The Contribution of Traffic to Atmospheric ... - ACS Publications

School of Chemistry, University of Birmingham,. Birmingham B15 2TT, United Kingdom. Impending legislation pertaining to atmospheric PAH concentrations...
2 downloads 0 Views 73KB Size
Environ. Sci. Technol. 1999, 33, 3538-3542

The Contribution of Traffic to Atmospheric Concentrations of Polycyclic Aromatic Hydrocarbons LEE H. LIM, ROY M. HARRISON, AND STUART HARRAD* Institute of Public & Environmental Health, School of Chemistry, University of Birmingham, Birmingham B15 2TT, United Kingdom

Impending legislation pertaining to atmospheric PAH concentrations has intensified efforts to identify their sources. Particulate-phase PAH were measured every 2 h between 06.00 and 20.00 on 12/5/96 at a heavily trafficked location in the center of Birmingham, U.K. Concentrations of 17 out of 18 PAH were significantly correlated (g 95% confidence level) to those of CO and NOx. Using the COPAH and NOx-PAH linear regression equations, it is estimated that during this sampling event 71% (59-87%) and 59% (4781%), respectively, of 4-7 ring PAH originated from traffic. At weekly intervals between 7/15/97 and 12/9/97, simultaneous monitoring of particulate-phase PAH was conducted over 24 h periods at the same city center location, and a campus site ca. 3 km southwest of the city center, and ca. 500 m from the nearest busy road. Intersite differences in PAH concentrations should thus arise primarily from differences in traffic emissions. Using B[e]P as an indicator, the contribution of traffic to concentrations of 4-7 ring PAH at the two sites was estimated at 80-82% and 61-67% for the city center and campus sites, respectively. Utilizing ratios of Σmethylphenanthrenes/phenanthrene, an estimated 60-84% of total PAH traffic emissions at the city center site originated from diesel vehicles.

Introduction Polycyclic aromatic hydrocarbons (PAH) are regarded as priority pollutants by both the United States Environmental Protection Agency and the European Community. Currently, the U.K. government’s Expert Panel on Air Quality Standards (EPAQS) has put forward for consultation a U.K. air quality standard for benzo[a]pyrene of 0.25 ng m-3. While this standard has yet to be adopted, it is of a challenging nature, given that current annual mean concentrations in U.K. cities are typically 0.56-1.8 ng m-3 (1). Against this backdrop, efforts to identify the sources of atmospheric PAH have intensified. Although there is inconsistency between inventories for the U.K. as a whole (2, 3), traffic emissions have been identified as the principal source of PAH in major conurbations such as Birmingham, U.K. (4). However, continuing changes in the size and composition of the U.K. motor fleetse.g. all new gasoline vehicles have been fitted with catalytic converters since 1/1/93, with the result that at the time of this study (1997), around 20% of the overall U.K. fleet were domestic vehicles fitted with catalytic converters, and U.K. new market * Corresponding author phone: +44 121 414 7298; fax: +44 121 414 3078; e-mail: [email protected]. 3538

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 33, NO. 20, 1999

penetration of diesel domestic vehicles increased from 4% in 1988 to 20% in 1993 (5)sprompted us to reevaluate the contribution of traffic emissions to concentrations of PAH in urban atmospheres. In particular, we wished to distinguish the contributions from gasoline and diesel-fueled traffic. To do so, we have applied source apportionment techniques originally applied in Copenhagen, Denmark (6) to Birmingham. Most of these techniques involved temporally consistent monitoring at two sites a few hundred meters apart, where one site was heavily trafficked, and the other not. Such strict criteria for site location potentially limits the generic applicability of these techniques. Given their success in Copenhagen, we wished to evaluate their utility when one of these criteria was relaxed, specifically, the intersite distance was 3 km.

Experimental Section Sampling Program. Two campaigns were conducted, with all measurements made on weekdays, thus representing a “worst-case scenario” for traffic emissions. Campaign I. Airborne concentrations of 18 PAH (particulate-phase only) were measured at 2 h intervals between 06.00 and 20.00 on 12/5/96. This sampling site (city site) was a heavily trafficked city center location in Birmingham, U.K. The sampler was sited 8.7 m above ground level, ca. 7 m from a busy road (30-35 000 vehicles d-1), and ca. 600 m from an Automatic Urban Network monitoring station from which measurements of CO and NOx were taken. Campaign II. A total of 12 PAH and five methyl phenanthrenes (MePh) were measured in 19 sample pairs (duration 24 h) taken simultaneously on a weekly basis between 7/15/ 97 and 12/9/97 at the same city center site as campaign I and at a site (campus site) on the campus of the University of Birmingham located ca. 500 m from the nearest busy road and ca. 3 km distant from the city site. Note that although both vapor and particulate-phase PAH concentrations were recorded, only the latter are reported in this paper. Air Sampling. Samples were collected using a GrasebyAndersen high-volume air sampler. Particulate-associated PAH were trapped on Teflon coated glass fiber filter (GFF) papers. Sampling flow-rates were maintained at ca. 0.650.8 m3 min-1 yielding sample volumes of ca. 100 and 1000 m3 for 2 and 24 h samples, respectively. GFFs were spiked with deuterated PAH prior to Soxhlet extraction for 24 h with dichloromethane. Following solvent removal, extracts were eluted through a silica gel solid-phase extraction column with hexane (15 mL) and concentrated to 100 µL prior to GC/MS analysis on a Fisons’ MD-800 instrument, fitted with an HP-5 trace analysis column (60 m × 0.25 mm i.d. × 0.25 µm film thickness) and operated in selected ion monitoring mode. One microliter of the sample extract was injected in the splitless mode. Injector and detector temperatures were both 250 °C. The He head pressure was 25 psi, and the oven temperature was programmed for 40 °C for 1 min and 8 °C min-1 to 300 °C and held for 37 min. Individual PAH were quantified via the internal standard method relative to the closest eluting deuterated standard. Compounds monitored in campaign I were as follows: acenaphthylene (Ac), acenaphthene (Ace), fluorene (Fl), phenanthrene (Ph), anthracene (An), dibenzothiophene (Diben), fluoranthene (Fluo), pyrene (Py), benz[a]anthracene (B[a]A), chrysene (Chry), benzo[b]naphtho[2,1-d]thiophene (BNT), benzo[b]fluoranthene (B[b]F), benzo[j+k]fluoranthene (B[j+k]F), benzo[a]pyrene (B[a]P), indeno[1,2,3-cd]pyrene (Ind), benzo[ghi]perylene (B[ghi]P), dibenzo[a,h]anthracene (D[a,h]A), and coronene (Cor). In campaign II, the following compounds 10.1021/es990392d CCC: $18.00

 1999 American Chemical Society Published on Web 09/11/1999

TABLE 1. Certified and Experimentally determined Concentrations of PAH in SRM 1941a (ng g-1) compd

certified mean

exptl mean

certified RSDa

exptl RSDb

Fl Ph An Fluo Py B[a]A Chry B[b]F B[j+k]F B[a]P B[ghi]P Ind D[a,h]A

97.3 489 184 981 811 427 592b 740 416 628 525 501 73.9

118 541 190 953 834 428 584 741 524 561 617 570 63.0

8.8 4.7 7.6 8.0 3.0 5.9 6.3 14.9 5.0 8.3 11.8 14.4 13.1

7.6 6.4 8.2 11.0 6.9 9.1 7.3 8.4 8.9 9.6 8.9 7.7 8.8

a Relative standard deviation. chrysene and triphenylene.

b

Certified concentration is the sum of

were monitored for their potential utility in quantifying diesel PAH emissions: 4,5-dimethylphenanthrene (4,5-MePh), 3-methylphenanthrene (3-MePh), 9-methylphenanthrene (9MePh), 1-methylphenanthrene (1-MePh), 2-methylphenanthrene (2-MePh), and Ph. In addition, the following 4-7 ring PAH were studied: benzo[e]pyrene (B[e]P), B[a]A, Chry, BNT, B[b]F, B[j+k]F, B[a]P, Ind, B[ghi]P, D[a,h]A, and Cor. Note that the purpose of this campaign was to estimate traffic emissions of PAH based on the assumption that the difference in PAH concentrations between the two sites is due to the difference in the traffic contribution and not factors such as preferential reaction/deposition of vapor and/or particulate phase PAH. The more volatile PAH are therefore not reported for this campaign, as they are more susceptible to variations in vapor/particle phase partitioning. Quality Control and Quality Assurance. Analyte losses during sampling were evaluated by adding a known quantity of a sampling efficiency standard (SES, d14-terphenyl) to the GFF prior to sampling. The mean SES recovery (which reflects analyte losses due to both sampling and analysis) was 95% (range 60-130%) for all samples. Mean recoveries of deuterated internal standards were 95% (80-120%), 90% (8398%), and 81% (73-91%) for d10-anthracene, d12-benz[a]anthracene, and d12-perylene, respectively. Analytical accuracy and precision were assessed by conducting five replicate analyses of SRM 1941a (Organics in Marine Sediment) (Table 1). The repeatability of air sampling and analysis combined was assessed by taking duplicate outdoor air samples at the campus site over an identical 24 h period, using two identical high volume samplers located adjacent to each other. This exercise was repeated four times. The results (calculated for particulate phase PAH only) revealed an average difference between duplicate analyses of 9.6% (range 7.0-12% for individual PAH). We then compared these differences, with the percentage differences between the mean concentrations for each individual PAH at the two sites. In all but one cases the exception being phenanthrenesthe differences between the mean concentrations for the two sites far exceeded that attributable to sampling and analytical error combined.

Results and Discussion Campaign I. Temporal Variation of PAH at City Site on 5/12/96. The concentrations of individual PAH, CO, and NOx determined during each 2 h period are shown in Table 2. Figure 1 illustrates the “traffic-related” temporal variation observed in concentrations of the five ring PAH and CO. With the exception of D[a,h]A, concentrations of each individual PAH were significantly correlated (g95% confi-

dence level) to those of both CO and NOx. In the U.K. West Midlands conurbationsof which Birmingham is the major citys98% of CO and 85% of NOx emissions arise from road traffic (7). This suggests that a high proportion of PAH at this site on this day are due to traffic emissions. Given these highly significant correlations, we estimated the proportion arising from vehicular emission of individual PAH using the linear regression equations obtained for plots of PAH against CO concentrationsFigure 2 shows the plots obtained for B[b]F and B[a]A. Taking B[a]A as an example, linear regression against CO yielded eq 1.

B[a]A ) 0.774 × CO + 0.657

(1)

Assuming that at the city site nearly all CO is traffic generated and that the background CO concentration in the absence of traffic is 0.1 ppm (8), substituting CO ) 0.1 ppm in eq 1, gives an estimate of the B[a]A concentration in the absence of traffic ) 0.73 ng m-3. On 12/5/96, the mean B[a]A concentration ) 2.53 ng m-3 (Table 2). Thus, the mean level due to traffic ) 1.80 ng m-3 (2.53-0.73), and the percentage of B[a]A due to traffic at this site on this day ) (1.80/2.53)*100% ) 71%. Applying the same technique to individual PAH, the percentage contribution of traffic ranges from 35 to 87% (mean ) 59%) for regression with CO and from 27 to 81% (mean ) 54%) for regression with NOx. However, only particulate phase PAH were monitored in this study. Thuss even at the low temperatures on this day; average temperature ) 1.1 °C, range ) -3.0 to 4.1 °Csparticulate phase concentrations of the lower molecular weight PAH are likely to be significantly influenced by changes in vapor/particle partitioning ratios caused by diurnal temperature variations. As a result, a more reliable estimation of the contribution of traffic is obtained by using only those PAH (B[a]A, Chry, B[b]F, B[j+k]F, Ind, B[ghi]P, and Cor) which are likely to be predominantly present in the particulate phase at the temperatures experienced during this campaign (1). For these compounds, the mean traffic contribution was 71% (range ) 59-87%) for regression with CO and 59% (range ) 4781%) for regression with NOx. Campaign II. Table 3 shows the mean particulate concentrations of individual PAH detected at the city and campus sites. Mean particulate phase Σ(PAH + MePh) concentrations at the city site were 1.7 times those at the campus site (9.44 ng m-3 cf. 5.69 ng m-3). This study assumes that the difference between concentrations of PAH at the two sites is due to the difference in the traffic contribution. Specifically, the city site represents a highly traffic-influenced site, whereas the campus site, which is 3 km southwest, is chosen to represent an urban background site with comparatively little traffic influence. Selecting two sites at close range is designed to minimize intersite differences in contributions from nontraffic sources such as long-range transport and home heating etc. as well as losses due to chemical reactions and deposition. Assuming this to be so, intersite differences in PAH profiles may be used to derive a traffic PAH emission profile which is defined as the PAH profile that would exist if traffic was the only PAH source. This approach was used successfully by Nielsen (6). However, in Nielsen’s study, the intersite distance was only a few hundred meters, and it is recognized that the greater distance (3 km) separating the two sites in this study introduces an increased level of uncertainty into the above assumption. To obtain a PAH traffic profile, δPAH was plotted against δB[e]P for each individual 4-7 ring PAH monitored, where δPAH represents the difference in concentrations of individual PAH between the two sites, and δB[e]P represents the difference in B[e]P concentrations between the two sites. B[e]P is used as a reference because of its relative atmospheric VOL. 33, NO. 20, 1999 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

3539

TABLE 2. Temporal Variation of PAH (ng m-3), NOx (ppb), and CO (ppm) Concentrations on 12/5/96 sampling period (h) compd

06-08

08-10

10-12

12-14

14-16

16-18

18-20

mean

Ac Ace Fl Diben Ph An Fluo Py B[a]A Chry BNT B[b]F B[j+k]F B[a]P Ind B[ghi]P D[a,h]A Cor NOx CO

0.03 0.03 0.05 0.03 1.35 0.16 3.63 4.44 1.49 1.74 0.14 1.86 0.53 1.33 1.13 3.08 0.09 2.67 189 1.60

0.11 0.04 0.14 0.05 3.79 0.62 5.01 7.71 4.37 4.62 0.41 8.69 2.29 6.04 3.09 7.33 0.17 9.25 727 5.15

0.09 0.04 0.08 0.03 2.03 0.35 4.70 5.47 4.08 4.74 0.41 9.06 2.21 5.53 3.45 7.85 0.09 5.73 544 4.00

0.05 0.03 0.07 0.03 0.95 0.17 2.85 3.41 1.59 2.22 0.19 3.00 0.75 1.50 1.46 3.26 0.09 3.30 178 1.25

0.04 0.02 0.04 0.02 0.69 0.11 2.68 3.43 1.24 1.62 0.13 2.10 0.60 1.14 1.09 2.11 0.09 1.98 101 1.15

0.05 0.03 0.05 0.03 1.08 0.16 3.93 4.53 2.33 2.78 0.22 4.17 1.04 2.47 1.97 4.39 0.14 5.09 168 2.10

0.06 0.02 0.09 0.04 0.86 0.21 3.03 3.51 2.59 3.14 0.33 4.57 1.18 2.50 2.23 4.66 0.21 5.12 160 1.65

0.06 0.03 0.08 0.03 1.54 0.25 3.69 4.64 2.53 2.98 0.26 4.78 1.23 2.93 2.06 4.67 0.12 4.73 295 2.41

TABLE 3. Mean Particulate Phase Concentrations of PAH during Campaign II (7/15/97-12/9/97)

FIGURE 1. Temporal variation of concentrations of CO and five ring PAH on 12/5/96.

compd

city site (ng m-3)

campus site (ng m-3)

Ph 4,5-MePh 3-MePh 9-MePh 1-MePh 2-MePh B[a]A Chry BNT B[b]F B[j+k]F B[e]P B[a]P Ind B[ghi]P D[a,h]A Cor total

0.32 0.03 0.08 0.12 0.16 0.07 0.58 0.95 0.10 0.81 0.80 0.72 0.80 1.12 1.49 0.22 1.07 9.44

0.28 0.02 0.04 0.05 0.08 0.04 0.37 0.57 0.07 0.55 0.58 0.45 0.53 0.60 0.80 0.12 0.54 5.69

FIGURE 2. Scatter plots of temporal variations of CO versus B[b]F and B[a]A concentrations on 12/5/96. stability (6), which ensures that differences in PAH profile between the two sites are predominantly due to differences in the influence of traffic and not losses due to atmospheric reactions etc. For this method to be applicable, Nielsen (6) considered that the following conditions should be met: (a) there should be significant positive linear correlation between δPAH with δB[e]P, and (b) traffic emissions should be the major atmospheric source of PAH at the city site. The first condition is clearly met for all individual PAH monitored with all correlations at the 99.9% confidence levelsFigure 3 3540

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 33, NO. 20, 1999

FIGURE 3. Relationship between δB[b]F and δB[e]P with y-intercept forced to zero. shows the relationship between δB[b]F and δB[e]P with the y-intercept forced to zero. Furthermore, good evidence that the second condition is met is provided by the results of campaign I. The traffic profile derived from individual δPAH versus δB[e]P plots is shown in Figure 4. Evaluation of PAH Traffic Contribution. We used two methods to estimate the percent traffic contribution of PAH at a given site, both adapted from Nielsen (6). Method 1

TABLE 4. B[ghi]P/B[e]P and Cor/B[e]P Ratios Observed at City and Campus Sites and the Calculated PAH Traffic Contribution (Using B[e]P as an Indicator)

FIGURE 4. Traffic emission profile - particulate phase PAH composition relative to B[e]P from traffic sources (error bars are (2 σ). works thus; the traffic profile (Figure 4) shows the ratio of Cor to B[e]P originating from traffic ) 1.80 (δCor/δB[e]P). Table 3 shows that mean particulate phase concentrations of Cor at the city and campus sites were 1.07 and 0.54 ng m-3, respectively, while those of B[e]P were 0.72 and 0.45 ng m-3, respectively. Assuming that all Cor is traffic-generated, the concentration of B[e]P which is traffic generated (B[e]Ptraffic) is given by

B[e]Ptraffic ) Cor concentration/ (δCor/δB[e]P) ) 1.07/1.80 ) 0.59 ng m-3 at the city site and

) 0.54/1.80 ) 0.30 ng m-3 at the campus site Therefore, the percentage of B[e]P that is traffic-related is 82% and 67% at the city and campus sites, respectively. Crucial to this method is the assumption that all Cor is trafficgenerated. Although this is likely to be an oversimplification, many studiessincluding our own temporal study in campaign Ishave shown a high correlation between Cor and vehicular emissions (8-10), and we believe this estimation method to be fundamentally sound, when used at heavily traffic-influenced sites. On the basis that he observed highly significant (g99% confidence level) correlation between concentrations of selected individual PAH and B[e]P, Nielsen (6) deemed it reasonable to assume that the fraction of B[e]P that is trafficgenerated was a reliable indicator of the fraction of those selected PAH arising from traffic. To verify the validity of this assumption for our data, we regressed concentrations of each individual 4-7 ring PAH studied against B[e]P at both sites. In all cases, we observed highly significant correlation (g99% confidence level). However, this does not necessarily imply that the percentage traffic contribution to concentrations of individual PAH is identical to that of B[e]P. We therefore used the equations obtained for regressions of individual PAH against B[e]P to estimate whether the traffic contribution estimate for B[e]P was likely to be an under- or overestimate for the other PAH monitored. To illustrate, for B[b]F at the city site the regression equation was B[b]F ) 1.17 × B[e]P 0.02. Our estimate of the traffic contribution to B[e]P concentrations at the city site is 82%, and the mean concentrations of B[e]P and B[b]F are 0.72 and 0.81 ng m-3, respectively (Table 3). Therefore, the concentration of B[e]P due to traffic ) 0.72 × 0.82 ) 0.59 ng m-3, and the concentration in the absence of traffic ) 0.72-0.59 ) 0.13 ng m-3. Substituting B[e]P ) 0.13 ng m-3 into the regression equation gives the concentration of B[b]F in the absence of traffic ) 0.132 ng m-3. The mean concentration of B[b]F in

site

B[ghi]P/B[e]P ratio

Cor/B[e]P ratio

% traffic contribution method 1

% traffic contribution method 2

city campus

2.07 1.78

1.49 1.20

82 67

80 61

the presence of traffic therefore ) 0.81-0.132 ) 0.678 ng m-3, and the percentage contribution of traffic to B[b]F at the city site ) 84%. Repeating this process for the other PAH at both sites revealed that in all cases the estimated traffic contribution closely matched that of B[e]Psslightly overestimating in 65% of cases (e.g. for B[ghi]P the estimated traffic contribution at the city site was 80%), slightly underestimating in the remaindersand confirmed the validity of using B[e]P as an indicator for the traffic contribution to concentrations of other 4-7 ring PAH. In method 2, the ratios of B[ghi]P:B[e]P and Cor:B[e]P are used as traffic indicators, since both B[ghi]P and Cor have been reported to closely correlate with traffic emissions (6). Fractional traffic contributions to concentrations of B[e]P at a given site are obtained by solving the following equations for fractional traffic contribution and taking the mean:

(B[ghi]P/B[e]P)site ) ((B[ghi]P/B[e]P)traffic × fractional traffic contribution) + ((B[ghi]P/B[e]P)nontraffic × (1 - fractional traffic contribution)) (Cor/B[e]P)site ) ((Cor/B[e]P)traffic × fractional traffic contribution) + ((Cor/B[e]P)nontraffic × (1 - fractional traffic contribution)) Whereby (B[ghi]P/B[e]P)traffic and (Cor/B[e]P)traffic are values obtained from the traffic profile (Figure 4)s2.37 and 1.80, respectively; (B[ghi]P:B[e]P)nontraffic and (Cor:B[e]P)nontraffics i.e. those that would be observed in the absence of traffics are assumed to be 0.8 and 0.3, respectively (6); and B[ghi]P: B[e]Psite and Cor:B[e]Psite values are given in Table 4. It is important to note that obtaining nontraffic B[ghi]P:B[e]P and Cor:B[e]P ratios is extremely difficult for an urban location and that the values used are thus subject to unquantifiable uncertainty. However, the close agreement between the estimates derived by methods 1 and 2 is encouraging. Table 4 shows the calculated traffic contribution to B[e]P concentrations to be 80% and 61% for the city and campus sites, respectively. As discussed above, the high degree of correlation between B[e]P and other 4-7 ring PAH means that these estimates may reasonably be extrapolated to all PAH monitored. The Contribution of Diesel Vehicle Emissions to Overall Traffic Emissions. Given the significance of traffic emissions as a PAH source in Birmingham and the growing contribution of diesel vehicles to the overall U.K. vehicle fleet, we employed a method to estimate the proportion of traffic-generated B[e]Psand thus Σ4-7 ring PAHsoriginating from diesel exhaust gases in each sample (6). This proportion (Diesel B[e]P), was estimated using the following relationship:

diesel B[e]P ) B[e]P concentrationsite × [(MePh:Ph)site - MePh:Phgasoline traffic)/ (MePh:Phdiesel traffic - MePh:Phgasoline traffic)] Whereby B[e]P concentrationsite ) B[e]P concentration at a site (MePh:Ph)site ) MePh/Ph ratio measured at a site ) ΣVOL. 33, NO. 20, 1999 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

3541

FIGURE 5. Relationship between δdiesel B[e]P and δB[e]P. (4,5-MePh, 3-MePh, 9-MePh, 1-MePh, and 2-MePh)/Ph

MePh:Phgasoline traffic ) MePh/Ph ratio from gasoline traffic ) 0.7 (6) MePh:Phdiesel traffic ) MePh/Ph ratio in diesel exhaust gas ) 5.5 (6) Note that the particulate phase MePh:Ph ratios used here aresin line with ref 6sassumed to equal the MePh:Ph ratio for the sum of both vapor and particle phases. We consider this a reasonable assumption, but it should be noted that any deviation from this would mean that any difference in ratio between the city and campus sites would not be wholly due to differences in traffic emissions, but at least in part attributable to preferential reaction/deposition of the vapor and/or particulate phase. Values of δB[e]P and δDiesel B[e]P, which is the difference between diesel B[e]P at the city site and that measured simultaneously at the campus site, were calculated for each sample and sample pair. A plot of δdiesel B[e]P against δB[e]P for the city site is shown in Figure 5. With the y-intercept forced to zero, the gradient gives the fractional contribution of diesel emissions to total traffic (i.e. gasoline and diesel) emissions of B[e]Psand thus a reasonable estimate of emissions of Σ4-7 ring PAH. Figure 5 shows diesel to contribute 60 ( 9% of overall traffic emissions at the city site, a value similar to that (64 ( 7%) obtained by Nielsen for Copenhagen in winter 1992 (6). Clearly, this method is highly dependent on the values used for MePh:Phgasoline traffic and MePh:Phdiesel traffic, and establishing representative emission factors for diffuse sources such as traffic is extremely difficult. Specifically, the MePh:Phdiesel traffic value of 5.5 used above was an average of emissions from five types of diesel engine ranging from 1500 to 10 400 cm3 running at zero engine load and a constant speed of 3000 rpm (9) as well as from a Swedish bus simulating typical driving conditions in a European city (10). We derived an alternative estimate of MePh:Phdiesel traffic by determining MePh/Ph ratios in 10 diesel exhaust particulate samples collected from a 4000 cm3 Perkins diesel engine running at 100%, 75%, 50%, and 25% and zero engine loads at both 2600 and 1600 rpm. When the mean MePh/Ph ratio of 3.89

3542

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 33, NO. 20, 1999

recorded in these experiments was used, the diesel contribution to overall traffic emissions increased to 84 ( 13%. This study uses three separate techniques to estimate the extent to which traffic emissions contribute to atmospheric concentrations of PAH at both a heavily trafficked city center location and an urban background site. Despite the distance between the two sites exceeding that used in an earlier study, the methods still functioned well, suggesting that these source apportionment techniques may have wide generic applicability. While each estimation method has an associated uncertainty, the narrow range of estimates obtained from different methods supports the overall conclusion that traffic emissions are the most significant source of atmospheric PAH in Birmingham, U.K. The contribution of diesel emissions to those of traffic overall is shown to be appreciable. It is clear that reducing traffic emissions of PAH is likely to play an important role in lowering atmospheric concentrations of these compounds.

Acknowledgments Most of the source apportionment techniques applied in this study were developed by Torben Nielsen of the Risø National Laboratory, Roskilde, Denmark. The authors gratefully acknowledge his inspiration and the gift of several MePh standards. We are also grateful to Ji Ping Shi for providing the samples of diesel exhaust particulates and to the Government of Brunei for providing a scholarship to Lee Hoon Lim.

Supporting Information Available Tables giving the regression parameters (slope and intercept) and percentage traffic contributions to atmospheric concentrations of the 4-7 ring PAH measured at both sites in this study are included. This material is available free of charge via the Internet at http://pubs.acs.org.

Literature Cited (1) Halsall, C. J.; Coleman, P. J.; Davis, B. J.; Burnett, V.; Waterhouse, K. S.; Harding-Jones, P.; Jones, K. C. Environ. Sci. Technol. 1994, 28, 2380-2386. (2) Wild, S. R.; Jones, K. C. Environ. Pollut. 1995, 88, 91-108. (3) Salway, A. G.; Eggleton, H. S.; Goodwin, J. W. L.; Murrells, T. P. UK Emissions of Air Pollutants 1970-1994; National Atmospheric Emissions Inventory Report Reference AEAT/RAMP/20090001/ R/003; 1996; pp 45-47. (4) Harrison, R. M.; Smith, D. J. T.; Luhana, L. Environ. Sci. Technol. 1996, 30, 825-832. (5) QUARG, Airborne Particulate Matter in the United Kingdom, third report of the Quality of Urban Air Review Group (QUARG), 1996. (6) Nielsen, T. Atmos. Environ. 1996, 30, 3481-3490. (7) London Research Centre, West Midlands Atmospheric Emissions Inventory, September 1996. (8) Deacon, A. R.; Derwent, R. G.; Harrison, R. M.; Middleton, D. R.; Moorcroft, S. Sci. Tot. Environ. 1997, 203, 17-36. (9) Tanaka, H.; Onda, T.; Ogura, N. Environ. Sci. Technol. 1990, 24, 1179-1186. (10) Westerholm, R. N.; Alme`n, J.; Li, H.; Rannug, J. U.; Egeba¨ck, K.-E.; Gra¨gg, K. Environ. Sci. Technol. 1991, 25, 332-338.

Received for review April 7, 1999. Revised manuscript received July 12, 1999. Accepted July 20, 1999. ES990392D