strategy
The approach was to design a direct impingement collection vessel that at high sampling rates promoted intimate contacting between sampled gas and large volumes ofa refrigerated, liquid solvent; rapid organics dissolution from sampled gas and particulates; prompt dilution and quenching of the resulting solution; and minimal solvent losses. DCM was chosen as the solvent because of its low freezing point, low flammability, high dissolution power for polycyclic and other organic compounds, and its high volatility to facilitate sample recovery.
Safety note: DCM is a toxicant and is hazardous. Proper safety precautions must be followed in storage, handling, use, and disposal of this compound.
One motivation for the sampler design is to inhibit artifactual reactions by preserving the catch in cold, dilute solution thrughout sampling, and by preventing sample concentration by filters or sorbent surfaces. Simplicity is a further motivation, since this design requires only one operation, solvent separation, for sample recovery. The present approach is somewhat reminiscent ofvanous sampling trains where the freshly diverted sampling stream was directed through one or more impinger vessels. For example, almost three decades ago Stenberg et al. (12) published on their sampling ofautomotive and incinerator emissions for benzo(a)pyrene by educting exhaust through cooled water bubblers and then through a particulate filter.

Sampler Description
The sampling train consists of two or more series-connected DCM contacting vessels (impingers) (Fig. 1), to collect, cool, dilute, and preserve the catch. Each vessel is a 0.45 cm wall, 4-L cylindrical pyrexreactionkettle( -30cmdeep, lengthtodiameter of about 2.4), divided into inlet (-1 L) and main (-3 L) chambers by a 0.64-cm thick fritted glass diskwith 150to 170 1tm orifices. A 5-cmdeeplayerofno. 4or6glassbeads isplacedabove the frits in impingers 1 and 2 to aid gas-liquid contacting and to break up aerosols formed by rapid cooling ofthe sampled stream.
The inlet chamber precools the sampled stream and separates residual moisture as ice in the first impinger. The fritted disk causes the sampled stream to enter the main chamber as a torrent oftiny bubbles carrying sufficient momentum to stir the solvent. Bubbling, sfirring, and the use ofa large solvent volume combine to promote organics dissolution from the carrier gas and entrained particulates by increasing gas-liquid contact areas, by providing a high concentration driving force for dissolution (by 306 -1 SAMPLER FOR 1DXYCOLOGICALASSAYS ON COMBUS77ONEFFLUEN7S FIGURE 2. Typical DCM sampling train. diluting and homogenizing the resulting solution of sample), and by reducing interphase transport lengths (by converting the sampled gas to bubbles). These design and operating features provide good sample collection efficiencies with a manageable size contacting vessel while cooling and diluting the sample to resist artifactual reactions, even for long collection times.
The solvent temperaure in the main chamber ofthe contacting vessel ( Fig. 1) is monitored by a Teflon-coated type-K thermocouple and is independently regulated to within ±20(C by electronically controlling delivery of cold N2 (from a liquid nitrogen tank) to a submerged heat exchange coil ( 1.83 m of 0.64-cm OD Teflon-coated copper tubing) using a feedback signal from the thermocouple. A baffle (a 11.4-cm OD, 0.16-cm thick Teflon sheet) inthe upper third ofthe main chamber inhibits solvent losses caused by entainment in the escaping carrier gas. Teflon components are used to prevent sample contamination from metallic ions leachable from copper and other metals by acidic gases (CO2, NO2, and SO2) in combustion effluents. During sampling, the first DCM reservoir is maintained at -30 ± 2°C to prevent plugging of its frit by ice. This temperature reflects compromises between preventing frit icing and minimizing solvent and sample losses at high gathering rates. The remaining DCM reservoirs are typically operated at -70 ± 2°C to increase organics solubility and to resist artifacUal reactions and solvent losses while preventng DCM solidification. These erars were selected based on operating experience and depend on the solvent, sampling conditions, and the collection vessel geometry.
In the present applications, a typical sampling train ( Fig. 2) consisted of an invasive probe to interrogate the medium of interest (materials, length, and internal diameter, here pyrex, -91 cm, and -2.54 cm, respectively, are chosen to be compatible with the temperature ofthe sampled medium, and to provide a well-defined sampling zone, convenient interfacing to downstream equipment, and desired sampling rates); a 65-cm long bulb condenser flushed with water thermostatted at 1°(C to precool and partially dehumidify the sampled stream; a condensate collector; two or more series-connected DCM contacting vessels to collect, dilute, and preserve the sample; a check valve to prevent backstreaming of educted effluent or atmospheric gases to the collector vessels; a manual metering valve to select the sampling rate; a vacuum pump; and a flow meter to measure the sampling rate. With proper attention to interfacing and sample acquisition design prcedures, the DCM sampler should be compatible with other probes.

Operating Procedure
BEefre sampling, each collection vessel is charged with about 2 L of DCM. The DCM reservoirs and the condenser are then cooledheirdesiredoperatingtemperatres bydirecngtheappropriatecoolants (N2orwater, respectively) through theheatexchangecoilsorwaterjacket. During sampling, theseteperatures are automatically regulated at the operating values specified above. Thesamling rateisthen setby activatingthevacuumpump andadjustingthemetering valve whilemonitoring the flowmeter (Fig. 2). Sampling then proceeds for the desired collection time by aspirating gasthrough theprobe and DCM impingers, and, if necessary, occasionally adjusting the metering valve to keep the sampling rate withinprescribed limits. Upon completing sampling, the DCM contacting vessels are disconnected from each other. The entrance and exit ports ofeach vessel are then lightly covered with aluminum foil, and each vessel is allowed to attain ambient temperature by natunal warming.
Safety note: This warming causes dramatic increases in the DCM vapor pressure in each contacting vessel. Disconnecting the DCM vessels and then lightly covering the entrance and exit port ofeach with aluminum foil allows each DCM reservoir to separately equilibrate with the ambient atmosphere, enabling these pressure increases to be relieved as they develop. If these precautions are not followed, there is serious danger that the accumulating DCM vapor will overpressurize the vessel to the point ofshatering, posing serious threatofpersonal injury. Thus, the DCM contacting vessels must never be tightly sealed before the DCM atains ambient temperature and before the pressure in the head space above the DCM liquid has become equilibrated with amospheric pressure. Disconnecig the DCM vessels also prevents intneixing ofvessel contents in case ofmore rapid warming and pressurization of individual DCM reservoirs.
After the DCM has reached ambient temperature, liquid water is separated from the condensate collector and from the first DCM vessel and extracted with DCM in a separatory funnel. The resulting DCM is then combined with the DCM from the first collector. The probe is rinsed with DCM, and the rinse is added to the first DCM collection vessel. The DCM from each ofthe collector vessels is then separately concentrated by Kuderna Danish (KD) evaporation. Typically, 6 hr are required for concentting 2 L ofDCM to 15 mL, using a six-ball Snyder column with a minal 1.5 L KD concentrator, heated in a hot water bath at -80°C. The resulting organic concentrates are then further worked up and subjected to chemical analysis and toxicological testing as desired.

Collection Reservoir Retention Efficiencies for Aromatic Compounds of Different Volatility
The two experiments just described were also used to assess the retention ofindividual reference polyaromatic hydrocarbon (PAH) compounds in the DCM impinger vessels. To this end, 200 mg each ofbromobenzene, 1-bromonaphthalene, 9-bromophenanthrene, and 1-bromopyrene were dissolved in the first collection reservoir before sampling began. Table 1 shows that, generally, about 70 to > 90% ofeach compound was retained in Collector 1. Collector 2 typically contained . 1% ofeach compound. Control (no flow) runs (Table 1) generally gave recovery efficiencies between about 80 and 90%, implying that about 10 to 20% of the initial charge of each compound is lost during sample concentration and analysis (here by gas chromatography) and not by escape from collector 1 during sampling. Thus, the true DCM reservoir retention efficiencies are about 80 to 90% for aromatics representing a broad range ofvolatility (boiling points of 156 to > 3600C).

Extent of Artifactual Nitration of Selected Aromatic Compounds
Nitrogen oxides, especially in the presence of water vapor and oxygen, are highly reactive and have been observed to cause nitration of organic compounds. Nitrogen-containing compounds, typically nitroarenes, can be formed under a remarkably wide range ofconditions. Furthermore, nitration oforganics has been shown to occur in gases (13)(14)(15), liquid solution (16), and on solid sorbents (7,17). Thus, artifact formation is a potential problem with any sampler operating in an NO,, environment and in the presence of oxygen and moisture.
To determine, for a set of conditions pertinent to combustion sampling, the possible extent of PAH nitration in the DCM sampler by NO,-containing gases, 200 standard mL/min of a moist (bubbled through organics-free water) N2/air mixture (1:1 v/v) containing 107 ppm NO and 1.5 ppm NO2 was bubbled through 500 mL of a DCM solution of naphthalene and pyrene (1.0 mg/mL each) at ambient temperature for 345 min. The contacting vessel was a l.5-L gas washing bottle (Lurex, Inc.) with a frit configuration similar to the DCM sampler in Figure 1 and fitted with a reflux condenser at its exit to reduce solvent loss.
After 345 min, the DCM solutionofpyrene and naphthalene was concentratedfrom500to5.0mLinaKDconcentratorandanalyzed by gas chromatography with NO2 selective pyrolysis/chemiluminescent detection (18)(19)(20)(21). No nitroarenes were detected. Given the detection limit ofthe instrument (0.01 ng/4L injected for 1-nitropyrene) and the concentration factor of 100:1, nitration of the arene surrogates would be less than 1 part in 107. A control run without naphthalene or pyrene in the contacting vessel showed no nitration ofthe DCM after about 420 min ofexposure but did reveal low levels of nitrogen-containing compounds, presumably generated by nitration ofthe DCM stabilizing agent (cyclohexene) or by nitration of DCM impurities.
Obviously, the formation of nitroarenes at levels below the detection limits ofour analytical instruments (GC/ND, GC/MS, GC/FTIR) cannot be ruled out. This fact has important consequences in environmental sampling for toxicological assessment since some nitroarenes exhibit especially potent genotoxic action in mutagenicity assays based on S. typhimurium (22,23). Even at very low concentration levels of 0.1 ng/mL or less (based on the present findings ofnitration of < 1 part in 107 ofarenes present at 1 mg/mL), it is still possible that artifactually nitrated PAH could cause misleading mutagenicity determinations. Furthermore, wide ranges of operating conditions can be of interest in environmental sampling. Thus, nitration and all artifacts must still be scrutinized when employing the DCM sampler. For nitration artifacts, one practical approach is to spike the DCM reservoirs with nitration detectors, e.g., organic compounds known to be absent from the medium being sampled, but ofcomparable nitration reactivity to compounds expected in the sample. For example, in combustion sampling, a fully deuterated aromatic compound such as naphthalene-di could be employed. The extent ofartifactual nitration wvuld then be measured by mass spectrometric determination ofthe yield ofdeuterated nitronaphthalene, which could only be generated by reactions within the DCM sampling train. For reliable measurements, the artifact formation behavior of a given sampler must always be established for the sampling conditions of interest. Table 2 shows the recoveries ofDCM solubles (identifed with organic complex mixtures) sampled from a turbulent premixed ethylene/air flame (T -1300°C) in a well-stirred combustor (WSR) described by Nenniger (24); and exhaust ofa residential oil burner fired under cyclic, generally low-smoke emission conditions, ofthe type described by Leary Table 2 suggest that for DCM collectors ofthe present size (Fig. 1), sampling rates above aboutone-third SCFM aretobe avoided whenquantitativedataon organics yields, compositions, and total bioactivity are required.

Collection Efficiencies for Combustion Effluents
Comparison ofExtract Yields in Side-by-Side Sampling with the DCM Sampler and a Filter/Polymeric Sorbent Bed Sampling Train Table 3 presents mean emitted yields of DCM extractables (solubles) measured by simultaneously sampling the fluegas from generally low smokedensity cyclic firing (5 minon, lOminoff) of a residential oil burner with the DCM sampler and with a filter/ polymeric sorbent bed sampler. Each sampler was connected to one arm ofan inverted y-shaped probe. With two or four seriesconnected collection vessels, the DCM sampler gaveroughly four times higher extract yields. The filter/sorbent bed results at 0.89 and 1.5 SCFM ( Separate catches of DCM extractables were collected by simultaneously sampling flue gas from cyclic (5 min on, 10 min off), generally low-smoke emissions firing of a residential oil burner using a four-vessel DCM sampling train, and a filter/ polymeric sorbent bed sampler. Each sampling train was connected to one arm of an inverted Y-shaped probe.. The products from each DCM bath and fromcombining the extracts ofthe filter, sorbent, and condensate were assayed for mutagenic activity to S. typhimurium without exogenous metabolizing agents, using the protocol ofSkopek et al. (25,26).
This assay measures thefraction ofbacterial cells killed (cytotoxicity) andthefractionofthe surviving cells, in excess ofthose mutated by natural background, mutated at different concentrations (doses) ofthe material being tested. One measureofspecific mutagenic activity useful forcomparisonpurposes isthe fraction, F; above the 95 % background level, ofsurviving cells mutated at some constant dose. The larger the value of F, the greater the specific mutagenic activity ofthe sample. Here, using dose-effect curves that generally involved testing at doses of 30, 100, and sometimes 300 ug/mL, the quantity Fwas determined ata dose of 501g/mL. Inthis manner, Fvalues wereobtained forthe material collected ineachofthe DCM contactingvessels, andforthe combined DCM extracts from the filter, sorbent, and condensate.
Itis also instructive to compare the amount ofmutagenicity, M, estimated to be emitted by the oil burner, using the data provided by each sampling train. The total emitted mutagenicity accounts for the specific mutagenic activity ofa given sample and the total amount ofthat sample emitted per weight of fuel combusted. The quantity Mwas computed from the information obtained with each sampler as follows. The corresponding quantity Fwas divided by the test dose (50 Ag/mL), multiplied by the corresponding weight ofsample collected, corrected for the fraction ofoil burner exhaust educted through the sampling train, and nornmalized to a basis ofunit weight of fuel fired, here 100kg. The resulting calculated quantity can be interpreted as an estimate of the number of mutated bacterial cells that would be obtained if all the exhaust from burning 100 kg offuel in the oil burner were directed at a suitable flow rate, see below) through the given sampling train, and all the extractables thus collected were tested in doses of50 ,g/mL. The larger the quantity M, the greater the number of emitted mutants detected by the given sampler. Table 4 presents results on specific and total mutagenicity from three different experiments. The data show that the DCM sampler collected more material and more mutagens (higher F values), than did the filter/sorbent bed sampling train. Because of these two effects, estimates of total emissions of bacterial mutagens (Mvalues) are significandy higher (factors of 3 to 20) when based on data from the DCM sampler.   (25,26). Higher values imply greater total mutagenicity.

Detection of High Volatile Organic Compounds in Atmospheric Pressure Combustion
Lafleur et al. (27) identified several highly volatile unsaturates including aliphatic poly-ynes, and mono-as well as poly-eneand poly-yne-substituted benzenes in extracts obtained with the DCM sampler from an atmospheric pressure, turbulent premixed ethylene-air flame. Pentadiyne, 1-hexen-3,5-diyne, mand p-diethynyl-benzene, and 1,3,5-hexatriyne were among the specific compounds detected. Compounds of this type are of great interest in determining the detailed chemistry of PAH formation at high temperatures (28). Further, some are suspect toxicants. A conventional filter/polymeric sorbent bed sampler designed for higher molcular weight PAHs may not detect these compounds because of their high volatility and chemical reactivity.

Operating Limits and Design Considerations for Other Applications
The performance features described were demonstrated for DCM collection vessels sized and configured as in Figures 1 and  2 and operated under narrowly defined conditions. When other vessel sizes and operating conditions are of interest, it is recommended that the design specifications and performance limitations described in the following paragraphs be carefully considered. Furthermore, in samplers of any size, solvent impurities may preclude reliable sampling of gaseous media with toxicant concentrations below a critical value. Procedures for estimating minimum acceptable toxicants concentrations are also discussed.
Detailed mathematical analysis of sampler design and operation was outside the present scope, although modern separations science provides excellent resources for such an endeavor (29,30). However, a global mathematical representation of sampler performance is useful in discussing criteria for reliable sampler performance. For example, under steady-state sampling conditions, i.e., for constant sampling rate, collection efficiency, and inlet concentration of sample, the total yield of sample in a DCM collection train, Y, can be approximately estimated from the relation: where S is the sampling rate in volume ofgas per unit time, C( is the concentration (in mass per unit volume) in the sampled stream at the inlet to the first collection vessel of material to be collected, and t, is the total time of sampling. The quantity n(S) is a sampling train efficiency, here defined as the fraction ofsample entering the collection train that is retained within the DCM reservoir(s). Ideally, n(S) would be unity and under certain conditions can be made to approach this limit. In general, however, n(S) is < 1, and may depend on sampling rate, chemical reactions, size, geometry, and temperature of the collection vessel; the fluid mechanics of sample stream-solvent contacting; and the concentrations (rigorously the chemical potentials) ofsample in the sampled stream and in the solvent (29,30). These complexities, as well as the fact that in general n(S), S, and Ci may each be time dependent, and ignored in the present, approximate treatment, but would require carefu scrutiny in formal mathematical modeling for sampler design, operation, and automatic control.
Effects on sampler design and performance ofthe other quantities on the right-hand side of Eq. (1) are now considered.

Samplig Rate
Low sampling rates will furnish inadequate sample yields in reasonable collection times, while excessive rates will produce intolerable solvent and sample losses (and potentially, frit plugging by icing). Table 2 shows that for sampling times ofabout 60 to 120 min, a sampling train with three series-connected DCM impinger vessels provides high sampling efficiencies, i.e., minimal extractables beakthough beyond collector 3, at sampling rates ofabout one-third SCFM (1.57 x 10 m3/s). At higher flow rates, the DCM sampler still yields more extractables (Tables 3 and 4) and mutagens (Table 4) than does the filter/sorbent bed sampler, but breakthough ofproduct to the downstream collectors ( Table 2 and 4) is excessive. Thus, sampling rates above about one-tird SCFM cannot be recommended for quantitative work with samplers sized and configured as shown in Figures 1 and 2, respectively. Acceptable sampling rates, and corresponding ng efficiencies, n(S), should be exerimentally determined for the specific collector vessel size and geometry, collection times, and gaseous media of interest.
ibtal Sampling Tlme At constant values of C,, S, and n(s), Eq. (1) implies that cumulative sample yield should increase linearly with total sampling time. For n(s) to be constant, the sample concentration in each DCM reservoir should be small compared to the maximum sample solubility in DCM at that DCM temperature. Excessive sampling times will obviously invalidate this condition and also cause intolerable carry over and loss ofDCM. At sampling rates of about one-third SCFM and the DCM reservoir temperatures given earlier, experience shows that the present design accommodates sampling times of 1 to 7 hr with minimal solvent carry over ( .-10-15%).

Toxicants Concentration in the Media To Be Sampled
Even ultra-high purity DCM can contain small levels of impurities. Residue concentrations stated by suppliers are about 2 to 3 ppm (w/w), but mauments in this laboratory found DCM backgounds of0.25 to 1 ppm to be more typical. When toxicants identification, characterization, or quantitation is an objective ofsampling, effects ofsolvent conutminants must be considered.
The obvious strategy of further purifying the solvent prior to sampling is not easily appLied here. Solvent decontamination into the tens ofparts per billion range would be necessary for some environmental health sampling measurements now of interest. Without expensive clean room facilities and labor-intensive cleanup protcols, stanard methods for liquids purification, such as treatment on adsorption columns or hard-cut distillation, will not meet this level of cleanup and may result in new impurities and greater overall contamination of the solvent.
Assuming no chemical reactions between the solvent impurities and the sample, an alternative strategy is to operate so that the cumulative weight of sample collected, Y, substantially exceeds the total weight of DCM contaminants: Y>> rp, VS (2) where r is the concentration ofthe solvent impurities in weight per weight of solvent, p, is the density of the solvent, and V, is the volume ofsolvent in the reservoir contning Y Assuming the validity of Eq. (1) and using it to substitute for Ygives n(S)SCA, >> rpsVs (3) Thus, in principle, effects of solvent contaminants can be countered by sampling at high rates (increasing S) or for longer times (increasing t) and by employing a smaller DCM impinger vessel (decreasing V). However, as discussed ealier, for a fixed sampler size and geometry, sampling rate and sampling time may already be prescribed by other design or operating constraints. Normally, collection efficiency [n(S)] is already at a high level, leaving little elasticity to help increase Y Furthermore, excessive reductions in V, will degrade the strong organics dissolution, cooling, and dilution powers ofthe sampler. Thus, for many applications, Eqs. Toxicological activity in DCM impurities may impose additional design or operating constraints in sampling gaseous media with dilute toxicant concentrations. For example, bacterial mutagenicity was detected in some residues from fresh DCM.
The sample yield must tere be so large that this background mutagenicity is negligible compared to the mutagenicity of the sanple. Calculations ofthe r ired sampling conditions wld, in general, require knowledge ofthe dose-response curves for the DCM residue and for the sample. Since the latter is nonnally one of the objectives of sampling, iterative estimation techniques would be necessary. Here a simplified case is considered to il-lusSte how the design calcaion would be performed. To avoid complicated algebra, zero threshhold, linear dose response curves are assumed for the DCM impurity and for the sample.
The quantities MFr and MF, are, respectively, the mutant fractions caused by the the DCM residue and the sample, i.e., the mutant fractions in excess ofthe 95 % background mutation fraction. These can be written as MFr = ArDr (6) MFs = A, Ds (7) where subscripts r and s denote DCM residue and sample, respectively, A is the slope ofthe dose-response curve (in mutant fraction/dose), and D is the dose (in mass/unit volume of test medium). (For orientation purposes, the quantity Fintroduced in the discussion of  (9) or Ds Dr >> ArlAs (10) Now D, ID, is diectly proportional to the ratio of cumulative sample yield to total amount of DCM impurity: Eq. (10) thus requires that Y >> (Ar/As ) rPsVs (12) Again assuming that Eq. (1) can be used to substitute for Yin Eq.

Discussion: Implications for Environmental Health Sciences
Complex mixtures ofwide boiling-range organic compounds are ubiquitous environmental contaminants. Assessment oftheir potential public health impacts requires knowledge oftheir composition, local concentrations, toxicological activity, sources, emissions factors, and formation mechanisms. For products of combustion and related high temperature processes such as incineration, this information must usually be obtained by off-line chemical analysis and toxicological testing of complex organic mixtures recovered from flames, exhausts, and other elevated temperature flows, and from ambient and indoor atmospheres. Reliable sampling apparatus is clearly essential to collecting representative, uncontaminated samples in sufficient quantities for in-depth characterization.
The DCM sampler is very well suited for combustion sampling. In sampling combustion exhaust side-by-side with a filter/polymeric sorbent bed sampler, the DCM equipment provided higher recoveries of S. typhimurium mutagens active without exogenous metabolizing agents and of total DCM solubles identified with complex organic mixtures. The DCM sampler exhibits high recovery and retention efficiencies for individual PAH compounds and for whole complex mixtures of DCM solubles. Minimal breakthough ofextractables beyond a second DCMcontaing vessel (< 5%; lible 2) wasdetecedin sampling ahigh t , ublent premixed flame. This implies that a two-collector DCM sampling train, sized and configured as in Figures 1 and 2, respectively, should provide high, absolute collection efficiencies (>90%) when sampling continuous combustors within the prescribed sampling rates and total sampling times (-1/3 SCFM and S 120 min). In general, however, DCM sampler perlbrnance, including collection efficiencies for whole mixtures, individual organic compounds, and various toxicants, will depend on collection vessel size and geometry, solvent type, tprature and impurities, sampling rate, samling time, and sample concentration in the medium being interrogd. Limitations for the equipment shown in Figures   1 and 2 and geneappaches to design and sizng for other applications are discussed above.
A test under conditions pertinent to combustion sampling resulted in < 1 part in 107 artifactual nitration ofpure aromatic compounds dissolved in DCM. However, because of the very higbacterial m icity ofsome nitroamatics and the wide rangeosfsping conditions ofenvr nental intrest, nitrion and all artifacs must still be scrutinized when employing the DCM samler. For nitration artifacts, one approach to this end would be to employ in siU nittion de s as discussed above.
The  (27) identified phenol, benzaldehyde, pentadiyne, l-hexen-3,5-diyne, m-and p-diethynyl-benzene, 1,3,5,hexatriyne, and odter high volatile, low molecular weight (< 126 amu) olefinicand acetylenic-substituted benzenes, in complex mixtures obtained with this sampler from an atmospheric pressure, turbulent premixed ethylene-air flame. These compounds were obtained in the same master sample dtat contained high molecular weight compounds including PAH of up to at least five ring. No special effort was made to prefereltily sample for low boiling-point compounds. Data on the identities and concentrations of volatile and reactive organic compounds in flames are important in determining the chemistry of PAH and soot formation and depletion, and thus in understanding how combustion generates bioactive effluents. For example, Bittner and Howard (28) de polyacetylenes in low pressre (-20 torr) laminar premixed benzene-oxygen-argon flames using online molecular beam mass spectrometry and assigned acetylene and diacetylene important mechanistic roles in hydrocarbon decay.
Little seems to be known about the bioactivity of volatile organic unsaturates of the types recovered with the DCM sampler. However, the well-established carcinogenicity ofat least one conjugated aliphatic diene, butadiene, suggests that aromatics with unsaturated aliphatic substituents, and aliphatic unsaturates themselves, warrant toxocological characterization.
The DCM sampler provides large quantities ( -100-250 mg) of DCM extractables by sampling flames or their effluents for relatively short times. This sampler may thus be able to supply traditionally unavailable organics in quantities sufficient for use as standards in chemical analysis and for detailed toxicological testing of individual compounds. PNlar organic compounds have been associated with bacterial cell mutagenicity in wood smoke (3) and residential oil burner effluents (1,2,32) and command increasing attention as combustion-generated toxicants. Be oftheir high ractivity, low boiling points, and (for polar compounds) poor affinity for certain sorbents, polar or unsaturated organic volatiles may be harder to detect and quantify with sampling trains using a filter and single type ofsorbent when simultaneous collection ofless volatile polycycic organics is also desired. A gas sampling loop should recover the more volatile of these compounds, but in small yields, making workup, chemical analysis, and toxicological characterization more difficult. Furthermore, separate, upstream collectors would typically be required for simultaneous collection of less volatile organics. The DCM sampler simplifies combustion sampling of organics of wideranging volatility and chemical functionality by providing, in a single collection vessel, high yields of organic compounds spanning broad ranges ofboiling point, polarity, and reaivity. By employing proper design and operating criteria, including characterization ofcollection efficiencies and artificts generation behavior for the conditions of interest, the above or appropriately modified versions ofthe DCM sampler are expected to be applicable to toxicological assessment ofcombustors, incinerators, and other high-temperature process stams and of ambient air sheds and indoor atmospheres. The sampler is expected to contribute to public health impact assessments by facilitating detailed determinations of the identities, compositions, concentrations, sources, formation mechanisms, and biological activity of environmental toxicants.