Ambient Air Measurement of Acrolein and Other Carbonyls at the

the San Francisco Bay Bridge toll plaza during rush hour traffic, which may be considered a “worst case scenario” for outdoor airborne carbonyls. ...
0 downloads 6 Views 144KB Size
Download PDF
Environ. Sci. Technol. 2002, 36, 2227-2235

Ambient Air Measurement of Acrolein and Other Carbonyls at the Oakland-San Francisco Bay Bridge Toll Plaza

together with the advancement of new catalysts and fleet turnover throughout the 1990s, are likely to account for part of the gap between our determination and the 1996 levels.

HUGO DESTAILLATS, REGGIE S. SPAULDING, AND M. JUDITH CHARLES* Department of Environmental Toxicology, University of California, Davis, California 95616

Acrolein (2-propenal, acrylic aldehyde), a contaminant commonly present in urban and rural atmospheres, is a product of incomplete combustion (e.g., wood burning, forest fires, motor vehicle emissions) (1). Motor vehicle emissions are considered a major source of acrolein to the environment, accounting for approximately 75% of its total emissions in the United States (2). In motor vehicle emissions, acrolein constitutes 3-10% of the total aldehydes (1). Acrolein is also generated by primary photooxidation of hydrocarbons, particularly 1,3-butadiene, a hydrocarbon prevalent in motor vehicle exhaust (3, 4). Laboratory studies establish that hydroxyl radical oxidation of 1,3-butadiene yields acrolein and that further oxidation of acrolein yields glyoxal, glycolaldehyde, and malonaldehyde (4). The mutagenic and genotoxic properties of R,β-unsaturated carbonyls and dicarbonyls, including acrolein, glyoxal, and malonaldehyde, are well-established. The major pathway by which these compounds induce toxicity is presumably by the formation of DNA adducts (5-13). Acrolein is a severe lung irritant that, at high acute exposures, can induce oxidative stress and delayed-onset lung injury, including asthma, congestion, and decreased pulmonary function (14, 15). Although little is known about the health effects resulting from chronic exposure to low levels of acrolein, the inhibition of cell proliferation and an enhancement of apoptosis in the presence of other toxins was reported (15). Because of concerns about adverse human health effects posed by acrolein in ambient air, the California Office of Environmental Health Hazard Assessment (OEHHA) stipulates a Chronic Inhalation Reference Exposure Level (REL) for acrolein of 0.06 µg/m3 (16). The only other carbonyls regulated by OEHHA are formaldehyde, acetaldehyde, and glutaraldehyde, with REL of 3.00, 9.00, and 0.08 µg/m3, respectively. Levels of acrolein exceeding 0.06 µg/m3 were reported by the U.S. Environmental Protection Agency (U.S. EPA) in its National Air Quality and Emissions Trend Report 1998 at urban and rural locations based on 1996 data (0.20 and 0.12 µg/m3, respectively) (17) in a study of on-road emissions (0.102-0.315 µg/m3) (18) and in ambient air in Rome, Italy (0.4-1.5 µg/m3) (19). By using the Assessment System for Population Exposure Nationwide Model (ASPEN), a Gaussian dispersion model, the U.S. EPA predicts median ambient air concentrations of acrolein higher than 0.06 µg/m3 in many urban areas (20). The results also predict that, in California, ambient air concentrations ranging from 0.012 to 0.28 µg/m3 will occur in the San Francisco Bay area (San Francisco, Alameda, Santa Clara, and San Mateo counties), the Los Angeles area (Los Angeles and Orange Counties), and in San Diego County. Moreover, of the 188 hazardous air pollutants, the highest noncancer health risk was attributed to exposures to acrolein (2, 20-23). Despite the known risks that R,β-unsaturated carbonyls and dicarbonyls pose to human health, ambient air measurements of acrolein and glyoxal are sparse, and to our knowledge, malonaldehyde has not been reported in ambient air. The lack of data to validate the ASPEN model is primarily due to the absence of a suitable analytical method. The traditional method to measure carbonyls relies on deriva-

Interest in ambient concentrations of acrolein and other R,β-unsaturated aldehydes and dicarbonyls (e.g., crotonaldehyde, methyl glyoxal, glyoxal, malonaldehyde (malondialdehyde)) is growing because either they exist at high levels in motor vehicle emissions or they arise from photooxidation of other hydrocarbons emitted from mobile sources. In addition, their mutagenic, genotoxic, or carcinogenic properties are well-established, and the results of a dispersion-modeling study regarding the health risks posed by the 188 hazardous air pollutants in California attributes the highest noncancer risk to exposure to acrolein. Such modeling studies, conducted by the U.S. Environmental Protection Agency (U.S. EPA), also predict median ambient air concentrations of acrolein higher than 0.06 µg/m3, the chronic inhalation reference exposure level stipulated by the California Office of Environmental Health Hazard Assessment in counties surrounding the Oakland-San Francisco Bay Bridge. We measured acrolein and other potentially toxic carbonyls in air sampled at the San Francisco Bay Bridge toll plaza during rush hour traffic, which may be considered a “worst case scenario” for outdoor airborne carbonyls. We identified 36 carbonyls in the sample extracts, including 14 saturated aliphatic carbonyls, six unsaturated carbonyls, four aromatic carbonyls, six dicarbonyls, and six hydroxy carbonyls. Structural information to support tentative identification of carbonyls and hydroxycarbonyls was obtained by using a method that involves O-(2,3,4,5,6-pentafluorobenzyl)hydroxylamine (PFBHA) and PFBHA/bis(trimethylsilyl)trifluoroacetamide (BSTFA) derivatization in concert with gas chromatography/ ion trap mass spectrometry. Most notably, we report for the first time the presence of malonaldehyde in the ambient atmospheric environment. A relatively linear relationship between retention time and the molecular weight of the derivatives was established to assist in obtaining structural information about chemicals for which authentic standards are not readily available. Levels of acrolein exceeded the California reference exposure level during morning rush hour traffic. The measured values, however, were significantly lower than estimates of county-wide average acrolein concentrations predicted by a U.S. EPA modeling study based on 1996 data. Successful regulatory efforts such as the introduction of reformulated gasoline, * Corresponding author phone: 530-754-8757; fax: 530-752-3394; e-mail: [email protected]. 10.1021/es011394c CCC: $22.00 Published on Web 04/17/2002

 2002 American Chemical Society

Introduction

VOL. 36, NO. 10, 2002 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

2227

tization with 2,4-dinitrophenylhydrazine (DNPH), followed by separation and detection of the hydrazones with HPLC and UV-visible absorption. However, R,β-unsaturated carbonyls are unstable when collected in stainless steel canisters (24, 25) or in water, both in its underivatized form (26) and as a DNPH derivative (26-28). Therefore, it is expected that the DNPH method underestimates acrolein, crotonaldehyde, and other carbonyls of environmental concern. In our laboratory, we utilize a method based on O-(2,3,4,5,6pentafluorobenzyl)hydroxylamine (PFBHA) derivatization in concert with gas chromatography/ion trap mass spectrometry (GC/ITMS) (29-32). To improve chromatography and sensitivity, the hydroxyl and carboxyl groups on the PFBHA derivatives are further derivatized with bis(trimethylsilyl)trifluoroacetamide (BSTFA). Advantages of this method, compared to the traditional DNPH method, are (1) the PFBHA derivatives of acrolein, crotonaldehyde, glyoxal, and methyl glyoxal are more stable than the DNPH derivatives of these compounds (26-28); (2) R-dicarbonyls can be distinguished from hydroxycarbonyls; and (3) unambiguous molecular weight determinations are afforded by interpretating the electron ionization, methane chemical ionization, and pentafluorbenzyl alcohol chemical ionization ion trap mass spectra. Herein, we apply the method to measure acrolein and other airborne carbonyls in ambient air at the Oakland-San Francisco Bay Bridge toll booth plaza. The measurements provide preliminary information on whether ambient air concentrations of acrolein predicted by the ASPEN model are likely to occur under what may be considered a “worst case scenario” and whether toll booth attendants may be a susceptible population at risk to exposure to acrolein.

Experimental Section Site Description and Sampling. We sampled at the OaklandSan Francisco Bay Bridge toll plaza during three periods of rush hour traffic: April 23, 2001, from 3:00 to 7:00 p.m. (sample A) and April 24, 2001, from 6:00 to 10:00 a.m. and from 3:00 to 7:00 p.m. (samples B and C, respectively). The toll plaza consists of 17 toll booths and an administration building. The Oakland-San Francisco Bay Bridge is characterized by the busiest commuter traffic in northern California, with nearly a quarter of a million vehicles traveling across the bridge per day. The toll booth plaza is located in Oakland, CA, at the east end of the bridge. The sampling was conducted on the sidewalk between the building and the first booth, with the air inlet at a distance of 50 cm from the passing traffic and 1 m from the ground. Air was sampled by using four impingers in series filled with 20 mL of PFBHA aqueous solution (0.25 mg/mL). The impingers were immersed in an ice/water bath during sampling. A potassium iodide (KI)-coated annular denuder was connected to the first impinger to remove ozone prior to collection of the carbonyls. The annular denuder consists of a 1 m × 6.2 mm internal diameter stainless steel tubing coated with KI(s). Previous work demonstrates that the KIcoated annular denuder removes 99.5% of ozone from a 1 ppm ozone standard sampled at a flow rate of 2 L/min (30). Ambient air was drawn through the impingers with a 3/4 HP pump at a fixed flow rate during 4 h in each experiment. The flow rate was controlled by means of two mass flow controllers (HFM 205 Hastings Instruments), with a precision better than 3%. Duplicate samples were collected in each experiment. The air flow rate ranged from 0.9 to 1 L/min in samples A and B and was reduced to 0.45 to 0.6 L/min for sample C. In each case, a field blank and matrix spike solution were prepared and kept in the ice bath aside the impingers. Internal standards were added to each sample, the blank, and the matrix spike solutions immediately after the sample was collected. 2228

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 36, NO. 10, 2002

Reagents. Authentic standards of formaldehyde, acetaldehyde, acetone, isobutyraldehyde, methyl ethyl ketone, 2-pentanone, hexanal, octanal, decanal, dodecanal, acrolein, methyl vinyl ketone, methacrolein, crotonaldehyde, p-tolualdehyde, glyoxal, methyl glyoxal, glycolaldehyde, hydroxyacetone, 3-hydroxy-3-methyl-2-butanone, 5-hydroxy2-pentanone, glutaraldehyde, 2,3-butanedione, 2,3-pentanedione, 2,4-pentanedione, 3-ethyl-2,4-pentadione, and 2-hexanone were purchased from Aldrich Chemical Co. (Milwaukee, WI). Benzaldehyde was purchased from Acros Organics (NJ). The internal standards 13C3-acetone and 4-fluorobenzaldehyde were purchased from Aldrich, and the internal standard 4-hydroxy-13C6-benzaldehyde was purchased from Cambridge Isotope Laboratories (Andover, MA). These standards were utilized without further purification. O-(2,3,4,5,6-pentafluorobenzyl)hydroxylamine was purchased from Aldrich and further purified by recrystallizing 2 times in distilled isopropyl alcohol (Fisher Scientific, Pittsburgh, PA). High purity HPLC-quality dichloromethane was purchased from Burdick and Jackson (Muskegon, MI) and distilled prior to use. The HPLC grade water (Fisher Scientific) employed in the sampling was distilled over KMnO4 to oxidize the organics in the water. The glassware was silanized by immersion in a 15% v/v solution of dichlorodimethylsilane in toluene to minimize analyte adsorption to the container walls. Derivatization and Extraction. All samples were kept in the dark at room temperature for at least 24 h to ensure complete derivatization. The PFBHA derivatives were then extracted in dichloromethane after acidification of the samples with H2SO4. Residual water in the extract was eliminated by passing the extracts through an anhydrous NaSO4(s) column (8 cm length, 5 mm i.d.). The eluate was collected and reduced to 100 µL by evaporation of the solvent under a mild N2(g) stream. The extract was split into two (50 µL) aliquots. The aliquot containing the PFBHA derivatives was analyzed directly. The PFBHA derivatives of carbonyls containing hydroxyl or carboxyl moieties in the second aliquot were silylated by using a 1:10 mixture of BSTFA and trimethylchlorosilane (TMCS, catalyst). High-Resolution Gas Chromatography/Ion Trap Mass Spectrometry (HRGC/ITMS). A Varian Star 3400 CX gas chromatograph interfaced to a Saturn 2000 ion trap mass spectrometer was utilized. The injector temperature was programmed to maintain a temperature of 280 °C during the first minute and then increased to 310 °C at 50 °C/min (splitless injection mode). High-resolution gas chromatographic separation was performed on a 30 m × 0.25 mm (i.d.) DB-XLB HRGC column with a 0.25 µm film thickness (Agilent Technologies, Folsom, CA). The capillary column has a liquid-phase formula equivalent to 12% (phenylmethyl)polysiloxane. The oven temperature was programmed to maintain a temperature of 70 °C for 1 min. The oven temperature was increased to 100 °C at a rate of 5 °C/min, then increased to 280 °C at a rate of 10 °C/min, and to 310 °C at a rate of 30 °C/min. The temperature was maintained at 310 °C/min for 10 min. The electron ionization (EI) mass spectra were acquired by using an ion trap temperature of 150 °C with automatic gain control. A maximum ionization time of 25 000 µs and ion target of 20 000 was employed, and the masses were scanned from 50 to 650 amu. Methane chemical ionization (methane CI) ion trap mass spectra were obtained with a filament current of 10 µA and an ion trap temperature of 150 °C. Automatic reaction control using a maximum ionization time of 500 µs, a maximum reaction time of 80 ms, and an ion target of 5000 was employed. Methane was introduced into the ion trap until a ratio of about 1:1 was obtained between the ions at m/z 17 and 29. The masses

were scanned from 50 to 650 amu. Vapor of pentafluorobenzyl alcohol (PFBOH) was introduced in the trap by means of an ampule containing solid PFBOH in a thermostatic bath according to a method previously described (29). The PFBOH chemical ionization (PFBOH CI) mass spectra was obtained by using automatic reaction control with maximum ionization time of 2000 µs, a maximum reaction time of 128 ms, and ion target of 5000. The masses were scanned from 230 to 650 amu. Identification and Quantification. The approach to identify the carbonyls was described in previous work (30). Briefly, the retention time and relative retention time of the carbonyls in the sample extract are obtained from the m/z 181 ion EI chromatogram. The molecular weight of the carbonyls is determined by interpreting the methane CI and PFBOH CI mass spectra. We identified all compounds whose (M + H)+ ion in the chemical ionization mass spectra was present at a signal-to-noise ratio greater than 3:1. Tentative identification was based on interpretation of the mass spectra. The identification was confirmed by comparing the relative retention time and ion trap mass spectra of the PFBHA or PFBHA/BSTFA derivative in the sample extract to those of an authentic standard. Quantification was accomplished by using internal standardization. Five solutions containing the native compounds in concentrations ranging from 50 to 1000 pg/µL and the internal standard were analyzed to construct a calibration curve. A linear regression equation was obtained by plotting the [(peak areaderivative of analyte/(peak areaderivative of internal standard)] versus the concentration of the analyte. Solutions used to construct the calibration curve were analyzed before and after each set of sample extracts, typically comprised of eight sample extracts. A solution containing 250 pg/µL (i.e., a midpoint standard) was analyzed after every four sample extracts to verify the stability of the calibration curve. 13C -Acetone (retention time t ) 9.65 min) was employed 3 r as the internal standard to quantify saturated and unsaturated aliphatic carbonyls. 4-Fluorobenzaldehyde (tr ) 17.65 and 17.77 min) was utilized as the internal standard to quantify aromatic carbonyls and dicarbonyls, and 4-hydroxy-13C6benzaldehyde (tr ) 21.61 min for the PFBHA derivative and 21.78 for the PFBHA/BSTFA derivative) was employed as the internal standard to quantify hydroxycarbonyls. In most cases, correction for blank was not necessary. When the analyte was present in the blank, the concentration in the blank sample extract was subtracted from the concentration measured in the sample extract.

Results and Discussion Identification of Carbonyls. We identified 14 saturated aliphatic carbonyls, six unsaturated aliphatic carbonyls, four aromatic carbonyls, six dicarbonyls, four saturated hydroxy carbonyls, and two unsaturated hydroxy carbonyls in the sample extracts. The compounds identified, the molecular weight of the underivatized and derivatized compounds, and the retention times and relative retention times are presented in Tables 1 and 2. The identities of formaldehyde, acetaldehyde, acetone, isobutyraldehyde, methyl ethyl ketone, 2-pentanone, hexanal, octanal, decanal, dodecanal, acrolein, methyl vinyl ketone, methacrolein, crotonaldehyde, benzaldehyde, p-tolualdehyde, glyoxal, methyl glyoxal, glycolaldehyde, hydroxyacetone, 3-hydroxy-3-methyl-2-butanone, and 5-hydroxy-2-pentanone were confirmed. We tentatively identified butanal, malonaldehyde, butenedial, and carbonyls with the following elemental formula: C6H12O, C9H18O, C6H8O, C8H12O, C8H8O, C4H6O2, C5H8O2, C4H6O2 and C7H10O2. Identification of Malonaldehyde in the Presence of Coeluting Interferences. The identification of malonaldehyde was possible by interpreting the EI, methane CI, and PFBOH CI mass spectra. We present these mass spectra in Figure 1

to illustrate the power of this approach. In the EI mass spectra, the presence of a carbonyl is evident by the m/z 181 ion. The ion at m/z 323 is the (M)+ ion of the PFBHA derivative of 4-hydroxy-13C6-benzaldehyde, the internal standard. Thus, in this case, one might identify the compound as the internal standard if identification was based solely on the retention time of the internal standard and interpreting the EI mass spectra. Insight into the possibility of the presence of other coeluting compounds was gleaned from interpretation of the methane CI mass spectra. In the CI mass spectra of PFBHA derivatives, the base peak is generally the (M + H)+ ion (29, 30). On the basis of this information, we would identify the m/z 324 ion as the (M + H)+ ion of 4-hydroxy-13C6benzaldehyde. The m/z 126 ion, would then be the (M - 197)+ ion, which is commonly observed in the EI and CI methane mass spectra of PFBHA carbonyl derivatives. We hypothesize that the m/z 380 and m/z 463 are (M + H)+ ions of compounds that coelute with the internal standard. This hypothesis can be evaluated by interpreting the PFBOH CI mass spectra. In the PFBOH CI mass spectra, (M)+, (M + H)+, and (M - H)+ ions can be observed, and the relationship between these ions and the (M + 181)+ ion facilitates molecular weight determinations of the compound (29-32). Accordingly, in the PFBOH CI mass spectra, we identify the ions at m/z 323 and 504 as the (M)+ and (M + 181)+ ions of PFBHA derivative of 4-hydroxy-13C6-benzaldehyde. The relationship between the ions at m/z 378 and 560 ions indicates that these ions are the (M - H)+ and (M + 181)+ ions, respectively, of a coeluting compound. Similarly, the relationship between the ions at m/z 463 and 643 indicate that these ions are the (M + H)+ and (M + 181)+ ions of another coeluting compound. Possible elemental formulas for these compounds would be C12H24O and C3H4O2. We identified the C12 carbonyl as dodecanal and tentatively identified the C3 dicarbonyl as malonaldehyde. Because methylglyoxal, which eluted earlier, is the only other possible alternative for a C3 dicarbonyl with that molecular formula these data and the retention time data (see following section) strongly support the identification of malonaldehyde. To our knowledge, this is the first measurement of malonaldehyde in the ambient atmospheric environment. Determination of Structural Characteristics on the basis of the Retention Time. Given the several structural possibilities for an elemental formula, we explored the relationship between the structure of the PFBHA and PFBHA/BSTFA derivatives and their retention times. Although the molecular weight is the major parameter determining the elution times of structurally similar compounds, molecules bearing π-electron systems (i.e., double bonds, aromatic rings) are likely to interact more strongly with the liquid phase, thus exhibiting higher retention times than saturated molecules of similar mass. In Figure 2, we plot the retention times of PFBHA and PFBHA/BSTFA derivatives versus the molecular weight of the derivative (the retention time and relative retention time of the analytes are presented in Tables 1 and 2). The data presented in Figure 2 also include the chromatographic information corresponding to the three internal standards employed in this work. A linear relationship between the retention time of the PFBHA derivatives of saturated carbonyls and molecular weight of the derivative is evident. A regression equation of tr ) (-15.82 + 0.101 MW) min with an R2 value of 0.992 was calculated for saturated carbonyls ranging from C1-C12. Accordingly, saturated carbonyls, for which there were no standards available, are indicated as tentatively identified if the retention time of the PFBHA and molecular weight fall within an acceptable error (5-10%) of this regression line. Although fewer unsaturated and aromatic carbonyls were measured, the data indicate a linear relationship between the retention time of these compounds and the molecular weight of the PFBHA derivatives. UnsatVOL. 36, NO. 10, 2002 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

2229

TABLE 1. Carbonyls Identified in Extracts of Air Sampled on the Oakland-San Francisco Bay

urated and aromatic carbonyl derivatives exhibit higher retention times than the corresponding saturated carbonyls at the same mass to charge (m/z) ratio. The regression equation for the line is tr ) (-26.44 + 0.145 MW) min with an R2 value of 0.977. If the retention time and molecular weight of an unknown PFBHA derivative falls within an acceptable range of this line, an unsaturated or aromatic carbonyl is indicated. Similarly, a linear relationship between the retention time and the molecular weight of the PFBHA/ BSTFA derivatives of saturated hydroxycarbonyls was pos2230

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 36, NO. 10, 2002

tulated (see Figure 2B). Although the data are limited, the aromatic and unsaturated hydroxycarbonyls elute at greater retention times than the extrapolated linear behavior of the saturated hydroxycarbonyl derivatives at molecular weights up to 400 amu. The derivatives corresponding to dicarbonyls, being the heaviest compounds analyzed, elute at longer times. For these compounds, the retention times are relatively shorter than the linear prediction based on the single carbonyl/PFBHA derivatives curve (Figure 2A). In summary, we used retention time data to obtain structural information

TABLE 1 (Continued)

TABLE 2. Hydroxycarbonyls Identified in Extracts of Air Sampled on the Oakland-San Francisco Bay Bridge

about the analyzed compounds. In all cases, the compounds reported as tentatively identified (white data points) follow the linear behavior exhibited by the standards of each category, which is an additional criterion for identity confirmation (see Figure 2, parts A and B). Many of the compounds identified are common photooxidation and combustion products of hydrocarbons previously identified in motor vehicle exhaust (33, 34). Acrolein, malonaldehyde, methyl glyoxal, glyoxal, formaldehyde, and glycoaldehyde are generated from hydroxyl radical oxidation of 1,3-butadiene (3, 4). Acrolein is also formed by the reaction of O3 with 1,3-butadiene (4). Hydroxy acetone and glycolaldehyde are common photooxidation products of alklybenzenes, constituents of gasoline (35). The highest level for

acrolein, methyl glyoxal, and glycolaldehyde were observed in the extract of air collected at 6:00-10:00 a.m., and the highest concentration of hydroxy acetone was obtained in the extract of air collected from 3:00-7:00 p.m. Photochemical oxidation is indicated by the high concentrations of these species, and specifically high concentrations of glycoladehyde and hydroxyacetone. Although it is feasible that OH radical oxidation of 1,3-butadiene could be a primary source of acrolein during daytime and ozone could be the primary source during nighttime and early morning hours, the data are insufficient to draw any conclusions regarding the relative importance of direct emissions versus photooxidation as a source of acrolein. VOL. 36, NO. 10, 2002 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

2231

FIGURE 1. Ion trap mass spectra of coeluting compounds in an extract of air sampled at the Oakland-San Francisco Bay Bridge. The internal standard (I.S.) indicated is the PFBHA derivative of 4-hydroxy-13C6-benzaldehyde. Quantification. We quantified carbonyls for which the maximum retention was observed in the first or the second impinger, assuming that at least 80% of the total ambient concentration was measured in all four impingers (see Table 3). For analytes detected in the fourth impinger, the concentration measured may be less than the true value due to breakthrough. For this reason, we estimated the lower and upper range of the concentration by considering the percent recovery (reported in Table 4) and the extent of breakthrough in each case. The minimum (min) value of this range corresponds to the lowest concentration reported in each duplicate, and the maximum (max) value was calculated as follows:

max value ) min value[1 + (1 - % recovery × 10-2) + f ] (1) where f ) 0.20, if the average retention in the fourth impinger is g20%; f ) 0.10 if this value lies between 10% and 20% (most of the cases); f ) 0.05 when the retention in the fourth impinger corresponds to less than 10% of the total sample; and f ) 0 in the absence of breakthrough. When the maximum concentration measured was higher than the value calculated by using eq 1, the measured value was also used as the maximum value. The estimated ranges of concentrations for the eight quantified carbonyls are reported in the last column of Table 3. The most abundant carbonyls generated by vehicle 2232

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 36, NO. 10, 2002

emissions (e.g., acetone, formaldehyde, acetaldehyde) were present in high concentrations in the fourth impinger. While we do not report quantitative information about these compounds, the amount quantified in the impingers correspond to ambient levels that are similar to previously reported values, nearly 1 order of magnitude higher than acrolein (18). The method provides reasonable precision, as indicated by a relative deviation between duplicate samples, generally less than 25% for six of the seven analytes quantified in sample extracts A and B and for five of the seven analytes quantified in sample extract C. The method detection limits for the analytes are presented in Table 4. The accuracy of the method, as indicated by the mean percent recovery was >72%. Linear calibration curves are indicated by mean relative response factors