Measurement of Polycyclic Aromatic Hydrocarbons Associated with

View: PDF | PDF w/ Links | Full Text HTML. Citing Articles; Related ... V. Kislov and Ralf I. Kaiser. Journal of the American Chemical Society 2008 13...
0 downloads 0 Views 342KB Size
Download PDF
Environ. Sci. Technol. 1996, 30, 1023-1031

Measurement of Polycyclic Aromatic Hydrocarbons Associated with Size-Segregated Atmospheric Aerosols in Massachusetts JONATHAN O. ALLEN, NAMEETA M. DOOKERAN, KENNETH A. SMITH, AND ADEL F. SAROFIM* Chemical Engineering Department, Massachusetts Institute of Technology, Cambridge, Massachusetts 02139

KOLI TAGHIZADEH AND ARTHUR L. LAFLEUR Center for Environmental Health Sciences, Massachusetts Institute of Technology, Cambridge, Massachusetts 02139

Size-segregated atmospheric aerosols were collected from urban and rural locations in Massachusetts using a micro-orifice impactor. The samples were analyzed for polycyclic aromatic hydrocarbons (PAH) with molecular weights between 178 and 302, using gas chromatography/mass spectrometry. Fifteen PAH were quantified in the urban samples and nine in the rural samples. The quantification results are in good agreement with available ambient monitoring data. In the urban samples, PAH were distributed among aerosol size fractions based on molecular weight. PAH with molecular weights between 178 and 202 were approximately evenly distributed between the fine (aerodynamic diameter 2 µm) aerosols. PAH with molecular weights greater than 228 were associated primarily with the fine aerosol fraction. In the rural samples, low and high molecular weight PAH were associated with both the fine and coarse aerosols. Slow mass transfer by vaporization and condensation is proposed to explain the observed PAH partitioning among aerosol size fractions.

Introduction Polycyclic aromatic hydrocarbons (PAH) are mutagenic air pollutants formed as byproducts of combustion. After formation and emission, PAH partition between the gas phase and atmospheric aerosols. Atmospheric particles are classically grouped into the ultrafine, accumulation, and coarse size modes (1). Ultrafine mode aerosols have an aerodynamic diameter (Dp) less than ≈0.1 µm; they are * Corresponding author e-mail address: [email protected]; fax: (617)253-2072.

0013-936X/96/0930-1023$12.00/0

 1996 American Chemical Society

emitted from combustors and formed in the atmosphere by homogeneous nucleation. Accumulation mode aerosols are in the approximate range 0.1 < Dp < 2.0 µm and are formed by coagulation and condensation. Coarse mode aerosols, those larger than ≈2 µm, are generated by mechanical attrition and as sea spray. The environmental fate of PAH depends, in part, on the distribution of the PAH among the aerosol size fractions. Particle size affects the removal rate of the associated PAH from the atmosphere by dry and wet deposition (2, 3). The mechanism and location of deposition of particulate phase PAH in the lung are also affected by particle size. The large particles tend to impact on the upper regions of the lung, and the small particles diffuse to the surface of the alveoli (4). Therefore, measurements of the amount of PAH associated with different aerosol size fractions are necessary for a complete understanding of the environmental fate of and human exposure to PAH. Since 1975, measurements of the distribution of PAH with particle size have been performed in and around Toronto, Los Angeles, Antwerp, Barcelona, and Paris (514). In these studies, PAH were found predominantly associated with fine aerosols (Dp < 2.0 µm). PAH of the same molecular weight were observed to partition similarly among atmospheric aerosols (10, 8). PAH were found to partition to larger aerosols in warmer periods (5, 7-9) and at sites away from emission sources (5, 10, 13). The objective of this work was to apply recent advances in aerosol sampling and PAH analysis techniques to develop an improved method for measuring PAH associated with size-segregated atmospheric aerosols. In the present study, aerosols were collected with a micro-orifice impactor (MOI) from sites in Boston and rural Massachusetts. PAH with molecular weights between 178 and 302 were identified and quantified by gas chromatography/mass spectrometry (GC/MS). A secondary objective of this study was to examine possible mechanisms of PAH partitioning in light of the measured distribution of PAH with particle size.

Experimental Methods Materials. The solvents used were glass-distilled OmniSolv dichloromethane (DCM), 99.99% purity, and HPLC grade Aldrich cyclohexane, 99.9% purity. Aldrich dibutyl phthalate, 99+% purity, was used. All glassware were cleaned in detergent and then rinsed three times with deionized water, methanol, and DCM. Newly purchased glassware was only rinsed with DCM. Polytetrafluoroethylene (PTFE) lined screw caps were used. Sample Collection. Urban and rural air samples were collected from the roofs of National Ambient Air Quality Standards monitoring stations operated by the Massachusetts Department of Environmental Protection (DEP). At both sampling sites, the air inlet was located 4 m above ground level. Urban samples were collected at the Kenmore Square sampling site, which is located 1.5 km west of downtown Boston, MA, at 590 Commonwealth Avenue (42°20′54′′ N, 71°05′57′′ W). This site is on a traffic island in the center of a divided six-lane street near a major intersection. A bus station is located 170 m away. Rural samples were collected at the Quabbin Summit site, which is located on the 81 000 acre Quabbin Reservation in central

VOL. 30, NO. 3, 1996 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

1023

TABLE 1

TABLE 2

Air Conditions during Sampling location Kenmore Square

date

June 16, 1994 June 18, 1994 June 20, 1994 June 22, 1994 June 24, 1994 Quabbin Summit July 19, 1994 July 21, 1994 July 25, 1994 July 29, 1994 July 31, 1994

tempa

Operating Characteristics of MOI a

a

(°C)

NOx O3 (ppm) (ppm)

23 29 24 25 21 24 26 25 22 23

0.071 0.015 0.066 0.033 0.100 0.008 0.007 0.005 0.005 0.006

PM10b (µg/m3)

33

0.063 0.059 0.062 0.045 0.036

23

a Daily average measured by Massachusetts DEP. b Monthly average measured by Massachusetts DEP.

Massachusetts (42°17′54′′ N, 72°20′05′′ W). The nearest urban center is Springfield, MA, located 30 km to the southwest. This site was chosen to sample rural air typical of that which enters the Boston airshed from the west. A total of five 24-h samples were collected at each sampling site on an alternate day schedule. Samples and field blanks were collected for 24 h from midnight to midnight. The Kenmore Square samples were collected on 5 days between June 16 and June 24, 1994. A total of 188 m3 of air was sampled over 120 h. The Quabbin Summit samples were collected on 5 days between July 19 and July 31, 1994. A total of 189 m3 of air was sampled over 119 h. All samples were retrieved the morning following the sampling period and stored in a freezer at -20 °C until they were analyzed. Table 1 shows the average air temperature and pollutant levels during the samplings period as measured by the Massachusetts DEP. The sampling train consisted of an inlet tube and a cascade impactor followed by a regulating valve, rotameter, and vacuum pump. The sampler inlet was a straight 0.6-m PTFE-lined tube with an inside diameter of 0.95 cm. The cascade impactor used was a micro-orifice impactor (MOI) manufactured by MSP Corporation (Minneapolis, MN) (15). The MOI collects size-segregated aerosols by impaction on nine stages. A 25-mm quartz after-filter downstream of the impactor collected particles that were not collected on the impactor stages. A Gast Manufacturing Corporation (Benton Harbor, MI) Model DAA-111-EB diaphragm pump was used. The entire sampling system was placed in a weatherproof enclosure. For the Kenmore Square sampling, the system was turned on and off manually; for the Quabbin Summit sampling, a 7-day timer controlled the system. An hour counter recorded the elapsed sample collection times. The inlet tube and all parts of the MOI were carefully cleaned with DCM before placement in the field. The impactor was tested for clogged nozzles in the laboratory by monitoring the downstream pressure. Before each day’s sampling, the system was checked for leaks in the field by sealing the inlet and monitoring the flow rate. The regulating valve was adjusted to a flow of 27 L/min before each sample collection. Rotameter readings were recorded before and after each sample collection. The sample flow rate was determined to be the average of the beginning and ending flow rates. The size of particles collected on the impaction plates is a function of the flow rate and operating pressures of the MOI. The diaphragm pump maintained an average sam-

1024

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 30, NO. 3, 1996

stage

no. of orifices

orifice diam (mm)

0 1 2 3 4 5 6 7 8

1 10 10 20 40 80 900 2000 2000

10.0 3.80 2.47 1.37 0.72 0.4376 0.1361 0.0594 0.0514

cutoff diameter (µm) design flow actual flow 18.0 5.62 3.16 1.78 1.00 0.585 0.320 0.131 0.080

19.2 6.00 3.38 1.90 1.07 0.626 0.343 0.141 0.087

pling flow rate of 26.3 L/min at standard temperature and pressure; less than the design flow rate of 30 L/min. This reduction in air flow and pressure drops across the impactor stages changed the sizes of particles collected. The aerodynamic diameter for which 50% of particles are collected on an impaction stage (Dp50) depends on the design and operation of the impaction jets as

Dp50 )

x

9µWSt50 FpCV0

(1)

where µ is the air viscosity; W is the nozzle diameter; St50 is the Stokes number corresponding to Dp50 [≈0.22 (16)]; Fp is the density of the particle; C is the slip correction; and V0 is the average velocity of the jet exiting the nozzle. Of these variables, only V0 depends on the volumetric flow rate (Q) and C depends on the operating pressure (P). Pressure drops for a flow rate of 26.3 L/min were measured, and actual Dp50 values were subsequently calculated as

Dp50actual ) Dp50design

(

)

C(Pdesign)Qdesign C(Pactual)Qactual

1/2

(2)

Operating the MOI with a 26.3 L/min flow rate increased the cutoff diameters to approximately 1.07 times the design values. Table 2 lists Dp50 for the design and actual flow rates. Changes in flow during and between sample collections would introduce variability in the size of particles collected on each impactor stage. The flow measured at the beginning and end of a run differed by less than 10%. This variation would cause a change of approximately 5% in the impactor stage cutoff sizes. Because the flow was not recorded during sampling and these sampling errors introduce only minor variations in Dp50, for the remainder of this work the air flow is taken to be constant. A PTFE membrane with an underlay of aluminum foil was placed on each stage of the MOI to collect particles for analysis. The PTFE membranes used in this work were Millipore (Bedford, MA) Mitex, 10-µm pore size membranes. The PTFE membrane was cleaned by sonication in DCM for 5 min. The aluminum was cleaned with DCM in a Soxhlet extractor for 24 h. Each impaction medium was coated with approximately 0.1 mL of 20% dibutyl phthalate solution in cyclohexane to reduce particle bounce (17). The after-filters were Pallflex (Putnam, CT) Tissuquartz 2500 QAT-UP quartz fiber filters that were baked by the manufacturer. The filters were sonicated in DCM for 5 min. Impaction and filter media blanks were prepared in the same manner as the sampling media.

TABLE 3

Selected Ion Monitoring Program eluting compounds

time after injection (min)

internal standard

naphthalene

naphthalene-d8

15.2-22.6

acenaphthylene acenaphthene fluorene

acenaphthene-d10

22.6-26.8

phenanthrene anthracene

phenanthrene-d10

26.8-32.2

fluoranthene pyrene

pyrene-d10

32.2-36.6

benzo[ghi]fluoranthene benzo[c]phenanthrene cyclopenta[cd]pyrene benz[a]anthracene chrysene triphenylene naphthacene benzo[b]fluoranthene benzo[k]fluoranthene benzo[j]fluoranthene benzo[e]pyrene benzo[a]pyrene perylene indeno[1,2,3-cd]fluoranthene indeno[1,2,3-cd]pyrene dibenz[a,j]anthracene dibenz[a,c]anthracene dibenz[a,h]anthracene benzo[b]chrysene picene benzo[ghi]perylene anthanthrene naphtho[1,2-k]fluoranthene naphtho[2,3-b]fluoranthene naphtho[2,3-k]fluoranthene dibenzo[b,k]fluoranthene naphtho[2,3-a]pyrene dibenzo[a,l]pyrene dibenzo[a,e]pyrene dibenzo[e,l]pyrene dibenzo[a,i]pyrene dibenzo[a,h]pyrene coronene

chrysene-d12

8.0-15.2

36.6-40.8

40.8-45.5

45.5a-55.0

a

PAH

monitored ions 128 129 136 137 152 153 154 155 164 165 166 167 178 179 188 189 202 203 212 213 226 227 228 229 240 241

perylene-d12

252 253 264 265

dibenz[a,h]anthracene-d14

276 277 278 279 292 293

coronene-d12

300 301 302 303 312 313

Changed to 44.5 for Quabbin Summit samples.

Sample Preparation. Samples from the impactor stages and the after-filter were split in half; then the halves from each of the 5 collection days were composited for analysis. To the composited sampling media, 10 µL of a DCM solution containing 4 µg/mL of eight deuterated PAH internal standards was added. The deuterated PAH internal standards were naphthalene-d8, acenaphthene-d10, phenanthrene-d10, pyrene-d10, chrysene-d12, perylene-d12, dibenz[a,h]anthracene-d14, and coronene-d12. The sampling media were then covered with DCM and sonicated for 30 min. The particle/liquid suspension and additional DCM used to rinse the jar were then filtered by syringe using a

0.2-µm pore size PTFE filter. The filtrate was evaporated under clean dry N2 at room temperature to a volume of approximately 1 mL. To test the efficacy of the extraction protocol, the Kenmore Square impactor stage 7 sample was extracted twice. For all quantified PAH, the second extraction contained less than 3% of the amount initially recovered. The PAH fraction of the samples was separated from dibutyl phthalate by high-performance liquid chromatography (HPLC). The HPLC system was a Beckman System Gold equipped with a Programmable Solvent Module 126 and a Programmable Detector Module 166 controlled by

VOL. 30, NO. 3, 1996 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

1025

System Gold software on a personal computer. The separation was accomplished by two Jordi-Gel poly(divinylbenzene) columns in series. The columns were 500 mm long, 10 mm i.d., and packed with 5 µm particle size, 500-Å pore size divinylbenzene polymer. They were purchased from Jordi Associates (Billingham, MA). The separation was performed isocratically with 1.0 mL/min flow of DCM at a pressure of approximately 700 psi. The UV detector was unable to detect the small amounts of PAH present in the samples, on the order of 10 ng. Therefore, HPLC effluent was collected from 57 to 77 min after injection. These times were chosen based on injections of PAH and dibutyl phthalate, which showed that naphthalene elutes at 57.2 and coronene elutes at 68 min after injection. The separation of phthalates and PAH using this type of column has been previously reported (18). Sample Analysis. PAH were identified and quantified with a Hewlett Packard GC/MS system consisting of an HP Model 5890 Series II Plus gas chromatograph (GC) and an HP Model 5972 mass selective detector (MSD). The MSD was operated in electron impact mode with electron energies of 70 eV. The GC column was a 30 m HP-5 0.25 mm i.d. capillary column coated with 0.25 µm film thickness 5% cross-linked phenyl methyl siloxane stationary phase. The GC temperature was held at 50 °C for 1.5 min and then ramped to 310 °C at 6 °C/min. The final temperature was held for 10 min. The injector temperature was maintained at 280 °C. The GC/MS was controlled and data were collected by HP ChemStation software running on a personal computer. The MSD was run in selected ion monitoring (SIM) mode. The SIM program was designed to monitor the molecular [M+] and 13C isotope [(M + 1)+] ions of a group of PAH that elute at times near one of the deuterated PAH internal standards. Table 3 shows the SIM program and PAH detected in each time window. The MSD dwell time for each mass was set to between 20 and 50 ms to maximize the dwell time while maintaining a scan rate of 200 scans per minute. This scan rate allowed collection of at least 20 data points for typical peaks. At least 10 data points were collected for peaks at the lower limit of detection. The MSD was tuned at the start of a series of runs using the HP ChemStation AutoTune routine with perfluorotributylamine. Additionally, the HP ChemStation QuickTune routine was run at the start of each day. A midrange calibration standard containing 3 or 10 ng/µL of the internal standards was also run at the start of each day to confirm system performance. Before each sample injection, DCM blanks were run until all monitored ion signals were reduced to background levels. At least three duplicate GC/MS injections were made for each sample. For the first GC/MS analysis of a sample, the HPLC column effluent was evaporated under clean dry N2 at room temperature to a volume of approximately 5 µL. One microliter of the sample solution was injected onto the GC/MS. Before subsequent injections, 5 µL of DCM was added to the samples because they evaporated to dryness when stored for 24 h. PAH were identified by comparing peak retention times and [(M + 1)+]/[M+] ion ratios to reference standards. The retention times were normalized by assigning retention indices relative to the deuterated PAH internal standards. Retention indices of the deuterated PAH were determined by GC/MS injections of standard solutions of PAH and deuterated PAH. Naphthalene, phenanthrene, chrysene,

1026

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 30, NO. 3, 1996

FIGURE 1. Chromatogram of all monitored ions for the Kenmore Square impactor stage 7 sample with identified compounds.

and benzo[ghi]perylene were assigned the retention indices used by Lee and co-workers of 200, 300, 400, and 501.32, respectively (19). Retention indices for naphthalene-d8, phenanthrene-d10, chrysene-d12, dibenz[a,h]anthracened14, and coronene-d12 were then calculated to be 199.49, 299.34, 399.15, 494.21, and 549.71. The retention indices in this work were very repeatable. Standard deviations of retention indices for PAH in 16 sample injections were less than 0.2 for all PAH except fluoranthene and pyrene. Retention index standard deviations for fluoranthene and pyrene were 1.8 and 0.8, respectively. This greater variability may be due to the elution of residual dibutyl phthalate a short time before fluoranthene. Fluoranthene and pyrene were readily identified because pyrene-d10 eluted between them and there were no interfering peaks with a mass to charge ratio of 202. PAH with retention indices differing by 1 unit are completely separated and individually identified. In cases where a number of PAH isomers have similar elution times, e.g., benzo[b]fluoranthene and benzo[k]fluoranthene, the isomers were quantified together. PAH were deemed identified if the [(M + 1)+]/[M+] ratio and retention time matched a reference standard. PAH were quantified if, in addition to being identified, the signal to noise ratio was greater than 20. Figure 1 is a representative chromatogram showing the abundance of all the monitored ions for the Kenmore Square impactor stage 7 sample. All of the prominent peaks are identifiable as either PAH or deuterated PAH. Each PAH was quantified by comparing the peak area for its molecular ion [M+] to the peak area for the molecular ion of the internal standard in the same SIM program time window. PAH to deuterated PAH responses were calibrated by triplicate injections of standards containing PAH at concentrations from 100 to 0.03 ng/µL and deuterated PAH at a fixed concentration of 2.2 ng/µL. The amount of PAH in a sample was calculated as

PAH (ng) ) 10b × deuterated PAH (ng) × PAH response deuterated PAH response

(

)

1/m

(3)

The constants m and b in eq 3 are the slope and intercept of a straight line fit to the logarithm of relative responses versus the logarithm of relative amounts of PAH and deuterated PAH. In all cases the linear fit was good, with R2 g 0.977.

TABLE 4

PAH Identified in Kenmore Square Samples PAH

mol wt

concn with aerosol (ng/m3)

naphthalene acenaphthylene acenaphthene fluorene phenanthrene anthracene fluoranthene pyrene benzo[ghi]fluoranthene cyclopenta[cd]pyrene benz[a]anthracene chrysene/triphenylene benzofluoranthenes benzo[e]pyrene benzo[a]pyrene perylene indeno[1,2,3-cd]pyrene dibenzanthracenes benzo[b]chrysene picene benzo[ghi]perylene dibenzo[a,l]pyrene/dibenzo[b,k]fluoranthene coronene

128 152 154 166 178 178 202 202 226 226 228 228 252 252 252 252 276 278 278 278 276 302 300

identified identified identified identified 13.28 ( 0.62a 1.28 ( 0.08 14.70 ( 0.41 8.07 ( 0.21 0.90 ( 0.04 identified 1.66 ( 0.08 1.78 ( 0.10 3.22 ( 0.10 1.34 ( 0.03 1.17 ( 0.02 0.22 ( 0.01 1.03 ( 0.05 0.15 ( 0.01 identified identified 0.82 ( 0.05 identified 0.11 ( 0.01

a

fraction with aerosol modes ultrafine accumulation coarse

0.03 0.06 0.06 0.07 0.10

0.36 0.43 0.48 0.50 0.61

0.61 0.51 0.46 0.43 0.29

0.11 0.10 0.16 0.19 0.18 0.14 0.25 0.17

0.72 0.71 0.74 0.71 0.74 0.79 0.68 0.75

0.17 0.19 0.10 0.10 0.08 0.07 0.07 0.08

0.36

0.58

0.06

0.51

0.44

0.05

One standard deviation.

It is generally accepted that isotopically labeled isomeric standards experience losses similar to the analyte throughout a workup procedure. However, the internal standards were added to the sampling medium prior to extraction, and they may have been extracted with a higher efficiency than compounds in the complex environmental matrix. A comparison of extraction efficiencies was not performed in this work due to the small sample size. Using urban air particulate matter (NIST SRM 1649), Burford and co-workers found extraction efficiencies of at least 80% for native PAH and 100% for spiked deuterated PAH using 30-min sonication in DCM (20). Based on these results, we assume that the extraction efficiencies of the native and spiked PAH were identical in this study. The sensitivity of the analytical equipment determines the lower limit of quantification. Peaks with a signal to noise ratio smaller than 20 could not be reliably integrated. From the response factor calibration standards, the quantifiable amount of PAH was approximately 0.03 ng in a 1-µL injection. Because at least 5% of the total sample was injected for each GC/MS run, the limit of quantification for PAH in a 200-m3 air sample was ≈0.003 ng/m3. Peaks with a signal to noise ratio between 10 and 20 could be identified but not quantified. Thus, the limit of detection was ≈0.0015 ng/m3. Blank samples were analyzed to quantify sample contamination from materials and equipment used in the collection and analysis of the samples. With each set of samples, one blank sample was carried to the field and three additional method blanks were made up in the laboratory. No systematic difference was noticed between the field and method blanks, indicating that contamination did not occur during sample collection and transport. The amount of PAH found in the blanks was low but variable. For example, for seven impaction media blanks the average amount of pyrene was 2.52 ng with a standard deviation of 1.45 ng. For the Kenmore Square samples, the amount found in the blank samples was generally less than 5% of

the amount in any sample. However, for some PAH in the Quabbin Summit samples, the amounts found on the blanks were comparable to the amounts found in the samples. All reported PAH concentrations have been blank-corrected by subtracting the mean blank concentration from the sample concentration and summing the sample and blank variances. Samples were stored in a freezer and analyzed within 90 days of collection. The time between final preparation and completion of the GC/MS injections was less than 2 weeks. In that time, however, there was a noticeable decrease in the amount of naphthalene and naphthalened8 in some samples. This was probably due to evaporation through the pierced septum of the sample vial. Some samples were reanalyzed after 60 days of further storage. Unlike the original samples, the aged samples did not have quantifiable amounts of PAH with molecular weights from 128 to 166. Therefore, naphthalene, acenaphthylene, acenaphthene, and fluorene could not be accurately quantified by this method. The amounts of other PAH in the aged samples were within the range of experimental results from the original GC/MS runs.

Results and Discussion Urban Boston. Deposits were visible on all the Kenmore Square impactor stages; fibers on stage 0, brownish deposits on stages 1-3, black deposits on stages 4-8, and gray deposits on the after-filter. A total of 23 PAH were identified in the Kenmore Square samples, of these 15 were quantifiable (see Table 4). PAH listed in Table 3 but not in Table 4 were not identified in these samples. The total measured PAH concentrations fall within the range of urban particulate phase PAH concentrations as measured by numerous filter sampling studies (21-35). The fraction of PAH collected on stage 8 and the after-filter (Dp < 0.14 µm) is referred to as the ultrafine fraction in Table 4. PAH collected on impactor stages 4-7 (0.14 < Dp < 1.9 µm) is the

VOL. 30, NO. 3, 1996 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

1027

FIGURE 2. Distribution of molecular weight 252 PAH with particle size in the Kenmore Square samples.

FIGURE 3. Distribution of phenanthrene, pyrene, indeno[1,2,3-cd]pyrene, and coronene with particle size in the Kenmore Square samples.

accumulation fraction. The sum of ultrafine and accumulation fractions is referred to as the fine fraction. PAH collected on stages 1-3 (1.9 < Dp < 19 µm) is the coarse fraction.

Figure 2 shows the normalized distribution of molecular weight 252 PAH with particle size. Dashed lines at Dp ) 0.14 and 1.9 µm show the ultrafine-accumulation and accumulation-coarse fraction divisions. The error bars

1028

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 30, NO. 3, 1996

FIGURE 4. Fraction of PAH associated with coarse particles (Dp > 1.9 µm) in the Kenmore Square (O) and Quabbin Summit (×) samples.

show one standard deviation from the mean analysis results. The lower limit of particle size collected on the after-filter, Dp ) 0.01 µm, has been arbitrarily selected. Calculations of the aspiration efficiency of our sampler using the semiempirical correlation of Tsai and Vincent show that for a wind velocity of 1 m/s, particles with Dp ) 19.2 µm have an aspiration efficiency of 47% (36). PAH collected on stage 0 are not reported because of the low collection efficiencies of particles larger than 19.2 µm. The distributions of benzofluoranthenes, benzo[e]pyrene, benzo[a]pyrene, and perylene with particle size are nearly identical. This trend is observed for all PAH of similar molecular weight, as the fractions of PAH found with each particle size range show (see Table 4). Figure 3 shows the distributions of four PAH of different molecular weights with particle size. These distributions show an increase in the fraction of PAH associated with larger aerosols as molecular weight decreases. Others have observed that PAH of the same molecular weight have similar distributions with aerosol size (10, 8). Recently, Venkataraman and Friedlander found a preferential accumulation of lower molecular weight PAH (MW e 228) compared to higher molecular weight PAH in larger particles (0.5 < Dp < 2.0 µm) in qualititative agreement with the present findings (8). PAH are mainly generated by combustion sources, which also emit mainly fine particles (37, 38). Therefore, PAH are emitted in the gas phase or associated with fine particles. PAH can become associated with coarse particles either by the growth of fine combustion-generated particles or by volatilization from fine particles followed by condensation onto coarse particles. If the main mechanism for PAH association with coarse particles were the growth of combustion-generated particles, the mixture of PAH in fine and coarse particles would be similar. Figure 4 demonstrates that this is not observed. Instead, the fraction of PAH associated with coarse urban particles monotonically decreases from approximately 0.55 for PAH of molecular weight 178 to less than 0.1 for PAH of molecular weight greater than 252. One explanation for the observed PAH partitioning is that PAH, especially higher molecular weight PAH, do not attain an equilibrium distribution among urban aerosols. The flux of PAH from the fine combustion-generated particles to the coarse particles by volatilization and

condensation is directly related to the gas phase concentration of PAH. It has been shown that experimentally measured gas phase concentrations of PAH in urban samples correlate with their subcooled liquid vapor pressures (39). PAH vapor pressures are strongly correlated with molecular weight. For example, the estimated sublimation pressures at 25 °C are 1.6 × 10-2, 6.0 × 10-4, and 1.9 × 10-10 Pa for phenanthrene, pyrene, and coronene, respectively (40, 41). Because high molecular weight PAH have much lower fluxes by volatilization and condensation, their time to partition to larger particles is much greater than that for the lower molecular weight PAH. Therefore, they tend to remain on the fine particles with which they were emitted. Other studies of PAH associated with size-segregated aerosols have found that PAH tend to partition to larger aerosols to a greater degree in warmer periods (5, 7-9). It has also been found that the fraction of PAH in the gas phase increases with temperature (24, 42). An increase in the fraction of PAH in the gas phase would increase the flux of PAH by volatilization and condensation, leading to faster equilibration of PAH among aerosol size fractions, in agreement with these experimental observations. It has been suggested that the observed partitioning of PAH as a function of molecular weight may be due to lower molecular weight PAH being co-emitted with larger particles, for example, in meat cooking emissions. While some authors have found that PAH emission profiles can be correlated with source types, the amount of PAH emitted has not been shown to vary monotonically with molecular weight for noncatalyst and catalyst-equipped automobiles, diesel trucks, or meat cooking (43, 44, 37). Unless a correlation between source type and PAH molecular weight is established, this hypothesis can be put aside. Individual PAH may be preferentially removed from the atmosphere by photooxidation. If PAH reactivity were correlated with molecular weight, photooxidation might explain the observed PAH partitioning behavior. Measurements of the rates of PAH disappearance in wood smoke and adsorbed on fly ash, however, show that PAH reactivity is not correlated with molecular weight (45, 46). Instead, Behymer and Hites found that PAH reactivity was correlated with the maximum net atomic charge which is independent of PAH molecular weight (46). Differences in chemical affinities between PAH and different size particles could also explain the observed PAH partitioning. As PAH molecular weight rises, PAH become more hydrophobic, as demonstrated by the n-octanolwater partition coefficient, which increases with PAH molecular weight (47). If we hypothesize that coarse mode aerosols contain a larger fraction of water, higher molecular weight PAH would tend to partition to the smaller particles. Similarly, if we hypothesize that high molecular weight PAH adsorb more strongly than lower molecular weight PAH, then high molecular weight PAH would tend to associate with the fine aerosol fraction, which has a much higher total surface area than the coarse fraction (1). If PAH do not attain equilibrium with urban aerosols due to slow mass transfer, there are important consequences for modeling the environmental fate of PAH. PAH partitioning in the urban atmosphere could not be assumed to be in equilibrium. Indeed, the time scale to attain equilibrium partitioning may be on the order of time scales for photooxidation, deposition, and coagulation, necessitating complex models that include all these mechanisms.

VOL. 30, NO. 3, 1996 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

1029

TABLE 5

PAH Identified in Quabbin Summit Samples PAH fluoranthene pyrene benzo[ghi]fluoranthene benz[a]anthracene chrysene/triphenylene benzofluoranthenes benzo[e]pyrene benzo[a]pyrene indeno[1,2,3-cd]pyrene dibenzanthracenes benzo[ghi]perylene a

mol wt

concn with aerosol (ng/m3)

202 202 226 228 228 252 252 252 276 278 276

0.507 ( 0.114 ( 0.046 0.078 ( 0.005 0.033 ( 0.006 0.136 ( 0.010 0.186 ( 0.021 0.080 ( 0.008 identified 0.034 ( 0.008 identified 0.016 ( 0.004

0.155a

ultrafine

fraction with aerosol modes accumulation

coarse

0.08 0.04 0.12 0.07 0.09 0.08 0.11

0.44 0.60 0.63 0.41 0.50 0.31 0.35

0.48 0.36 0.25 0.52 0.41 0.61 0.54

0.01

0.28

0.71

0.11

0.32

0.57

One standard deviation.

aerosols vary greatly with location, season, and weather. However, the observed differences in partitioning at urban and rural sites are consistent with the hypothesis that PAH mass transfer among urban particles is incomplete. On time scales relevant to urban aerosols, higher molecular weight PAH would remain with the particles with which they were emitted. But on time scales relevant to regional emissions, all PAH would distribute more closely toward equilibrium. With this explanation, the rural data indicate that at equilibrium at least 50% of PAH will be associated with coarse rural aerosols.

Acknowledgments FIGURE 5. Distribution of benzo[e]pyrene with particle size in the Quabbin Summit samples.

Rural Massachusetts. The deposits on the rural impactor samples were visibly lighter that those collected in Kenmore Square, and there were no visible deposits on the after-filter. The total concentration of each PAH found in the Quabbin Summit samples was at least 1 order of magnitude lower than the concentration found at Kenmore Square. The amounts of many PAH in the rural samples were comparable to the amounts found in the method blanks, making accurate quantification impossible. PAH were considered quantified only if the sum of the blankcorrected concentrations over all stages was positive by at least two standard deviations. Table 5 lists the PAH identified and quantified along with the total concentrations and distributions among the aerosol size fractions. The distribution of PAH among aerosol sizes is qualitatively different for the urban and rural samples. Figures 2b and 5 show the urban and rural distributions of benzo[e]pyrene. PAH in the rural samples were associated with the coarse aerosols to a greater degree than in the urban samples. This is in qualitative agreement with other impaction sampling studies, which found that PAH collected at sites away from emissions sources tend to partition to larger particles (5, 8, 10, 13). Figure 4 shows that the fraction of PAH associated with coarse particles in the rural samples is nearly constant at approximately 0.55, with the exception of benzo[ghi]fluoranthene. It should be noted that direct comparisons of partitioning between different aerosol samples cannot be made. This is because the nature (e.g., sources, chemical compositions, and size distributions) and histories (e.g., temperature, humidity, and photooxidation profiles) of the

1030

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 30, NO. 3, 1996

We thank A. Rana Biswas for his assistance in assembling and testing the sampler. We also thank the staff of the Massachusetts Department of Environmental Protection for access to the sampling sites and ambient data. This research was supported by the National Institute of Environmental Health Sciences and the Environmental Protection Agency. One of us (J.O.A.) was partially supported by a grant provided by the S. C. Johnson Wax Company.

Literature Cited (1) Whitby, K. T. Atmos. Environ. 1978, 12, 135-159. (2) Main, H. H.; Freidlander, S. K. Atmos. Environ. 1990, 24A, 103108. (3) Slinn, W. G. N. Atmospheric Sciences and Power Productions1979; U.S. Department of Energy: Washington, DC, 1983. (4) U.S. Environmental Protection Agency. Technical Report EPA600/8-82-029, 1982. (5) Pierce, R. C.; Katz, M. Environ. Sci. Technol. 1975, 9, 347-353. (6) Katz, M.; Chan, C. Environ. Sci. Technol. 1980, 14, 838-843. (7) Miguel, A. H.; Friedlander, S. K. Atmos. Environ. 1978, 12, 24072413. (8) Venkataraman, C.; Friedlander, S. K. Environ. Sci. Technol. 1994, 28, 563-572. (9) Van Vaeck, L.; Van Cauwenberghe, K. Atmos. Environ. 1978, 12, 2229-2239. (10) Van Vaeck, L.; Broddin, G.; Van Cauwenberghe, K. Environ. Sci. Technol. 1979, 13, 1494-1502. (11) Van Vaeck, L.; Van Cauwenberghe, K. A. Environ. Sci. Technol. 1985, 19, 707-716. (12) Aceves, M.; Grimalt, J. O. Environ. Sci. Technol. 1993, 27, 28962908. (13) Sicre, M. A.; Marty, J. C.; Saliot, A.; Aparicio, X.; Grimalt, J.; Albaiges, J. Atmos. Environ. 1987, 21, 2247-2259. (14) Pistikopoulos, P.; Wortham, H. M.; Gomes, L.; Masclet-Beyne, S.; Bon Nguyen, E.; Masclet, P. A.; Mouvier, G. Atmos. Environ. 1990, 24A, 2573-2584. (15) Marple, V. A.; Rubow, K. L.; Behm, S. M. Aerosol Sci. Technol. 1991, 14, 434-446. (16) Rao, A. K.; Whitby, K. T. J. Aerosol Sci. 1978, 9, 77-86. (17) Turner, J. R.; Hering, S. V. J. Aerosol Sci. 1987, 18, 215-224.

(18) Lafleur, A. L.; Wornat, M. J. Anal. Chem. 1988, 60, 1096-1102. (19) Lee, M. L.; Vassilaros, D. L.; White, C. M.; Novotny, M. Anal. Chem. 1979, 51, 768-774. (20) Burford, M. D.; Hawthorne, S. B.; Miller, D. J. Anal. Chem. 1993, 65, 1497-1505. (21) Gordon, R. J. Environ. Sci. Technol. 1976, 10, 370-373. (22) Lunde, G.; Bjorseth, A. Nature 1977, 268, 518-519. (23) Cautreels, W.; Van Cauwenberghe, K. Atmos. Environ. 1978, 12, 1133-1141. (24) Yamasaki, H.; Kuwata, K.; Miyamoto, H. Environ. Sci. Technol. 1982, 16, 189-194. (25) Grosjean, D.; Fung, K.; Harrison, J. Environ. Sci. Technol. 1983, 17, 673-679. (26) Steinmetzer, H. C.; Baumeister, W.; Vierle, O. Sci. Total Environ. 1984, 36, 91-96. (27) Muel, C.; Saguem, S. Int. J. Environ. Anal. Chem. 1985, 19, 111131. (28) Cretney, J. R.; Lee, H. K.; Wright, G. J.; Swallow, W. H.; Taylor, M. C. Environ. Sci. Technol. 1985, 19, 397-404. (29) Jaklin, J.; Krenmayr, P. Int. J. Environ. Anal. Chem. 1985, 21, 33-42. (30) Guicherit, R.; Schulting, F. L. Sci. Total Environ. 1985, 43, 193219. (31) Colmsjo¨, A. L.; Zebu ¨ hr, Y. U.; O ¨ stman, C. E. Atmos. Environ. 1986, 20, 2279-2282. (32) Pyysalo, H.; Tuominen, J.; Wickstro¨m. K.; Skytta¨, E.; Tikkanen, L.; Salomaa, S.; Sorsa, M.; Nurmela, T.; Mattila, T.; Pohjola, V. Atmos. Environ. 1987, 21, 1167-1180. (33) Rosell, A.; Grimalt, J. O.; Rosell, M. G.; Guardino, X.; Albaige´s, J. Fresenius J. Anal. Chem. 1991, 339, 689-698. (34) Barale, R.; Giromini, L.; Ghelardini, G.; Scapoli, C.; Loprieno, N.; Pala, M.; Valerio, F.; Barrai, I. Mutat. Res. 1991, 249, 227-241. (35) Bodzek, D.; Luks-Betlej, K.; Warzecha, L. Atmos. Environ. 1993, 27A, 759-764.

(36) Tsai, P.-J.; Vincent, J. H. J. Aerosol Sci. 1993, 24, 919-928. (37) Rogge, W. F.; Hildemann, L. M.; Mazurek, M. A.; Cass, G. R.; Simoneit, B. R. T. Environ. Sci. Technol. 1993, 27, 636-651. (38) Hildemann, L. M.; Markowski, G. R.; Jones, M. C.; Cass, G. R. Aerosol Sci. Technol. 1991, 14, 138-152. (39) Pankow, J. F. Atmos. Environ. 1987, 21, 2275-2283. (40) Murray, J. J.; Pottie, R. F.; Pupp, C. Can. J. Chem. 1974, 52, 557563. (41) Sonnefeld, W. J.; Zoller, W. H.; May, W. E. Anal. Chem. 1983, 55, 275-280. (42) McVeety, B. D.; Hites, R. A. Atmos. Environ. 1988, 22, 511-536. (43) Westerholm, R. N.; Alsberg, T. E.; Frommelin, A. B.; Strandell, M. E.; Rannug, U.; Winquist, L.; Grigoriadis, V.; Egeba¨ck, K. Environ. Sci. Technol. 1988, 22, 925-930. (44) Rogge, W. F.; Hildemann, L. M.; Mazurek, M. A.; Cass, G. R.; Simoneit, B. R. T. Environ. Sci. Technol. 1991, 25, 1112-1125. (45) Kamens, R. M.; Guo, Z.; Flucher, J. N.; Bell, D. A. Environ. Sci. Technol. 1988, 22, 103-108. (46) Behymer, T. D.; Hites, R. A. Environ. Sci. Technol. 1988, 22, 13111319. (47) Schwarzenbach, R. P.; Gschwend, P. M.; Imboden, D. M. Environmenal Organic Chemistry; John Wiley & Sons: New York, 1993.

Received for review July 12, 1995. Revised manuscript received November 3, 1995. Accepted November 3, 1995.X ES950517O X

Abstract published in Advance ACS Abstracts, January 15, 1996.

VOL. 30, NO. 3, 1996 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

1031