Article pubs.acs.org/est
Modern and Fossil Contributions to Polycyclic Aromatic Hydrocarbons in PM2.5 from North Birmingham, Alabama in the Southeastern U.S. Li Xu,† Mei Zheng,*,‡,§ Xiang Ding,§,∥ Eric S. Edgerton,⊥ and Christopher M. Reddy# †
National Ocean Sciences Accelerator Mass Spectrometry Facility, Department of Geology and Geophysics, Woods Hole Oceanographic Institution, Woods Hole, Massachusetts 02543, United States ‡ State Key Joint Laboratory of Environmental Simulation and Pollution Control, College of Environmental Sciences and Engineering, Peking University, Beijing 100871, China § School of Earth and Atmospheric Sciences, Georgia Institute of Technology, Atlanta, Georgia 30332, United States ∥ State Key Laboratory of Organic Geochemistry, Guangzhou Institute of Geochemistry, Chinese Academy of Sciences, Guangzhou 510640, China ⊥ Atmospheric Research & Analysis, Inc., Cary, North Carolina 27513, United States # Department of Marine Chemistry and Geochemistry, Woods Hole Oceanographic Institution, Woods Hole, Massachusetts 02543, United States S Supporting Information *
ABSTRACT: Analyzing the radiocarbon (14C) content of polycyclic aromatic hydrocarbons (PAHs) in atmospheric particulate matter can provide estimates on the source contributions from biomass burning versus fossil fuel. The relative importance of these two sources to ambient PAHs varies considerably across regions and even countries, and hence there is a pressing need to apportion these sources. In this study, we advanced the radiocarbon analysis from bulk carbon to compound class specific radiocarbon analysis (CCSRA) to determine Δ14C and δ13C values of PAHs in PM2.5 samples for investigating biomass burning and fossil fuel source contributions to PAHs from one of the Southeastern Aerosol Research and Characterization (SEARCH) sites in North Birmingham (BHM), Alabama during winter (December 2004-February 2005) and summer (June-August 2005) by accelerator mass spectrometry. To compare our ambient samples to known sources, we collected and analyzed fenceline samples from the vicinity of a coke plant in BHM. As expected, PAHs from the coke plant fenceline samples had very low radiocarbon levels. Its Δ14C varied from −990 to −970‰, indicating that 97 to 99% were of fossil source. PAHs in the ambient PM2.5 had Δ14C from −968 to −911 ‰, indicating that 92−97% of PAHs were from fossil fuel combustion. These levels indicated the dominance of fossil sources of ambient PAHs. The radiocarbon level of ambient PAHs was higher in winter than in summer. Winter samples exhibited depleted δ13C value and enriched Δ14C value because of the increased contribution of PAHs from biomass burning source. However, biomass burning contributed more to heavier PAHs (modern source accounting for 6−8%) than lighter ones with a modern contribution of 3%.
■
INTRODUCTION PAHs often with three to six rings are produced from incomplete combustion processes of any carbon-containing materials such as fossil fuel and biomass and hence frequently are observed in ambient aerosols. These compounds are ubiquitous in the environment including air, water, and sediment.1 PAHs in PM2.5 (particles with an aerodynamic diameter less than 2.5 μm), such as benzo[a]pyrene, are known to have carcinogenic and mutagenic effects.2 To formulate effective control strategies for PAHs and reduce the potential toxic effects of PM2.5, the knowledge of its sources in the atmosphere is required. © 2011 American Chemical Society
There are several ways to study sources of PAHs including ratios between specific PAHs as well as distribution patterns of PAHs. For example, the ratio of fluoranthene to the sum of fluoranthene and pyrene (Fluo/(Fluo+Py)) and the ratio of indeno[1,2,3-cd]pyrene to the sum of indeno[1,2,3-cd]pyrene and benzo[ghi]perylene (Ind-Py/(Ind-Py+BghiP)) have been used to diagnose sources such as unburned petroleum, liquid Received: December 12, 2011 Accepted: December 22, 2011 Published: December 22, 2011 1422
dx.doi.org/10.1021/es2043189 | Environ. Sci. Technol. 2012, 46, 1422−1429
Environmental Science & Technology
Article
could be occasionally influenced by emissions from coke plants (contributed 0.52 μg/m3 average to fine organic carbon), and thus two fenceline samples from the vicinity of coke plants were also included for radiocarbon measurements of its PAHs in this study to compare ambient samples to known source. This will undoubtedly provide helpful information for particulate matter research especially when to formulate control plans for particulate PAHs with carcinogenic and mutagenic effects in the southeastern U.S. In this study, the relationships between PAH radiocarbon level and organic carbon to elemental carbon ratio as well as the summer/winter variation are also examined.
fossil fuel combustion, and combustion from solid fuel (e.g., grass, wood, and coal). The distribution of alkyl-substituted PAHs relative to its parent PAHs has been used to distinguish pyrogenic (combustion) source from petrogenic (petroleum) source.1 Dimethylphenanthrene (DMP) ratio 1,7/(2,6 + 1,7)DMP can be also used to distinguish fossil and biomass burning contributions.3,4 However, measuring radiocarbon content in bulk carbon or in individual organic compounds is a direct way to deduce whether carbon is from fossil fuel combustion or modern biomass burning. This is because fossil fuel does not contain any 14C due to its half-life of 5730 years (14C free, i.e., Δ14C = −1000‰). Biomass, on the other hand, has a 14C content which reflects the recent and current atmosphere. Current atmospheric CO2 and freshly produced biomass have a Δ14C value of +70‰,5 while contemporary wood fuel sources have Δ14C between +218‰ and +225‰.6−8 Therefore, 14 C measurements can provide information of the relative contribution of fossil fuel and biomass combustion to PAHs in PM2.5. The apportionment of fossil fuel-derived PAHs and biomass-derived PAHs through the radiocarbon composition measurement in aerosol samples is indeed a more reliable way than other methods. This technique has been applied more in sediments than in aerosols.9,10 Even with limited measurements of 14C in aerosols, most studies focus on bulk carbon in PM2.5.11,12 The application of CCSRA is relative uncommon due to the complexity of the method, including the isolation and quantification of carbon addition during extraction, isolation, purification, and subsequent conversion to CO2 and graphite. The minimal material requirement (roughly 20 μg of carbon per sample) for accelerator mass spectrometry (AMS) requires extensive air sampling. For example, the measurement of 14C for total PAHs (not individual PAH) in Aspvreten, Sweden was based on the two- or three-year aerosol composite samples (two samples corresponding to time periods of 1995− 1997 and 1998−2001, respectively).13 To our knowledge, examining radiocarbon content in individual PAH compounds has been only conducted for the National Institute of Standards and Technology (NIST) reference material for urban dust SRM1649a,14 household soot,15 and airborne particulate matter.16,17 The SEARCH program is a multiyear and multisite air quality-monitoring program in the southeastern U.S., including four pairs of urban-rural sites in Georgia, Alabama, Mississippi, and Florida.18 By now, more than 100 papers have been published on findings related to air quality from this program, including 14C measurements in bulk carbon.12 Since PAHs are toxic compounds in PM2.5 and important primary tracers for source apportionment of carbonaceous aerosols, it is essential to move the radiocarbon analysis from bulk carbon to classes of PAHs in order to enhance our understanding of sources of PAHs in PM2.5 in the southeastern U.S. In the current study, the sources of PAHs were examined through CCSRA for the first time to get insight of PAH origins in ambient PM2.5 SEARCH samples. This is a valuable addition to our understanding of PAH sources through a large air quality monitoring platform. By taking advantage of CCSRA technique, it will provide a clear, direct, and quantitative answer to the two important sources of PAHs in the southeastern U.S.: fossil fuel vs biomass burning. The urban site BHM was selected for this study because the highest PAH levels have been consistently observed at this site since this program started back in 1998.19 In a previous study by Zheng et al.,12 it was shown that this site
■
MATERIALS AND METHODS PM2.5 aerosol samples were collected using a Tisch (Cleves, OH) high volume sampler with a flow rate of 1.13 m3/min for 24 h. Samples were collected on quartz filters. Four composite samples were prepared for the radiocarbon analysis including a summer sample with low organic carbon (OC) to elemental carbon (EC) ratio (1.43), a summer sample with high OC/EC ratio (2.81), a winter sample with low OC/EC ratio (1.72), and a winter high sample with a high OC/EC ratio (3.27) (see Table S1 in the Supporting Information). The sample with low OC/EC value is more likely dominated by primary emissions rather than secondary formation compared to the sample with higher OC/EC ratio. The summer low sample (SL), the summer high sample (SH), and the winter high sample (WH) are composed of 8, 7, and 6 individual 24 h PM2.5 samples, respectively (see Table S1). Detailed information on sample treatment is described elsewhere.19 Prior to extraction, portions of a perdeuterated internal standard composite, including acenaphthene-d10, chrysene-d12, and dibenzo[a,h]anthracene were spiked into the samples. Ambient samples were extracted twice by mild sonication with 40 mL of hexane, followed by three successive extractions with 40 mL of mixed solvent (benzene:propanol 2:1, v/v). After extraction, samples were filtered, combined, and concentrated. Since these samples were also analyzed for polar compounds like fatty acids, methylation with diazomethane was done before gas chromatography−mass spectrometry (GC/ MS) analysis at the Georgia Institute of Technology (GT). Samples were analyzed by an Agilent 6890 GC/5973 MS in the scan mode with a 30 m HP-5 MS capillary column (i.d. 0.25 mm, 0.25 μm film thickness). Splitless injection of a 1 μL sample was performed with a four-minute solvent delay. The GC temperature was initiated at 65 °C (held for 2 min) and increased to 300 at 10 °C/min (held for 20 min). PAHs were quantified using an internal calibration method by authentic standards. The detection limit of PAHs ranged from 0.01 ng/ m3 (BghiP) to 0.03 ng/m3 (benzo[b]fluoranthene). Recoveries of PAHs in matrix spiked samples (PAH standards spiked into pre-extracted filters, n = 3) and NIST SRM1649a reference samples (n = 3) ranged from 90 to 112% and 80 to 120%, respectively, demonstrating that our analytical processes, including extraction, purification, and GC-MS analysis, were well-controlled. Table 1 lists the abbreviations and measured concentrations of PAHs. The remaining filters were sent to Woods Hole Oceanographic Institution (WHOI) for 14C preparation and analysis. PM2.5 samples on filters were extracted using a pressurized fluid extraction system (ASE 200, Dionex, USA) at 100 °C and 1000 PSI with dichloromethane/methanol (9:1 v/v). The extract was saponified with 0.5 M KOH in methanol/water (4:1 v/v) for 2 h at 80 °C to hydrolyze the esters in the sample for further fatty 1423
dx.doi.org/10.1021/es2043189 | Environ. Sci. Technol. 2012, 46, 1422−1429
1.10 ± 0.33
Abbreviation in brackets. Sum of benzo[b]fluoranthene, benzo[j]fluoranthene, and benzo[k]fluoranthene. T1 as the sum of Fluo+Py+BaA+Chry. T2 as the sum of BbjkF+BeP+BaP. T3 as the sum of DBahA +Ind-Py+BghiP. fDBahA was excluded.
e
0.85 ± 0.23 0.71 ± 0.18 0.94 ± 0.25 1.01 ± 0.27
23.2 21.1 173 188 44.3 405
b a
0.62 0.77 4.35 4.79 1.39 10.5 0.38 0.44 1.75 2.40 0.82 4.97 1.09 1.18 2.67 5.39 2.27 10.3
0.82 0.69 1.11 1.73 2.84 1.23 0.73
WL SH
0.39 0.33 0.42 0.61 1.35 0.61 0.44 0.55 0.48 0.65 0.99 3.00 1.56 0.83
SL coke
29.6 16.0 68.1 96.7 135 44.4 51.5 34.3 11.7 29.3 210 231 41.0 482 2.72 1.79 1.84 2.00 3.29 1.15 1.26 0.94 0.25 1.12 8.34 5.69 1.37 15.4
WH WL
0.61 0.41 0.85 1.21 2.32 0.80 0.66 0.66 0.18 0.71 3.08 3.78 0.90 7.75 0.41 0.27 0.45 0.51 1.55 0.49 0.53 0.44 0.14 0.53 1.63 2.57 0.67 4.87
SH SL
0.66 0.53 0.69 0.97 3.48 1.26 1.00 1.42 0.36 1.42 2.85 5.74 1.78 10.4
WHOI
Table 1. PAH Concentrations (ng/m3) Measured at Two Laboratories
fluoranthene (Fluo)a pyrene (Py) benzo[a]anthracene (BaA) chrysene (Chry) benzo[b+j+k]fluoranthene (BbjkF)b benzo[e]pyrene (BeP) benzo[a]pyrene (BaP) dibenzo[a,h]anthracene (DBahA) indeno[1,2,3-cd]pyrene (Ind-Py) benzo[ghi]perylene (BghiP) T1c T2d T3e total PAHsf average
GT
c
0.70 1.04 9.97 6.02 1.74 17.7
acid analysis (not reported in this study) and to decompose some interfering compounds in the PAH fraction to improve the final purity of the targeted PAHs. The neutral fraction was extracted three times with hexane and separated into three fractions using silica gel column chromatography (Fisher 100− 200 mesh, deactivated with 5% water). Aliphatic and aromatic hydrocarbon (PAH) fractions were eluted with n-hexane and nhexane/dichloromethane (1:1 v/v) respectively. The polar fraction was eluted with dichloromethane/methanol (9:1 v/v). The PAH fraction was cleaned up with n-pentane/dimethylformamide partitioning20 and dried with sodium sulfate anhydrous. Solvent was reduced to proper volume under gentle nitrogen stream. The PAHs were quantified with a gas chromatography with flame ionization detector (GC-FID). Details of the preparative capillary gas chromatography (PCGC) method used for PAH isolation have been reported previously.21,22 PAH fractions were repeatedly injected into a preparative capillary gas chromatography. Compounds were separated on a Chrompak CP-Sil 5 CB capillary column (50 m length, 0.53 mm i.d., 1 μm film thickness). About 1% of the effluent was diverted to the FID for signal monitoring. The rest of the material was collected in the glass U-tubes. The four-ring PAHs (fluoranthene, pyrene, benzo[a]anthracene, and chrysene) were trapped in the first trap (T1). Five-ring benzo[b+j +k]fluoranthene, benzo[e]pyrene, and benzo[a]pyrene were trapped in the second trap (T2). Dibenzo[a,h]anthracene, indeno[1,2,3-cd]pyrene, and benzo[ghi]perylene were trapped in the third trap (T3). Since the low molecular weight PAHs, such as naphthalene and acenaphthylene, are dominant in the gas phase and their abundance in PM2.5 is quite low so that radiocarbon analysis could not be performed with these much lighter and volatile PAHs. Trapped PAHs were transferred by 4 times rinsing with dichloromethane to 4 mL vials. The samples were passed through a silica gel column (4 cm × 0.5 cm i.d.) to remove column bleed from PCGC. A small aliquot from each trap was taken to check the purity of the PAHs by GC-FID. Carbon isotope analysis was performed at the National Ocean Science Accelerator Mass Spectrometry (NOSAMS) facility at WHOI. The rest of PAHs were transferred to prebaked quartz tubes and solvent was evaporated. After the addition of 100 mg of copper(II) oxide, the quartz tubes were sealed under vacuum and combusted at 850 °C for 5 h. About 10% of the generated CO2 was utilized for δ13C analysis, and the remainder was reduced to graphite for AMS as described by Pearson et al.23 The radiocarbon analysis results were reported as fraction modern compared to the NBS oxalic acid I (NIST SRM 4990) after δ13C correction. The Δ14C values were calculated based on the sampling year.24 The reported Δ14C error includes the internal error (statistical uncertainty by the number of 14C counts) and external uncertainty (uncertainty from reproducibility of the same sample).25 For small samples in the present study, the correction of combustion blank may cause some error. The filters were not combusted directly in our method. They were solvent extracted instead. The PAH blank from solvent and baked filters was negligible, and other impurities in the solvent or other contaminating compounds in the sample handling processes were removed in the PCGC step. The column bleed from PCGC was cleaned by column chromatography. The blank from the solvent in the final bleed cleanup (less than 4 mL) was also negligible. Without enough sample, we were unable to do duplicate CCSRA in this study. Methyl hexacosanoate from Sigma was tested if the steps from PCGC will cause a bias on the Δ14C. We observed no
d
0.51 1.38 0.36 1.08 0.30 0.93 0.38 1.20
coke
Article
0.33 1.20
0.88 0.74 1.01 0.75 1.00 0.70 1.16
WH WL
0.74 0.60 0.76 0.70 0.82 0.65 0.91 30.0 26.2 52.8 63.7 106 41.2 41.3 3.09 2.42 1.81 2.66 3.29 1.65 1.08
WH
coke
1.19 1.10 1.07 0.98 1.16 0.80 1.21
SL
1.05 0.81 1.07 0.83 1.14 0.81 1.20
SH
WHOI/GT ratio
0.98 0.61 1.29 1.52 1.28 1.08 1.25
Environmental Science & Technology
1424
dx.doi.org/10.1021/es2043189 | Environ. Sci. Technol. 2012, 46, 1422−1429
Environmental Science & Technology
Article
Figure 1. Comparison of PAH concentrations at different SEARCH sites.
difference (Δ14C was −409‰ after PCGC, compared to −410‰, −410‰, and −412‰ from original samples). Based on the Δ14C data and the fact that fossil fuel contains no 14C while biomass has modern levels of radiocarbon, the relative contribution of these two sources to PAHs can be estimated from the following equation
previous source apportionment studies based on chemical mass balance (CMB) model12,19,30 demonstrated that there were seven major primary sources including emissions from dieselpowered vehicles, gasoline-powered vehicles, biomass burning, paved road dust, meat cooking, vegetative detritus, and coke facilities. As a site located approximately 2 km from a coke plant, the BHM site can be occasionally impacted by emissions from coke facilities, especially when it is downwind of the coke facilities and under special atmospheric conditions (e.g., when strong atmospheric inversion occurs). However, this source only accounted for ∼2% of OC in PM2.5 samples on average and in a few cases reached 10% (Figure 2a). The influence of coke plants on PAHs at this site can be clearly seen from a significantly positive correlation between concentrations of PAHs and source impacts from coke facilities estimated from CMB (Figure 2b). Distribution and Ratios of PAHs. About twelve PAHs from the same filter were measured at the Georgia Institute and Technology (GT) and WHOI independently. Benzo[b]fluorantherene, benzo[j]fluoranthene, and benzo[k]fluoranthene are lumped together in Table 1 as B[bjk]F since these three compounds cannot have baseline separation in GC chromatograms. It can be seen in Table 1 that the concentrations measured at two laboratories (GT and WHOI) are comparable, except for Ind-Py. The averaged ratio of WHOI to GT was 1.10 ± 0.33, 1.01 ± 0.27, 0.94 ± 0.25, 0.85 ± 0.23, and 0.71 ± 0.18 for the coke sample, the SL sample, the SH sample, the WH sample, and the WL sample, respectively. As reported by Zheng et al.,19 the uncertainty of the organic compounds measurement was evaluated as approximately 20% which generally covered the difference from our interlaboratory comparison. The data agreed better for summer samples compared to winter samples. A lower PAH level was consistently seen for Ind-Py from WHOI measurement, only about one third or half of the levels reported at GT, indicating the difference is primarily due to analytical methods. In the following text, the discussion of PAH concentrations, compositions, and ratios were only based on the data from GT. It is interesting to note that total PAH concentration in the SL sample was twice of that of the SH sample, while in winter, higher total PAH concentration was seen in the WH sample (also by 2-fold), suggesting combustion products were more enriched in the SL and
Δ14 CPAH = (Δ14 Cbiomass)(f biomass) + (Δ14 Cfossil fuel )(1 − f biomss)
(1)
where Δ CPAH is the measured C content of each PAH fraction, fbiomass is the fraction of PAH derived from biomass burning, and 1-fbiomss is the fraction of PAH from fossil fuel combustion (ffossil fuel). The characteristic radiocarbon abundance of fossil fuel (Δ14Cfossil fuel) is −1000‰. There are at least two components for the modern end. One source is the wood burning of trees which grew in the elevated atmospheric 14C over the past 30 to 50 years. This component may have a Δ14C level of +225‰.7,26 The other component is from burning of materials produced during current year such as leaves and annual grass with a Δ14C of +70 ‰.5 Since no biomass burning sample was analyzed in the current study, we adapted a Δ14C of +152‰8 as the modern end for the calculation of fbiomass in this study. 14
14
■
RESULTS AND DISCUSSION Highest PAH Level at the North Birmingham Site. Based on data from 2001 to 2005, the highest average PM2.5 concentration among SEARCH sites (18.3 μg/m3) was found at BHM. Organic matter contributed about 39% of PM2.5 mass, followed by sulfate (24%), and elemental carbon (10%).27,28 BHM aerosol also contained more EC than other SEARCH sites (usually less than 10% of total PM2.5). Since EC is mainly from diesel exhaust and wood burning,29 the high level of EC at BHM indicated the importance of combustion sources at this site. The summed PAH concentration at this site was about 10 times higher than that measured at three other sites, including the paired rural site in Alabama (Centreville, CTR), the urban (Jefferson Street, Atlanta, JST) site in Georgia, and the urban site in Florida (Pensacola, PNS),19,28 and there were episodic events with significantly elevated concentrations of PAHs (close to 400 ng/m3) (Figure 1). In the southeastern US, our 1425
dx.doi.org/10.1021/es2043189 | Environ. Sci. Technol. 2012, 46, 1422−1429
Environmental Science & Technology
Article
Figure 2. (a) OC contributions from coke facility emission and (b) correlation between OC from coke facility emission and sum of PAHs at BHM site.
WH samples. The concentrations of PAHs seemed to be inversely correlated to OC/EC in summer but they were correlated in winter. This is reasonable because higher OC/EC ratio in the SH sample was likely due to higher contribution from secondary organic aerosol31 which was formed without combustion procedure; while in winter, OC at BHM was mainly from primary combustion sources,12 resulting in a higher OC/EC ratio and high PAH level (as also seen in the WH sample).19 Several PAHs with similar molecular weight and structure were pooled for AMS analysis, mainly because of the need for enough material. T1, T2, and T3 fractions contain the mixture of Fluo+Py+BaA+Chry (MW 202 and 228), of BbjkF+BeP+BaP (MW 252), and of DBahA+Ind-Py+BghiP (MW 276 and 278), respectively (Table 1). The relative contribution of each fraction is shown in Figure 3. PAHs in BMH were clearly dominated by lighter PAHs, with the majority PAHs present in the T1 and T2 fractions (40 ± 10% and 45 ± 6%, respectively). In winter, the contributions of the lightest PAHs (T1 fraction) to total PAHs were higher (56% in WH and 41% in WL) as compared to those (26% in SL and 35% in SH) in summer.
Figure 3. Relative contribution of each fraction to total PAHs.
Ratios between specific PAHs have been used in some studies to assist the identification of PAH emission sources. Yunker et al.32 used the ratio of Fluo/(Fluo+Py) to distinguish sources including unburned petroleum (ratio
