Daily Trends and Source Apportionment of Ultrafine Particulate

Figure 3 represents the factor profiles obtained though PMF modeling using 352 of 374 .... Truck Rule set forth by the State of California in 2006 and...
0 downloads 0 Views 4MB Size
Download PDF
Article pubs.acs.org/est

Daily Trends and Source Apportionment of Ultrafine Particulate Mass (PM0.1) over an Annual Cycle in a Typical California City Toshihiro Kuwayama, Chris R. Ruehl, and Michael J. Kleeman* Department of Civil and Environmental Engineering, University of California at Davis, One Shields Avenue, Davis, California 95616, United States S Supporting Information *

ABSTRACT: Toxicology studies indicate that inhalation of ultrafine particles (Dp < 0.1 μm) causes adverse health effects, presumably due to their large surface area-to-volume ratio that can drive heterogeneous reactions. Epidemiological associations between ultrafine particles and health effects, however, have been difficult to identify due to the lack of appropriate long-term monitoring and exposure data. The majority of the existing ultrafine particle epidemiology studies are based on exposure to particle number, although an independent analysis suggests that ultrafine particle mass (PM0.1) correlates better with particle surface area. More information is needed to characterize PM0.1 exposure to fully evaluate the health effects of ultrafine particles using epidemiology. The present study summarizes 1 year of daily PM0.1 chemistry and source apportionment at Sacramento, CA, USA. Positive matrix factorization (PMF) was used to resolve PM0.1 source contributions from old-technology diesel engines, residential wood burning, rail, regional traffic, and brake wear/ road dust. Diesel PM0.1 and total PM0.1 concentrations were reduced by 97 and 26%, respectively, as a result of the adoption of cleaner diesel technology. The strong linear correlation between PM0.1 and particle surface area in central California suggests that the adoption of clean diesel engines reduced particle surface area by similar amounts. PM0.1 sulfate reduction occurred as a result of reduced primary particle surface area available for sulfate condensation. The current study demonstrates the capability of measuring PM0.1 source contributions over a 12 month period and identifies the extended benefits of emissions reduction efforts for diesel engines on ambient concentrations of primary and secondary PM0.1.

■

INTRODUCTION Recent toxicological studies have linked ultrafine particle exposure to adverse health effects such as respiratory disease and cardiopulmonary morbidity. 1−4 Despite the robust toxicology results, the majority of epidemiological studies carried out to date have not been able to identify strong independent health effects for ultrafine particles.5−7 The original hypothesis about potential health effects of ultrafine particles centered on the high surface area-to-volume ratio that can provide numerous sites for heterogeneous reactions.8 Most epidemiological studies have used commercially available instruments to measure particle number concentration as a surrogate for particle surface area. Particle number concentrations are dominated by particles with aerodynamic diameter ≪0.1 μm with strong fluctuations in response to dynamic nucleation conditions that exist continuously downwind of freeways and occasionally over broad regions.9,10 In contrast, particle surface area is typically associated with particles that have aerodynamic diameter around 0.1 μm, at which behavior is much more stable. Particle number and surface area are potentially dominated by different sources and atmospheric © 2013 American Chemical Society

processes. Other metrics for particle surface area besides particle number may be necessary to relate atmospheric ultrafine particle toxicity and epidemiology. PM0.1 mass is one possible candidate for a new ultrafine particle metric that is closely aligned with particle surface area.11 PM0.1 is dominated by particles with aerodynamic diameter between 0.05 and 0.1 μm, which is where most of the particle surface area resides under typical ambient conditions. The majority of the PM0.1 in urban areas is composed of the tail of the combustion mode particles that dips below the 0.1 μm diameter threshold.12−14 PM0.1 mass concentrations (and particle surface area) do not fluctuate as strongly as particle numbers in response to atmospheric nucleation events.15 Linear regression analysis of PM0.1 mass-to-surface area and particle number-to-surface area in both urban and rural regions in central California during both summer and winter seasons Received: Revised: Accepted: Published: 13957

July 21, 2013 November 15, 2013 November 18, 2013 November 18, 2013 dx.doi.org/10.1021/es403235c | Environ. Sci. Technol. 2013, 47, 13957−13966

Environmental Science & Technology

Article

Figure 1. Map of downtown Sacramento, CA, USA, with wind rose, sampling site location, and major transportation paths including highways and railways.

L min−1. Each MOUDI was equipped with an upstream AIHLdesign cyclone to remove particles with diameter >1.8 μm from the sample stream for their tendency to bounce off impaction stages.18 Particles with aerodynamic diameter 95% (see Table S2 in the Supporting Information for R2 and MDL values). The wind rose shown in Figure 1 represents the frequency of the observed wind directions in 30° wind sectors at Sacramento International Airport (approximately 16 km from sample location) between October 2009 and November 2010. The predominant wind direction in Sacramento was from the southwest, where major highways are located (I-5 and I-80). The wind pattern indicates that sources located directly to the west and toward the northeast of the sampling site did not contribute strongly to the PM0.1 concentrations on most days of the year unless they were regionally mixed prior to capture.

shows that PM0.1 mass is strongly correlated to surface area with R2 of ∼0.97, whereas particle number correlates to surface area with R2 of ∼0.70 (refer to Table S1 in the Supporting Information).16 PM0.1 mass, however, has not been the first choice in previous epidemiological studies due to the additional cost and effort required to make accurate PM0.1 measurements. The few data sets with PM0.1 measurements that do exist involve point measurements at a few sites, and fewer studies involve chemical composition information.14,17 These data sets provide a valuable starting point for PM0.1 analysis, but they cannot support source apportionment calculations that may reveal important exposure trends. The objective of this study is to summarize 24 h average PM0.1 concentrations, chemical composition, and source contributions in Sacramento, CA, USA, over a 1 year period during which large changes occurred in emissions from heavyduty diesel vehicles. Daily chemical speciation of PM0.1 over the full year was analyzed using positive matrix factorization (PMF) to identify major source contributions. The results were compared to complementary studies carried out at the Port of Oakland (California’s third busiest port for goods movement) to confirm the regional extent of the trends. The PM0.1 measurements and source apportionment information provide a useful data set for future model predictions of PM0.1 exposure. The implications for public health based on the identified trends are discussed.

■

EXPERIMENTAL METHODS

PM0.1 samples were collected in the center of Sacramento, CA, USA, on the roof of the California Air Resources Board (CARB) facility (38°34′06.5″ N, 121°29′35.5″ W) at a height of 10 m above ground. The site is approximately 500 m north from the closest highway center (I-80) in proximity to commercial and residential neighborhoods as shown in Figure 1. Measurements were made on 374 days with 24 h resolution between noon and noon starting October 26, 2009, until November 3, 2010. PM0.1 samples were collected on the final stage (0.056 μm < Dp < 0.1 μm) of two nonrotating MicroOrifice Uniform Deposit Impactors (MOUDIs) operating at 30

■

RESULTS Figure 2 summarizes monthly trends during the year of PM0.1 measurements at Sacramento for several major chemical species 13958

dx.doi.org/10.1021/es403235c | Environ. Sci. Technol. 2013, 47, 13957−13966

Environmental Science & Technology

Article

concentrations of a few nanograms per cubic meter that peaked in the spring months when wind speeds were highest. Figure 3 represents the factor profiles obtained though PMF modeling using 352 of 374 available data points that passed quality control checks with 26 chemical species that account for the majority of the PM0.1 mass. PMF is a multivariate receptor model used to determine factor profiles and contributions that can be compared to known source profiles. The details of the model are described by Paatero and Tapper24 and Paatero.25 Guidelines for PMF use are available through the U.S. Environmental Protection Agency (EPA).26 In this present study, signal-to-noise (S/N) ratios were assessed to categorize individual species into proper variable groups described by Paatero and Hopke.27 Categorization of species with S/N < 0.2 as “bad” and 0.2 < S/N < 2 as “weak” reduced the weight of the species during factor identification by increasing their uncertainty or removing them from further analyses. Residual analysis and bootstrap runs were also performed to determine the model fit and stability of the solution. Matrix rotation was performed with different FPEAK values to assess the range of optimal solutions.26,28 Sensitivity tests were performed for variable model uncertainty ranges to quantify changes in the object function Q (sum of the residual/uncertainty for all species). Also included in the analysis were three gas phase species measured at the CARB Sacramento−Del Paso Manor monitoring site approximately 11 km from downtown Sacramento. Water-soluble ions such as ammonium and nitrate that contribute to PM0.1 fractions were not analyzed. Multiple studies in California during both winter and summer seasons have determined that 0.1 μm).12,16 Condensation of semi-volatile organic vapors and secondary organic aerosol formation at reduced temperatures are likely not dominant factors in the seasonal trends. PM0.1 sulfur concentrations were found to be highest in the late winter months, a trend exactly opposite from what is observed in the eastern United States, which experiences sulfate peaks in the summer months. A study of PM2.5 in cities across California’s San Joaquin Valley similarly measured higher sulfate concentrations during the winter months.23 The higher relative humidity in the California atmosphere during the winter months promotes gas partitioning to the aqueous phase, in which sulfate formation is rapid. Lower relative humidity in the summer months shuts down the aqueous formation pathway. Trace metals such as Mg and Fe that are known to be present in windblown dust had very low

where mΔθ is the number of events from wind sector Δθ that are in the upper 50 percentile of daily mass contributions and nΔθ is the total number of events from wind sector Δθ. Analysis of the conditional probability value for all wind sectors under each factor can be used to determine the most probable direction for source(s).32,33 Due to the lack of data for factors with intermittent signals (rail and brake wear/road dust), concentrations that exceeded one standard deviation of the daily mean concentrations were considered as meeting the criterion. This approach effectively removes the nonsignificant signal in the intermittent factors while preserving apparent source contributions during peak events. Peaks in the source contribution time series indicate conditions when both meteorology and intermittent emissions align for sources such as rail or when wind speeds reach a critical threshold for dust resuspension for sources such as brake wear/road dust. Sampling days with wind speed 0.04 μg m−3. The sporadic nature of the signal is caused by the relatively infrequent train activity (compared to on-road traffic) combined with a predominant wind direction that did not frequently originate from the northeast (direction of rail lines). PM0.1 concentrations associated with on-road mobile sources were consistently above detection limits at the Sacramento monitoring site due to the proximity to major freeways with significant traffic volume. The majority of the PM0.1 mass contribution from traffic sources was composed of EC and OC. PM0.1 concentrations and composition from traffic sources were relatively constant throughout the annual cycle, with slightly higher concentrations during the winter months when atmospheric mixing was reduced due to reduced boundary layer heights and when cold-start emissions contributed to higher PM loads. The PM0.1 brake wear and road dust signal at the Sacramento monitoring site experienced peak concentrations >0.1 μg m−3 during high wind events but were otherwise detected at an annual average concentration of 0.006 μg m−3. The brake wear and road dust signal was highest when wind speeds were approximately 22 mph and wind direction was from the southwest or west-northwest (both in the direction of major highways). The majority of the PM0.1 mass from brake wear and road dust can be attributed to trace metals found in soil near on-road traffic.

Factor 5 was identified as on-road brake wear and road dust characterized by Fe, Cu, Zn, and Pb. Zn is a common additive in motor oil and may have been expected to appear in the regional traffic signal as well as the road dust signal. Recent tests indicate that motor oil has intermediate volatility; therefore, the PM0.1 OC associated with motor oil may have partially largely evaporated prior to reaching the sampling site in the current study.47 Vehicle fuels in the United States were reformulated to remove Pb in the 1970s, but road dust around major highways still contains significant amounts of Pb.48 These metals are widely used in brake pads and tire weights that contribute to road dust.49 The profile also contains trace contributions from OC and EC, which may have resulted from residual overlap between factors.50 Figure 4 illustrates the daily PM0.1 time series between November 2009 and November 2010 decomposed into major sources and chemical components at downtown Sacramento (see Supporting Information Figures S1−S4 for detailed daily PM0.1 time series of major chemical components). Factor 1 accounted for the majority of the PM0.1 sulfur and 12% of the PM0.1 EC + OC concentrations from November 2009 until mid-April 2010, after which the contributions from this source decreased dramatically. This time frame is consistent with the reduction in EC concentration observed at the Port of Oakland after the Phase 2 implementation of the CTMP.34 The Port of Oakland implemented the CTMP to comply with the Emissions Reduction Plan for Ports and Goods Movement in California and the Drayage Truck Rule set forth by the State of California in 2006 and 2007, respectively. Phase 2 of the CTMP prohibited Port entrance to drayage trucks with engine models predating 1994 and ensured that engine models dating between 1994 and 2003 had CARB-verified level 3 diesel particle filters (DPF) properly installed. These regulations were designed to reduce the emissions related to goods movement to 2001 levels by the year 2010 and to reduce adverse health effects related to goods movement by 85% by the year 2020. The DPFs employed by the drayage trucks are able to remove particles in the size range 0.02 μm < Dp < 3 μm at efficiencies >99%.51,52 Removal of these primary particles reduces the surface area available for condensation as the exhaust cools to ambient temperature. The reduction in ambient PM0.1 sulfur during mid-April 2010 (Figure 4) is likely caused by the removal of surface area from the primary emissions, leading to the condensation/coagulation of sulfuric acid on a smaller number of primary particles that grew out of the PM0.1 size range or the condensation/coagulation of sulfuric acid directly onto ambient particles larger than PM0.1. The comparison of the sulfur concentrations in the PM2.5 and PM0.1 size fractions (Figure S5 in the Supporting Information) supports the hypothesis that the sulfur mass is approximately constant even though PM0.1 concentrations decrease strongly. These findings suggest that after-treatment devices may control secondary PM0.1 that forms from atmospheric processes in addition to primary PM0.1 emitted directly from the tailpipe by removing surface area in the PM0.1 size range. A large reduction in ambient SO2 concentration was observed at the same time that ambient PM0.1 concentrations decreased (Figure S6 in the Supporting Information). Most after-treatment devices pre-installed in new diesel trucks or installed aftermarket in older trucks were designed to remove +85% of primary PM emissions by using diesel oxidation catalysts (DOCs) and DPFs in tandem or by using catalystcoated DPFs. Laboratory tests have shown that these devices 13962

dx.doi.org/10.1021/es403235c | Environ. Sci. Technol. 2013, 47, 13957−13966

Environmental Science & Technology

Article

Figure 5. Factor contributions to daily average PM0.1 mass concentration for pre- and post-drayage truck regulation implementation at Sacramento, CA, USA. PM0.1 mass concentration decomposed into major mass contributing species EC, OC, and S. “Other” chemical components include Na, Mg, P, Cl, K, Ti, V, Cr, Mn, Fe, Ni, Cu, Zn, As, Br, Sr, Mo, Ag, Cd, Sn, Sb, and Pb.

ratio than what has been measured in direct source emissions measurements of residential wood combustion.41,55 The differences between the composition of the PMF factor profile and the measured composition of wood smoke particles may indicate evaporation of semi-volatile OC after dilution and aging of emissions in the atmosphere. A long-term (2000− 2008) regional model calculation for PM0.1 source contributions in the central San Joaquin Valley, California, predicts that diesel engines account for 4% of total PM0.1 OC and residential wood burning accounts for 46% total PM0.1 OC,56 which are very close to the measurements in the current study. The differences between the model predictions and the observations may be attributed, but not limited to, differences between spatial domains, emission profiles, and meteorology. Regional traffic accounted for 27% of total PM0.1 measured at Sacramento prior to implementation of the goods movement measures and 81% of PM0.1 after implementation of the measures. The absolute concentration contributions (0.11 and 0.14 μg m−3, respectively) show that PM0.1 mass from regional traffic did not fluctuate strongly. The regional traffic EC/TC ratio before (0.339) and after (0.443) the regulation implementation at Sacramento suggests slight changes in the local on-road fleet composition. The old diesel trucks associated with Factor 1 account for a small number of total trucks on local highways that contribute to a disproportionately large amount of emissions from diesel truck traffic prior to April

Figure 5 compares PM0.1 concentrations before and after April 1, 2010, which generally corresponds to the date for adoption of additional goods movement controls. Old diesel engines accounted for approximately 27% of total PM0.1 prior to adoption of new controls and approximately 2% after implementation. PM0.1 EC from old diesel sources was reduced by 95% (from 0.012 to 0.00064 μg m−3), PM0.1 OC was reduced by 96% (from 0.027 to 0.00095 μg m−3), PM0.1 S was reduced by 96% (from 0.064 to 0.0025 μg m−3), and PM0.1 trace metals were reduced by 97% (from 0.0092 to 0.00028 μg m−3). As discussed previously, some of this material may have condensed on particles with Dp > 0.1 μm, where it still contributed to PM2.5 concentrations. PM0.1 from residential wood smoke accounted for approximately 44% of total PM0.1 at the Sacramento monitoring site prior to the implementation of additional goods movement controls and 11% of total PM0.1 after additional controls were implemented. This change is associated with different winter months sampled (November−February vs November). If a full annual cycle was analyzed after implementation of the goods movement controls, wood smoke would almost certainly account for an increased fraction of the PM0.1 mass in the post-control time period. Wood burning contributed 49% of the annual average PM0.1 OC measured at Sacramento. Approximately 6% of the total carbon in the PM0.1 wood smoke signal was composed of EC, which is a higher EC/OC 13963

dx.doi.org/10.1021/es403235c | Environ. Sci. Technol. 2013, 47, 13957−13966

Environmental Science & Technology

Article

1, 2010. This results from the fact that each old diesel truck emits >100 times as much PM as a new or retrofitted diesel truck (DPFs remove 99% of the diesel PM exhaust51,52). Replacing the old diesel trucks removes a small number of highemitting vehicles and adds the same small number of lowemitting vehicles to the regional traffic factor. This change does not strongly affect the chemical profile of the regional traffic signal because this factor already includes a large number of clean diesel engines on local highways. The increase in PM0.1 concentrations after the implementation period (0.03 μg m−3) is therefore the maximum contribution that can be attributed to the new/retrofitted heavy-duty diesel engine fleet on major highways. Differences in boundary layer heights, activity patterns, and meteorological conditions should be considered as uncertainty for further analyses in understanding the fleet turnover. A recent study suggests new diesel engines, and diesel engines with DPF retrofits contribute to a negligible fraction of ultrafine PM emissions when compared to old-technology diesel engines.57 The ultrafine PM contribution from diesel engines may not have been removed completely, but the regulations implementations led to a significant improvement in the air quality as shown in the Factor 1 contribution reduction.

controls. Emissions of nucleation mode particle number may still increase due to the adoption of diesel emissions controls, but particle surface area is reduced in the real atmosphere. SO2 concentrations at Sacramento−Del Paso Manor decreased significantly when emissions controls were adopted for diesel engines. SO2 concentrations “rebounded” to historical levels within 15 months, presumably because the adsorption surfaces within the emissions control devices became saturated, which allowed gas-phase SO2 to once again pass through. It is expected that ambient PM0.1 sulfur/sulfate concentrations did not “rebound” like the SO2 concentrations because the emissions control technology continued to remove primary particle surface area. Additional ambient measurements should be made at Sacramento to verify that PM0.1 concentrations from diesel engines were permanently reduced due to the adoption of new emissions control technology.

DISCUSSION The Emission Reduction Plan for Ports and Goods Movement in California motivated the Port of Oakland to implement the CTMP to improve air quality and public health. The results of the current study indicate that the CTMP reduced PM0.1 concentrations associated with old-technology on-road diesel engines by 97% (0.011 μg m−3 less EC, 0.026 μg m−3 less OC, 0.062 μg m−3 less S, and 0.0089 μg m−3 less other trace metals) in Sacramento. Previous work has shown that a 75 ± 15% reduction in PM1.8 mass associated with old-technology diesel engines was observed at the Port of Oakland and a 49 ± 15% reduction in PM2.5 EC was measured just outside the Port of Oakland34,58 during the same time frame. The linear correlation between PM0.1 mass and particle surface area made in central California suggests that the old diesel engines contributed 26.9% of the total PM0.1 surface area on average in downtown Sacramento before the regulation implementation and contributed 3.5% of the surface area after the regulation implementation. The cleaner diesel engines contribute 94.9% less PM0.1 surface area than the old diesel engines. These results indicate that the benefits of emissions controls for goods movement are realized both within the immediate proximity of the port and adjacent to major highways leading away from the port. This regional effect may help explain PM0.1 concentration patterns and associated health effects in Sacramento and in other locations that adopt goods movement emissions control programs in the future. Several laboratory studies have suggested that the adoption of diesel after-treatment devices with oxidation catalysts could increase ultrafine sulfur emissions in the nucleation mode.54,59 The reduction in PM0.1 sulfur observed in the current study shows that the opposite effect occurred in the real atmosphere. The control technology applied to diesel trucks reduced primary PM emissions and the primary particle surface area. This in turn reduced secondary PM0.1 concentrations possibly because the secondary material condensed in larger size fractions. The implications include, but are not limited to, reduced exposure to condensed sulfates and semi-volatile organics (not studied) due to the adoption of emissions

Corresponding Author

■

ASSOCIATED CONTENT

S Supporting Information *

Additional experimental details. This material is available free of charge via the Internet at http://pubs.acs.org.

■

■

AUTHOR INFORMATION

*(M.J.K.) Phone: +1 (530) 752-8386. Fax: +1 (530) 752-7872. E-mail: [email protected]. Notes

The authors declare no competing financial interest.

■

ACKNOWLEDGMENTS This research was funded by the U.S. Environmental Protection Agency under Contract R832414-010 to the University of California, Davis. Although the research described in the paper has been funded by the U.S. Environmental Protection Agency, it has not been subject to the Agency’s required peer and policy review and therefore does not necessarily reflect the reviews of the Agency, and no official endorsement should be inferred. We thank Jack Romans from the California Air Resources Board for moderating the research project and former University of California at Davis undergraduates Benjamin Bow, Andrew Burton, Hilary Mann, and Elizabeth (Flores) Quilici for their field and laboratory assistance.

■

REFERENCES

(1) Donaldson, K.; et al. Ultrafine particles. Occup. Environ. Med. 2001, 58 (3), 211−216. (2) Donaldson, K.; et al. The pulmonary toxicology of ultrafine particles. J. Aerosol Med. 2002, 15 (2), 213−220. (3) Oberdorster, G. Toxicology of ultrafine particles: in vivo studies. Philos. Trans. R. Soc. London Ser. A−Math. Phys. Eng. Sci. 2000, 358 (1775), 2719−2739. (4) Kreyling, W. G.; Semmler, M.; Moller, W. Dosimetry and toxicology of ultrafine particles. J. Aerosol Med. 2004, 17 (2), 140−152. (5) Brown, D. M.; et al. Increased inflammation and intracellular calcium caused by ultrafine carbon black is independent of transition metals or other soluble components. Occup. Environ. Med. 2000, 57 (10), 685−691. (6) Tran, C. L.; et al. Inhalation of poorly soluble particles. II. Influence of particle surface area on inflammation and clearance. Inhalation Toxicol. 2000, 12 (12), 1113−1126. (7) Renwick, L. C.; et al. Increased inflammation and altered macrophage chemotactic responses caused by two ultrafine particle types. Occup. Environ. Med. 2004, 61 (5), 442−447. 13964

dx.doi.org/10.1021/es403235c | Environ. Sci. Technol. 2013, 47, 13957−13966

Environmental Science & Technology

Article

(8) Seaton, A.; et al. Particulate air-pollution and acute health-effects. Lancet 1995, 345 (8943), 176−178. (9) Zhang, K. M.; et al. Evolution of particle number distribution near roadways. Part II: the ‘road-to-ambient’ process. Atmos. Environ. 2004, 38 (38), 6655−6665. (10) Kulmala, M.; et al. Formation and growth rates of ultrafine atmospheric particles: a review of observations. J. Aerosol Sci. 2004, 35 (2), 143−176. (11) Warheit, D. B.; et al. Pulmonary instillation studies with nanoscale TiO2 rods and dots in rats: toxicity is not dependent upon particle size and surface area. Toxicol. Sci. 2006, 91 (1), 227−236. (12) Kleeman, M. J.; et al. Source apportionment of fine (PM1.8) and ultrafine (PM0.1) airborne particulate matter during a severe winter pollution episode. Environ. Sci. Technol. 2009, 43 (2), 272−279. (13) Zhu, Y. F.; et al. Study of ultrafine particles near a major highway with heavy-duty diesel traffic. Atmos. Environ. 2002, 36 (27), 4323−4335. (14) Daher, N.; et al. Seasonal and spatial variability in chemical composition and mass closure of ambient ultrafine particles in the megacity of Los Angeles. Environ. Sci.−Processes Impacts 2013, 15 (1), 283−295. (15) Kittelson, D. B. Engines and nanoparticles: a review. J. Aerosol Sci. 1998, 29 (5−6), 575−588. (16) Ham, W. A.; Kleeman, M. J. Size-resolved source apportionment of carbonaceous particulate matter in urban and rural sites in central California. Atmos. Environ. 2011, 45 (24), 3988−3995. (17) Wehner, B.; Wiedensohler, A. Long term measurements of submicrometer urban aerosols: statistical analysis for correlations with meteorological conditions and trace gases. Atmos. Chem. Phys. 2003, 3, 867−879. (18) John, W.; Reischl, G. A cyclone for size-selective sampling of ambient air. J. Air Pollut. Control Assoc. 1980, 30 (8), 872−876. (19) Herner, J. D.; Green, P. G.; Kleeman, M. J. Measuring the trace elemental composition of size-resolved airborne particles. Environ. Sci. Technol. 2006, 40 (6), 1925−1933. (20) Birch, M. E.; Cary, R. A. Elemental carbon-based method for monitoring occupational exposures to particulate diesel exhaust. Aerosol Sci. Technol. 1996, 25 (3), 221−241. (21) NIOSH. NIOSH Manual of Analytical Methods, 4th ed.; National Institute for Occupational Safety and Health: Cincinnati, OH, USA, 1996. (22) Ham, W. A.; et al. Size distribution of health-relevant trace elements in airborne particulate matter during a severe winter stagnation event: implications for epidemiology and inhalation exposure studies. Aerosol Sci. Technol. 2010, 44 (9), 753−765. (23) Chow, J. C.; et al. PM10 and PM2.5 compositions in California’s San Joaquin Valley. Aerosol Sci. Technol. 1993, 18 (2), 105−128. (24) Paatero, P.; Tapper, U. Positive matrix factorization − a nonnegative factor model with optimal utilization of error-estimates of data values. Environmetrics 1994, 5 (2), 111−126. (25) Paatero, P. Least squares formulation of robust non-negative factor analysis. Chemom. Intell. Lab. Syst. 1997, 37 (1), 23−35. (26) EPA. EPA Positive Matrix Factorization (PMF) 3.0 Fundamentals & User Guide; U.S. Environmental Protection Agency: Washington, DC, 2008. (27) Paatero, P.; Hopke, P. K. Discarding or downweighting highnoise variables in factor analytic models. Anal. Chim. Acta 2003, 490 (1−2), 277−289. (28) Song, X. H.; Polissar, A. V.; Hopke, P. K. Sources of fine particle composition in the northeastern US. Atmos. Environ. 2001, 35 (31), 5277−5286. (29) Herner, J. D.; et al. Dominant mechanisms that shape the airborne particle size and composition distribution in central California. Aerosol Sci. Technol. 2006, 40 (10), 827−844. (30) 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 (17), 6469−6475.

(31) Chow, J. C.; et al. Size-resolved aerosol chemical concentrations at rural and urban sites in Central California, USA. Atmos. Res. 2008, 90 (2−4), 243−252. (32) Ashbaugh, L. L.; Malm, W. C.; Sadeh, W. Z. A residence time probability analysis of sulfur concentrations at Grand Canyon National Park. Atmos. Environ. 1985, 19 (8), 1263−1270. (33) Kim, E.; et al. Source identification of PM2.5 in an arid northwest US city by positive matrix factorization. Atmos. Res. 2003, 66 (4), 291−305. (34) Kuwayama, T.; et al. Particulate matter emissions reductions due to adoption of clean diesel technology at a major shipping port. Aerosol Sci. Technol. 2013, 47 (1), 29−36. (35) Robert, M. A. Size and composition quantification of particulate matter emissions from motor vehicles. In Environmental Engineering; University of California, Davis: Ann Arbor, MI, USA, 2008; p 197. (36) Schauer, J. J.; et al. ACE-Asia intercomparison of a thermaloptical method for the determination of particle-phase organic and elemental carbon. Environ. Sci. Technol. 2003, 37 (5), 993−1001. (37) Chow, J. C.; et al. Source profiles for industrial, mobile, and area sources in the Big Bend Regional Aerosol Visibility and Observational Study. Chemosphere 2004, 54 (2), 185−208. (38) Kim, E.; Hopke, P. K. Improving source apportionment of fine particles in the eastern United States utilizing temperature-resolved carbon fractions. J. Air Waste Manage. Assoc. 2005, 55 (10), 1456− 1463. (39) Chow, J. C.; et al. Evaluation of organic markers for chemical mass balance source apportionment at the Fresno Supersite. Atmos. Chem. Phys. 2007, 7 (7), 1741−1754. (40) McDonald, J. D.; et al. Fine particle and gaseous emission rates from residential wood combustion. Environ. Sci. Technol. 2000, 34 (11), 2080−2091. (41) Kleeman, M. J.; Schauer, J. J.; Cass, G. R. Size and composition distribution of fine particulate matter emitted from wood burning, meat charbroiling, and cigarettes. Environ. Sci. Technol. 1999, 33 (20), 3516−3523. (42) Burkhardt, M.; Rossi, L.; Boller, M. Diffuse release of environmental hazards by railways. Desalination 2008, 226 (1−3), 106−113. (43) Joe, D. K., et al. Implementation of a high-resolution sourceoriented WRF/chem model at the Port of Oakland. Atmos. Environ. 2014, 82, 351−363. (44) Lee, J. H.; Hopke, P. K. Apportioning sources of PM2.5 in St. Louis, MO using speciation trends network data. Atmos. Environ. 2006, 40, S360−S377. (45) Gildemeister, A. E.; Hopke, P. K.; Kim, E. Sources of fine urban particulate matter in Detroit, MI. Chemosphere 2007, 69 (7), 1064− 1074. (46) Kim, E.; Hopke, P. K.; Edgerton, E. S. Source identification of Atlanta aerosol by positive matrix factorization. J. Air Waste Manage. Assoc. 2003, 53 (6), 731−739. (47) May, A. A.; et al. Gas-particle partitioning of primary organic aerosol emissions: (1) gasoline vehicle exhaust. Atmos. Environ. 2013, 77, 128−139. (48) Young, T. M.; et al. Resuspension of soil as a source of airborne lead near industrial facilities and highways. Environ. Sci. Technol. 2002, 36 (11), 2484−2490. (49) Lough, G. C.; et al. Emissions of metals associated with motor vehicle roadways. Environ. Sci. Technol. 2005, 39 (3), 826−836. (50) Dutton, S. J.; et al. Source apportionment using positive matrix factorization on daily measurements of inorganic and organic speciated PM2.5. Atmos. Environ. 2010, 44 (23), 2731−2741. (51) Vaaraslahti, K.; et al. Nucleation mode formation in heavy-duty diesel exhaust with and without a particulate filter. Environ. Sci. Technol. 2004, 38 (18), 4884−4890. (52) Mayer, A., et al. Reliability of DPF-Systems: Experience with 6000 Applications of the Swiss Retrofit Fleet; SAE Paper 2004-01-0076; SAE International: Warrendale, PA, USA, 2004. (53) Kittelson, D. B.; et al. On-road evaluation of two diesel exhaust aftertreatment devices. J. Aerosol Sci. 2006, 37 (9), 1140−1151. 13965

dx.doi.org/10.1021/es403235c | Environ. Sci. Technol. 2013, 47, 13957−13966

Environmental Science & Technology

Article

(54) Herner, J. D.; et al. Effect of advanced aftertreatment for PM and NOx reduction on heavy-duty diesel engine ultrafine particle emissions. Environ. Sci. Technol. 2011, 45 (6), 2413−2419. (55) Kleeman, M. J.; et al. Size distribution of trace organic species emitted from biomass combustion and meat charbroiling. Atmos. Environ. 2008, 42 (13), 3059−3075. (56) Hu, J. L.; et al. Identifying primary PM2.5 and PM0.1 sources for epidemiological studies in California. Environ. Sci. Technol. 2013. (57) Preble, C., et al. Effects of Truck Retrofit/Replacement Program on Diesel Engine Emissions at the Port of Oakland; American Association for Aerosol Research: Minneapolis, MN, USA, 2012. (58) Dallmann, T. R.; Harley, R. A.; Kirchstetter, T. W. Effects of diesel particle filter retrofits and accelerated fleet turnover on drayage truck emissions at the Port of Oakland. Environ. Sci. Technol. 2011, 45 (24), 10773−10779. (59) Swanson, J. J.; et al. Influence of storage and release on particle emissions from new and used CRTs. Atmos. Environ. 2009, 43 (26), 3998−4004.

13966

dx.doi.org/10.1021/es403235c | Environ. Sci. Technol. 2013, 47, 13957−13966