Lubricating Oil and Fuel Contributions To Particulate Matter Emissions

Nov 17, 2007 - ... MaldonadoSatya SardarRobert Jay WeberAllen L. RobinsonAllen H. ... Harley , Jose L. Jimenez , André S. H. Prévôt , and Allen L. ...
0 downloads 0 Views 238KB Size
Environ. Sci. Technol. 2008, 42, 235–242

Lubricating Oil and Fuel Contributions To Particulate Matter Emissions from Light-Duty Gasoline and Heavy-Duty Diesel Vehicles M I C H A E L J . K L E E M A N , * ,† SARAH G. RIDDLE,‡ MICHAEL A. ROBERT,† AND CHRIS A. JAKOBER§ Department of Civil and Environmental Engineering, Department of Chemistry, and Agriculture and Environmental Chemistry Graduate Group, University of California, Davis, 1 Shields Avenue, Davis, California 95616

Received May 04, 2007. Revised manuscript received October 08, 2007. Accepted October 15, 2007.

Size-resolved particulate matter emissions from heavy-duty diesel vehicles (HDDVs) and light-duty gasoline vehicles (LDGVs) operated under realistic driving cycles were analyzed for elemental carbon (EC), organic carbon (OC), hopanes, steranes, and polycyclic aromatic hydrocarbons. Measured hopane and sterane size distributions did not match the total carbon size distribution in most cases, suggesting that lubricating oil was not the dominant source of particulate carbon in the vehicle exhaust. A regression analysis using 17R(H)-21β(H)-29-norhopane as a tracer for lubricating oil and benzo[ghi]perylene as a tracer for gasoline showed that gasoline fuel and lubricating oil both make significant contributions to particulate EC and OC emissions from LDGVs. A similar regression analysis performed using 17R(H)-21β(H)-29-norhopane as a tracer for lubricating oil and flouranthene as a tracer for diesel fuel was able to explain the size distribution of particulate EC and OC emissions from HDDVs. The analysis showed that EC emitted from all HDDVs operated under relatively high load conditions was dominated by diesel fuel contributions with little EC attributed to lubricating oil. Particulate OC emitted from HDDVs was more evenly apportioned between fuel and oil contributions. EC emitted from LDGVs operated under fuel-rich conditions was dominated by gasoline fuel contributions. OC emitted from visibly smoking LDGVs was mostly associated with lubricating oil, but OC emitted from all other categories of LDGVs was dominated by gasoline fuel. The current study clearly illustrates that fuel and lubricating oil make separate and distinct contributions to particulate matter emissions from motor vehicles. These particles should be tracked separately during ambient source apportionment studies since the atmospheric evolution and ultimate health effects of these particles may be different. The source profiles for fuel and lubricating oil contributions to EC and OC emissions derived in this study provide a foundation for future source apportionment calculations.

* Corresponding author telephone: (530) 752-8386; fax: (530) 7527872; e-mail: [email protected]. † Department of Civil and Environmental Engineering. ‡ Department of Chemistry. § Agriculture and Environmental Chemistry Graduate Group. 10.1021/es071054c CCC: $40.75

Published on Web 11/17/2007

 2008 American Chemical Society

Introduction Motor vehicles have been identified as a major source of airborne particulate matter (PM) with an aerodynamic diameter smaller than 2.5 µm (PM2.5) (1–3). Recent studies show that motor vehicles also emit significant quantities of ultrafine particles with an aerodynamic diameter smaller than 0.1 µm (PM0.1) (4–9). Both PM2.5 and PM0.1 have been implicated as potential sources of negative health effects (10–15). The accurate quantification of motor vehicle contributions to ambient PM2.5 and PM0.1 concentrations is necessary to identify potential threats to public health. The majority of the PM emitted from motor vehicles is composed of carbonaceous material with a mass distribution that peaks between 0.1 and 0.3 µm and a tail that extends into the PM0.1 size range (4–6, 8). The only possible feedstocks for these carbonaceous PM emissions are (a) the fuel that powers the motor vehicles and (b) the lubricating oil used in the engines. The chemical signature of lubricating oil has been identified in the PM exhaust from both gasoline- and diesel-powered vehicles (16, 17), and heavy polycyclic aromatic hydrocarbons (PAHs) thought to be derived from gasoline have also been identified in light-duty vehicle exhaust (18, 19). It therefore seems likely that both fuel and lubricating oil contribute to the motor vehicle tailpipe emissions. The health effects of exhaust particles derived from lubricating oil, gasoline, and diesel fuel may be quite different if chemical composition affects toxicity. The purpose of the current study is to separately quantify fuel and lubricating oil contributions to PM emissions from five classes of lightduty gasoline vehicles (LDGVs) and four heavy-duty diesel vehicles (HDDVs) operated under realistic driving cycles. Organic compounds are identified that can act as unique tracers for lubricating oil, gasoline, and (approximately) diesel fuel. Multiple regression analysis is used to derive source profiles that list the amount of PM emitted per unit of each tracer compound. The contribution that fuel and oil make to the size distribution of PM emitted from LDGVs and HDDVs is also presented.

Methods Sample Collection. Size-resolved concentrations of elemental carbon (EC), organic carbon (OC), and individual organic compounds were measured in PM emitted from HDDVs and LDGVs operated under transient driving cycles on chassis dynamometers (4, 5, 20, 21). Four HDDVs were used in six tests that employed partial or complete versions of the Heavy Heavy Duty Diesel Test (HHDDT) driving cycle created by the California Air Resources Board. Test HDD-1 used only the idle + creep portion of the HHDDT cycle, while all other tests were conducted using the full cycle. Test HDD-4 used a simulated intertial weight of 66 000 lbs, while all other tests used a weight of 56 000 lbs. A complete summary of the HDDV tests is provided in Table S1 (Supporting Information) of the current manuscript. A total of 24 LDGVs were tested using the Federal Test Procedure (FTP) driving cycle including cold-start, hotrunning, and hot-start sections (see ref 4 for a summary comparison to other driving cycles). Vehicles were grouped on the basis of their emissions control technology: lowemission vehicles (LEVs), three-way catalyst (TWC), oxidation catalyst (OCAT), noncatalyst (NCAT), and vehicles emitting visible smoke (SMKR). A complete summary of the LDGV tests is provided in Table S2 (Supporting Information) of the current manuscript. VOL. 42, NO. 1, 2008 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

235

FIGURE 1. Normalized size distribution of total carbon, 17r(H)-21β(H)-29-norhopane, and fluoranthene emitted from diesel vehicles operated on a chassis dynamometer under transient driving cycles. Normalized concentrations in each size fraction are calculated by dividing the mass by the total measured PM1.8 mass. A dilution sampling system was employed to mix hot tailpipe exhaust with clean air to mimic the atmospheric dilution process without contaminating the emissions signature with material from other sources. Dilution ratios employed in each test were specifically chosen to be as large as possible to create realistic particle size distributions while still concentrating the sample sufficiently to collect enough mass for subsequent chemical analysis. Dilution factors ranged from 129 to 584 during diesel testing (5) and 124 to 393 during gasoline testing (4). PM samples were collected with reference ambient air samplers (RAAS; Andersen, Smyrna, GA) and micro-orifice uniform deposit impactors (MOUDIs; MSP Corporation, Shoreview, MN). Sample Extraction and Analysis for Organic Compounds. The concentrations of EC and OC in all RAAS and MOUDI samples were measured using the NIOSH thermal optical analysis method with a Sunset Laboratories carbon analyzer (Sunset Laboratories, Tigard, OR) (4, 5). The remaining portion of each sample was then extracted from the collection media with ∼15 mL of dichloromethane and concentrated using nitrogen evaporation to a final volume of 50 µL. A known amount of an isotopically labeled sterane (RRR-20R-cholestane-d4) and two isotopically labeled PAHs (chrysene-d12 and 236

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 42, NO. 1, 2008

dibenz[ah]anthracene-d14) was spiked onto each sample before extraction to serve as recovery and quantification internal standards. All sample quantification was carried out relative to these internal standards to increase the precision and accuracy of the results. All sample analysis was performed with gas chromatography (GC) and ion trap mass spectrometry (ITMS) (20, 21).

Results and Discussion Regression Analysis. Figures 1 and 2 plot normalized particle size distributions of total carbon and trace organic species emitted from LDGVs and HDDVs during transient driving cycles. Normalized concentrations are calculated by dividing the compound mass in each size bin by the compound mass across all size bins to show the shape of each size distribution on a similar scale. EC and OC size distributions were highly correlated in all tests such that the normalized total carbon size distribution also represents the EC and OC size distributions. A statistical analysis of all of the GC-ITMS measurements revealed that organic compounds could be generally classified into two groups on the basis of their size distributions. Two hopanes (17R(H)-21β(H)-29-norhopane and 17R(H)-21β(H)-hopane) and two steranes (Rββ-20R-stig-

FIGURE 2. Normalized size distribution of total carbon, 17r(H)-21β(H)-29-norhopane, and benzo[ghi]perylene emitted from gasoline vehicles operated on a chassis dynamometer under transient driving cycles. Normalized concentrations in each size fraction are calculated by dividing the mass by the total measured PM1.8 mass. mastane and Rββ-20S-stigmastane) had nearly identical size distributions in all samples. The likely source for these compounds was lubricating oil (16, 17, 22). Size distributions of heavy PAHs benzo[ghi]perylene and coronene were highly correlated in LDGV tests and were likely derived from the incomplete combustion of gasoline (18). Tests on LDGVs and HDDVs have shown that heavy PAHs are emitted primarily from LDGVs while light three- and four-ring PAHs are found at very high levels in HDDV exhaust (20, 21). Size distributions of light PAHs fluoranthene and pyrene were highly correlated during HDDV tests and were likely derived from the incomplete combustion of diesel fuel. The size distribution of 17R(H)-21β(H)-29-norhopane is not highly correlated with the measured size distribution of total carbon emissions in Figure 1c-f or Figure 2a,b,d,e. Likewise, the size distributions of fuel-derived PAHs and total particulate carbon are different in several important cases. Oil and fuel contributions to total carbon emissions appear to be separate and distinct in 8 of 11 size distributions plotted in Figures 1 and 2. One technique to separately identify the influence of multiple independent variables (fuel and lubricating oil) on a dependent variable (carbon emissions) is multivariate regression using the different particle size

fractions as separate observations within each test. A simple regression model with the form y ) ax1 + bx2 can be fit to each sample, with parameters “a” and “b” found to minimize the residual concentration. In this approach, “x1” and “x2” represent organic tracer compounds present in lubricating oil and fuel, respectively, while “a” and “b” represent the amount of carbon per unit of tracer compound. The intercept for the equation is forced through zero. Hopanes and steranes will be used as lubricating oil tracers while PAHs will be used as fuel tracers. Some PAHs may partition into the lubricating oil over time (23), but the dominant source of PAHs in the emissions is expected to be from the fuel (18, 23). A multiple regression analysis was performed twice for each sample set in the current study to separately identify oil and fuel contributions to EC and OC. The amount of carbon (either EC or OC) in each of the six PM samples collected by the MOUDI was fit using the equation carbon ) (µg C ⁄ ng oil tracer) × oil tracer + (µg C ⁄ ng fuel tracer) × fuel tracer (1) The inclusion of multiple hopanes and steranes in the regression analysis did not add information to the analysis because all species within this compound class had size VOL. 42, NO. 1, 2008 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

237

TABLE 1. Results of the multivariate regression analyses of fuel and oil tracers with both elemental and organic carbon for heavy duty diesel vehicles and light duty gasoline vehicles operated on chassis dynamometersa elemental carbon (µg EC/ng tracer) oil tracer HDD-1 HDD-2 HDD-3 HDD-4 HDD-5 HDD-6

2.6 ( 0.64 5.6 ( 8.7 8.7 ( 0.37 0.24 ( 0.51 2.2 ( 4.7 9.8 ( 4.6

LEV TWC OCAT NCAT SMKR

0.82 ( 0.28 1.1 ( 0.79 1.45 ( 0.39 1.1 ( 1.2 0.20 ( 0.055

fuel tracer

organic carbon (µg OC/ng tracer) R2

Diesel Vehicles 0.97 0.93 1.00 0.99 0.96 0.97 Gasoline Vehicles 6.3 ( 0.11 1.00 1.7 ( 0.13 0.99 0.036 ( 0.20 0.99 0.34 ( 0.026 0.99 0.047 ( 0.014 0.97 7.1 ( 9.8 4.95 ( 4.92 5.4 ( 0.62 24.0 ( 1..5 12.3 ( 1.6 8.5 ( 1.6

oil tracer

fuel tracer

R2

5.6 ( 2.8 4.6 ( 3.2 7.9 ( 0.48 0.0 ( 0.45 0.71 ( 0.94 0.045 ( 0.69

11.2 ( 42.1 2.2 ( 1.8 2.7 ( 0.81 15 ( 1.4 1.95 ( 0.31 2.6 ( 0.23

0.94 0.97 0.99 0.99 0.98 0.99

1.0 ( 0.28 1.50 ( 0.70 2.9 ( 1.3 3.5 ( 1.6 3.7 ( 0.46

1.4 ( 0.11 0.97 ( 0.12 4.0 ( 0.66 0.25 ( 0.034 0.91 ( 0.12

0.99 0.99 1.00 0.98 0.99

a Oil tracers concentrations are 17R(H)-21β(H)-29-norhopane, 17R(H)-21β(H)-hopane, Rββ-20R-stigmastane, and Rββ-20S-stigmastane normalized to the concentration 17R(H)-21β(H)-29-norhopane. Fuel tracer concentrations for gasoline vehicles are benzo[ghi]perylene and coronene normalized to the concentration of benzo[ghi]perylene. Fuel tracers for diesel vehicles are fluoranthene and pyrene normalized to the concentration of fluoranthene. Uncertainty estimates are 95% confidence intervals.

distributions that were highly correlated. Furthermore, including multiple hopanes and steranes dilutes the true power of the regression analysis since each sample set is limited to six observations (different MOUDI size fractions). In the present study, the average concentrations of all hopanes and steranes were expressed as an equivalent mass concentration of 17R(H)-21β(H)-29-norhopane for the concentration of “oil tracer”. Likewise, statistical analysis showed that PAH size distributions emitted from LDGVs and HDDVs had very similar shapes. The concentrations of two heavy PAHs (coronene and benzo[ghi]perylene) were expressed as the equivalent mass concentration of benzo[ghi]perylene for the gasoline “fuel tracer”, and the concentrations of the two light PAHs (fluoranthene and pyrene) were expressed as the equivalent mass concentration of fluoranthene for the diesel “fuel tracer”. Equivalent concentrations of each compound were calculated by multiplying the actual compound mass with the ratio of the surrogate compound mass to actual compound mass determined from regression across all six size fractions. This approach gives equal weighting to the information in all trace compounds even though the absolute value of some compound concentrations may be larger than others. Table 1 contains the results of the regression analysis for lubricating oil and fuel contributions to EC and OC emitted from LDGVs and HDDVs. Test results for individual LDGVs are not available in the current study since the use of realistic dilution factors required that the exhaust from multiple vehicles be composited in each sample to collect sufficient mass for chemical analysis. The uncertainty ranges shown in Table 1 are 95% confidence intervals. Some of the coefficients shown in Table 1 have uncertainty ranges that are comparable to the predicted value. This is caused by similarity in the size distributions for oil and fuel tracers during some of the chassis dynamometer tests. In tests HDDV-3, -4, -5, and -6 and LDGV-LEV, -TWC, -NCAT, and -SMKR, there were large differences between oil and fuel tracer size distributions, clearly emphasizing the fact that oil and fuel contributions to EC and OC emissions are separate. The agreement between the linear fit described by eq 1 to the concentration of EC and OC in each sample is summarized using the correlation coefficient (R2). The twoparameter regression model with separate oil and fuel contributions had R2 g 0.93 for all tests. Attempts to fit the measured EC size distributions from HDDVs using only oil tracers produced correlation coefficients (R2) ranging from 238

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 42, NO. 1, 2008

0.67 to 0.98. Attempts to fit the measured OC size distributions from HDDVs using only fuel tracers produced correlation coefficients (R2) ranging from 0.73 to 0.99. Single-variable fits for LDGVs were even less accurate, emphasizing the fact that both fuel and oil made contributions to EC and OC emissions from all vehicles. Figures S1-S4 in the Supporting Information plot reconstructed carbon concentrations measured in each particle size fraction using eq 1 versus measured concentrations. Carbon concentrations in different size fractions span a continuous range of values during each chassis dynamometer test. Predicted carbon concentrations show good agreement with measurements across the entire range of concentrations. Regression statistics are usually not dominated by a single large observation. PM0.1 is denoted by an open circle in Figures S1-S4 (Supporting Information) and usually falls in the middle of the measured range of values. On the basis of these results, eq 1 and the values listed in Table 1 appear to explain the size distribution of EC and OC emitted from LDGVs and HDDVs. The contributions that lubricating oil and fuel make to PM0.1 concentrations should be accurately represented by this analysis. The amount of EC associated with each unit of oil tracer ranged from 0.24 to 9.8 µg EC/ng tracer for HDDVs depending on the vehicle age, driving cycle, and simulated inertial load (Table 1). The test with the highest simulated load (HDD-4) had the lowest amount of EC associated with each unit of oil tracer (0.24 µg EC/ng tracer). The same vehicle tested with the identical driving cycle but with a 15% lower inertial load exhibited one of the highest EC concentrations per unit of oil tracer (8.7 µg C/ng tracer). Vehicle load is correlated with exhaust temperature, which appears to modify either the amount of tracer compound or the fraction of motor oil that is measured as EC in the vehicle exhaust. A previous analysis has revealed that higher loads shift the size distribution of motor oil tracer compounds to smaller sizes with less size shift observed for EC (20). The amount of EC associated with each unit of oil tracer emitted from LDGVs was more constant than that observed for HDDVs. LDGVs emitting visible smoke (SMKR) had the lowest ratios of EC per unit of oil tracer at approximately 0.2 µg C/ng tracer. It seems likely that these vehicles emitted motor oil with minimal chemical modification (i.e., no conversion of lubrication oil OC to EC). All other gasoline vehicle categories emitted approximately 1 µg EC/ng oil tracer

under the FTP driving cycle, indicating that at least some of the motor oil was chemically altered by the combustion process. The amount of EC associated with fuel tracers varied from 5 to 24 µg EC/ng tracer for HDDVs depending on engine load conditions. Test HDD-4, which operated under the highest inertial load conditions (66 000 lbs inertial weight), had the highest EC emissions with 24 µg EC/ng fuel tracer. Test HDD-5 was a vehicle operated at the standard inertial load (56 000 lbs) under the standard five-mode transient driving cycle, but this vehicle was only equipped with a 280 hp engine compared to the 460 hp (HDD-1 and HDD-2) and 510 hp (HDD-3 and HDD-4) engines used in other tests. The increased production of soot from diesel engines under relatively high load conditions is well documented, and the ability of the regression analysis to identify this feature in the measurements increases confidence in the regression approach. The amount of EC associated with fuel tracers emitted from LDGVs spanned the range from ∼0 to 6.3 µg EC/ng tracer. The LEVs emitted the greatest amount of EC per unit of fuel tracer, while the smoking vehicles (SMKR) emitted the smallest amount of EC per unit of fuel tracer. Timeresolved analysis showed that LEVs operated under the FTP driving cycle emitted most of their PM during the cold-start phase of the test when the combustion conditions were fuelrich and the exhaust system was relatively cold (4). The enhanced emission of soot from LDGVs under cold fuel-rich conditions has been identified in previous studies (24), once again building confidence in the regression analysis method. Smoking vehicles emitted unburned motor oil with little chemical modification. The low amount of EC per unit of fuel tracer from the SMKRs suggests that little pyrolized fuel was emitted from these vehicles. The amount of OC associated with each unit of oil tracer ranged from ∼0 to 7.9 µg OC/ng tracer for HDDVs and 1.1 to 3.7 µg OC/ng tracer for LDGVs. In most cases, the amount of OC associated with each unit of fuel tracer was comparable in magnitude, with values ranging from ∼2.2 to 15 µg OC/ng tracer for HDDVs and 0.25 to 4.0 µg OC/ng tracer for LDGVs. Figure 3a plots the total particulate carbon () EC + OC) per unit of 17R(H)-21β(H)-29-norhopane emitted from HDDVs and LDGVs. Lubricating oil that passes through the engine should retain a characteristic ratio of total carbon to 17R(H)-21β(H)-29-norhopane that reflects the relative concentration of the hopane in the oil. The degree to which the lubricating oil chars in the combustion chamber (converting OC to EC) should not affect the total carbon concentration. All of the LDGVs illustrated in Figure 3a exhibit a relatively constant ratio of total carbon per unit of 17R(H)-21β(H)29-norhopane with values ranging from 2.5 to 4.5 µg C/ng 17R(H)-21β(H)-29-norhopane. The uncertainty range around the HDDV predictions is greater than that for LDGVs because hopane size distributions are similar to light PAH size distributions in the HDDV exhaust. Generally speaking, the HDDVs tested in the current study appear to emit ∼3–10 µg C/ng 17R(H)-21β(H)-29-norhopane. The enhancement of carbon emissions per unit of norhopane in HDDVs versus LDGVs may be due to the use of heavier-grade oils in the HDDVs. The total carbon to norhopane concentrations derived from a regression analysis of exhaust particles in the current study fall within the range of values directly measured in new motor oil and used motor oil collected from in-use vehicles by Fujita et al. (23), shown on the far right side of Figure 3a. Fujita et al. hypothesize that the ratio of total carbon to hopanes decreases for used oil because lighter components are lost with use. The agreement between the direct oil measurements made by Fujita et al. and the regression results for exhaust particles calculated in the current study suggests

FIGURE 3. Quantity of total particulate carbon () EC + OC) associated with each unit of norhopane derived from motor oil (panel a) and PAH derived from fuel (panel b) during HDDV and LDGV testing. Note that the PAH used for diesel fuel was fluoranthene and the PAH used for gasoline was benzo[ghi]perylene. The uncertainty range corresponds to 95% confidence intervals for the regression analysis summarized in Table 1. Direct measurements of new and used oil shown in panel a are adapted from Fujita et al. (23). that the category averages shown in Figure 3 are representative of real-world conditions. The large difference between test HDD-3 and HDD-4 shown in Figure 3a is unexpected since both tests used the same vehicle and driving cycle, but with an 18% higher load in HDD-4. Higher loads result in hotter exhaust temperatures, which may have caused the more volatile fraction of the lubricating oil to partition to the gas phase while the less volatile fraction containing the hopane and sterane tracers remained in the particle phase. This trend is consistent with the shift of hopane and sterane size distributions to smaller particle sizes during test HDD-4 (see Figure 1d). Figure 3b shows that the apparent concentration of total carbon to PAH is significantly enhanced in HDD-4 compared to HDD3, suggesting that the volatilized fraction of the oil may have condensed onto soot particles derived from diesel fuel rather than the residual oil particles as the exhaust cooled. Figure 3b plots the total particulate carbon emitted per unit of fluoranthene (HDDVs) and per unit of benzo[ghi]perylene (LDGVs) during chassis dynamometer tests. The amount of PM produced from each unit of fuel depends on engine technology, emissions control technology, and driving cycle, as expected. Most HDDVs emitted ∼8–19 µg C/ng fluoranthene under normal load conditions, with ∼40 µg C/ng fluoranthene observed during test HDD-4, which used higher load conditions in which some fraction of lubricating oil may have condensed onto fuel-derived soot particles. LDGVs emitted ∼1–10 µg C/ng benzo[ghi]perylene, with the highest relative concentrations observed in relatively “clean” vehicles that released the majority of their PM emissions during the cold-start phase of the FTP driving cycle. VOL. 42, NO. 1, 2008 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

239

FIGURE 4. Reconstructed source profiles for 1 µg of particulate matter emitted from heavy duty diesel vehicles (HDDVs) operated under HHDDT driving cycles. Each panel represents emissions from a different HDDV. Test HDDV-1 used only the idle + creep portion of the HHDDT driving cycle, while all other tests used the full five modes of the HHDDT including transients and high-speed cruises. All tests used a simulated inertial weight of 56 000 lbs except for HDDV-4 which used a simulated weight of 66 000 lbs. Size-Resolved Source Profiles. Figure 4 plots source contributions to each microgram of particulate carbon emitted from HDDVs predicted using the regression factors shown in Table 1 and the size-resolved emissions rates of each trace organic compound as discussed in previous studies (20). These source profiles can be multiplied by the PM1.8 emissions rate to predict size-resolved source contributions to absolute carbon emissions. EC and OC associated with oil and fuel tracers are shown separately in each panel. Residual concentrations (positive and negative) are shown as “unknown” in each figure. All reconstructed diesel source profiles had a peak in the size bin between an aerodynamic diameter of 0.1 and 0.18 µm, but source contribution trends depend strongly on vehicle age and load. The tests with the highest loads relative to engine horsepower (HDDV-4 and HDDV-5) had the greatest amount of EC associated with diesel fuel and the least amount of EC associated with lubricating oil. This matches the expected behavior of compression ignition engines operating under fuel-rich conditions. The test with 240

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 42, NO. 1, 2008

the lowest load (HDDV-1) had the most OC associated with lubricating oil. Fuel contributions to OC size distributions were relatively consistent across all tests except HDDV-3. Figure 5 plots source contributions to each microgram of particulate carbon emitted from LDGVs predicted using the regression factors shown in Table 1 and the measured emissions rates of tracer compounds (21). All reconstructed LDGV source profiles peak between a particle diameter of 0.1 and 0.18 µm except SMKRs, which peak between a particle diameter of 0.18 and 0.32 µm. Fuel contributions to EC dominate the carbonaceous emissions from LEVs, TWCs, and NCATs. Each of these samples was dominated by emissions released without a functioning catalyst. In the case of LEVs and TWCs, a significant fraction of the emissions were released when the engine was operating under “coldstart” conditions in a fuel-rich mode and before the catalyst reached the operating temperature. Enhanced EC production from spark-ignition engines is expected under these condi-

FIGURE 5. Reconstructed source profiles for 1 µg of particulate matter emitted from light duty gasoline vehicles (LDGVs) operated under the Federal Test Procedure driving cycle. Each panel represents an average of multiple vehicles within different emissions control technology classes: low-emission vehicles (LEVs), three-way catalyst vehicles (TWC), oxidation catalyst vehicles (OCAT), noncatalyst vehicles (NCAT), and smoking vehicles (SMKR). tions. Fuel contributions to OC are also significant in all LDGV samples, but little trend is observed between the different vehicle categories. Oil contributions to OC are most significant from the smoking vehicles (SMKRs) as would be expected since the majority of the smoke emissions are likely motor oil. Oil OC emissions are also significant from vehicles equipped with oxidation catalysts (OCATs), although the absolute emissions rates from these vehicles (micrograms per kilometer) are much lower than emissions from SMKRs (4). Disclaimer. This report was prepared by the University of California at Davis as an account of work partially sponsored by the Coordinating Research Council (CRC). Neither the CRC, members of the CRC, the University of California at Davis, nor any person acting on their behalf (1) makes any warranty, express or implied, with respect to the use of any information, apparatus, method, or process disclosed in this report, or (2) assumes any liabilities with respect to use, or damages resulting from the use or inability

to use, any information, apparatus, method, or process disclosed in this report.

Acknowledgments This research was supported by the California Air Resources Board under contract 01-306, the Coordinating Research Council under contract E-55/59-1.5a, and the National Science Foundation’s Nanophases in the Environment, Agriculture, and Technology - Integrative Graduate Education, Research, and Training (NEAT-IGERT) initiative award DGE-9972741. The authors thank the staff at the HagenSmit Laboratory and the staff operating the WVU portable dynamometer for assistance with vehicle testing. The authors thank Dr. Kimberly Prather, David Sodeman, and Stephen Toner of UCSD for assistance with sample collection. The authors thank Dr. James Schauer at the University of Wisconsin, Madison, for help with quantification standards and Nehzat Motallebi at the California Air Resources Board for help with project coordination. VOL. 42, NO. 1, 2008 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

241

Supporting Information Available Figures S1 to S4 illustrate the agreement between predicted vs measured carbon concentrations in each particle size fraction. This material is available free of charge via the Internet at http://pubs.acs.org.

Literature Cited (1) Schauer, J. J.; Rogge, W. F.; Hildemann, L. M.; Mazurek, M. A.; Cass, G. R. Source apportionment of airborne particulate matter using organic compounds as tracers. Atmos. Environ. 1996, 30, 3837–3855. (2) Gertler, A. W.; Gillies, J. A.; Pierson, W. R. An assessment of the mobile source contribution to PM10 and PM2.5 in the United States. Water, Air, Soil Pollut. 2000, 123, 203–214. (3) Fraser, M. P.; Yue, Z. W.; Buzcu, B. Source apportionment of fine particulate matter in Houston, TX, using organic molecular markers. Atmos. Environ. 2003, 37, 2117–2123. (4) Robert, M. A.; Jakober, C. A.; VanBergen, S.; Kleeman, M. J. Size and Composition Distribution of Particulate Matter 1. Lightduty Gasoline Vehicles. J. Air Waste Manage. Assoc. 2007,in press. (5) Robert, M. A.; Jakober, C. A.; VanBergen, S.; Kleeman, M. J. Size and composition distribution of particulate matter 2. Heavyduty diesel vehicles. J. Air Waste Manage. 2007, in press. (6) Kleeman, M. J.; Schauer, J. J.; Cass, G. R. Size and Composition Distribution of Fine Particulate Matter Emitted from Motor Vehicles. Environ. Sci. Technol. 2000, 34, 1132–1142. (7) Phuleria, H. C.; Sheesley, R. J.; Schauer, J. J.; Fine, P. M.; Sioutas, C. Roadside measurements of size-segregated particulate organic compounds near gasoline and diesel-dominated freeways in Los Angeles, CA. Atmos. Environ. 2007, 41, 4653–4671. (8) Kittelson, D. B.; Watts, W. F.; Johnson, J. P. On-road and laboratory evaluation of combustion aerosols - Part 1: Summary of diesel engine results. J. Aerosol Sci. 2006, 37, 913–930. (9) McDonald, J. D.; Barr, E. B.; White, R. K.; Chow, J. C.; Schauer, J. J.; Zielinska, B.; Grosjean, E. Generation and characterization of four dilutions of diesel engine exhaust for a suhchronic inhalation study. Environ. Sci. Technol. 2004, 38, 2513–2522. (10) Dockery, D. W.; Pope, C. A., III; Xu, X.; Spengler, J. D.; Ware, J. H.; Fay, M. E.; Ferris, B. G., Jr.; Speizer, F. E. An association between air pollution and mortality in six U.S. cities. N. Engl. J. Med. 1993, 329, 1753–1759. (11) Samet, J. M.; Dominici, F.; Curriero, F. C.; Coursac, I.; Zeger, S. L. Fine particulate air pollution and mortality in 20 US Cities, 1987–1994. N. Engl. J. Med. 2000, 343, 1742–1749. (12) Oberdorster, G.; Gelein, R. M.; Ferin, J.; Weiss, B. Association of Particulate Air-Pollution and Acute Mortality - Involvement of Ultrafine Particles. Inhalation Toxicol. 1995, 7, 111–124.

242

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 42, NO. 1, 2008

(13) Penttinen, P.; Timonen, K. L.; Tiittanen, P.; Mirme, A.; Ruuskanen, J.; Pekkanen, J. Ultrafine particles in urban air and respiratory health among adult asthmatics. Eur. Respir. J. 2001, 17, 428–435. (14) Seagrave, J.; McDonald, J. D.; Gigliotti, A. P.; Nikula, K. J.; Seilkop, S. K.; Gurevich, M.; Mauderly, J. L. Mutagenicity and in vivo toxicity of combined particulate and semivolatile organic fractions of gasoline and diesel engine emissions. Toxicol. Sci. 2002, 70, 212–226. (15) 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, 455–460. (16) Schauer, J. J.; Kleeman, M. J.; Cass, G. R.; Simoneit, B. R. T. Measurement of emissions from air pollution sources. 5. C-1C-32 organic compounds from gasoline-powered motor vehicles. Environ. Sci. Technol. 2002, 36, 1169–1180. (17) Schauer, J. J.; Kleeman, M. J.; Cass, G. R.; Simoneit, B. R. T. Measurement of emissions from air pollution sources. 2. C-1 through C-30 organic compounds from medium duty diesel trucks. Environ. Sci. Technol. 1999, 33, 1578–1587. (18) Marr, L. C.; Kirchstetter, T. W.; Harley, R. A.; Miguel, A. H.; Hering, S. V.; Hammond, S. K. Characterization of Polycyclic Aromatic Hydrocarbons in Motor Vehicle Fuels and Exhaust Emissions. Environ. Sci. Technol. 1999, 33, 3091–3099. (19) Riddle, S.; Jakober, C. A.; Robert, M. A.; Cahill, T. M.; Charles, J. M.; Kleeman, M. J. Large PAHs detected in fine particulate matter emitted from light-duty gasoline vehicles. Atmos. Environ. 2007,in press. (20) 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, 1962–1969. (21) 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, in press. (22) Rogge, W. F.; Hildemann, L. M.; Mazurek, M. A.; Cass, G. R.; Simoneit, B. R. T. Sources of fine organic aerosol. 2. Noncatalyst and catalyst-equipped automobiles and heavy-duty diesel trucks. Environ. Sci. Technol. 1993, 27, 636–651. (23) Fujita, E.; Campbell, D.; Zielinska, B. Chemical Analysis of Lubrication Oil Samples from a Study to Characterize Exhaust Emissions from Light-Duty Gasoline Vehicles in the Kansas City Metropolitan Area; Desert Research Institute: Reno, NV, 2005. (24) Watson, J. G.; Fujita, E.; Chow, J. C.; Zielinska, B.; Richards, L. W.; Neff, W. D.; Dietrich, D. Northern Front Range Air Quality Study Final Report; Desert Research Institute: Reno, NV, 1998.

ES071054C