Environ. Sci. Technol. 2010, 44, 2315–2321
Size-Resolved Organic Speciation of Wintertime Aerosols in California’s Central Valley THOMAS M. CAHILL* Division of Mathematical and Natural Sciences, Arizona State University at the West Campus, P.O. Box 37100, Phoenix, Arizona 85069
Received September 26, 2009. Revised manuscript received February 15, 2010. Accepted February 20, 2010.
Size-resolved aerosol samples, including the entire ultrafine fraction, were simultaneously collected along a transect in California’s Central Valley during the winter of 2009. The samples were analyzed for PAHs, alkanes, organic acids, and sugars. The results showed that the organic constituents of aerosols did not follow the same pattern as PM10, thus indicating that simple PM measurements are not good indicators of trace toxic organic chemicals. Levoglucosan, a tracer of wood smoke, was the most abundant organic chemical detected, thus demonstrating the predominance of wood smoke in the valley. The size profile of levoglucosan showed a maximum in the 0.34-0.56 µm size mode, which is larger than published emission profiles. This suggests that wood smoke aerosols increased in size as they aged in the environment. Some chemicals, such as benzo[a]pyrene, had similar aerosol size profiles as levoglucosan and likely arose from the same source. Other chemicals, such as coronene and sugars, had very different size profiles, indicating that they have different sources. One unexpected result was the relatively large fraction of certain chemicals present in the ultrafine fraction, which highlights the importance of collecting the entire ultrafine fraction.
Introduction The southern part of the Central Valley of California is the largest contiguous area of the United States simultaneously in violation of federal air quality standards for ozone (summer) and PM2.5 fine mass (winter) despite minimal heavy industry and relatively low population density (60 people/ km2). In winter, particulate mass increases steadily from low levels in the northern valley to high levels and frequent PM2.5 violations in the southern valley. These air pollution violations have prompted a series of detailed studies in the southern Central Valley, also called the San Joaquin Valley, that characterized the aerosols and estimated their sources (1-6). In contrast, the northern valley has received little attention, since the particulate concentrations are generally in compliance. However, limited toxics monitoring by the California Air Resources Board (CARB) (7) demonstrates that some northern cities have concentrations of toxic organics, such as benzo[a]pyrene, that exceed those in the southern Central Valley. Despite the high concentrations of selected toxics, no detailed aerosol studies have been conducted in the Northern Central Valley. Studies reporting ambient size-resolved organic aerosol concentrations are still rare, but they are increasing in the * Author e-mail:
[email protected]. 10.1021/es902936v
2010 American Chemical Society
Published on Web 03/05/2010
literature (e.g., refs 3, 8). The advantages of size-resolved data are considerable; the size profiles can be used to assess which aerosols are most likely to be captured by the lung and are the most hazardous. Size profiles also identify which aerosols are responsible for visibility reduction, because certain size ranges scatter light. Unfortunately, size-resolved aerosol sampling requires more effort, since each sample is subdivided into several size fractions. Another challenge with size-resolved sampling is that the smaller subsamples have less mass, which makes the detection of trace chemicals more difficult. The objective of this research was to determine the organic constituents of size-resolved aerosol samples along a transect from north to south covering the entire Central Valley during high wintertime aerosol mass. This research builds upon previous research by conducting (1) sampling in the northern valley regions, which have not been previously studied in detail, and (2) size-resolved sampling that divides aerosols into nine particle size fractions, including all of the ultrafine fraction that is critical to potential health effects due to high lung capture efficiencies. The size-resolved organic data are particularly useful for both identifying the potential sources of the aerosols and assessing their impacts.
Materials and Methods Ambient particulate matter was simultaneously collected along a north-south transect within the Central Valley of California between January 5 and 22, 2009. The sampling was conducted at CARB rooftop air quality monitoring sites in Redding, Chico, Sacramento, Fresno, and Bakersfield (Figure 1). The meteorological conditions were largely dominated by stagnation with two to four rain events. The average temperature ranged from 9.0 to 11.3 °C and the relative humidity averaged 56-73%. Full meteorological data are presented in the Supporting Information. The aerosols were collected by two eight-stage DRUM impactor samplers (9) at each site. One sampler used baked aluminum foil substrate to collect particulate matter for organic analysis, while the second sampler used greased Mylar substrates to collect aerosols for elemental analysis. The sampler divides the particulate matter into eight size
FIGURE 1. A map showing the locations of the five sampling sites along a transect in the Central Valley of California, which is outlined with a dotted line. The southern part of the Central Valley is typically referred to as the San Joaquin Valley, while the northern area is called the Sacramento Valley. VOL. 44, NO. 7, 2010 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
2315
fractions, namely, 10-5.0, 5.0-2.5, 2.5-1.15, 1.15-0.75, 0.75-0.56, 0.56-0.34, 0.34-0.26, and 0.26-0.09 µm aerodynamic diameter. An after-filter (Whatman stretched PTFE 46.2 mm filter) was placed after the final impaction stage to collect particulates 70% average recovery) for all the samples (full data presented in the Supporting Information). The extraction and derivatization of the labeled organic acids, sugars, and cholesterol was effective for most internal standards except butanedioic acid-13C2 (37.0%), ribose-13C5 (63.3%), and glucose-13C6 (66.9%) for samples collected at sites other than Bakersfield. Of the nine Bakersfield samples, six showed a complete lack of derivatization of the internal standards and the other three greatly inhibited derivatization rates. The samples were processed in aerosol size groups rather than site groups, so the systematic derivatization failure appears to be associated with the samples rather than processing. Subsequent tests indicate that the poor derivatization was likely the result of high concentrations of sulfates 2316
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 44, NO. 7, 2010
FIGURE 2. Average ((SD) PM10 mass (A) in January 2009 along the sample transect as determined by the California Air Resources Board (7), where the organic samplers were located. Only sampling times where all sites reported data were used to generate this graph, and data during one rain event was excluded. The sum of different organic chemical classes (B) show a different pattern than PM10 mass, indicating that mass is not necessarily a good indication of trace organic chemicals. which are present in large quantities in the wintertime aerosols (1) (see Supporting Information section S2 for details). Two samples from Fresno also showed inhibited derivatization, but these data were still useable since the internal standards used for quantification adjusted for the lowered derivatization rates. The blank aluminum and filter substrates demonstrated that they were largely free of contamination, although hexadecanoic and octadecanoic acids showed erratic concentrations, which led to higher minimum detection limits. Mass Data. The CARB stations reported PM10 mass at all sites (Figure 2) (7). The PM10 mass clearly showed increasing concentrations from north (Redding) to south (Bakersfield) along the transect. However, the organic chemicals did not show this trend or the trend was much weaker (Figure 2, Tables 1-3). The alkane concentrations (Table 1) showed increasing concentration heading south, but the increases were not as large as for PM10. The PAH concentrations did not show any north-south trend, and the highest concentrations were in Chico near the north end of the transect (Table 2). The polar chemicals showed the highest concentrations in Chico followed by Fresno (Table 3). Overall, the organic compounds did not follow the same trend as PM10 mass, which was dominated by ammonium nitrate and ammonium sulfate (1, 2). This data demonstrates that PM10 is not necessarily a good predictor of organic chemicals in aerosols. Levoglucosan and Sugars. Levoglucosan, which is a tracer of biomass combustion (11-13), was the most abundant chemical detected at all fully characterized sites. The concentrations ranged from 213 to 497 ng/m3 (Table 3). These concentrations were lower than those measured during the State of California’s IMS95 air quality study in January 1996 by Schauer and Cass (2130 ng/m3 (6)) and Nolte et al. (4) (2980 ng/m3). Samples collected in the winter of 2000/2001 by Kleeman et al. (3) showed wintertime Sacramento daytime and nighttime concentrations to be 61 and 2050 ng/m3,
TABLE 1. Concentration of Particulate Alkanes (in ng/m3) along a North to South Transect in the Central Valley of Californiaa compound
Redding
Chico
Sacramento
Fresno
Bakersfield
C20 n-alkane C21 n-alkane C22 n-alkane C23 n-alkane C24 n-alkane C25 n-alkane C26 n-alkane C27 n-alkane C28 n-alkane C29 n-alkane C30 n-alkane C31 n-alkane C32 n-alkane C33 n-alkane C34 n-alkane C35 n-alkane C36 n-alkane sum of identified n-alkanes carbon preference indexb
ND 0.040 ND ND ND ND ND 0.540 0.183 2.56 0.147 0.635 ND ND ND 0.069 ND
ND 0.143 0.174 0.510 0.975 2.85 2.65 5.21 2.77 11.0 3.33 4.42 ND 0.255 0.179 0.414 0.119
ND 0.188 0.215 0.616 1.39 3.41 3.76 5.01 3.13 9.29 3.41 6.55 0.842 0.940 0.574 0.339 0.113
0.116 0.313 ND ND 2.23 4.48 6.52 7.86 6.62 10.1 5.29 5.34 1.44 0.897 1.06 1.01 0.302
0.092 0.592 1.47 3.07 4.28 4.16 4.66 4.54 3.65 7.92 3.45 5.57 0.664 1.26 0.959 0.509 0.437
4.2
35.6
41.3
55.7
49.9
10.3
2.4
2.0
1.3
1.4
a The values presented are the sum of the eight impactor stages and the after-filter. C37 to C47 n-alkanes were sporadically detected at low concentrations (see Supporting Information Table S7). These alkanes are included in the sum of identified alkane calculation. b The carbon preference index (CPI) is defined as ∑(odd carbon lengths)/∑(even carbon lengths) (24).
TABLE 2. Concentration of Particulate PAHs (in ng/m3) along a North to South Transect in the Central Valley of Californiaa compound
Redding
Chico
Sacramento
Fresno
Bakersfield
fluoranthene pyrene benzo[c]phenanthrene benz[a]anthracene chysene + triphenylene benzo[ghi]fluoranthene cyclopenta[cd]pyrene benzo[b+j+k]fluoranthene benzo[a]fluoranthene benzo[e]pyrene benzo[a]pyrene perylene indo[123,cd]pyrene benzo[ghi]perylene anthanthrene dibenz[a,h]anthracene picene + benzo[b]chrysene coronene sum of PAHs
0.0070 ND 0.0027 0.0264 0.0174 ND ND 0.336 0.0614 0.180 0.191 0.0585 0.223 0.268 0.0809 0.0261 0.0761 0.128 1.84
0.0380 0.0402 0.0796 0.465 0.552 0.107 0.283 2.67 0.712 1.13 1.77 0.367 1.59 1.87 1.59 0.163 0.411 0.663 15.7
0.0267 0.0419 0.0180 0.0651 0.0551 0.0218 0.0116 0.526 0.0970 0.276 0.310 0.0746 0.434 0.747 0.309 0.0220 0.0333 0.411 3.89
0.0439 0.0629 0.0097 0.0794 0.0989 0.0347 0.0406 0.974 0.204 0.446 0.606 0.127 0.773 1.17 0.548 0.0748 0.180 0.516 6.67
0.0296 0.0580 ND 0.0620 0.0471 0.0155 0.0159 0.424 0.0734 0.249 0.262 0.0738 0.369 0.619 0.253 0.0180 0.0198 0.356 3.26
a The values presented are the sum of the eight impactor stages and the after-filter. An additional seven PAHs with molar masses of 302 were detected at low concentrations and are presented in the Supporting Information.
respectively. Rinehart et al. (5) reported annual average levoglucosan concentrations for Sacramento and Fresno as 62.6 and 201.9 ng/m3, respectively. These annual averages are lower than the wintertime values, since the winter has higher particulate matter concentrations. Unlike the total aerosol mass, the concentrations of levoglucosan did not show a systematic increase from north to south. The highest concentrations were detected in Chico in the northern valley, which was an area not sampled in previous studies that focused on the Southern Central Valley. These unexpectedly high concentrations indicate that more attention needs to be focused on this region. The high levoglucosan concentrations demonstrate the large wood smoke contribution to winter time organic aerosols in the valley, which agrees with previous source apportionment studies (5, 6, 14, 15).
The size distribution of levoglucosan showed that the maximum concentrations were in the 0.34-0.56 µm size fraction (Figure 3), which agrees with a prior study in the Central Valley (3). However, levoglucosan was also present in the very fine to ultrafine size fractions, with the two northern-most sites having the greatest proportion of levoglucosan in the smaller size modes. The size profile of aerosols from source sampling has shown the maximum concentration of aerosols in the 0.1-0.2 µm size mode (16), which suggests that two northern sites may be a mixture of “fresh” emissions and regional background. Therefore, the observed size profiles of levoglucosan may provide insight into local and regional wood smoke contributions. In contrast to levoglucosan, the sugars were predominantly present in the coarse size fraction between 1.15 and 10 µm (Figure 3). These sugars are largely expected to arise VOL. 44, NO. 7, 2010 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
2317
TABLE 3. Concentration of Particulate Sugars, Levoglucosan and Organic Acids (in ng/m3)a compound levoglucosan fructose glucose sucrose hexadecanoic acid octadecanoic acid nonadecanoic acid eicosanoic acid docosanoic acid tetracosanoic acid hexacosanoic acid octacosanoic acid butanedioic acid nonanedioic acid 9-octadecenoic acid dehydroabietic acidc cholesterol sum of polar chemicals
Redding 213 5.42 15.8 4.12 5.95 ND 2.19 3.64 17.8 32.8 14.7 3.05 1.77 2.44 ND 18.6 ND 368
Chico
Sacramento
Fresno
Bakersfield
407 11.4 21.6 4.32
b
68.2 87.3 1.64 8.74 20.0 27.8 14.2 4.69
31.0 49.5 2.20 6.89 26.3 39.2 23.1 5.68
b
Miscellaneous Compounds 11.0 8.00 12.3 14.3 37.8 42.6 46.8 16.8 0.157 ND 1076 665
4.04 6.66 40.3 45.7 ND 794
b
Sugars and Related Compounds 497 255 7.12 10.1 16.4 19.1 2.38 7.06 Alkanoic Acids 17.5 30.1 7.24 26.4 74.7 112 75.8 13.4
b b b
b b b b b b b
b b b b
a The values presented are the sum of the eight impactor stages and the after-filter. Another 26 analytes are presented in the Supporting Information. These analytes are included in the sum of polar chemical calculation. b No concentrations are reported due to poor derivatization. c Quantification based on a self-synthesized standard using methods in ref 31.
FIGURE 3. Chemical concentration as a function of particulate. The smallest size bin represents the material collected on the after-filter. Polar chemicals (levoglucosan, glucose, and tetracosanoic acid) are not reported for Bakersfield due to poor derivatization. from primary biological materials (plant detritus, spores, etc.), but biomass combustion component (17) could generate sugars in the fine aerosol size fractions. Although the total aerosol loading of sugars was relatively low (3-7% of identified organic mass), they represented the largest fraction of identified organic material in the coarse size mode, where the concentrations of other organic chemicals were low. PAHs. The sum of quantified PAHs were the lowest in Redding (1.84 ng/m3) and the highest at Chico (15.7 ng/m3) (Figure 2, Table 2). These concentrations are generally lower than previously reported wintertime PAH concentrations 2318
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 44, NO. 7, 2010
(excluding methyl PAHs) in Fresno by Schauer and Cass (6) (35.2 ng/m3) and in Sacramento day and night samples (1.74 and 35.16 ng/m3 respectively) by Kleeman et al. (3). The PAH concentrations did not show the same pattern as PM10 mass, since the maximum concentrations were in Chico in the northern Central Valley. Air toxics monitoring by CARB has shown both similar patterns and concentrations of benzo[a]pyrene at Chico (1.99 ng/m3), Fresno (1.40 ng/m3), Bakersfield (0.48 ng/m3), and Sacramento/Roseville (0.45 ng/ m3) (ref 7, January data from 2000 to 2005, which are the most recent years available). Therefore, it appears that this
geographic pattern is consistent between years. The high concentrations of PAHs at Chico, which were missed in prior detailed aerosol studies, highlight the importance of future studies conducting sampling in the northern valley. The aerosol size profiles of PAHs showed that not all the PAHs have the same size distribution. Benzo[a]pyrene had the highest concentrations in the 0.34-0.56 µm size mode, which matches the peak in the wood smoke profile (Figure 3), and the size profile correlated well (R2 ) 0.83) with levoglucosan. These data strongly suggest that benzo[a]pyrene was largely a result of wood smoke, which is supported by source experiments (18-20). In contrast, coronene showed the highest concentrations in the ultrafine (
