Environ. Sci. Technol. 2009, 43, 272–279
Source Apportionment of Fine (PM1.8) and Ultrafine (PM0.1) Airborne Particulate Matter during a Severe Winter Pollution Episode M I C H A E L J . K L E E M A N , * ,† SARAH G. RIDDLE,‡ MICHAEL A. ROBERT,† CHRIS A. JAKOBER,§ PHILLIP M. FINE,| MICHAEL D. HAYS,⊥ JAMES J. SCHAUER,# AND MICHAEL P. HANNIGAN∇ Department of Civil and Environmental Engineering, University of California, Davis, 1 Shields Avenue, Davis California 95616, Department of Chemistry, University of California, Davis, 1 Shields Avenue, Davis California 95616, Agriculture and Environmental Chemistry Graduate Group, University of California, Davis, Department of Civil and Environmental Engineering, University of Southern California, Los Angeles, California, National Risk Management Research Laboratory, U.S. Environmental Protection Agency, Research Triangle Park, North Carolina, Department of Civil and Environmental Engineering, University of Wisconsin, Madison, Wisconsin, and Department of Mechanical Engineering, University of Colorado, Boulder, Colorado
Received February 09, 2008. Revised manuscript received October 17, 2008. Accepted October 29, 2008.
Size-resolved samples of airborne particulate matter (PM) collected during a severe winter pollution episode at three sites in the San Joaquin Valley of California were extracted with organic solvents and analyzed for detailed organic compounds using GC-MS. Six particle size fractions were characterized with diameter (Dp) < 1.8 µm; the smallest size fraction was 0.056 < Dp < 0.1 µm which accounts for the majority of the mass in the ultrafine (PM0.1) size range. Source profiles for ultrafine particles developed during previous studies were applied to the measurements at each sampling site to calculate source contributions to organic carbon (OC) and elemental carbon (EC) concentrations. Ultrafine EC concentrations ranged from 0.03 µg m-3 during the daytime to 0.18 µg m-3 during the nighttime. Gasoline fuel, diesel fuel, and lubricating oil combustion products accounted for the majority of the ultrafine EC concentrations, with relatively minor contributions from biomass combustion and meat cooking. Ultrafine OC concentrations ranged from 0.2 µg m-3 during the daytime to 0.8 µg m-3 during the nighttime. Wood combustion was found to be the largest source of ultrafine OC. Meat cooking was also identified as a * Corresponding author phone: (530) 752-8386; fax: (530) 7527872; e-mail:
[email protected]. † Department of Civil and Environmental Engineering, University of California. ‡ Department of Chemistry. University of California. § Agriculture and Environmental Chemistry Graduate Group, University of California. | University of Southern California. ⊥ U.S. Environmental Protection Agency. # University of Wisconsin. ∇ University of Colorado. 272
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 43, NO. 2, 2009
significant potential source of PM0.1 mass but further study is required to verify the contributions from this source. Gasoline fuel, diesel fuel, and lubricating oil combustion products made minor contributions to PM0.1 OC mass. Total ultrafine particulate matter concentrations were dominated by contributions from wood combustion and meat cooking during the current study. Future inhalation exposure studies may wish to target these sources as potential causes of adverse health effects.
Introduction Increasing evidence suggests that airborne ultrafine particles may pose a danger to human health (1). The shear number of ultrafine particles may overwhelm the alveolar macrophages that clear foreign objects from the lungs (2). Alternatively, ultrafine particles may cross cell membranes where they can interfere with the internal cell functions (3). The distribution of chemicals in ultrafine particles may make them more toxic than other size fractions. Different sources release ultrafine particles with different chemical composition (see, for example, refs 4-9) which could lead to variable health effects. Source contributions to health effects via atmospheric ultrafine particle exposure can be tested during during a series of simultaneous health effects and source apportionment experiments. This approach properly accounts for important atmospheric transformations and simultaneously considers multiple sources, which may reveal interactions that influence health effects. The key to this approach is a technique for the accurate source apportionment of airborne ultrafine particulate matter. Numerous studies have performed source apportionment calculations for coarse (Dp < 10 µm) and fine (Dp < 2.5 µm) airborne particle mass (PM) fractions (see, for example, refs 10-12), but very little is currently known about source contributions to airborne ultrafine particle mass (PM0.1). The focus of the current paper is to demonstrate a technique for PM0.1 source apportionment and to quantify PM0.1 source contributions during a severe air quality episode that occurred in California’s San Joaquin Valley (SJV) between December 2000 and January 2001. The SJV routinely experiences some of the highest particulate matter concentrations in the United States (13, 14), and the population in the region is growing rapidly. Source contribution information for PM01 will better characterize potential threats to public health in this region. Methods. Previous studies (13, 15) have described the PM0.1 measurement methods in great detail and so only a brief summary is provided here. PM samples were collected over 7 days during the period December 15-28, 2000 at Sacramento and over 11 days during the period December 15, 2000 to January 7, 2001 at Modesto and Bakersfield, California (15). Separate samples were collected during daytime (10 am to 6 pm) and nighttime (8 pm to 8 am) hours. Size-resolved samples were collected with a micro orifice uniform deposit impactor (MOUDI) loaded with aluminum foil substrates, whereas bulk PM1.8 samples were collected with a reference ambient aerosol sampler (RAAS) loaded with quartz fiber filters. All sampling media were prebaked at 500 °C for 48hrs to remove background carbon and stored at -18 °C before and after collection in Petri dishes lined with baked aluminum foil sealed with Teflon tape. Aluminum substrates were not coated with any kind of grease due to the interference that this would cause in the analytical methods. Samples were extracted in organic solvents and quantified using gas chromatagraphy-mass spectrometry (GC-MS). Each of the molecular markers was quantified using an isotopically labeled internal standard. Elemental carbon (EC) and organic 10.1021/es800400m CCC: $40.75
2009 American Chemical Society
Published on Web 12/11/2008
carbon (OC) concentrations were quantified using a thermal optical method following the NIOSH temperature protocol. Colocated MOUDI and RAAS samples were in strong agreement with correlation slopes >0.7 for all species and correlation coefficients (R (2)) > 0.99 for OC, > 0.9 for PAHs (excluding a few of the lightest semivolatile compounds), > 0.8 for hopanes and steranes, and > 0.99 for levoglucosan and cholesterol. Strong agreement between colocated impactor and filter samples suggests that volatilization losses were not a major problem since these losses are known to be influenced by the surface area of the collection medium and the flow rate through the collection device, both of which differ between the MOUDI and RAAS. Traditional molecular marker chemical mass balance (CMB) studies employ “source profiles” that describe the concentration of elements and molecules normalized by the concentration of particulate OC. Although numerous elements and molecules may be carried through this procedure, the majority of the source attribution information is contained in a set of “core” tracer compounds that are unique to individual source categories. These core tracer compounds are chosen for their specificity and for their stability in the particle phase once they are emitted to the atmosphere. Numerous semivolatile or reactive compounds are specific to certain sources but they are not conserved in the particle phase and so they do not carry useful source apportionment information. These elements and compounds are carried through the calculation primarily to corroborate the final CMB result with mass conservation checks. The difficulties encountered during the collection of ultrafine particles (low pressure, bounce artifacts, adsorption/ desorption artifacts) and the low concentration of ultrafine particles in the atmosphere make it difficult to obtain useful information about semivolatile and/or reactive tracer compounds for ultrafine source apportionment calculations. The core tracer compounds specific to individual source categories were used as the basis for ultrafine source apportionment calculations in the current study. These compounds include levoglucosan (biomass combustion), hopanes and steranes (lubricating oil), and heavy PAHs (gasolinefuel). PM released from coal combustion also contains significant quantitites of PAHs, hopanes, and steranes but this source is not present in California. The amount of EC and OC associated with each unit of these tracers emitted from sources present in California is discussed in previous manuscripts (4, 16). Cholesterol has been used historically as a tracer for meat cooking (17, 18) but recent evidence suggests that other “unknown” sources of cholesterol may be present in the atmosphere (19, 20). The concentration of meat cooking particles in the atmosphere identified with cholesterol is considered tentative in the present study. Diesel fuel combustion products contained in heavy duty diesel vehicle (HDDV) PM exhaust were quantified during the emissions source testing (5) using light PAHs fluoranthene and pyrene (16). These compounds are both semivolatile and reactive in the atmosphere (21) making their use for ambient source apportionment studies uncertain anywhere outside the immediate roadside environment. Traditional CMB calculations effectively use EC as a tracer for diesel fuel contributions to ambient PM because other possible sources of EC such as biomass combustion are constrained by specific molecular markers such as levoglucosan. EC that is not associated with other sources is therefore attributed to diesel engines through mathematical necessity to minimize the difference between predicted vs measured concentrations. This methodology was also adopted in the current study to estimate diesel fuel contributions to ambient PM. The source apportionment algorithm devised for the current study prioritizes sources in order from most certain to least certain based the level of confidence in the tracer
compounds. Conservation of mass constraints are employed separately for EC and OC so that uncertain sources do not apportion more carbon than the measured concentrations. The individual steps in the algorithm to predict source contributions to EC and OC are summarized as follows: (1) Calculate biomass contributions to EC using levoglucosan as a tracer. Certainty ) high. (2) Calculate motor oil combustion product contributions to EC using hopanes and steranes (expressed as equivalent concentration of 17R(H)-21β(H)-29-norhopane) as a tracer. Certainty ) high. (3) Calculate gasoline-fuel combustion product contributions to EC using benzo[ghi]perylene and coronene (expressed as equivalent concentration of benzo[ghi]perylene) as a tracer. Certainty ) high. (4) Calculate meat cooking contributions to EC using cholesterol as a tracer. Certainty ) low but predicted EC associated with meat cooking emissions is minimal. This step does not introduce significant uncertainty to calculated EC source contributions. (5) Assign residual EC to diesel-fuel. Certainty ) medium. (6) Calculate diesel-fuel combustion product contribution to OC using diesel-fuel EC as a tracer. Certainty ) medium. (7) Calculate biomass contributions to OC using levoglucosan as a tracer. Certainty ) high. (8) Calculate motor oil combustion product contributions to OC using hopanes and steranes (expressed as equivalent concentration of 17R(H)-21β(H)-29-norhopane) as a tracer. Certainty ) high. (9) Calculate gasoline-fuel combustion product contributions to OC using benzo[ghi]perylene and coronene (expressed as equivalent concentration of benzo[ghi]perylene) as a tracer. Certainty ) high. (10) Calculate meat cooking contributions to OC using cholesterol as a tracer. Certainty ) low. Constrain meat cooking contributions to the amount of residual OC not already accounted for by other sources. The simplified source apportionment algorithm was used to reanalyze the data set for the SJV described by Schauer and Cass (10) yielding good agreement with the original chemical mass balance (CMB) solution. Additional details are provided in the Supporting Information. Source Apportionment Profiles. Table 1 summarizes the source profiles used in the current study. Cholesterol and levoglucosan were used as unique tracers for meat cooking and biomass combustion, respectively, with ratios of tracer/ OC on MOUDI stages as described by Kleeman et al. (4). Benzo[ghi]perylene and coronene were used as a unique markers for gasoline-fuel PM, and 17R(H)-21β(H)-29-norhopane, 17R(H)-21β(H)-hopane, and Rββ20R-C29-ethylcholestane were used as a unique marker for lubricating oil PM as discussed by Kleeman et al. (16). Splitting the fuel and oil contributions to PM allows all source profiles to be expressed independent of particle size. Lubricating oil and gasoline-fuel contributions to EC and OC depend on vehicle technology. Lubricating oil profiles and gasoline-fuel profiles that were weighted by the distribution of vehicles expected in the vicinity of each sampling site were developed in the present study to quantify the PM contribution from the onroad gasoline fleet. The California Emissions Factor model (EMFAC) was used to estimate the distribution of PM emissions from different classes of gasoline vehicles in the counties surrounding Sacramento, Modesto, and Bakersfield during December 2000 and January 2001. EMFAC estimated that 95-96% of the PM2.5 emissions from gasoline-powered motor vehicles were produced by vehicles equipped with three way catalysts (TWC). Noncatalyst equipped vehicles (NCAT) were assumed to account for 3-4% of the PM2.5 emissions from gasoline-powered vehicles with the remaining 1% of the on-road gasoline vehicle PM2.5 emissions attributed VOL. 43, NO. 2, 2009 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
273
TABLE 1. Source Profiles Used for Simple Apportionment Calculations. All Source Profiles Are Independent of Size source diesel fuel gasoline gasoline lubricating oil lubricating oil lubricating oil wood burning meat cooking
tracer
tracer/OC
Tracer/EC
EC benzo[ghi]peryleneb,c coronenec 17R(H)-21β(H)-29-norhopanec 17R(H)-21β(H)-hopanec Rββ20R-C29-ethylcholestanec levoglucosan cholesterol
2.56 0.0011 ( 0.00013 0.0008 ( 0.000096 0.0006 ( 0.00027 0.0006 ( 0.00027 0.0003 ( 0.00013 0.15 ( 0.045 0.0015 ( 0.00075
1 0.0006 ( 0.000048 0.0004 ( 0.000032 0.0008 ( 0.00068 0.0008 ( 0.00068 0.0004 ( 0.00034 5.0 ( 1.5 0.15 ( 0.075
a
a Residual EC not explained by other sources was used as a tracer for diesel-fuel contributions to OC. Diesel fuel uncertainty was estimated using the combined uncertainty from all other source contributions in each sample. See text discussion for source apportionment algorithm. b Gasoline engines were found to be the dominant source of benzo[ghi]perylene during source tests. c Coal combustion may also release these compounds, but this source is not present in significant quantitites in California. See text discussion for additional caveats.
to vehicles emitting visible smoke (SMKR). All of the LDGV profiles are discussed in greater detail by Kleeman et al. (16). The amount of total carbon per unit of hopanes associated with lubricating oil emitted from HDDVs ranged from ∼0 to 17 (µg C/ng hopane) (16). Different grades of lubricating oil and, to a certain extent, different operating conditions result in different amounts of hopanes and steranes per unit of carbon in the lubricating oil particle-phase emissions (16). The uncertainty range associated with lubricating oil profiles from HDDVs was relatively large because the size distributions of light PAHs used to identify diesel fuel contributions to exhaust PM were similar to the size distributions of hopanes and steranes. Test HDDV-5 produced a lubricating oil profile that was consistent with LDGV tests within relatively tight uncertainty estimates. The lubricating oil profile developed for HDDV-5 will be used for all HDDVs in the current study. A complete description of the HDDV profiles is presented by Kleeman et al. (16).
Results Size-Resolved Source Contributions. Carbonaceous material accounts for the majority of the airborne particle mass in the PM0.1 size fraction, with reduced contributions at larger sizes due to the prevalence of secondary ammonium nitrate during the study period (see Figure 2 of ref (15)). Figure 1 of the current paper illustrates the size distribution of predicted source contributions to particulate OC concentrations at Sacramento, Modesto, and Bakersfield using the source profiles illustrated in Table 1 and the measured concentrations of molecular markers at the receptor locations (15). Wood burning and meat cooking are predicted to dominate the OC size distribution at all locations. Contributions to OC concentrations from lubricating oil, gasoline fuel, and diesel fuel are predicted to be relatively minor at the sampling sites during the present study. This reflects the fact that the sites were located in residential and light commercial areas. OC concentrations are generally not overpredicted in any size fraction with the exception of overpredictions to ultrafine OC concentrations at Modesto at night. Positive residual OC concentrations were detected during all daytime sampling periods and at Bakersfield during the nighttime. The residual OC may be produced by some primary source such as natural gas combustion that was not included in the calculation. Extensive oil and gas refining operations in the vicinity of Bakersfield may be one example of such an unknown source. It is also possible that secondary organic aerosol (SOA) formation occurs in the atmosphere, leading to the formation of additional OC in the particle-phase that cannot be attributed to a primary source. Molecular markers for SOA source apportionment (22) were not measured in the current study. Figure 2 shows the size distribution of predicted source contributions to particulate EC concentrations at Sacramento 274
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 43, NO. 2, 2009
between December 15 and 28, 2000 and Modesto, and Bakersfield between December 15, 2000 and January 7, 2001. Both predicted and measured EC concentrations are significantly lower than OC concentrations. EC overpredictions are observed during the evening hours at Sacramento but are otherwise small. Gasoline-fuel is the single largest source of predicted EC concentrations at all locations and times, with a peak between 0.18 and 0.32 µm particle diameter and a tail extending into the ultrafine mode. Predicted contributions to EC concentrations from diesel engines are observed during the day but generally not at night. This trend is consistent with the increased use of diesel vehicles during working hours and the increased transport of diesel exhaust to the sampling sites during the daytime when wind speeds and atmospheric mixing are greater than during the night. Lubricating oil contributions to EC are relatively small at all locations, with much of this material predicted to occur at sizes larger than 0.1 µm particle diameter. Predicted wood combustion and meat cooking contributions to EC concentrations are minor. Fine and Ultrafine Source Contributions. Table 2 and Figure 3 illustrate predicted source contributions to particulate carbon (organic + elemental) concentrations in the PM1.8 size fraction at Sacramento, Modesto, and Bakersfield during the current study. Uncertainty estimates reflect uncertainty in source profiles, uncertainty in ambient measurements, and uncertainty in colocated measurements of molecular markers thought to originate from the same source. All uncertainties are propogated through the calculation using standard techniques. For a sum or a difference (ie. y ) a + b - c) the absolute error in y ) σy ) (σa2 + σb2 + σc2)0.5. For a multiplication or a division (ie. y ) a × b/c) the relative error in y ) σy/y ) ((σa/a)2 + (σb/b)2 + (σc/c)2)0.5. Uncertainty for wood smoke and food cooking source contributions are dominated by uncertainty in the source profiles (see Table 1) while uncertainty for other sources is more evenly distributed between uncertainty for source profiles and ambient measurements. Total carbonaceous PM1.8 concentrations ranged from 7.5 to 27.3 µg m-3. Wood smoke accounted for ∼50% of PM1.8 carbon during the nighttime at Sacramento and Modesto and ∼20% of nighttime carbonaceous PM1.8 at Bakersfield. Gasoline fuel contributions to nighttime carbonaceous PM1.8 ranged from 27 to 41%. Diesel fuel made little contribution to nighttime PM1.8 concentrations, which is not surprising since the sampling sites were not located close to major highways and the majority of the diesel traffic on surface streets occurs during normal working hours during the day. Motor oil contributions to PM1.8 concentrations ranged from 4 to 8% during the nighttime hours. Predicted meat cooking contributions to PM1.8 concentrations were surprisingly large at Modesto and Bakersfield during the evening hours, ranging from 12 to 19%. This level of meat cooking is consistent with
FIGURE 1. Predicted source contributions to size-resolved particulate organic carbon (OC) at Sacramento, Modesto, and Bakersfield between December 15, 2000 and January 7, 2001. Residual concentrations of OC (positive or negative) are illustrated as unknown. the cholesterol concentrations measured in the airborne particles while simultaneously considering carbon mass conservation constraints. Nevertheless, the quantification of meat cooking contributions to airborne particle concentrations using cholesterol as a tracer should be considered preliminary, since it is possible that some other source of cholesterol is present in the atmosphere. Wood smoke contributions to PM1.8 concentrations are predicted to decrease during the daytime hours. This trend is consistent with increased temperatures during the day and the normal workday diurnal pattern. Predicted meat cooking contributions to PM1.8 are significantly enhanced during the day, ranging from 40 to 57%. Once again, this level of meat smoke contribution is consistent with the measured cholesterol concentrations in the airborne particle PM1.8 measurements, but it is higher than the level suggested by previous source apportionment studies for central California during winter stagnation events (10). The quantification of meat cooking contributions to airborne particle mass
should be considered preliminary at the present time. Predicted gasoline fuel contributions to PM1.8 concentrations ranged from 14 to 24%. Diesel fuel contributions to PM1.8 were predicted to be 13% at Sacramento during the day but were less than 1% at Modesto and Bakersfield. Lubricating oil (from either gasoline or diesel engines) was predicted to account for ∼6-7% of the PM1.8 at Modesto and Bakersfield but could not be detected at Sacramento during the day. Table 3 and Figure 4 illustrate predicted source contributions to ultrafine (PM0.1) total particulate carbon (organic + elemental) concentrations at Sacramento, Modesto, and Bakersfield. Total PM0.1 concentrations ranged from 0.38 µg m-3 at Modesto during the daytime to 1.0 µg m-3 at Modesto during the nighttime. Wood combustion was the single largest source of ultrafine particles at the Sacramento, Modesto, and Bakersfield during the evening hours. Wood smoke contributions to ultrafine (PM0.1) particle concentrations during the nighttime ranged from 54% at Bakersfield to 79% at Modesto. Gasoline-fuel accounted for approximately VOL. 43, NO. 2, 2009 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
275
FIGURE 2. Predicted source contributions to size-resolved particulate elemental carbon (EC) at Sacramento, Modesto, and Bakersfield between December 15, 2000 and January 7, 2001. Residual concentrations of EC (positive or negative) are illustrated as unknown. 19-24% of the ultrafine (PM0.1) particle mass during the nighttime sampling periods at all locations. Diesel fuel contributions to PM0.1 mass were predicted to be 1-3% during the night. Ultrafine particle concentrations during daytime sampling periods contained a greater amount of material from unknown sources. Wood smoke was still the single largest source of known ultrafine particle mass with contributions ranging from 3 to 42%. Estimated meat cooking contributions to ultrafine particle mass ranged from 11 to 37%. Given the surprisingly large predictions for meat cooking contributions to PM1.8, the meat cooking contribution to PM0.1 must also be viewed with caution. Diesel fuel contributions to PM0.1 mass were predicted to be 21% at Sacramento and 11% at Bakersfield during the day. Diesel fuel contributions to PM0.1 at Modesto were still predicted to be less than 1% during the day. Gasoline fuel was predicted to account for 2-9% of the PM0.1 mass, whereas motor oil was predicted to account for 276
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 43, NO. 2, 2009
3-5%. Unknown sources accounted for 58% of the PM0.1 mass at Modesto and 24% of the PM0.1 mass at Sacramento during the day.
Discussion The results of the current study illustrate the capability to carry out size-resolved source apportionment calculations that extend into the ultrafine size range. Most combustion sources emit particle mass distributions that peak between 0.1 and 0.3 µm aerodynamic diameter with tails in the ultrafine range (4, 16). The size distributions of primary source contributions identified at receptor sites in the current study are shifted to larger sizes and broadened. Primary emissions explain the majority of the OC mass, ruling out SOA condensation as a growth mechanism. Previous modeling studies have identified coagulation as a possible growth mechanism for primary emissions under the severely polluted conditions encountered in the SJV (23).
TABLE 2. Predicted Source Contributions to PM1.8 Carbon Concentrations at Sacramento, Modesto, and Bakersfield between December 15, 2000 and January 7, 2001a PM1.8 (ng m-3)
SAC day
SAC night
MOD day
diesel EC diesel OC gasoline EC gasoline OC lubricating Oil EC lubricating Oil OC wood smoke EC wood smoke OC meat cooking EC meat cooking OC unknown EC unknown OC total measured carbon
703 ( 365 274 ( 142 675 ( 154 390 ( 95 ND ND 12.1 ( 4 405 ( 121 43.3 ( 22 4330 ( 2165 0 ( 365 700 ( 2200 7530 ( 467
0 ( 1770 0 ( 690 7210 ( 1402 4160 ( 890 965 ( 1048 1330 ( 1072 410 ( 123 13700 ( 4104 134 ( 67 0 ( 0.004 0 ( 1770 0 ( 4394 23100 ( 311
0 ( 515 0 ( 201 1730 ( 191 999 ( 142 305 ( 346 389 ( 338 12.7 ( 4 423 ( 127 47.3 ( 24 4730 ( 2367 0 ( 515 2670 ( 2429 10900 ( 467
MOD night 0 ( 1192 0 ( 465 4910 ( 1022 2840 ( 642 492 ( 547 627 ( 526 416 ( 125 13900 ( 4163 234 ( 117 5400 ( 2701 0 ( 1192 0 ( 5058 27300( 311
BFK day
BFK night
0 ( 468 0 ( 183 1250 ( 164 722 ( 114 289 ( 288 396 ( 270 46.7 ( 14 1560 ( 467 44.5 ( 22 4450 ( 2227 0 ( 468 2490 ( 2325 11100 ( 467
0 ( 818 0 ( 319 4430 ( 650 2560 ( 440 441 ( 444 604 ( 420 132 ( 40 4390 ( 1318 26.9 ( 13 2690 ( 1345 0 ( 818 7240 ( 2017 21100 ( 311
a Uncertainty estimates represent the combined uncertainty of the ambient measurements and the profiles used to calculate source contributions. ND indicates all tracers were below detection for the indicated source.
FIGURE 3. Predicted source contributions to carbonaceous PM1.8 concentrations at Sacramento, Modesto, and Bakersfield between December 15, 2000 and January 7, 2001. VOL. 43, NO. 2, 2009 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
277
TABLE 3. Predicted Source Contributions to PM0.1 Carbon Concentrations at Sacramento, Modesto, And Bakersfield between December 15, 2000 and January 7, 2001a PM0.1 (ng m-3)
SAC day
SAC night
MOD day
MOD night
BFK day
BFK night
diesel EC diesel OC gasoline EC gasoline OC lubricating oil EC Lubricating oil OC wood smoke EC wood smoke OC meat cooking EC meat cooking OC unknown EC unknown OC total measured carbon
58.3 ( 110 22.8 ( 43 8.17 ( 8 4.71 ( 5 ND ND 3.11 ( 1 104 ( 31 0.877 ( 0.4 87.7 ( 44 0 ( 110 89.7 ( 130 379 ( 156
16.1 ( 161 6.27 ( 63 144 ( 144 83.1 ( 84 ND ND 20.5 ( 6 684 ( 205 2.63 ( 1 0 ( 0.0007 0 ( 161 0 ( 241 954 ( 99
0 ( 118 0 ( 46 60.3 ( 43 34.8 ( 25 5.36 ( 7 6.84 ( 7 0.33 ( 0.2 11 ( 3 0.421 ( 0.2 42.1 ( 21 0 ( 118 227 ( 124 387 ( 156
0 ( 188 0 ( 73 267 ( 174 154 ( 101 ND ND 54.3 ( 16 1810 ( 543 3.31 ( 2 0 ( 0.0004 0 ( 188 0 ( 562 995 ( 99
15.6 ( 110 6.08 ( 43 2.75 ( 3 1.59 ( 2 3.59 ( 4 4.92 ( 5 2.49 ( 1 83 ( 25 4.23 ( 2 72.3 ( 36 0 ( 110 0 ( 126 196 ( 156
11.5 ( 96 4.48 ( 37 65 ( 65 37.5 ( 38 4.65 ( 5 6.38 ( 5 7.87 ( 2 262 ( 79 3.76 ( 2 82.8 ( 41 0 ( 96 0 ( 125 486 ( 99
a Uncertainty estimates represent the combined uncertainty of the ambient measurements and the profiles used to calculate source contributions. ND indicates all tracers were below detection for the indicated source.
FIGURE 4. Predicted source contributions to carbonaceous PM0.1 concentrations at Sacramento, Modesto, and Bakersfield between December 15, 2000 and January 7, 2001. 278
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 43, NO. 2, 2009
Nighttime concentrations of particulate carbon are generally more than double the daytime concentrations because stagnant atmospheric conditions trap local emissions close to the ground at night (23). The difference between daytime and nighttime source contributions to PM0.1 and PM1.8 concentrations reflects the mix of sources around the sampling locations. Sites were generally located in light commercial (Modesto and Bakersfield) or residential (Sacramento) areas and so the relative contribution from transportation sources decreases at these sites during the evening hours. Identical sampling conducted adjacent to a major transportation corridor would likely reveal the opposite trend (see, for example, ref 24). The diurnal variation of source contributions suggests that inhalation exposure studies operating at a fixed location could examine the health effects of different airborne particle sources during the day vs night.
Acknowledgments This research was supported by the California Air Resources Board, Research Division under contract no. 01-306, the San Joaquin Valleywide Study Agency under contract no. 2000 05PM, and the United States Environmental Protection Agency under grant no. RD-83241401-0. The research described in the article has not been subject to the Environmental Protection Agency’s required peer and policy review and therefore does not necessarily reflect the views of the Agency and no official endorsement should be inferred. We thank Dan Chang and Jorn Herner for help with sample collection, Hector Maldonado for help with EMFAC analysis, and Nehzat Motallebi for help with project coordination.
Supporting Information Available A detailed description of the comparison between the simplified source apportionment algorithm and a full CMB solution is provided. This material is available free of charge via the Internet at http://pubs.acs.org.
Literature Cited (1) Pope, C. A. Review: Epidemiological basis for particulate air pollution health standards. Aerosol Sci. Technol. 2000, 32 (1), 4–14. (2) Oberdorster, G.; Ferin, J.; Gelein, R.; Soderholm, S. C.; Finkelstein, J. Role of the alveolar macrophage in lung injurysStudies with ultrafine particles. Environ. Health Perspect. 1992, 97, 193–199. (3) Li, N.; Sioutas, C.; Cho, A.; Schmitz, D.; Misra, C.; Sempf, J.; Wang, M. Y.; Oberley, T.; Froines, J.; Nel, A. Ultrafine particulate pollutants induce oxidative stress and mitochondrial damage. Environ. Health Perspect. 2003, 111 (4), 455–460. (4) Kleeman, M. J.; Robert, M. A.; Riddle, S. G.; Fine, P. M.; Hays, M. D.; Schauer, J. J.; Hannigan, M. P. Size distributions of trace organic compounds emitted from biomass combstion and meat charbroiling. Atmos. Environ. 2008, 42, 3059–3075. (5) Riddle, S.; Robert, M. A.; Jakober, C. A.; Hannigan, M. P.; Kleeman, M. J. Size distribution of trace organic species emitted from heavy-duty diesel vehicles. Environ. Sci. Technol. 2007, 41 (6), 1962–1969. (6) Zielinska, B.; Sagebiel, J. C.; Arnott, W. P.; Rogers, C. F.; Kelly, K. E.; Wagner, D. A.; Lighty, J. S.; Sarofim, A. F.; Palmer, G. Phase and size distribution of polycyclic aromatic hydrocarbons in diesel and gasoline vehicle emissions. Environ. Sci. Technol. 2004, 38, 2557–2567. (7) Hays, M. D.; Smith, N. D.; Kinsey, J.; Dong, Y. J.; Kariher, P. Polycyclic aromatic hydrocarbon size distributions in aerosols from appliances of residential wood combustion as determined by direct thermal desorption - GC/MS. J. Aerosol Sci. 2003, 34 (8), 1061–1084.
(8) Phuleria, H. C.; Geller, M. D.; Fine, P. M.; Sioutas, C. Size-resolved emissions of organic tracers from light-and heavy-duty vehicles measured in a California roadway tunnel. Environ. Sci. Technol. 2006, 40 (13), 4109–4118. (9) Riddle, S.; Robert, M. A.; Jakober, C. A.; Hannigan, M. P.; Kleeman, M. J. Size distribution of trace organic species emitted from light-duty gasoline vehicles. Environ. Sci. Technol. 2007, 41 (21), 7464–7471. (10) Schauer, J. J.; Cass, G. R. Source apportionment of wintertime gas-phase and particle-phase air pollutants using organic compounds as tracers. Environ. Sci. Technol. 2000, 34 (9), 1821– 1832. (11) Subramanian, R.; Donahue, N. M.; Bernardo-Bricker, A.; Rogge, W. F.; Robinson, A. L. Contribution of motor vehicle emissions to organic carbon and fine particle mass in Pittsburgh, Pennsylvania: Effects of varying source profiles and seasonal trends in ambient marker concentrations. Atmos. Environ. 2006, 40 (40), 8002–8019. (12) Cao, J. J.; Wu, F.; Chow, J. C.; Lee, S. C.; Li, Y.; Chen, S. W.; An, Z. S.; Fung, K. K.; Watson, J. G.; Zhu, C. S.; Liu, S. X. Characterization and source apportionment of atmospheric organic and elemental carbon during fall and winter of 2003 in Xi’an, China. Atmos. Chem. Phys. 2005, 5, 3127–3137. (13) Herner, J. D.; Aw, J.; Gao, O.; Chang, D. P.; Kleeman, M. J. Size and composition distribution of airborne particulate matter in northern California: I-particulate mass, carbon, and watersoluble ions. J. Air Waste Manage. Assoc. 2005, 55 (1), 30–51. (14) Chow, J. C.; Chen, L. W. A.; Watson, J. G.; Lowenthal, D. H.; Magliano, K. A.; Turkiewicz, K.; Lehrman, D. E. PM2.5 chemical composition and spatiotemporal variability during the California Regional PM10/PM2.5 Air Quality Study (CRPAQS). J. Geophys. Res., [Atmos.] 2006, 111(D10) (15) Kleeman, M. J.; Riddle, S. G.; Jakober, C. A. Size distribution of particle-phase molecular markers during a severe winter pollution episode. Environ. Sci. Technol. 2008, 42, 6469–6475. (16) Kleeman, M. J.; Riddle, S. G.; Robert, M. A.; Jakober, C. A. Lubricating oil and fuel contributions to particulate matter emissions from light-duty gasoline and heavy-duty diesel vehicles. Environ. Sci. Technol. 2008, 42 (1), 235–242. (17) Rogge, W. F.; Hildemann, L. M.; Mazurek, M. A.; Cass, G. R.; Simonelt, B. R. T. Sources of fine organic aerosol 0.1. Charbroilers and meat cooking operations. Environ. Sci. Technol. 1991, 25 (6), 1112–1125. (18) Schauer, J. J.; Kleeman, M. J.; Cass, G. R.; Simoneit, B. R. T. Measurement of emissions from air pollution sources 1. C1 through C29 organic compounds from meat charbroiling. Environ. Sci. Technol. 1999, 33 (10), 1566–1577. (19) Sheesley, R.; Schauer, J. J.; Bean, E.; Kenski, D. Trends in secondary organic aerosol at a remote site in Michigan’s upper peninsula. Environ. Sci. Technol. 2004, 38, 6491–6500. (20) Fine, P. M.; Chakrabarti, B.; Krudysz, M.; Schauer, J. J.; Sioutas, C. Diurnal variations of individual organic compound constituents of ultrafine and accumulation mode particulate matter in the Los Angeles Basin. Environ. Sci. Technol. 2004, 38, 1296– 1304. (21) Arey, J.; Zielinska, B.; Atkinson, R. The formation of nitro-PAH from the gas-phase reactions of fluoranthene and pyrene with the OH Radical in the presence of NOx. Atmos. Environ. 1986, 20 (12), 2339–2345. (22) Lewandowski, M.; Jaoui, M.; Offenberg, J. H.; Kleindienst, T. E.; Edney, E. O.; Sheesley, R. J.; Schauer, J. J. Primary and secondary contributions to ambient PM in the midwestern United States. Environ. Sci. Technol. 2008, 42 (9), 3303–3309. (23) Herner, J. D.; Ying, Q.; Aw, J.; Gao, O. H.; Chang, D. P.; Kleeman, M. J. Dominant mechanisms that shape the airborne particle size and composition distribution. Aerosol Sci. Technol. 2006, 40, 827–844. (24) Riddle, S. G.; Robert, M. A.; Jakober, C. A.; Hannigan, M.; Kleeman, M. J. Size resolved source apportionment of airborne particle mass in a roadside environment. Environ. Sci. Technol. 2008, 42, 6580–6586.
ES800400M
VOL. 43, NO. 2, 2009 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
279
