Environ. Sci. Technol. 2007, 41, 7464-7471
Size Distribution of Trace Organic Species Emitted from Light-Duty Gasoline Vehicles SARAH G. RIDDLE,† MICHAEL A. ROBERT,‡ CHRIS A. JAKOBER,§ MICHAEL P. HANNIGAN,| AND M I C H A E L J . K L E E M A N * ,‡ Department of Chemistry, University of California, Davis, 1 Shields Avenue, Davis, California 95616, Department of Civil and Environmental Engineering, University of California, Davis, 1 Shields Avenue, Davis, California 95616, Agriculture and Environmental Chemistry Graduate Group, University of California, Davis, 1 Shields Avenue, Davis, California 95616, and Department of Mechanical Engineering, University of Colorado, Boulder, Colorado 80309
Size distributions for particulate hopanes+steranes and nonvolatile polycyclic aromatic hydrocarbons (PAHs) emitted from five classes of light-duty gasoline-powered vehicles were measured using the federal test procedure (FTP), unified cycle (UC), and correction cycle (CC) driving cycles. 17R(H)-21β(H)-29-norhopane, 17R(H)-21β(H)-hopane, Rββ20R-stigmastane, and Rββ-20S-stigmastane were highly correlated and behaved consistently across sampling methods. Coronene and benzo[ghi]perylene were the most ubiquitous heavy PAHs detected in the vehicle exhaust. The emission rates of hopanes, steranes, and PAHs contained in particles with aerodynamic diameters of less than 1.8 µm varied by 2 orders of magnitude between the lowest- and highest-emitting vehicle classes. Hopane+sterane size distributions emitted from vehicles without an operating catalyst (including “cold-start” emissions from catalystequipped vehicles) were bimodal with one mode between 0.10 and 0.18 µm and the second mode >0.32 µm particle diameter. Hopane+sterane emissions released from vehicles with a catalyst at operating temperature had a single mode between 0.1 and 0.18 µm diameter. Hopane+sterane emissions from visibly smoking vehicles had a single mode between 0.18 and 0.32 µm diameter. Heavy PAH size distributions for all vehicle classes consistently had a single mode between 0.10 and 0.18 µm particle diameter (0.1-0.32 µm diameter for smoking vehicles). The geometric standard deviations for PAH size distributions were generally smaller than the corresponding hopane+sterane distributions. These trends suggest that hopanes+steranes and heavy PAHs act as tracers for separate processes of particulate organic carbon formation. PAH and hopane+sterane emissions shifted to smaller sizes during the more aggressive UC and CC * Corresponding author phone: (530) 752-8386; fax: (530) 7527872; e-mail:
[email protected]. † Department of Chemistry, University of California, Davis. ‡ Department of Civil and Environmental Engineering, University of California, Davis. § Agriculture and Environmental Chemistry Graduate Group, University of California, Davis. | University of Colorado. 7464
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 41, NO. 21, 2007
driving cycles relative to the FTP. The fraction of PAH and hopane+sterane emissions in the ultrafine (Dp < 0.1 µm) range more than doubled during “warm-start” UC and CC cycles vs the FTP cycle. The enhancement of ultrafine PAHs during “cold-start” UC driving cycles was less pronounced.
Introduction Many studies have identified adverse health effects associated with high levels of fine (Dp < 2.5 µm) airborne particulate matter (PM) (1, 2). Particles in the ultrafine size range (Dp < 0.1 µm) have been linked to pulmonary and cardiovascular health concerns (2-6). Light-duty gasoline vehicles (LDGVs) are a major source of fine PM in highly populated areas (79) with a significant portion of this particle mass in the ultrafine size range (10, 11). A recent study identified emissions from gasoline-powered motor vehicles as highly toxic to human lung tissue (12). The ability to recognize the health effects of size-resolved airborne particles emitted from gasoline-powered motor vehicles is limited by our ability to identify these particles in the atmosphere. A size-resolved chemical “fingerprint” or source profile is needed to quantify gasoline exhaust particle concentrations under realistic conditions. Several investigators have quantified polycyclic aromatic hydrocarbon (PAH), hopane, and sterane concentrations that are emitted from LDGVs in the fine size range (Dp < 2.5 µm) (13-16). These compounds were studied because they are useful for the construction of source profiles. Hopanes and steranes are known to be present in petroleum and are used as marker compounds by geochemists to assess the geological maturity of crude oil (17, 18). Hopanes and steranes are present in motor oil but are absent in gasoline (14, 19), presumably because they are removed during refining. Heavy PAHs such as coronene have been suggested as tracers for gasoline exhaust (20, 21). Certain PAHs have been shown to be both carcinogenic and mutagenic in epidemiological studies; thus, assessment of the size distribution of these compounds is vital to assess their potential health implications (22-24). One study has reported the concentration of PAHs, hopanes, and steranes in particles smaller than 0.18 µm diameter adjacent to a roadway (25). There is no study to date which simultaneously determines the size distributions of these compounds from in-use LDGVs in the fine and ultrafine size fractions as a function of driving cycle. The purpose of this study is to report the size distributions of 28 particle-phase hopane, sterane, and PAH compounds contained in tailpipe emissions from 5 categories of lightduty gasoline-powered vehicles spanning a range of engine types, emission-control technologies, and driving cycles. Measurements are reported for six size fractions between 0.056 and 1.8 µm particle aerodynamic diameter. Potential tracers for ultrafine source apportionment studies are identified, and ultrafine source profiles are presented.
Methods Sample Collection. Size-resolved and bulk PM1.8 emissions from LDGVs were collected at the California Air Resources Board (CARB) Haagen-Smit Laboratory in El Monte, CA, during August and September of 2002 using a chassis dynamometer combined with a dilution sampling system (26). Bulk PM1.8 was collected using Andersen reference ambient air samplers (RAASs) (Andersen Instruments, Smyrna, GA). Each sample leg of the RAAS was operated at a flow 10.1021/es070153n CCC: $37.00
2007 American Chemical Society Published on Web 10/03/2007
TABLE 1. PM1.8 Ratio of Analyte Mass to Total Organic Carbon Mass (µg g-1) for LEV, TWC PC, OCAT, NCAT, and SMOKERS Vehicles Operated under the FTP Driving Cycle LEV
TWC PC hopanes error
compound
ratio
error
ratio
17R(H)-21β(H)-29-norhopane 17R(H)-21β(H)-hopane
142 107
4 4
153 100
64 28
3 3
163 191 83 509 130 40 238 222
4 6 4 13 3 4 6 6
Rββ-20R-stigmastane Rββ-20S-stigmastanea
OCAT
NCAT
SMOKERS
ratio
error
ratio
error
ratio
error
1 2
136 162
4 5
205 184
6 7
98 72
0.7 0.5
steranes NDb 45 1
25 NDb
2
165 89
6 4
35 20
0.3 0.2
99 39 29 318 40 NDb 198 50
3 3 2 10 1
801 692 137 4206 419 188 4309 442
24 21 6 124 12 7 127 13
135 154 33 495 45 58 296 40
1 2 0.4 4 0.4 0.5 3 0.3
PAHs benzo[e]pyrene benzo[a]pyrene perylene benzo[ghi]perylene indeno[1,2,3-cd]fluoranthenea dibenzo[def,mno]chrysenea coronene MW 302 isomersa a
300 346 169 1050 265 104 1077 457
4 4 4 8 2 4 7 4
6 2
Analyte identification based on comparisons to relative retention times in the literature. b ND ) not detected
rate of 10 L min-1, yielding a total flow rate of 30 L min-1 through each sample manifold (attached to three sample legs) and a particle aerodynamic diameter cut size of 1.8 µm from the upstream AIHL-design cyclone (27). Size-resolved PM was collected using three micro-orifice uniform deposit impactors (MOUDIs) (MSP Corporation, Shoreview, MN), in six size fractions between 0.056 and 1.8 µm in aerodynamic diameter. Twenty-eight test LDGVs were grouped into five categories based on emissions control technology: lowemission vehicles (LEVs), three-way catalyst (TWC) vehicles, oxidation catalyst (OCAT) vehicles, noncatalyst (NCAT) vehicles, and visibly smoking vehicles (SMOKERs). LEVs have more-efficient catalysts than older TWC vehicles, increased engine durability, advanced electronic engine management, on-board diagnostic systems, and near-zero evaporative emissions. One of the smoking vehicles was equipped with a catalyst while the other was not. Within these categories, vehicles could be further classified as passenger cars (PCs) or light-duty trucks (LDTs)/sport utility vehicles (SUVs). Table S1 in the Supporting Information section summarizes the vehicle test matrix with more details provided by Robert et al. (28). The fuels used in the testing program were “tank fuels” present in the vehicles when they were procured for testing. Detailed analysis of the fuels shows that they have a sulfur content of 10-20 ppm. Likewise, the motor oil used in each test was procured along with the vehicle. Variation in the chemical composition of the fuel and oil may influence the resulting emissions of organic compounds from individual vehicles. Composite PM samples were collected from each LDGV category in order to average the chemical composition in each vehicle class and to collect enough mass for chemical analysis. All LDGVs were tested using the federal test procedure (FTP) driving cycle, and selected vehicle classes were also tested using the unified cycle (UC) and the correction cycle (CC) (29). The FTP cycle has moderate transient sections with the lowest top speed of all three cycles. The UC has the highest acceleration and greatest speed of all the cycles tested. The CC has a higher average speed section than the FTP cycle with some transient driving (30). Sample Extraction and Analysis for Organic Compounds. Extraction and analysis methods were identical to those employed during a previous analysis of diesel engine exhaust (31), and so only a brief summary is provided here. Samples
were spiked with an isotopically labeled sterane and two labeled PAHs and then sonicated twice in 15 mL of dichloromethane. Extracts were combined and then concentrated to a 50 µL volume via nitrogen evaporation (heavily loaded samples were only reduced in volume to 200 µL). Analysis was performed using an Agilent J&W DB-XLBMSD capillary gas chromatograph column followed by a Varian 2000 iontrap mass spectrometer. Analytes were identified by comparison to authentic standards or by comparison of relative retention times and mass spectra to those in the literature (32). Further analysis details are discussed in the Supporting Information.
Results and Discussion Quality-Assurance Checks. Figure S1 in the Supporting Information illustrates agreement between co-located MOUDI and filter measurements in the current study as a qualityassurance check on the size distribution measurements. Most PAH species identified behave consistently when collected with MOUDI and filter samplers. Within the hopane class of compounds, 17R(H)-21β(H)-29-norhopane and 17R(H)-21β(H)-hopane appear to exhibit the best correlation between MOUDI and filter measurements as demonstrated by their presence in almost all samples with correlation slopes ) 0.75-0.92 and correlation coefficients R2 ) 0.96-0.98. Hopanes 22R-17R(H)-21β(H)-30-homohopane and 22S-17R(H)-21β(H)-30-bishomohopane measured with MOUDIs and filters also agree strongly, but these compounds were not observed in all vehicle classes tested and so they are lessdesirable tracers for size-resolved lubricating oil PM emissions. Within the sterane class of compounds, Rββ-20Rstigmastane and Rββ-20S-stigmastane exhibit the most consistent behavior between MOUDI and filter measurements with correlation slopes ) 0.81-1.28 and correlation coefficients R2 ) 0.84-0.94. Lubricating oil tracers 17R(H)21β(H)-29-norhopane, 17R(H)-21β(H)-hopane, Rββ-20Rstigmastane, and Rββ-20S-stigmastane appear to have the greatest promise as size-resolved tracers for lubricating oil PM emitted from LDGVs. These same four hopanes and steranes were identified in a study of diesel exhaust particle size distributions (31) suggesting that they are general tracers for size-resolved lubricating oil particles emitted from motor vehicles. Only the four hopanes+steranes and eight PAHs with the best agreement between MOUDI and filter measurements were used to calculate representative size distributions in this study. Table 1 summarizes the list of VOL. 41, NO. 21, 2007 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
7465
FIGURE 1. (a) Comparison of TWC PM1.8 trace organic emission rates in the current study with previous measurements and (b) summary of PM1.8 emission rates for all vehicle classes tested under the FTP driving cycle. compounds considered in the analysis. A more detailed discussion of the quality-assurance checks is provided in the Supporting Information. Figure 1a shows the PM1.8 hopane and selected PAH emission rates from TWC vehicles operated under the FTP driving cycle measured in the current study along with measurements reported by Schauer et al. (13). 17R(H)-21β(H)-hopane, benzo[k]fluoranthene, and benzo[e]pyrene were found at almost exactly the same emission rates in both studies. Perylene and indeno[1,2,3-cd]fluoranthene were found at twice the emission rate in the current study, and emission rates of 17R(H)-21β(H)-29-norhopane and benzo[a]pyrene measured in the current study were 7-8 times higher than those measured by Schauer et al. All of the measured emission rates were much lower than those reported by Rogge et al. (14) (not shown in Figure 1), reflecting the general decrease in TWC emission rates over the last 10-20 years. Figure 1a also compares the dynamometer results measured in the current study to roadside measurements (33). Strong agreement is observed between the dynamometer measurements and the roadside measurements for heavy PAHs (perylene, indeno[1,2,3-cd]fluoranthene, and coronene) when results are expressed in units of (µg analyte/L fuel burned). Emission rates of hopanes and 7466
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 41, NO. 21, 2007
lighter PAHs measured in the roadside are higher than those measured during dynamometer tests, possibly because fuel and oil contributions to PM emissions vary independently (34). Figure 1b illustrates the PM1.8 emission factors of hopanes and selected PAHs for all the vehicle categories tested in the current study using the FTP driving cycle. PM1.8 emission rates were progressively larger per unit of fuel burned for LEVs < TWCs 0.32 µm. Hopanes and steranes are high molecular weight compounds with low vapor pressure that are expected to act as nonvolatile tracers for primary motor oil (14, 19). The presence of hopanes and steranes in particles larger than 0.32 µm indicates that a significant fraction of motor oil was also shifted to those particle sizes in the exhaust from these vehicles. Continuous measurements during each test show that the majority of the PM emitted from LEV and TWC vehicles was released during the “cold-start” portion of the FTP driving cycle before the catalyst reached operating temperature (28). It is therefore not surprising that NCAT vehicles (Figure 2d) also have a significant fraction of hopanes and steranes contained in particles larger than 0.32 µm even though measurements show that NCAT emissions were much higher and occurred more uniformly throughout the test (28). Once the catalyst reaches operating temperature, it appears to remove hopanes and steranes contained in particles larger than 0.32 µm. This trend can also be observed in Figure 2i, which shows emissions from TWC vehicles operated under the first two modes of the FTP driving cycle under “warm-start” conditions. The fraction of hopanes+ steranes in particles with diameters greater than 0.32 µm emitted by TWC LDT/SUV warm-start vehicles (Figure 2i) is greatly reduced relative to that of the TWC PC vehicles that were dominated by coldstart emissions (Figure 2a,b). Figure 2c shows that hopane+sterane emissions from OCAT vehicles are dominated by a single mode between 0.1 and 0.18 µm particle diameter. It is expected that OCAT vehicles also emit hopanes+steranes in size fractions larger than 0.32 µm diameter before the catalyst reaches operating temperature, but continuous measurements indicate that the OCAT vehicles emitted significant amounts of PM throughout the entire test, not just during the cold start (28). Figure 1 and Table S1 (Supporting Information) show that the total PM1.8 emissions from OCAT vehicles were an order of magnitude larger than the emissions from TWC vehicles. The OCAT vehicles tested in the current study had an average model year of 1979 while the average model year of TWC vehicles ranged from 1992 to 2000. Engine technology improvements over time and newer engine conditions reduce emissions from the newer vehicles. Figure 2e shows that hopane+sterane emissions from SMOKER vehicles are dominated by a single mode between 0.1 and 0.56 µm. Figure 1 and Table S1 (Supporting Information) show that emissions rates from SMOKER vehicles were 1 order of magnitude larger than those from OCAT vehicles and 2 orders of magnitude larger than those of TWC vehicles. One of the two SMOKER vehicles tested in the current study was equipped with a catalyst while the other was not. Continuous measurements show that the noncatalyst-equipped SMOKER vehicle dominated the emissions (28). The fact that SMOKER hopane+sterane size distributions had a single mode instead of the more typical bimodal distribution for vehicles operating without a catalyst suggests that the formation mechanism for the lubricating oil particles in the SMOKER exhaust was atypical, as might be expected for a vehicle emitting visible smoke. Figure 2g-i illustrate the effect of operating cycle on the size distribution of particulate hopanes+steranes emitted from TWC LDT/SUVs. Both the UC (2g) and the CC (2h) have higher average speed and sharper accelerations than the first two modes of the FTP (Figure 2i). All tests involved warmstart operation where the catalyst was close to operating temperature at the beginning of the test. The fraction of
hopane+sterane emissions in the 0.056-0.1 µm size fraction increases from ∼0.05 at lower FTP loads (Figure 2i) to ∼0.3 at the higher UC and CC loads (Figure 2g,h). This trend qualitatively matches the behavior of hopane+sterane size distributions emitted from diesel vehicles under different load conditions (31). Park et al. (35) showed that the effective density of diesel particles decreases as engine load increases. This reduces the apparent aerodynamic diameter of the exhaust particles, transferring more surface area to the ultrafine size range, and causing a greater fraction of the hopanes and steranes to appear in particles with aerodynamic diameters
