Environ. Sci. Technol. 2002, 36, 4608-4612
where cp is the concentration of condensed SOC in air (ng/m3), and cg is the concentration of SOC in the gas phase (ng/m3). Partitioning ratios are used to predict the transport and fate of SOCs in the environment (1). The uptake of airborne contaminants into the human body is partially described by Kp, as the mechanisms by which inhaled particles deposit in the respiratory system are different from gases (2). Pankow developed equations to predict gas-particle partitioning of SOC under both adsorptive and absorptive phenomena (3, 4). For absorptive partitioning, Kp is defined as
Electrostatic Sampler for Semivolatile Aerosols: Chemical Artifacts JOHN VOLCKENS* AND DAVID LEITH Department of Environmental Sciences and Engineering, University of North Carolina at Chapel Hill, CB 7431 Rosenau Hall, Chapel Hill, North Carolina 27599-7431
Electrostatic precipitators (ESPs) show promise as an alternative sampling method for semivolatile aerosols because they are less susceptible to adsorptive and evaporative artifacts than filter based methods. However, the corona discharge may alter the chemical composition of a sampled aerosol. Chemical artifacts associated with electrostatic precipitation of semivolatile aerosols were investigated in the laboratory. ESPs and filters sampled both particles and vapors of alkanes, polycyclic aromatic hydrocarbons, and alkenes across varying concentrations. Gravimetric measurements between the two sampling methods were well correlated. Ozone generated by the ESP corona was the primary cause of alkene reactions in the gas phase. Particles collected within the corona region were vulnerable to irradiation by corona ions over time. Particles collected outside the corona region did not react. Vapors passing through the corona reacted to a lesser extent. Vapors captured after passing through the ESP reacted with ozone that was not removed by the vapor trap. Chemical speciation of highly reactive compounds (i.e., alkenes or other compounds with relatively short half-lives outdoors) is not appropriate with ESPs. Electrostatic precipitation of these compounds is appropriate, however, when total organic carbon is of interest as the ESP does not alter the amount of mass measured gravimetrically. ESPs can make accurate measurements of more persistent semivolatile compounds, such as alkanes and PAHs.
Introduction Semivolatile organic compounds (SOCs) have vapor pressures between 10-4 and 10-11 atm over the ambient temperature range (1). As a result, airborne SOCs can exist simultaneously in both gas and particle phases. In the particle phase, SOCs can either adsorb to the particle surface or absorb within the particle matrix. Many classes of environmentally relevant pollutants include semivolatiles, including alkanes, alkenes, aldehydes, acids, polychlorinated biphenyls, polycyclic aromatic hydrocarbons (PAHs), Nitro-PAHs, polychlorinated dibenzo-p-dioxins and dibenzofurans. At equilibrium, the distribution of mass between the gaseous and condensed states can be described by a dimensionless partitioning ratio, Kp
Kp )
cp cg
(1)
* Corresponding author phone: (919)966-7337; fax: (919)966-7911; e-mail:
[email protected]. 4608
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 36, NO. 21, 2002
Kp )
(TSP)fomRT MomγiP0106
(2)
where TSP is the mass of total suspended particulate per unit volume of air (µg/m3), fom is the mole fraction of particulate available for absorption, R is the gas constant (8.2 × 10-5 m atm K-1 mol-1), T is the ambient temperature (K), Mom is the average molecular weight of the absorbing liquid phase (g/mol), γi is the activity coefficient of the SOC absorbed in the particle, P0 is the subcooled, liquid vapor pressure of the SOC in question (atm), and the final term in the denominator is a conversion factor (µg/g) (3). Equation 2 can predict Kp successfully when all its parameters are known. However, data on published values of SOC vapor pressures (P0) and activity coefficients (γi) vary considerably depending on the reference or model used (5, 6). Furthermore, TSP, fom, and Mom will vary with the time of day, season, and sampling location (7, 8). Consequently, Kp is usually determined empirically via field measurements. Unfortunately, accurate measurements of SOC partitioning ratios are not easy to make (9). Filters in conjunction with a vapor trap (e.g. activated carbon, polyurethane foam) are often used to measure Kp. Ideally, the filter retains 100% of the particle-bound SOC, while the gas-phase SOC passes through the filter and is captured on an vapor trap. Filters, however, are prone to both adsorptive and evaporative sampling artifacts. Because of the large effective surface area of the filter matrix, gasphase semivolatiles may adsorb to the filter during sampling. Gas to filter adsorption, a positive artifact, results in an overestimation of the particle-phase concentration and an underestimation of the gas-phase concentration. Alternatively, semivolatile particles collected on the filter may evaporate during sampling, creating a negative artifact. Particle evaporation occurs whenever the gas-phase concentration surrounding a particle drops below the equilibrium level dictated by Kp. Positive and negative artifacts can introduce substantial bias in Kp measurements when filters are used (10-13). Sampling Technology. Electrostatic precipitation has been used with some success to sample semivolatile aerosols indoors (13). Electrostatic precipitators (ESPs) use corona discharge to charge particles and a strong electric field to collect them on a conducting surface. Physical artifacts, such as vapor adsorption and particle evaporation, are reduced in an ESP compared to a filter because the particle collection surface area in an ESP is significantly smaller than the effective surface area of a filter. However, unlike filters, ESP samplers are subject to chemical artifacts. Ozone is a byproduct of corona discharge in air and may alter the chemical composition of the collected aerosol. In addition, the ion stream generated by the corona, which consists mainly of O2+ as 10.1021/es0207100 CCC: $22.00
2002 American Chemical Society Published on Web 09/28/2002
TABLE 1. Compounds Selected for Chemical Artifact Tests chemical class
relative reactivity
particle phase
gas phase
alkane PAH alkene
low medium high
docosane, C22 chrysene 1-docosene, C22
dodecane, C12 naphthalene 1-dodecene, C12
well as smaller amounts of O+, N2+, N+, NO+, and H3O+ ions (14), can react with both particles and vapors that enter the plasma region. These ions can oxidize the substrate surface, causing an artificial increase in substrate mass that can bias gravimetric measurements. Kaupp and Umlauff compared an ESP to a filter and cascade impactor for atmospheric particulate samples (15). They detected smaller amounts of PCBs and PAH within the ESP and concluded that ozone degradation of these compounds precluded the use of this device as a sampler of atmospheric particulate matter. Because sampling times were on the order of days, prolonged contact occurred between the ozone and the collected samples. In addition, their sampler generated two to three times more ozone than the current ESP (16). For the less reactive semivolatile compounds, their ESP collected more mass than other samplers used in the study. For these reasons, electrostatic precipitation shows promise as a sampling method for semivolatile aerosols. This work evaluates the potential of electrostatic precipitation as a sampling method for SOC and describes the chemical artifacts associated with electrostatic precipitation of semivolatile aerosols. A second manuscript describes physical artifacts, adsorption and evaporation, that occur when sampling with filters but to a lesser extent with electrostatic precipitation (17). A prudent choice between these two sampling methods requires understanding the artifacts associated with both procedures.
Experimental Section Chemical Artifact Experiments. ESPs and filters sampled aerosols of varying reactivity and concentration in a chamber. Lower particle concentrations approximate atmospheric conditions, whereas higher concentrations (6) match those normally encountered in aerosol chamber experiments. Compounds were selected from three chemical classes to simulate a broad range of aerosol reactivity, as shown in Table 1. Two compounds were selected from each class, one to represent the gas phase and another to represent the particle phase. Kp was estimated for each compound using eq 2 to ensure that at least 95% of the mass would reside in the desired phase under the conditions of each test (18). This criterion was established to prevent any physical artifacts (e.g., gas-filter adsorption) that would bias a filter measurement. By artificially eliminating the possibility of physical artifacts, the filter sampler was suitable as a reference for comparison with the ESP. The data and results of these calculations are provided in the Supporting Information. Aerosols were generated in a 1.0 m3 acrylic chamber by a Collison nebulizer (BGI Inc., Waltham, MA), as shown in Figure 1. Solutions of alkanes and alkenes were made from three parts vapor compound, C12, and one part particle compound, C22. PAH-coated particles were created by evaporating a saturated solution of chrysene/dichloromethane onto polydisperse glass beads less than 2.5 µm in diameter. A 200 mL aliquot of distilled water was added to the coated particles, and the resulting solution was sonicated for 30 min to break up agglomerates. The coated glass-beads were then nebulized to create particle-phase PAH. For gas-phase PAH, a similar process was employed to coat naphthalene onto a column of 5 mm glass beads (Fisher Scientific,
FIGURE 1. Schematic of the aerosol generation and sampling chamber. Arrows indicate direction of airflow.
FIGURE 2. Diagram of ESP sampler. Pittsburgh, PA) within a bubbling tube (ACE Glass Inc, Vineland, NJ). Nebulizer solutions were stored in sealed glass jars prior to use. A timing switch (model 451, Dimco Gray Co., Centerville, OH), in conjunction with a solenoid valve, issued bursts of filtered, compressed air to the nebulizer at regular intervals. Aerosols exiting the nebulizer passed though an aerosol neutralizer (model 3012, TSI Inc., St. Paul, MN) into a dilution cylinder. A diffusion dryer, inserted between the nebulizer and neutralizer, removed water vapor from the aerosol stream during the experiments with PAH. A mixing fan distributed the aerosol throughout the chamber. A variable-flow Hi-Vol fan (Graseby GMW, Village of Cleves, OH) pushed room air through a filter and an activated carbon adsorption bed (Fisher Scientific, Pittsburgh, PA) and into the chamber. An additional Hi-Vol fan exhausted air through a filter and into a laboratory ventilation hood. Steady concentrations between 10 and 10 000 µg/m3 were maintained by adjusting the nebulizer pulse time, the dilution cylinder flow, and the HiVol air exchange rate. A DustTRAK aerosol photometer (TSI, Inc. St. Paul, MN), previously calibrated by gravimetric analysis, monitored the aerosol concentrations in the chamber in real time. The ESP (Aerosol Associates, Chapel Hill, NC) collection efficiency as a function of particle size and operating characteristics have been described in detail elsewhere (16). A diagram of the ESP is shown in Figure 2. A voltage of +5.9 kV was applied to a 0.5 mm silver wire to produce a corona current of 15 µA. At these settings, the ESP collects 99.9% of particles larger than 1 µm aerodynamic diameter and at least 95% of particles from 0.05 to 1 µm at a flow of 4 L/min (16). The ozone concentration exiting the precipitator was apVOL. 36, NO. 21, 2002 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
4609
proximately 125 ppbv (16). ESP collection substrates, measuring 6 × 8 cm, were cut from thin aluminum foil sheets (Reynolds Metals Co, Richmond, VA) and rolled into cylinders to fit within the ESP. Forty-seven millimeter TFE-coated glass fiber filters (Pallflex Fiberfilm T60A20, Pall Gelman Sciences, Ann Arbor, MI) were placed in Teflon sampling cassettes (URG, Chapel Hill, NC). Both the ESP and the Teflon filter substrates were cleaned by Soxhlet extraction in a 1:1 (v/v) n-pentane/dichloromethane mixture for 4 h, baked at 100 °C for 4 h, and stored under room conditions prior to sampling. Gas-phase SOCs were captured on 20 cm, XAD-4 coated, annular glass denuders (URG, Chapel Hill, NC); one denuder was located immediately behind each sampler. A total of 12 experiments were conducted using four concentrations for each of the three aerosol types. For each run, an array of four ESPs and four filter cassettes were aligned in random order in two rows at one end of the chamber. To ensure equal aspiration efficiencies, the ESP samplers were fitted with a similarly sized inlet as the filter cassettes (0.45 cm diameter, 2 cm long stainless steel tube, SS-4-TA-1-8RS, Swagelok Inc., Solon, OH). Flow was set at 4 L/min and monitored with calibrated rotameters (Dwyer Instruments, Michigan City, IN). Sampling times depended on the particulate concentration in the chamber; the goal was to collect at least 200 µg of particulate mass per sampler. This goal was set to improve weighing precision and to suppress the biases associated with gravimetric analysis (19), as the weight of filter blanks varied by as much as ( 10 µg. Substrates were discharged on an Po210 strip and weighed on a microbalance (AT20, Mettler-Toledo, Columbus, OH) to the nearest 2 µg immediately before and after sampling. After the final weighing, substrates were placed in glass centrifuge tubes, spiked with a deuterated internal standard mixture of C12 and C22 alkanes, desorbed with 15 mL of n-pentane, and sonicated for 30 min. Laboratory blanks, treated in the same manner as the substrates, were carried throughout each test. Denuders were spiked with a similar internal standard mixture, extracted with successive rinses of 20 and 10 mL of n-pentane, respectively, and stored in clean, 40 mL amber glass jars. PAH samples were desorbed with dichloromethane. All solvents, standards, and internal standards were GC-MS grade and purchased from Fisher Scientific (GCMS grade, Pittsburgh, PA), Sigma-Aldrich Chemical Co. (St. Louis, MO), and Cambridge Isotope Laboratory (Cambridge, MA), respectively. Samples were analyzed via GC-MS (HP 5890 Series II GC, DB-5 60m column, HP 5792 MS, Agilent Technologies, Palo Alto, CA) in both scan and single ion modes for product identification and quantification, respectively. Carrier gas (UHP-He, Scientific Supply, UNC Chapel Hill) flow was set at 1.5 mL/min. Inlet and detector temperatures were set at 295 and 300 °C, respectively. Oven temperature was set at 50 °C prior to injection and ramped up to 300 °C at a rate of 15 °C/min. Substrate Oxidation. A separate experiment was conducted to measure the effect of the corona on the mass of the ESP’s foil substrate when sampling clean air. Purified, dry, filtered air was humidified with distilled water and passed through five ESPs for 20 h. Four ESPs sampled with the corona turned on, while one was turned off to serve as a control. Substrates were weighed before and after sampling. The oxidation rate was calculated by dividing the average increase in substrate mass by the sampling time. Because work by Boetler et al. (20) showed that the presence of water vapor can reduce ozone concentrations in ESPs, this experiment was repeated at relative humidities of 0, 5, 25, 50, 75, and 95%, respectively. Data were analyzed with Stata software (Intercooled Stata 6.0, Stata Corp., College Station, TX) using variance weighted least squares regression, weighted analysis of variance, and 4610
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 36, NO. 21, 2002
FIGURE 3. Oxidation rate of the aluminum foil substrate as a function of relative humidity. Error bars represent 1 standard deviation.
FIGURE 4. Flattened aluminum foil substrate after sampling 50% humidified air for 20 h. Solid line shows point of corona generation. Dotted line indicates direction of flow. two-sample, unpaired t-tests. Ninety-five percent confidence intervals (C.I.) were estimated (R ) 0.05).
Results and Discussion Substrate Oxidation. The aluminum foil substrates gained mass after sampling clean, humidified air for several hours, as shown in Figure 3. After several hours, a white band appeared on the substrates across from the corona generation point, as seen in Figure 4. SEM-EDS scans of the substrate surface showed the presence of oxygen, which is indicative of aluminum oxide formation. Scans of unused substrates did not show oxygen at the surface. The oxidative loading to the aluminum foil substrate increased with increasing relative humidity (P ) 0.001), indicating that water vapor enhanced the rate of substrate oxidation. H3O+ ions, a strong oxidant formed from water vapor in the corona, were the likely cause
FIGURE 5. ESP vs filter concentrations by gravimetry. Error bars represent 1 standard deviation. Error bars not shown are within the size of data point.
FIGURE 6. ESP vs filter concentrations for vapor phase (open points) and particle phase (closed points) compounds by GC-MS. Error bars represent 1 standard deviation. Error bars not shown are within the size of data point. of this bias. No increase in substrate mass was observed during corona operation in dry air. Particle-Phase Measurements. Gravimetric concentrations for the high molecular weight hydrocarbons present in the particle phase as measured by the ESPs and filters are shown in Figure 5. The data have been corrected for oxidation of the ESP substrate using the exponential fit from Figure 3 in conjunction with the relative humidity and sampling time for each test. The ESP measurements show good correlation with the filter data, indicating that the ESP collected the same amount of particulate mass as the filter. A regression through the origin for Figure 5 gives a slope of 1.002 (C.I. 0.9901.014) which is not significantly different from unity. The average coefficient of variation between samplers for each test was 0.05, indicating a uniform concentration distribution within the sampling chamber. Figure 6 shows a similar comparison between the ESPs and filter samplers for both vapor-phase and particle-phase compounds except that GC-MS analysis was used rather than gravimetric. The ESPs measured less of the C22 alkene
particles (i.e., 1-docosene) than the filter sampler across all concentrations (P < 0.0001). Furthermore, the discrepancy between the ESPs and filters increased with decreasing alkene concentration (P < 0.0001). The ESPs measured significantly less of the C22 alkane particles (i.e., docosane) at the lower concentrations only (P < 0.0001); no significant difference was detected between sampler types at docosane concentrations above 60 µg/m3. No reaction products were detected in the GC-MS chromatograms for the missing particle mass on the ESP substrates for either docosane or 1-docosene. Chrysene concentrations measured by the ESPs were 5% higher than filter measurements (C.I. 1.0532-1.0533). Gas-Phase Measurements. Concentrations of the C12 alkane vapor (i.e., dodecane) exhausted by the ESPs were consistently lower than those following the filter samplers by about 9% (C.I. 7-11%). However, no reaction products were detected for the missing dodecane mass. A substantial fraction of incoming C12 alkene vapor (i.e., 1-dodecene) was reacted by the ESP, as can be seen from Figure 6. Furthermore, the amount of 1-dodecene vapor reacted in the ESP was strongly correlated with concentration (P ) 0.0001). The major reaction product for 1-dodecene vapor was undecanal, which is consistent with gas-phase reactions of ozone with 1-substituted alkenes (21). The average reaction yield for undecanal in this study was 0.51 ( 0.18, a value that closely corresponds with those of Grosjean (21). Concentrations of naphthalene vapor measured behind the ESPs were less than behind the filter sampler at all concentrations except the highest (P ) 0.005). The discrepancy in naphthalene vapor between the ESP and filters ranged from 1% to 25% and also increased with decreasing concentration (P ) 0.0001). Mass spectra from the ESP-denuder for naphthalene showed as many as seven new product peaks and evidence of ring cleavage and carbonyl formation. The reaction products and yields were not investigated further since commercially available standards do not exist for these compounds. Reaction Mechanisms in the ESP. Ozone and coronarelated ions are the major reactants created by electrostatic precipitation. Each must reach the substrate surface to react with collected particles. Ozone generated at the corona is probably swept away by the bulk flow before it reaches the substrate. On the other hand, corona ions have high electrical mobility so are better able to reach and react with particles on the foil substrate. Analysis of particle deposition along the substrates indicates that most particles collect within or near the oxidation band shown in Figure 4. Particles collected in this vicinity are bombarded by corona ions even after collection. Since the corona provides a constant source of ions, particle reactions may be modeled by pseudo-firstorder kinetics. Figure 7 shows a semilog plot of C/C0 vs time for docosane concentrations measured by the ESP. The initial concentration, C0, was assumed to equal the filter concentration measured at the corresponding sampling time. The goodness of fit, R2 ) 0.93, supports a pseudo-first-order reaction of collected particles. A similar fit for the reaction of 1-docosene particles gave a rate constant of 6.3 × 10-5/s with an R2 ) 0.91. Gas-phase SOCs may react with the ion stream or with ozone generated by the corona. At a flow of 4 L/min, residence times in the corona region and the collection tube are 0.03 and 0.22 s, respectively. Undecanal is the primary product expected from the gas-phase reaction of ozone with 1-dodecene. However, gas-phase kinetics of ozone with 1-alkenes are too slow to account for the undecanal concentrations measured in the denuders behind the ESPs (22). Alternatively, if ozone is not removed by the denuder, then hydrocarbons adsorbed to the denuder walls remain vulnerable to ozone attack. A posterori measurements showed that the denuder removed approximately 40% of the VOL. 36, NO. 21, 2002 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
4611
denuder or a more efficient vapor trap, such as activated carbon, is recommended. Reaction artifacts can be minimized by minimizing the sampling time. Chemical artifacts can also be minimized through the use of a corona-free electrostatic precipitator, with a small sacrifice in collection efficiency (23).
Acknowledgments The authors would like to thank Russell Wiener and Gary Norris of the U.S. EPA for their support of this research and Maryanne Boundy and the Baity Lab of the University of North Carolina for their comments and suggestions. This work was supported in part by grant T32ES07018 from the National Institute for Environmental Health Sciences and a U.S. EPA - NNEMS fellowship (U-91567401-0).
Supporting Information Available FIGURE 7. Semilog plot of C/C0 vs time for docosane concentrations measured within the ESP and 1-dodecene concentrations measured on the XAD-4 coated denuder behind the ESP. Error bars represent 1 standard deviation. ozone generated by the ESP. Thus, ozone is also in excess, so the reactions of SOCs captured on the ESP denuder should also follow pseudo-first-order kinetics. A semilog plot of C/C0 vs time for 1-dodecene measured in the denuder behind the ESP gives a rate constant of 1.45E-4 ( 7.22E-6 s-1 (Figure 7). Arnold et al. report that reactions of O2+, O+, N2+, and N+ with n-octane result in charge transfer and dissociation of the alkane into smaller carbon chains (14). For example, at 300 K, the reaction of O+ with n-octane
O+ + n-C8H18 f C3H7+ + C5H11 + O
9
Literature Cited (1) (2) (3) (4) (5) (6) (7) (8)
(3)
produces the propyl-cation 42% of the time and a pentylcation 21% of the time. Many branched products are reported, with dissociation occurring primarily at the secondary carbons. No such reaction products were detected in the GC-MS chromatograms from the ESP substrates or denuders. However, alkane-cation reactions similar to eq 3 may produce products too volatile to be retained by the denuder or too polar to be solvated during the n-pentane extraction. Gas to Particle Conversion. If vapors transformed by the corona become less volatile, their reaction products may nucleate or adsorb to the particles or to the foil substrate. This artifact would result in an overestimation of particlephase mass. However, gas to particle conversion does not appear to bias the gravimetric measurements, seen in Figure 5, regardless of the discrepancy in measured vapor concentrations between the ESPs and filters, shown in Figure 6. Semivolatile Sampling. These results show that electrostatic precipitation can successfully sample aerosols that are moderately resistant to ozone oxidation. A small proportion of particles and vapors react as they pass through the corona region. However, particles are more susceptible to the corona ions when they collect within the ion impact zone on the ESP substrate. Vapors appear to react after collection on a denuder that follows an ESP. Although annular denuders are effective removers of SOC vapors they do not appear to fully remove ozone from the sampled airstream. In the case where sampled aerosols react readily with ozone, an ozone-removal
4612
Estimation of Kp and the percent of mass in either the vapor or particle phase for the chemicals listed in Table 1 (Table S1). This material is available free of charge via the Internet at http://pubs.acs.org.
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 36, NO. 21, 2002
(9) (10) (11) (12) (13) (14) (15) (16) (17) (18) (19) (20) (21) (22) (23)
Bidlemann, T. F. Environ. Sci. Technol. 1988, 22(4), 361-367. Pankow, J. F. Chem. Res. Toxicol. 2001, 14(11), 1465-1481. Pankow, J. F. Atmos. Environ. 1994, 28(2), 185-188. Pankow, J. F. Atmos. Environ. 1987, 21(11), 2275-2283. Mackay, D.; Shiu, W. Y.; Ma, K. C. Illustrated handbook of physical-chemical properties and environmental fate for organic chemicals; Lewis Publishers: Boca Raton, FL, 1992; Vol. 2. Jang, M., et al. Environ. Sci. Technol. 1997, 31, 2805-2811. Turpin, B. J.; Huntzicker, J. J. Atmos. Environ. 1991, 25A(2), 207-215. Hornbuckle, K. C.; Eisenreich, S. J. Atmos. Environ. 1996, 30(23), 3935-3945. Turpin, B. J.; Saxena, P.; Andrews, E. Atmos. Environ. 2000, 34(18), 2983-3013. Turpin, B. J.; Huntzicker, J. J. Atmos. Environ. 1994, 28(19), 30613071. McDow, S. R.; Huntzicker, J. J. Atmos. Environ. 1990, 24A(10), 2563-2571. Mader, B. T.; Pankow, J. F. Environ. Sci. Technol. 2001, 35(17), 3422-3432. Volckens, J.; et al. Am. Ind. Hyg. Assoc. J. 1999, 60(5), 684-689. Arnold, S. T.; Viggiano, A. A.; Morris, R. A. J. Phys. Chem. A 1997, 101(49), 9351-9358. Kaupp, H.; Umlauf, G. Atmos. Environ. 1992, 26A(13), 22592267. Cardello, N., et al. Aerosol Sci. Technol. 2002, 36(2), 162-165. Volckens, J.; Leith, D. Environ. Sci Technol. 2002, 36, 46134617. CRC. Handbook of Physical Properties of Organic Chemicals; Howard, P. H., Meylan, W. M., Eds.; CRC Press: Boca Raton, FL, 1997. Willeke, K.; Baron, P. Aerosol Measurement: Principles, Techniques, and Applications; John Wiley & Sons: New York, 1993. Boelter, K. J.; Davidson, J. H. Aerosol Sci. Technol. 1997, 27, 689-708. Grosjean, E.; Grosjean, D. Environ. Sci. Technol. 1997, 31(8), 2421-2427. Grosjean, E.; Grosjean, D. Intl. J. Chem. Kinetics 1995, 27, 10451054. Liu, S.; Dasgupta, K. Anal. Chem. 1996, 68, 3638-3644.
Received for review April 26, 2002. Revised manuscript received August 20, 2002. Accepted August 27, 2002. ES0207100
