Source Apportionment of Daily Fine Particulate Matter at Jefferson Street, Atlanta, GA, during Summer and Winter

Transcription

1 Journal of the Air & Waste Management Association ISSN: (Print) (Online) Journal homepage: Source Apportionment of Daily Fine Particulate Matter at Jefferson Street, Atlanta, GA, during Summer and Winter Mei Zheng, Glen R. Cass, Lin Ke, Fu Wang, James J. Schauer, Eric S. Edgerton & Armistead G. Russell To cite this article: Mei Zheng, Glen R. Cass, Lin Ke, Fu Wang, James J. Schauer, Eric S. Edgerton & Armistead G. Russell (2007) Source Apportionment of Daily Fine Particulate Matter at Jefferson Street, Atlanta, GA, during Summer and Winter, Journal of the Air & Waste Management Association, 57:2, , DOI: / To link to this article: Published online: 29 Feb Submit your article to this journal Article views: 240 View related articles Citing articles: 52 View citing articles Full Terms & Conditions of access and use can be found at

2 TECHNICAL PAPER ISSN J. Air & Waste Manage. Assoc. 57: Copyright 2007 Air & Waste Management Association Source Apportionment of Daily Fine Particulate Matter at Jefferson Street, Atlanta, GA, during Summer and Winter Mei Zheng and Glen R. Cass School of Earth and Atmospheric Sciences, Georgia Institute of Technology, Atlanta, GA Lin Ke School of Earth and Atmospheric Sciences, Georgia Institute of Technology, Atlanta, GA; and Department of Biological Science, California State University, Los Angeles, CA Fu Wang School of Earth and Atmospheric Sciences, Georgia Institute of Technology, Atlanta, GA; and East China University of Science and Technology, Shanghai, People s Republic of China James J. Schauer Environmental Chemistry and Technology Program, Wisconsin State Laboratory of Hygiene, University of Wisconsin-Madison, Madison, WI Eric S. Edgerton Atmospheric Research & Analysis, Inc., Cary, NC Armistead G. Russell School of Civil and Environmental Engineering, Georgia Institute of Technology, Atlanta, GA ABSTRACT The primary emission source contributions to fine organic carbon (OC) and fine particulate matter (PM 2.5 ) mass concentrations on a daily basis in Atlanta, GA, are quantified for a summer (y 3 to August 4, 2001) and a winter (uary 2 31, 2002) month. Thirty-one organic compounds in PM 2.5 were identified and quantified by gas chromatography/mass spectrometry. These organic tracers, along with elemental carbon, aluminum, and silicon, were used in a chemical mass balance (CMB) receptor model. CMB source apportionment results revealed that major contributors to identified fine OC concentrations include meat cooking (7 68%; average: 36%), gasoline exhaust (7 45%; average: 21%), and diesel exhaust (6 41%; average: 20%) for the summer month, and wood combustion (0 77%; average: 50%); gasoline exhaust IMPLICATIONS Understanding the sources of carbonaceous aerosol, a major constituent in PM 2.5, as well as PM 2.5, is essential for formulating effective control strategy for PM 2.5 emissions. This study provides a detailed analysis of PM 2.5 sources with 56 daily samples from Jefferson Street during y 2001 and uary This site was the former Atlanta Supersite. The CMB results reveal that primary emissions are important in winter, especially wood burning, as well as vehicular exhaust, and suggest that secondary aerosol formation dominates in summer in Atlanta. (14 69%; average: 33%), meat cooking (1 14%; average: 5%), and diesel exhaust (0 13%; average: 4%) for the winter month. Primary sources, as well as secondary ions, including sulfate, nitrate, and ammonium, accounted for 86 13% and % of the measured PM 2.5 mass in summer and winter, respectively. INTRODUCTION Recent epidemiological studies suggest that exposure to atmospheric fine particulate matter (PM 2.5 ) can cause adverse effects, including cough, respiratory stress in asthmatics, and reduced lung function. 1 In 1997, U.S. Environmental Protection Agency (EPA) implemented new national ambient air quality standards for ground-level PM 2.5 of 15 g m 3 for an annual average and 65 g m 3 for a daily average (3-yr average of the 98 percentile daily concentration). 2 Carbonaceous aerosol is a major component of PM 2.5, which is composed of hundreds of organic compounds with a variety of chemical and physical properties and emitted from a wide variety of sources Organic matter has been reported to contribute 10 70% of the total atmospheric dry fine particle mass. 12 However, the concentration, composition, source attribution, and transformation mechanism of organic aerosol in the atmosphere are not totally understood. A better understanding of the major sources and their contributions to organic aerosol and PM 2.5 is, therefore, a principal step to identify effective controls to reduce PM 2.5 in the atmosphere. 228 Journal of the Air & Waste Management Association Volume 57 February 2007

3 The chemical mass balance receptor modeling method has been proven to be an effective tool to apportion source contributions of airborne particulate matter. 13 Earlier applications of chemical mass balance (CMB) receptor modeling method used elements to apportion source contributions in airborne particles. However, some important sources do not produce emissions that have unique elemental tracers; instead, they emit some specific organic tracers. For example, the emission from meat cooking is largely organic carbon (OC) with cholesterol as its unique tracer but lacking the unique elements that could help identify this source. To improve this technique and include these important sources, Schauer et al. 14 included organic compounds in CMB for source apportionment. In their study, the contributions of up to nine primary particle source types were identified in ambient samples. Source apportionment of PM 2.5 using organic compounds as fitting species has been applied in a few studies Most of the studies focused on the spatial and/or seasonal comparisons using monthly composite samples 15,17,18 or on episodic events. 16 However, comprehensive study using organic tracers on a daily basis has not been reported. The lack of such daily investigation is, in part, because of the extensive sampling, and complex organic speciation analysis required. Here, organic tracers are used in CMB source apportionment to quantify daily contributions of the emission sources to OC and fine particle mass concentrations at the Jefferson Street monitoring site in Atlanta, GA (JST). This site was the location of the former Atlanta Supersite, 19 and data from this location are being used in ongoing health effects studies. Based on the spatial variability analysis of air pollutants at several sites in Atlanta, Jefferson Street was found as representative as any other site in Atlanta for both primary 20 and secondary pollutants. 21 EXPERIMENTAL WORK Sampling Daily PM 2.5 samples were collected on quartz fiber filters with the high volume sampler at JST. JST is located 4.2 km northwest of downtown Atlanta, in a light industrial and commercial area. This site is one of the urban sites in the Southeastern Aerosol Research and Characterization (SEARCH) air quality monitoring network. 22 Sampling of PM 2.5 was conducted daily during y 3 to August 4, 2001, and uary 2 31, 2002, representing summertime and wintertime periods, respectively. y 2001 (or ESP01 intensive) and uary 2002 (or ESP02 intensive) are two periods when EPA Supersites were conducting intensive measurements in the eastern United States. For the summer ambient samples, 24-hr ambient PM 2.5 samples were collected on quartz fiber filters (102 mm diameter) using a California Institute of Technologydeveloped high-volume dichotomous virtual impactor. 23 For the winter samples, PM 2.5 was collected on a quartz fiber filter (8 10 in.) with the Thermo Anderson high volume sampler (Andersen Instruments, Inc.). Other data, including OC, elemental carbon (EC), Al, Si, sulfate, nitrate, and ammonium were obtained from PM 2.5 samples taken with the Particulate Composition Monitor (PCM) sampler (Atmospheric Research & Analysis, Inc.) as described by Hansen et al., 22 which was operated in parallel with the high-volume sampler. The PCM sampler is a three-channel filter-based sampling system with inlets 5 m above ground level, designed to collect 24-hr samples. Each channel has a 10- m cyclone followed by a Well Impactor Ninety-Six, which has a 2.5- m cutoff size in particle aerodynamic diameter. A Harvard-Brigham Young University carbon paper denuder is equipped upstream of the quartz filter (channel 3) to remove gas-phase organic aerosol. PM 2.5 mass was measured on the 47-mm diameter Teflon filter from channel 1. If data from the filter were unavailable, mass from a continuous monitor (tapered element oscillating microbalance) measured at 30 C was used. A portion of quartz filter from channel 3 was analyzed by Desert Research Institute for OC and EC using the thermal optical reflectance (TOR) method. 24 Organic Speciation Analysis A total of 56 daily ambient samples (28 for summertime and 28 for wintertime) along with two field blanks were analyzed. Organic species in PM 2.5 were analyzed using the standardized method described by Zheng et al. 17 Briefly, after being spiked with deuterated internal standard (IS) mixtures, the quartz filter was extracted under mild sonication with successive additions of hexane (Fisher Optima Grade) and benzene/isopropyl alcohol (2:1, v/v; benzene: E&M Scientific; isopropyl alcohol: Fisher Optima Grade). Extracts were filtered, combined, and concentrated to the original volume of the IS mixture spiked. An aliquot of the concentrated extract was subject to derivatization with diazomethane to convert organic acids to their methyl ester analogues. Derivatized extract was then analyzed by gas chromatography (GC)/mass spectrometry. Remaining extract was stored in a freezer for future use. Target organic species in the extracts were identified and quantified by GC/mass spectrometry using a Hewlett- Packard (HP) 6890 GC equipped with an HP mass selective detector using a 30-m length 0.25-mm i.d m film thickness HP-5 mass spectrometry capillary column coated with 5% phenyl methyl siloxane. The GC/mass spectrometry operating conditions were as follows: oven temperature: 65 C for 2 min, 10 C min 1 to 300 C, hold at 300 C for 20 min; GC injector and GC/ mass spectrometry interface temperature: 300 C; carrier gas: ultrapure helium; flow rate: 1 ml min 1 ; injection mode: splitless; scan range: amu; and electron ionization mode: 70 ev. Organic tracers were identified and quantified. Most of the organic species were identified by comparing their mass spectra and elution times with those in the authentic primary standards (PMSTD). Identification of the target compounds that are not found in PMSTD was achieved by referring to secondary standards, such as candle wax and source emission samples. Quantification of the target compounds was done using the relative response factors of the target compounds to their corresponding IS obtained from the PMSTD. The concentrations of target compounds are presented in Tables 1 and 2. Volume 57 February 2007 Journal of the Air & Waste Management Association 229

4 Table 1. Atmospheric concentrations of fine particle-phase organic compounds in the summer daily samples (ng m 3 ). Compounds Aug 2 Aug 3 Aug 4 Aug Pentacosane a Hexacosane a Heptacosane a Octacosane b Nonacosane a Triacontane a Hentriacontane a Dotriacontane b Tritriacontane c (H)-21 (H) Norhopane a 17 (H)-21 (H)-Hopane b ,29,30-Trisnorneohopane a S,R-5 (H),14 (H),17 (H) Cholestanes b 20R-5 (H),14 (H),17 (H) Cholestane b 20S,R-5 (H),14 (H),17 (H) Ergostanes b 20S,R-5 (H),14 (H),17 (H) Sitostanes b Benzo(b)fluoranthene b Benzo(k)fluoranthene b Benzo(e)pyrene c Indeno(cd)fluoranthene c Indeno(cd)pyrene c Benzo(ghi)perylene c Octadecenoic acid b Levoglucosan b Cholesterol b Nonanal a Notes: a Identified using mass spectra and quantified using primary authentic standard with similar structure and volatility; b Identified and quantified using PMSTD; c Identified using secondary standards (candle wax, source emission samples, etc.) and quantified using primary authentic standard with similar structure and volatility. 230 Journal of the Air & Waste Management Association Volume 57 February 2007

5 Table 2. Atmospheric concentrations of fine particle-phase organic compounds in the winter daily samples (ng m 3 ) Compounds Pentacosane a Hexacosane a Heptacosane a Octacosane b Nonacosane a Triacontane a Hentriacontane a Dotriacontane b Tritriacontane c (H)-21 (H) Norhopane a 17 (H)-21 (H)-Hopane b ,29,30-Trisnorneohopane a S,R-5 (H),14 (H),17 (H) Cholestanes b 20R-5 (H),14 (H),17 (H) Cholestane b 20S,R-5 (H),14 (H),17 (H) Ergostanes b 20S,R-5 (H),14 (H),17 (H) Sitostanes b Benzo(b)fluoranthene b Benzo(k)fluoranthene b Benzo(e)pyrene c Indeno(cd)fluoranthene c Indeno(cd)pyrene c Benzo(ghi)perylene c Octadecenoic acid b Levoglucosan b Cholesterol b Nonanal a Notes: uary. a Identified using mass spectra and quantified using primary authentic standard with similar structure and volatility; b Identified and quantified using PMSTD; c Identified using secondary standards (candle wax, source emission samples, etc.) and quantified using primary authentic standard with similar structure and volatility. Volume 57 February 2007 Journal of the Air & Waste Management Association 231

6 The present analytical method was subject to a series of quality control procedures. All of the glassware was prebaked at 550 C for 12 hr, and the solvents used for extraction were Optima grade (hexane and isopropyl alcohol) or distilled (benzene). Before the GC/mass spectrometry analysis of ambient samples, a sensitivity test was conducted to ensure that the instrument performance meets quality assurance criteria for GC/mass spectrometry analysis. 25 The standard mixture for sensitivity test contains coronene (0.05 ng L 1 ), pyrene (0.02 ng L 1 ), and cholesterol (0.20 ng L 1 ). The diagnostic criteria of the instrumental sensitivity include coronene to pyrene ratio (m/z 300 to m/z 202; target: 0.25), coronene abundance (target: 10,000), and cholesterol abundance (target: 1,000,000). CMB Analysis The CMB receptor modeling method using organic markers as mass balance species has been applied recently in source apportionment studies In CMB, the ambient concentrations of chemical species are expressed as the sum of the products of source contributions and source compositions. 26 The concentration of chemical species i at the receptor site k, c ik, is calculated as: m c ik a ij s jk (1) j 1 where a ij is the relative concentration of chemical species i in the fine OC emission from source j, and s jk is the contribution to fine OC from source j at the receptor site k. Equation 1 states that the ambient concentration of mass balance species must be from the m sources included in the model and that no selective loss or gain occurs during transport from the source to the receptor site. Therefore, the selection of mass balance species must be limited to species for which all of the major sources are included in the model and species that are conserved in the atmosphere with no significant removal through dry and wet depositions or formation by chemical reactions over time during transport from the source to the receptor site. In the present study, CMB analysis was computed using the CMB7.0 software developed by Watson et al. 26 The diagnostics of CMB include the percentage of mass explained (target: %), R 2 (target: 0.8 1), 2 (target: 0 4), t test (target: 2), degrees of freedom (target: 5), no cluster sources, and calculated-to-measured ratio for fitting species (target: 0.5 2). 26 The selection criteria for particle-phase organic compounds used as tracers in the present study are based on the previous work by Schauer et al. 14 The list of organic tracers adopted by Schauer et al. 14 has been extended for the recent development in quantification of some molecular markers in the source testing studies, such as levoglucosan 28 and cholesterol. 29 In the present study, a total of 31 individual organic tracers along with three additional chemical constituents were applied in the CMB model to quantify the contribution of up to seven primary emission source categories to fine OC. The three additional chemical constituents are EC, silicon, and aluminum; the organic tracers are nine n-alkanes (with the number of carbon atoms between 25 and 33), seven hopanes and steranes, three resin acids, six polycyclic aromatic hydrocarbons (PAHs), two unsaturated fatty acids, and four other organic tracers. Seven source emission profiles were identified for use in the present study and obtained from previous source testing studies. The seven source profiles applied here include emissions from diesel-powered vehicles, 30 combined catalyst- and noncatalyst-equipped gasoline-powered vehicles, 31 wood combustion, 32 paved road dust, 33 meat cooking, 34 vegetative detritus, 6 and natural gas combustion 7. Two of the source profiles, including wood combustion and paved road dust, were reconstructed by Zheng et al. 17 specially for the southeastern United States. The denuded OC data obtained from previous work by Schauer et al. 30,31,34 were used in the source profiles of gasoline exhaust, diesel exhaust, and meat cooking in this study. The lack of denuded OC data for the source of wood combustion is because of an overload problem in the denuder during collection of the exhaust from wood combustion. The other three sources (road dust, vegetative detritus, and natural gas combustion) do not have denuded OC data. However, they are minor sources of ambient fine particulate OC in the southeastern United States. 17 EC and OC in the above source profiles were measured using the thermal optical transmittance (TOT) method. However, the ambient samples in this study were measured using the TOR method. Bias will be introduced in CMB if EC and OC in source profiles and ambient samples are measured by different protocols. Therefore, the ambient EC and OC data by the TOR method were converted to the equivalent TOT EC and OC data using the equations presented as the footnote in Tables 3 and 4. These equations were developed by analyzing the same filter with both protocols, and different equations were applied for summer and winter samples. The converted EC and OC data are used in CMB modeling and in the following discussions. RESULTS AND DISCUSSION Solvent Extractable Organic Compounds Thirty-one organic compounds were identified and quantified in the daily ambient samples (Tables 1 and 2). Here, the authors use these tracers to quantify the daily source contributions to fine OC and particle mass concentrations at JST for the ESP01/02 periods. Most compounds are not present in field blanks. However, some alkanes were detected in field blanks ( 5% on average). Octadecanoic acid is the only species that exhibited a significant level in field blanks ( 30%). Field blank was subtracted. The average uncertainty of organic tracer analysis was estimated as 16 4% from seven replicate analyses of standard reference material of urban dust (SRM1649a). By far, only several PAHs in SRM1649a have certified values, and the recovery of these PAHs with certified values ranges from 70% for fluoranthene to 144% for benzo(k)fluoranthene. Distribution of EC and OC Concentrations of EC and OC fluctuated significantly during both summer and winter sampling periods (Figure 1). The highest ambient concentrations of EC and OC for the summer days were measured on y 17 (0.91 gm 3 ) 232 Journal of the Air & Waste Management Association Volume 57 February 2007

7 Table 3. Source apportionment of fine organic carbon at Jefferson Street, Atlanta in the summer (mean SD in g m 3 ). Sampling Date Diesel Exhaust Gasoline Exhaust Meat Cooking Wood Combustion Road Dust Natural Gas Combustion Vegetative Detritus Other OC Identified OC Measured OC (TOR) OC (Converted % Measured Degrees of to TOT OC) e OC r 2 2 Freedom 7/3/ /4/ NA d /5/2001 a 7/6/ NA /7/ c NA /8/ NA /9/ /10/ /11/2001 a 7/12/ /13/ /14/ /15/ /16/ /17/ /18/ /19/ /20/ /21/ /22/2001 a 7/23/ /24/2001 a 7/25/ /26/ /27/ /28/2001 b 7/29/ /30/ /31/ /1/ /2/ /3/ /4/ Average Notes: a No sample; b OC data were not available; c Insignificantly different from zero with 95% confidence; d Concentrations of aluminum and silicon were not available; e OC is converted using the equation TOT *TOR and EC is converted by TOT *TOR Volume 57 February 2007 Journal of the Air & Waste Management Association 233

8 Table 4. Source apportionment of fine organic carbon at Jefferson Street, Atlanta in the winter (mean SD in g m 3 ). Sampling Date Diesel Exhaust Gasoline Exhaust Meat Cooking Wood Combustion Road Dust Natural Gas Combustion Vegetative Detritus Other OC Identified OC Measured OC (TOR) OC (Converted % Measured Degrees of to TOT OC) f OC r 2 2 Freedom 1/2/ NA d /3/ NA /4/ NA /5/ NA /6/2002 c NA /7/ /8/ NA NV e /9/ NA /10/ NA /11/2002 a 1/12/ NV /13/ NA /14/ /15/ /16/ NV /17/ /18/ /19/ /20/2002 b 1/21/ NV /22/ /23/ /24/ /25/ /26/ NV /27/ /28/ /29/ /30/ /31/ Average Notes: a No sample; b OC data were not available; c Insignificantly different from zero with 95% confidence; d Concentrations of aluminum and silicon were not available; e Negative value; f OC is converted using the equation TOT *TOR and EC is converted by TOT 0.481*TOR Journal of the Air & Waste Management Association Volume 57 February 2007

9 0.20 Concentration of EC and OC (µg m -3 ) (a) Summer (. 3 - Aug. 4, 2001) EC OC EC/OC EC/OC 0 * * * * ** 0.00 Wed 02 Fri 04 Sun 06 Tue 08 Thu 10 Sat 12 Mon 14 Wed 16 Tue 03 Thu 05 Sat 07 Mon 09 Wed 11 Fri 13 Sun 15 Tue 17 Thu 19 Sat 21 Mon 23 Wed 25 Fri 27 Sun 29 Tue 31 Thu 02 Sat (b) Winter (. 2-31, 2002) Fri 18 Sun 20 Tue 22 Thu 24 Sat 26 Mon 28 Wed 30 Concentration of EC and OC (µg m -3 ) EC/OC 0 * ** 0.00 Sampling Date Figure 1. Daily distribution of EC and OC at Jefferson Street, Atlanta: (a) summer period (y 3 to August 4, 2001) and (b) winter period (uary 2 31, 2002). *Sample was not collected; **Concentrations of EC and OC were not available. and y 13 (10.14 g m 3 ), respectively. uary 27 (1.58 g m 3 ) and uary 28 (11.70 g m 3 ) samples exhibited the highest levels of EC and OC, respectively, in the winter. Higher EC and OC concentrations were found in the winter ( g m 3 EC and g m 3 OC) as compared with the summer ( g m 3 EC and g m 3 OC). In urban areas, EC is believed to be largely from diesel engine exhaust, which has significantly higher EC/OC ratios (1.24) than other available source categories as tested by Hildemann et al., 35 including meat cooking (0), wood combustion ( ), gasoline exhaust with (0.45) and without (0.12) a catalyzed converter, road dust (0.08), and vegetative detritus (0.03). Coal combustion emission in China was found to have a high EC/OC ratio (1.03) in a recent study by Zheng et al. 36 Emission from fuel oil combustion is an exception, which has an extremely high EC/OC ratio of EC and OC in the above source profiles were all Volume 57 February 2007 Journal of the Air & Waste Management Association 235

10 OC Concentration (µg m -3 ) Diesel Exhaust Gasoline Exhaust Meat Cooking Wood Combustion Road Dust Natural Gas Combustion Vegetative Detritus Other Organic Carbon 2 * 0 Tue 03 Thu 05 Sat 07 Mon 09 Wed 11 Fri 13 Sun 15 Tue 17 Thu 19 Sat 21 Mon 23 Wed 25 Fri 27 Sun 29 Tue 31 Thu 02 Sat 04 * * * ** Sampling Time Figure 2. Source contributions to fine OC at Jefferson Street, Atlanta in the summer (y 3 to August 4, 2001). *Sample was not collected. **Concentrations of EC and OC were not available. Propagated errors of the identified source contributions are shown. measured by the TOT method. The average EC/OC ratio at Jefferson Street is in summer and in winter (Figure 1). Primary emissions from coal and fuel oil combustion are expected to be very small in most locations in the United States because of less use of these fuels and effective control strategies on their emissions. 37 Coal and fuel oil combustions are not the major contributors of primary particulate matter (PM) in the United States, but coal combustion can be a major contributor to fine OC ( 13%) and PM during the cold heating season in Beijing, China. 36 Thus, the higher EC/OC ratio found in ambient samples is expected to be primarily from automobile exhaust and wood combustion. However, in winter, gasoline exhaust may contribute to the high EC/OC ratios observed in some samples, because EC emissions from gasoline-powered vehicles may vary greatly at different ambient temperatures. A recent study by Kittelson et al. 38 showed that EC emissions from gasoline-powered vehicles under a cold-cold start condition (engine oil at 0 C) were unexpectedly high, accounting for 30 60% of PM mass. Cold-cold started gasolinepowered vehicles were also found to generate emissions with a higher EC/OC ratio. 37 In the Atlanta area, the daily average of the low and high temperature in uary is 0.6 C and 11.1 C, respectively. Therefore the cold-cold start condition (ambient temperature 0 C) is likely to occur. Source Apportionment of Fine OC Source contributions to OC concentrations in PM 2.5 were estimated using the CMB modeling method by integrating the source profiles and the ambient concentrations of fitting species determined from the ambient daily PM 2.5 samples. For each source category, CMB provides a source contribution estimate and associated standard error uncertainty, which are estimated with the uncertainties of chemical tracers in ambient samples and source profiles. Contributions from seven emission sources and their uncertainties were determined, including diesel exhaust, gasoline exhaust, meat cooking, wood combustion, road dust, natural gas combustion, and vegetative detritus (Figures 2 and 3 and Tables 3 and 4). Seven primary emission sources can explain 29 13% and 73 25% of the total measured OC in the summer and winter ambient PM 2.5 samples, respectively. For the summer daily samples, meat cooking, gasoline exhaust, and diesel exhaust were the three largest primary sources contributing, respectively, at , , and g m 3 to OC. The other four sources of wood combustion, road dust, natural gas combustion, and vegetative detritus contributed less ( g m 3 on average). That a large portion of OC cannot be attributed to the primary sources was observed in the summer samples ( g m 3 ). This portion of OC is defined as other OC (obtained by subtracting the apportioned fine particle OC concentration from the total fine particle OC concentration) and likely consists of secondary OC, other primary OC not included in the model, and various uncertainties involved. More other OC or unexplained OC in summer ( 70%) suggests that secondary organic aerosol (SOA) contributes significantly to fine OC in summer in Atlanta. For the winter daily samples, the major primary emission sources were wood combustion, gasoline exhaust, meat cooking, and 236 Journal of the Air & Waste Management Association Volume 57 February 2007

11 12 Diesel Exhaust Gasoline Exhaust Meat Cooking Wood Combustion Road Dust Natural Gas Combustion Vegetative Detritus Other Organic Carbon Wed 02 Fri 04 Sun 06 Tue 08 Thu 10 Sat 12 Mon 14 Wed 16 Fri 18 Sun 20 Tue 22 Thu 24 Sat 26 Mon 28 Wed 30 OC Concentration (µg m -3 ) * ** Sampling Time Figure 3. Source contributions to fine OC at Jefferson Street, Atlanta in the winter (uary 2 31, 2002). *Sample was not collected; **Concentrations of EC and OC not available. Propagated errors of the identified source contributions are shown. diesel exhaust, contributing , , , and g m 3 to OC, respectively. The other two sources of natural gas combustion and vegetative detritus contributed equally to OC (0.13 g m 3 on average). Other OC was much less in the winter ( g m 3 ) and varied greatly ( g m 3 ). Gasoline exhaust and wood combustion exhibited the most obvious seasonal changes in their contributions to fine OC. For gasoline exhaust, the average contributions increased dramatically by 1 gm 3 from 0.20 gm 3 in the summer to 1.30 g m 3 in the winter. Similar results were found in the study by Zheng et al., 17 in which the CMB-calculated source contributions of gasoline exhaust at JST were 0.09 g m 3 and 0.83 g m 3 for the y and uary samples, respectively. When examining the results from the other SEARCH sites, a similar trend of larger contribution from gasoline exhaust in uary and smaller in y was found. 17 Activities of diesel- and gasoline-powered vehicles are likely similar between seasons. This was found for diesel exhaust ( g m 3 in y vs g m 3 in uary). The much larger contribution of gasoline exhaust to fine OC concentration during the wintertime is possibly because of higher emissions of PM and associated organic tracers (hopanes and steranes) from gasoline-powered vehicles in winter. For wood combustion, the average contribution to fine OC increased significantly from 0.07 g m 3 in the summer to 2.14 g m 3 in the winter. This is consistent with higher wood combustion activities in the winter (e.g., there are burn bans in the summer and more home heating/fireplace use of wood, as well as prescribed burning in the winter). PM 2.5 Mass Apportionment PM 2.5 mass apportionment can be calculated from fine OC source apportionment results based on the ratios of OC to PM 2.5 obtained from source tests. Sulfate, nitrate, and ammonium are mostly from secondary formation, because the amount from the primary sources identified in the present study is very small. The identified primary emissions of sulfate, nitrate, and ammonium were calculated according to the ratios of those ions to total OC reported in the primary particle source testing studies. Contributions from the primary sources identified plus secondary ions, on average, accounted for 86 13% and % of the measured PM 2.5 mass during the summer and winter days, respectively (Figures 4 and 5 and Tables 5 and 6). Other organic matter in both winter and summer samples was obtained by multiplying the other OC by a factor of 1.4. Because SOA and oxygenated species are abundant in summer, a higher factor could be used when converting other OC to other OM. This will lead to a higher percentage of explained mass in summer. Major contributors to PM 2.5 mass concentrations were secondary sulfate ( g m 3 ), other OM ( g m 3 ), secondary ammonium ( gm 3 ), meat cooking ( gm 3 ), diesel exhaust ( g m 3 ), secondary nitrate ( gm 3 ), road dust ( gm 3 ), and gasoline exhaust ( g m 3 ) for the summer samples and wood combustion ( g m 3 ), gasoline exhaust ( g m 3 ), secondary sulfate ( g m 3 ), other OM ( g m 3 ), secondary nitrate ( g m 3 ), secondary Volume 57 February 2007 Journal of the Air & Waste Management Association 237

12 PM2.5 mass concentrations (µg m -3 ) Diesel Exhaust Gasoline Exhaust Meat Cooking Wood Combustion Road Dust Natural Gas Combustion Vegetative Detritus Other Organic Matter Secondary Sulfate Secondary Nitrate Secondary Ammonium Others 0 * * * * ** Tue 03 Thu 05 Sat 07 Mon 09 Wed 11 Fri 13 Sun 15 Tue 17 Thu 19 Sat 21 Mon 23 Wed 25 Fri 27 Sun 29 Tue 31 Thu 02 Sat 04 ammonium ( g m 3 ), diesel exhaust ( g m 3 ), and meat cooking ( g m 3 ) for the winter samples. Sampling Date Figure 4. Source contributions to PM 2.5 at Jefferson Street, Atlanta in the summer (y 3 to August 4, 2001). *Sample was not collected; **Concentrations of EC and OC were not available. PM 2.5 mass apportionment results clearly show that secondary sulfate, secondary ammonium, and other OM or unexplained organic matter are the major components of Wed 02 Fri 04 Sun 06 Tue 08 Thu 10 Sat 12 Mon 14 Wed 16 Fri 18 Sun 20 Tue 22 Thu 24 Sat 26 Mon 28 Wed 30 PM2.5 mass concentrations (µg m -3 ) Diesel Exhaust Gasoline Exhaust Meat Cooking Combined Wood Combustion Road Dust Natural Gas Combustion Vegetative Detritus Other Organic Matter Secondary Sulfate Secondary Nitrate Secondary Ammonium Others 0 * ** Sampling Date Figure 5. Source contributions to PM 2.5 at Jefferson Street, Atlanta in the winter (uary 2 31, 2002). *Sample was not collected; **Concentrations of EC and OC were not available. 238 Journal of the Air & Waste Management Association Volume 57 February 2007

13 Table 5. Source apportionment of PM 2.5 at Jefferson Street, Atlanta in the summer ( g m 3 ). Sampling Date Diesel Exhaust Gasoline Exhaust Meat Cooking Wood Combustion Road Dust Natural Gas Combustion Vegetative Detritus Other OM Secondary Sulfate Secondary Nitrate Secondary Ammonium Identified Mass PM 2.5 Mass Others % Mass Explained 7/3/ /4/ NA d NV e /5/2001 a 7/6/ NA /7/ c NA /8/ NA /9/ /10/ /11/2001 a 7/12/ /13/ NV /14/ /15/ /16/ /17/ /18/ /19/ /20/ /21/ /22/2001 a 7/23/ /24/2001 a 7/25/ /26/ /27/ /28/2001 b 7/29/ /30/ /31/ /1/ /2/ /3/ /4/ Average Notes: a No sample; b OC data were not available; c Insignificantly different from zero with 95% confidence; d Concentrations of aluminum and silicon were not available; e Negative value. Volume 57 February 2007 Journal of the Air & Waste Management Association 239