Environ. Sci. Technol. 2001, 35, 26-32
In-Use Light-Duty Gasoline Vehicle Particulate Matter Emissions on Three Driving Cycles STEVEN H. CADLE,* PATRICIA MULAWA, PETER GROBLICKI, AND CHRIS LAROO† General Motors R&D Center, MD 480-106-269, Warren, Michigan 48090-9055 RONALD A. RAGAZZI, KEN NELSON, AND GERALD GALLAGHER‡ Colorado Department of Public Health and Environment, Air Pollution Control Division, 15608 East 18th Avenue, Aurora, Colorado 80011 BARBARA ZIELINSKA Atmospheric Sciences Center, Desert Research Institute, 2215 Raggio Parkway, Reno, Nevada 89512-1095
Twenty-four properly functioning and six high carbon monoxide emission light-duty gasoline vehicles were emission tested in Denver, CO, using the Federal Test Procedure (FTP), a hot start Unified Cycle (UC), and the REP05 driving cycles at 35 °F. All were 1990-1997 model year vehicles tested on both an oxygenated and a nonoxygenated fuel. PM10 emission rates for the properly functioning vehicles using oxygenated fuel averaged 6.1, 3.6, and 12.7 mg/mi for the FTP, UC, and REP05, respectively. The corresponding values for the high emitters were 52, 28, and 24 mg/mi. Use of oxygenated fuel significantly reduces PM10 on the FTP, with all the reduction occurring during the cold start. MOUDI impactor samples showed that 33 and 69% of the PM mass was smaller than 0.1 µm for the FTP and REP05 cycles, respectively, when collected under standard laboratory conditions. Particle number counts were much higher on the REP05 than the FTP. Counts were obtained using secondary dilution of samples drawn from the standard dilution tunnel. FTP PM10 was mostly carbonaceous material, 36% of which was classified as organic. For the REP05, as much as 20% of the PM10 was sulfate and associated water. Forty-five percent of the REP05 PM carbon emissions was classified as organic. Driving cycle had a significant impact on the distribution of the emitted polynuclear aromatic hydrocarbons.
Introduction Emission inventories and receptor modeling have been used to estimate the contribution of mobile sources to the winter ambient PM burden in the metro Denver area (1). The metro Denver area emission inventory developed by the Regional Air Quality Council (RAQC) estimates that mobile source * Corresponding author phone: (810)986-1603; fax: (810)986-1910; e-mail:
[email protected]. † Present address: U.S. Environmental Protection Agency, National Vehicle & Fuel Emissions Lab, Assessment & Standards Division, 2000 Traverwood Dr., Ann Arbor, MI 48105. ‡ Present address: 6093 E. Briarwood Dr., Englewood, CO 80112. 26
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 35, NO. 1, 2001
exhaust particles contribute 31% of the primary PM2.5 [airborne particulate matter (PM) having an aerodynamic diameter smaller than or equal to 2.5 µm] emissions. The ratio of gasoline vehicle to diesel vehicle exhaust PM2.5 emissions in this inventory is 0.27. The Northern Front Range Air Quality Study (NFRAQS) conducted receptor modeling for ambient PM2.5 at Welby for the 1997 winter. It was estimated that mobile source exhaust particle emissions contributed 58% of the primary PM2.5 at that site and that the ratio of gasoline exhaust to diesel exhaust PM2.5 was 2.8. The factor of 10 difference in estimates of the relative contributions of primary gasoline and diesel exhaust emissions to the ambient PM2.5 burden indicates that more work is needed to determine the actual contribution of these sources to the ambient winter Denver area PM2.5 concentrations. A chassis dynamometer study of exhaust PM emissions from in-use, light-duty gasoline and diesel vehicles was conducted in the summer of 1996 and the winter of 1997 in the Denver area as part of the NFRAQS (2, 3). A total of 197 vehicles were tested, including 86 at winter ambient temperature. All vehicles were tested using the Federal Test Procedure Urban Dynamometer Driving Schedule (FTP). Exhaust PM emission rates from this study were used by the RAQC in their metro Denver emission inventory. The PM chemical composition from a subset of vehicles was used to create source profiles used in the receptor modeling (4). While this is the most complete in-use vehicle PM study conducted to date, significant questions remain regarding in-use light-duty gasoline vehicle (LDGV) PM emissions. First, only nine 1991-1996 LDGVs were tested at winter ambient temperatures. These vehicles had an average exhaust PM10 (PM having an aerodynamic diameter smaller than or equal to 10 µm) emission rate of 2.8 mg/mi at 60 °F and of 24.9 mg/mi at the average 38 °F ambient temperature. The cold temperature PM emission rate was greatly influenced by two light-duty trucks, whose removal lowered the average PM10 emission rate to 7.1 mg/mi. Thus, the average cold PM emission rate for in-use 1991-1998 vehicles is highly uncertain. Second, average PM emission rates from LDGVs tested in the winter at 60 °F were lower than those from the summer testing at 72 °F. This difference is likely due to the use of a winter oxygenated fuel (5). However, since the vehicle fleets were not matched in the two portions of the study, this hypothesis needs verification. Third, it is well-documented that the FTP is deficient in the high-speed, high-load driving that occurs on the road. Vehicles operated under high-speed and high-load conditions can experience short periods of fuel enrichment that are likely to increase PM emission rates. Therefore, PM emission rates need to be determined for driving cycles other than the FTP. In addition, particle size, particle number, and PM chemical composition need to be determined as a function of driving cycle for comparison to the NFRAQS data. The Colorado Department of Public Health and Environment (CDPHE) conducted an oxygenated fuel performance evaluation study during 1998 (6). The study determined the impact of oxygenated fuels on the regulated emissions from 24 properly functioning, late-model LDGVs and six highemission LDGVs. The vehicles were tested on a chassis dynamometer in a cold cell at 35 °F using the standard cold start FTP, a hot-stabilized Unified Cycle, and the REP05. All vehicles were tested with an oxygenated and a nonoxygenated fuel typical of that used in the Denver area. PM10 emission rates were determined for all tests. In addition, under the Coordinating Research Council sponsorship, more detailed PM measurements were made for 12 of the vehicles, including 10.1021/es0010554 CCC: $20.00
2001 American Chemical Society Published on Web 11/22/2000
particle size distribution and PM chemical composition. Thus, this program was able to address the uncertainties from the NFRAQS study. Results from the PM portion of the study are presented in this paper.
Experimental Section The study was conducted at the CDPHE Aurora Emissions Technical Center in the eastern Denver, CO, metropolitan area starting in May 1998. Vehicle Recruitment and Testing. This program focused on in-use, late model (1990 or later model year) LDGVs. To represent as large a fraction of the in-use fleet as practical, the enhanced I/M database was examined to identify the most abundant makes and models of vehicles in the Denver metropolitan area. Letters were then sent to vehicle owners seeking voluntary participation in the program in exchange for a loaner vehicle and a monetary incentive. Twenty-four properly functioning vehicles were recruited: 12 in the 19901995 model year (MY) range with Federal tier 0 emissions packages and 12 in the 1994-1997 MY range with Federal tier 1 emission packages. Both groups of vehicles were further divided into 8 cars and 4 trucks. Six high emitters were recruited from I/M 240 test lanes, based on their CO emission rate. A minimum FTP CO emission rate of 30 g/mi was chosen for the program, based on the enhanced I/M cut point of 20 g/mi. Three of these vehicles were trucks. Five were tier 0 vehicles, and one was a tier 1 vehicle. Upon arrival at the laboratory, each vehicle was given a complete inspection to ensure that it was safe to test. In addition, all vehicles received an oil and filter change to ensure that adequate lubricant was present and to standadize the composition of the oil, which can contribute to exhaust PM. Last, the on-board fuel was drained and changed to the initial test fuel according to the program test design matrix. Vehicle preconditioning on the fuels followed the Auto/Oil Air Quality Improvement Research Program procedure (7) and are available in the final reports for this program (6, 8). After preconditioning, vehicles were soaked at 35 °F overnight. A cold start FTP was followed by a hot, running start Unified Cycle (UC), and a hot running start REP05. The Unified Cycle is normally run as a cold start test. However, the time required to soak the vehicle for a cold start UC test precluded its use in this program. Therefore, the UC began immediately upon completion of the measurement of the FTP phase 3 bag sample. The vehicle was left idling during the measurement period. The order in which the fuels were tested was randomized to minimize any conditioning effects on the fuel comparisons. The FTP driving cycle has been used for vehicle certification since 1975. This cycle represents urban driving. It has an average speed of 19.6 mph, a top speed of 56.7 mph, and maximum acceleration and deceleration rates of 3.3 mph/s. The FTP has three phases that represent cold start, hot stabilized, and hot start operation. Emission results are calculated by weighting the results obtained from the three phases. The REP05 driving cycle was developed to represent most of the driving that occurs outside of FTP conditions. The average speed is 51.5 mph, the maximum speed is 80.3 mph, and the peak acceleration rate is 8.48 mph/s. The California Air Resources Board has developed the Unified Cycle, which incorporates aspects of both the FTP and the REP05. When this cycle is conducted with a cold start, the emissions are a weighted average from three phases. In this study, however, since the UC was a hot running start, the emission rates are straight averages of total grams emitted per mile driven. The UC has an average speed of 24.6 mph, a maximum speed of 67.2 mph, and a maximum acceleration rate of 6.9 mph/s.
Fuel. Single batches of commercially available mid-grade winter fuels were supplied by Conoco. The oxygenated fuel used ethanol, since it was anticipated that 95% of the in-use fuel in the Front Range area would use ethanol. The nonoxygenated fuel was selected to be as close as practical in composition to the oxygenated fuel. Fuel properties are available in the final report for this program (8). The fuel sulfur concentration for both fuels was 0.019 wt %. Emission Measurements. The constant volume sampler was operated at a flow rate of 600 cfm to accommodate the high exhaust flows that occur in the REP05 and the UC. Regulated emissions (HC, CO, and NOx) and CO2 were measured in bag samples during all tests. Each properly functioning vehicle was tested at least twice on each cycle/ fuel combination. High emitters were tested a minimum of three times. PM samples were collected from all vehicles through two isokinetic probes inserted into the dilution tunnel. These probes were connected to University Research Glassware PM10 cyclones, which operated at a flow rate of 28.3 L min-1. Each cyclone outlet was attached to a 26.5-cm-long straight tube for flow straightening. The tube, in turn, was connected to a Y fitting, which accommodated two filter holders. Solenoid valves activated by the test computer systems were used to switch between filters. PM10 samples were collected simultaneously on both 37 mm diameter, 2.0 µm pore size Gelman Teflo and 37 mm diameter Pallflex Tissue Quartz 2500 QAT-UP filters. Quartz filters were prefired at 900 °C for 3 h to remove any carbon. Separate Teflo filters were collected for the three FTP phases, while a single quartz filter was collected over all three phases. A single filter sample was collected for each of the UC and REP05 tests. For some tests, the FTP and REP05 quartz filters were backed by a PUF/XAD vapor-phase trap to capture volatile polynuclear aromatic hydrocarbons (PAHs), oxygenated PAHs, steranes, and hopanes. Particle size distributions were determined with MSP Corp. Micro Orifice Uniform Deposit Impactors (MOUDI) on the FTP and the REP05. The impactor was run continuously for the entire FTP rather than by phase. Aluminum foils were used as collection media. The top stage foil was coated with grease to remove the largest particles. This was followed by ungreased stages with cutpoints of 12.2, 3.0, 1.2, and 0.12 µm and the final filter. This reduced set of stages was employed to maximize the mass collected per stage. Particle size distributions may be influenced by dilution ratio. The samples in this study were drawn directly from the dilution tunnel and thus do not represent real-world dilution conditions. Particle number distributions were obtained with a Dekati Inc. electrical low-pressure impactor (ELPI). This instrument has a measurement range from 0.03 to 10 µm using 13 impactor stages. The response from all stages is scanned once per second, thereby providing a near-continuous measurement of particle size distribution during the tests. Continuous particle count readings were obtained with a TSI model 3025A-S ultrafine condensation particle counter (UCPC). This instrument has a response time of less than 1 s and has a 50% detection of 3-nm size particles and a 90% detection of 5-nm particles. It was necessary to dilute the sample drawn from the dilution tunnel to prevent saturation of this instrument. This was accomplished by using diluters made from cartridge filters with different size “pinholes”. The dilution ratio of each diluter was determined by placing it in the output stream of an aerosol generator and using an electrical aerosol analyzer to determine the change in particle number. Dilution ratios were 2.2:1, 18.5:1, and 242:1. Higher dilution was obtained by using diluters in series. During the study, the diluters did not provide a constant dilution ratio at the low flows used by the UCPC. Thus, all particle number results have to be considered approximate. Particle number VOL. 35, NO. 1, 2001 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
27
TABLE 1. Average Age, Mileage, and FTP Emission Rate on the Nonoxygenated Fuel emission class tier 0 tier 0 tier 1 tier 1 high emitter high emitter
vehicle av age av HC type no. (yr) mileage (g/mi) car truck car truck car truck
8 4 8 4 3 3
5.9 5.8 2.0 1.5 5.3 6.0
CO NOx (g/mi) (g/mi)
72 800 0.72 11.3 66 700 1.02 18.2 30 300 0.33 4.68 24 600 0.33 2.34 101 000 4.04 65.5 78 800 12.9 211
0.94 1.3 0.36 0.46 0.87 0.36
measurements can be highly dependent on dilution conditions. A standard means of diluting to best represent realworld conditions has not been developed to date. Thus, these numbers may not reflect real-world emission rates. Sample Analysis. PM samples were analyzed at the Desert Research Institute (DRI). “Organic” carbon (OC) and “elemental” carbon (EC) were determined on 0.512 cm2 punches removed from the quartz filters using the DRI thermal optical reflectance (TOR) technique (9). Sulfate and nitrate were determined by ion chromatography (IC). Teflon filters were extracted in 15 mL of 40:60 water/isopropyl alcohol in an ultrasonic bath for 60 min and then in a mechanical shaker for 60 min. The ultrasonic bath temperature was maintained below 27 °C. Teflon and Pallflex filters were analyzed for 38 elements by energy-dispersive X-ray fluorescence (XRF) analysis (10). Seventy-five PAHs and 10 oxy-PAHs were determined from the quartz filter/PUF/XAD samples by GC/MS in selected ion monitoring mode. Samples have deuterated internal standards added before extraction to measure the recovery efficiency of the process. Hopanes and steranes were analyzed from these extracts by GC/MS as well. Authentic standards were available for five of the hopanes and steranes. The remaining hopanes and steranes were identified based on their mass spectra and retention time comparison with data available in the literature (11, 12). Deuterated PAHs were used as surrogate internal standards.
Results The number of vehicles tested, their average age, average odometer reading, and average FTP-regulated emissions rates using the nonoxygenated fuels are given in Table 1. The data should not be compared to the FTP tier 0 and tier 1 standards since these apply to room temperature emissions only. Tier 1 vehicles are also required to meet a low-temperature (20 °F) CO standard, which may account for some of the difference in tier 0 and tier 1 FTP CO emissions in this study. Three of the high emission vehicles had emission rates near the 30 g/mi CO minimum (33.4, 35.0, and 37.9 g/mi), two of which were cars. The other three high emitters had CO emission rates of 125, 243, and 354 g/mi. PM Emission Rates. Table 2 gives the average PM emission rates for each vehicle/fuel combination on the three driving cycles. The average FTP PM emission rate for the 24 properly functioning, 1990-1997 vehicles operated on the oxygenated fuel at 35 °F was 6.1 mg/mi. This is considerably lower than the 24.9 mg/mi PM emission rate found for 9 vehicles in the NRAQS program (2) but is in agreement with the 7.1 mg average obtained with two high-emitting trucks removed from the 9 vehicle average. The 33 vehicle average from both programs, including the two high-emitting trucks, is 11.2 mg/mi. The properly functioning tier 1 vehicles had significantly lower FTP PM emission rates than the tier 0 vehicles, 3.8 vs 8.3 mg/mi for the oxygenated fuel. It cannot be determined from this study if the difference is due to the improved emission control technology utilized by newer 28
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 35, NO. 1, 2001
TABLE 2. Average PM Emission Rates for All Vehicles emission class
vehicle type
tier 0
car
tier 0
truck
tier 1
car
tier 1
truck
high emitter
car
high emitter
truck
a
driving cycle
nonoxy fuel PM (mg/mi)
oxy fuel PM (mg/mi)
FTP hot UC REP05 FTP hot UC REP05 FTP hot UC REP05 FTP hot UC REP05 FTP hot UC REP05 FTP hot UC REP05
10.3 4.01 14.7 19.1 6.93 25.3 4.43 1.72 7.46 6.58 2.54 11.4 27.4 26.2 56.9 268 69.0 52.6
7.00 5.18 15.3 10.9 5.50 18.8 3.34 1.20 6.23 4.72 2.92 14.0a 27.2 27.2 49.6 76.5 28.9 24.2
High value omitted from average.
vehicles or is due to in-use emissions deterioration with the older vehicles. The average PM emission rate on the REP05 was 12.7 mg/mi using the oxygenated fuel. Since the REP05 is a hot running test, it is best compared to phases 2 and 3 of the FTP, which averaged only 1.4 and 2.3 mg/mi, respectively. Clearly, the high speed, high load conditions in the REP05 lead to increased PM emissions. This indicates that the FTP driving cycle under predicts real-world PM emissions, to the extent that it lacks the high speed and high load driving incorporated into the REP05. Assuming that 20% of the onroad vehicle miles traveled occur under REP05 type conditions, The REP05 results suggest that the FTP underpredicts in-use PM emission rates by approximately 1 mg/mi for these vehicles. The UC cycle integrates the FTP and REP05 driving modes into a single cycle in a manner that represents in-use driving in southern California. The average hot UC PM emissions for the 24 properly functioning vehicles operated on the oxygenated fuel was 3.6 mg/mi: 5.3 mg/mi for the tier 0 vehicles and 1.8 mg/mi for the tier 1 vehicles. Comparing the UC to the FTP phase 2 emission rate suggests that, for emission inventory purposes, the FTP emission rate should be raised by approximately 2 mg/mi to incorporate the effects of high-speed, high-load driving. This adjustment results in an average in-use PM emission rate of approximately 8 mg/ mi for properly functioning tier 0 and tier 1 vehicles operated on oxygenated fuel at 35 °F. This is much lower than the 24.9 mg/mi used to estimate the Northern Front Range inventory in the Denver area (2). The NFRAQS study suggested that oxygenated fuels significantly lowered FTP PM emissions. This study confirms that observation. Oxygenated fuel use lowered the FTP PM average emission rate of the properly functioning vehicles 34%. The effect of fuel was examined for each phase of the FTP. The average phase 1 PM emission rates were 36.0 and 22.6 mg/mi for the nonoxygenated and oxygenated fuels, respectively, for properly functioning vehicles. A t-test using paired data from individual vehicles showed that this difference is significant at the 99% confidence level. FTP phase 2, fleet-average PM emission rates were 1.4 mg/mi for both fuels. FTP phase 3, fleet-average PM emission rates were 3.8 and 2.3 mg/mi for the nonoxygenated and oxygenated fuels, respectively. The difference is due to two tier 0 trucks. The t-test showed that fuel differences were not statistically significant for either phase 2 or phase 3. The high PM emitters fell into two categories. Three vehicles that had moderate
FTP PM emission rates had most of the PM emissions in phase 1. For these vehicles, phase 1 FTP emissions were reduced when using the oxygenated fuel. However, the impact of oxygenated fuel on phases 2 and 3 was not consistent. The other three high CO emitters had high PM emission rates. Use of the oxygenated fuel reduced PM emissions in all three FTP phases for these vehicles. It should also be emphasized that most of the FTP PM emissions from the properly functioning vehicles occurs during the cold start, an average of 22.6 mg/mi for the oxygenated fuel tests. Thus, for emission inventory purposes, it is important that the proper number of cold starts be known. Switching from the nonoxygenated to the oxygenated fuel had modest effect on PM emissions on the hot UC and the REP05. On the UC, four of the vehicles showed significant decreases in the PM emission rate, one vehicle had no change, and one vehicle had an increased PM emission rate. There was considerable variability in run-to-run PM emission rates on the hot UC, in part due to the generally low emission rates. For the REP05, four vehicles showed a significant decrease in PM emission rate when using oxygenated fuel vs the nonoxygenated fuel. It may be that the oxygenated fuel reduces PM emissions during the brief enrichment events that occur during high-speed, high-load driving. PM emissions in the replicate tests generally were lower on the repeat test than the initial test. This may obscure small emission rate fuel effects. Particle Size. The FTP PM12 mass emission rate determined from MOUDI impactor samples was compared to the PM10 mass determined from the Teflon filters for 28 runs. The correlation was excellent, with an R 2 of 0.96. The slope of 1.23 indicates that the MOUDI consistently collected less mass than the Teflon filters. Results were similar for the 25 REP05 samples, which had an R 2 of 0.99 and a slope of 1.24. The reason for the difference in emission rates is not known but may be due to a combination of particle losses within the impactor and volatilization of some of the particulate from the impactor at the reduced pressure. The fraction of the PM mass collected by the MOUDI impactor that was below 12.2, 3.0, 1.2, and 0.12 µm was determined for both the FTP and the REP05 as a function of emission class and fuel. For the FTP, an average of 97.4, 94.3, and 92.2% of the mass was present in particles smaller than 12.2, 3.0, and 1.2 µm, respectively. An average of 33.2% of the mass was present in the ultrafine particle size range, i.e., below 0.12 µm. Switching between fuels had little effect on the particle size distribution. The REP05 particle size distribution is significantly different from that of the FTP. An average of 95.1, 88.7, and 83.6% of the mass was smaller than 12.2, 3.0, and 1.2 µm, respectively. The biggest difference is in the ultrafine percentage, which averaged 69.2% for the REP05. As with the FTP, switching fuels had little impact on the particle size distribution. The t-tests were performed on the 24 paired FTP and REP05 tests to determine if the observed differences in percent mass in the four size ranges were significant. The differences were significant at a greater than 95% confidence level for all four size ranges. The most significant finding is that the majority of the REP05 PM mass was present in the ultrafine particle range. The dilution ratio can have a major impact on the formation of particles in the ultrafine size range. Since the vehicle emissions tunnel used in this work has a much lower dilution ratio than would occur behind a moving vehicle, the fraction of ultrafine particle mass may not be representative of what enters the atmosphere. On the other hand, it will be shown below that sulfate emissions were much higher on the REP05 than the FTP and may account for much of the ultrafine PM. The ELPI average particle number distributions for stages 1-9 (0.03-1.7 µm) for the three test cycles are shown in Figure 1. There were 58, 60, and 53 FTP, hot UC, and REP05
FIGURE 1. ELPI particle number distribution. tests, respectively. The ELPI operates by counting the number of particles collected on each stage. A few small particles collected on upper stages can lead to very large errors in the mass determination. For this reason, mass distributions will not be discussed. It should also be kept in mind that no particles smaller than 0.03 µm are counted. Thus, most if not all of the nanoparticle size range is missed. While the ELPI indicates that the REP05 has a higher fraction of particles smaller than 0.11 µm than does the FTP, the difference is modest as compared to the differences suggested by the MOUDI results. This could be due to differences in particle bounce between driving cycles for the MOUDI, or it may be due to the inability of the ELPI to detect the nanoparticle mode. CPC Particle Number. Given the uncertainties in both the dilution ratio during the measurement and the proper way of diluting exhaust samples to represent real-world particle size distributions, these results should be interpreted with caution. The variability in the particle number measurement was examined by ratioing the particle number emission rate for the first and last repeat tests of various vehicle/fuel combinations. The average ratios for the FTP, hot UC, and REP05 were 1.32, 4.0, and 1.47, respectively. The fleet average particle number emission rates for the nonoxygenated fuel were 2.7 × 1013, 1.1 × 1013, and 38 × 1013 particles/mi for the FTP, hot UC, and REP05, respectively. The corresponding fleet averages for the oxygenated fuel were 3.1 × 1013, 0.63 × 1013, and 19 × 1013 particles/mi. Since the REP05 mass emission rate averaged 1.5 times that of the FTP, the higher particle counts on the REP05 cannot be attributed solely to increases in the particle mass. This suggests that the REP05 did have a larger fraction of particles in the smallest size ranges than the FTP, consistent with the MOUDI mass data. During the NFRAQS study, particle number was determined with an electrical aerosol analyzer. The average particle count for FTP tests of low PM emission vehicles (average PM emission rate of 3.3 mg/mi) at a test temperature of 75 °F was 1.8 × 1013. This is similar to the 2.7 × 1013 particles/mi found in this study for FTP emissions of vehicles that averaged 9.2 mg/mi PM. Diesel vehicles with mass PM emission rates of 157-930 mg/mi were found to emit 3.2-9.7 × 1014 VOL. 35, NO. 1, 2001 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
29
TABLE 3. Carbon Emission Rates emission class
vehicle type
tier 0
car
tier 0
truck
tier 1
car
tier 1
truck
high emitter
car
a
drive cycle
nonoxy PM (mg/mi)
nonoxy TC (mg/mi)
nonoxy %OC
oxy PM (mg/mi)
oxy TC (mg/mi)
oxy %OC
FTPa REP05 FTPa REP05 FTPa REP05 FTPa REP05 FTPa REP05
17.0 21.1 26.7 29.7 7.12 5.09 21.1 9.78 35.6 84.9
16.9 14.1 27.3 19.1 6.69 3.72 21.4 4.53 29.4 20.7
36.6 51.8 23.6 32.5 32.2 39.6 28.1 59.4 25.7 36.0
11.2 14.1 22.4 22.8 4.58 3.40 15.9
11.8 9.30 20.2 13.8 3.68 2.43 15.7 5.32 22.7 7.52
39.4 40.6 29.0 47.0 41.9 41.1 24.7 62.7 34.0 38.8
FTP emission rates are for the three phases integrated into one sample without weighting.
particles/mi during the NFRAQS study. It is clear that large numbers of particles are emitted by light-duty gasoline vehicles under laboratory conditions. Particle Composition. PM composition was determined for 12 vehicles: six tier 0, four tier 1, and two high emitters. Only one high emitter had data on the REP05. Analysis of UC samples was limited to anions and elements for the oxygenated fuel tests since it was hypothesized that cycle related differences in composition, if present, would be found by comparing the FTP to the REP05. Carbon. The average total carbon (TC) emission rate and the average OC percent of the TC are given in Table 3 for the FTP and the REP05. These results are for the integrated FTP and thus are not directly comparable to the weighted FTP emission rates. For this reason, the integrated PM mass emission rate is given as well. The average uncertainties in the OC and EC measurements were equivalent to 0.25 and 1.0 mg/mi, respectively. Average TC emission rates for the FTP are very close to the PM mass emission rates. A linear regression between the FTP PM mass emission rate and the TC emission rate had an R 2 of 0.97. The slope, 1.21, suggests that most of the mass is accounted for by TC. Approximately one-third of the FTP TC is present as OC. Since OC contains mass due to a variety of elements (i.e., hydrogen and oxygen), it is commonly assumed that source sample OC measurements should be multiplied by a factor of 1.2. The correlation between the REP05 PM mass emission rate and the TC mass emission rate was also excellent with an R 2 of 0.97. The slope was 1.48, indicating that TC accounted for considerably less of the mass for the REP05 than for the FTP. Some of the difference may be due to sulfate and the associated water. This is discussed below. Table 3 also shows that the average %OC is almost always higher for the REP05 than the FTP for both fuels. For the properly functioning vehicles on the oxygenated fuel, the TC emission rate was 12.3 and 8.18 mg/mi for the FTP and the REP05, respectively. The corresponding fractions of organic carbon were 36 and 45%, respectively. A paired t-test for both oxygenated and nonoxygenated fuels indicated that the difference in OC fraction was statistically significant at the 99% level. The 36% OC on the FTP is a lower percentage than found for most low PM emitting vehicles in the previous Denver study but agrees with the results from 75 °F tests conducted in the SCAQMD (South Coast Air Quality Management District), which also used an oxygenated fuel (3, 13, 14). This split between OC and EC may be important since it is one of the differences in emissions that help distinguish between gasoline and diesel vehicle emissions. The 36% OC found in this study is similar to the %OC used for diesels. Anions. Chloride, nitrate, and sulfate PM emission rates were determined. No differences were found in the rates for the two fuels. Thus, only the results for the oxygenated fuel 30
9
29.1 21.9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 35, NO. 1, 2001
are discussed. Chloride emission rates were low, averaging 0.035, 0.025, and 0.066 mg/mi for the properly functioning vehicles operated on the oxygenated fuel over the FTP, hot UC, and REP05, respectively. Uncertainties in these measurements averaged 0.034, 0.007, and 0.010 mg/mi, respectively. The corresponding nitrate emission rates were 0.043, 0.015, and 0.017 mg/mi with uncertainties of 0.038, 0.010, and 0.006 mg/mi, respectively. Sulfate emission rates for the properly functioning vehicles were higher, averaging 0.070, 0.10, and 0.86 mg/mi for the FTP, hot UC, and REP05 with corresponding uncertainties of 0.033, 0.011, and 0.030 mg/ mi. Sulfate emission rates were consistently higher on the REP05 than the FTP or the hot UC. The one high emitter had a sulfate emission rate of 5.38 mg/mi on the REP05. The sum of the mass emission rates of the three anions was compared to the PM mass emission rate. For the FTP, they averaged only 2.1% of the total mass. However, for the REP05, the average was 10.5%. If the sulfate is emitted as sulfuric acid, it will have associated with it seven water molecules for each sulfate ion at the temperature and humidity used in filter weighing rooms. This raises the mass emission rate by a factor of 2.3. Since most of the anion mass is associated with sulfate, the sulfate and associated water may average as much as 20% of REP05 PM mass. Because sulfuric acid tends to nucleate very rapidly forming small particles, it may account for the increase in particle number observed with the CPC for the REP05 and the relatively high fraction of the mass in the ultrafine size range found by the MOUDI. Since the REP05 also had higher OC than the FTP, the possibility that OC is contributing to the higher number of particles found for the REP05 under these laboratory test conditions cannot be ruled out. Elements. The average FTP, REP05, and hot UC emission rates for the 14 elements detected by XRF are given in Table 4 for the tests conducted with oxygenated fuel. Results for the nonoxygenated fuel were similar and thus will not be discussed. In Table 4, the average tier 0 and tier 1 emission rates have been averaged. Tier 0 vehicle rates consistently averaged higher than those from tier 1 vehicles. For the FTP, Si, S, Fe, and Mg were the four most abundant elements, accounting for an average of 71% of the elemental XRF mass. For the REP05, sulfur was the most abundant element, followed by Si, Fe, and Zn. These four elements averaged 80% of the XRF mass. For the hot UC, Fe, S, Si, and Zn accounted for 63% of the XRF element mass. The sum of the mass emission rates of the elements is given near the bottom of the tables, followed by the corresponding FTP PM mass emission rate and the fraction of the mass accounted for by the XRF elements. Considering both fuels and tier 0 and tier 1 vehicles separately, the elements accounted for 2.7-4.0% of the FTP PM emissions, 12.3-24.8% of the REP05 PM emissions, and 7.2-31.7% of the hot UC emissions. Increased reentrainment of particles from the vehicle and the sampling
TABLE 4. FTP, UC, and REP05 Emission Rates of XRF Elements for the Oxy Fuel FTP (mg/mi)
REP05 (mg/mi)
UC (mg/mi)
element
tiers 0 & 1
high emitter
tiers 0 & 1
high emitter
tiers 0 & 1
high emitter
magnesium aluminum silicon phosphorus sulfur chlorine calcium chromium iron nickel copper zinc bromine lead
0.030 0.017 0.061 0.018 0.037 0.010 0.015 0.0008 0.026 0.0016 0.0034 0.019 0.00029 0.0020
0.044 0.015 0.099 0.051 0.065 0.006 0.033 0.0131 0.182 0.0028 0.0080 0.084 0.00003 0.011
0.0185 0.015 0.719 0.053 0.3385 0.092 0.075 0.00575 0.2405 0.0045 0.00815 0.1015 0.0057 0.005
0.011 0.008 1.457 0.038 1.681 0.081 0.070 0.0299 0.359 0.0044 0.0086 0.143 0.0289 0.016
0.0215 0.0085 0.052 0.0265 0.0575 0.014 0.0315 0.0032 0.083 0.00275 0.0037 0.0395 0.00065 0.002
0.006 0.025 0.144 0.038 0.084 0.010 0.043 0.0176 0.232 0.0032 0.0043 0.065 0.0039 0.009
total PM (mg/mi) PM fraction
0.24 8.54 0.028
0.346 6.06 0.057
0.684 5.150 0.133
0.615 20.500 0.030
1.68 10.9 0.154
3.936 21.900 0.180
TABLE 5. Average Emission Rates of Selected PAH Compounds for the Oxy Fuel Tests tier 0 (mg/mi)
tier 1 (mg/mi)
high emitter (mg/mi)
compound
FTP
REP05
FTP
REP05
FTP
REP05
naphthalene 2-methylnaphthalene 1-methylnaphthalene 1,3+1,6+1,7-dimethylnaphthalene dibenzofuran acenaphthylene acenaphthene fluorene phenanthrene 9-fluorenone xanthone acenaphthenequinone perinaphthenone methylphenanthrenes anthraquinone dimethylphenanthrenes anthracene fluoranthene pyrene 9-anthraldehyde benz[a]anthracene chrysene benzanthrone 1,4-chrysenequinone benzo[b+j+k]fluoranthene benzo[e]pyrene perylene benzo[a]pyrene 9,10-dihydrobenzo[a]pyrene indeno[123-cd]pyrene dibenz[ah+ac]anthracene benzo[b]chrysene
9.394 3.787 2.006 1.332 0.0835 3.2740 0.5332 0.3797 0.4690 0.0679 0.0122 0.0205 0.0689 0.2476 0.0269 0.08793 0.07832 0.08116 0.08873 0.03186 0.01114 0.01217 0.02223 0.02567 0.01383 0.00509 0.00311 0.00947 0.05167 0.01515 0.01946 0.00730
1.552 0.409 0.211 0.122 0.0317 0.0845 0.0054 0.0299 0.1216 0.0890 0.0068 0.0128 0.0220 0.0657 0.0160 0.02887 0.01750 0.03515 0.03564 0.01828 0.00358 0.00563 0.01097 0.01499 0.00425 0.00233 0.00144 0.00145 0.02583 0.00560 0.01080 0.00341
4.327 1.831 0.962 0.582 0.0242 0.9746 0.0463 0.0602 0.1286 0.0327 0.0070 0.0207 0.0238 0.038 0.0174 0.00645 0.02111 0.00995 0.01119 0.03020 0.00396 0.00581 0.01023 0.02527 0.00422 0.00291 0.00257 0.00288 0.04034 0.00837 0.01784 0.00217
0.898 0.298 0.160 0.096 0.0144 0.0195 0.0020 0.0078 0.0411 0.0308 0.0037 0.0117 0.0087 0.0177 0.0103 0.005 0.00656 0.00533 0.00504 0.01792 0.00121 0.00300 0.00609 0.01495 0.00083 0.00159 0.00135 0.00106 0.02330 0.00391 0.01043 0.00111
18.982 5.839 3.157 1.346 0.2846 0.3386 0.0517 0.2221 0.5206 0.0807 0.0208 0.0498 0.0502 0.2022 0.0463 0.02168 0.10384 0.05579 0.05635 0.07247 0.00847 0.01429 0.04402 0.05780 0.01315 0.00716 0.00533 0.00507 0.08980 0.02078 0.04172 0.00730
8.095 2.482 1.353 0.637 0.1113 0.0625 0.0203 0.0958 0.0852 0.0910 0.0133 0.0274 0.0263 0.0679 0.0286 0.01779 0.01711 0.01641 0.01510 0.03993 0.00362 0.00784 0.02407 0.03269 0.00410 0.00369 0.00292 0.00219 0.04967 0.00945 0.02268 0.00378
system under the high-flow conditions encountered during REP05 and UC testing may account for some of the differences in elemental abundances between test cycles. Comparisons were made between the XRF sulfur and the extractable sulfate sulfur. The linear regression correlation had an R 2 of 0.74. The slope of 0.51 indicates that only half of the sulfur present is accounted for by the sulfate. Results were similar for the hot UC, which had a linear regression R 2 of 0.83 and a slope of 0.56. Interestingly, the linear regression for the REP05, had an R 2 of 0.97 and a slope of 1.02. Apparently, the increased sulfur emission on the REP05 as compared to the FTP is primarily due to sulfate. The XRF chlorine measurement was compared to the IC chloride
measurement for all instances where both values were significantly above the uncertainty. The linear regression had an R 2 of 0.79 and a slope of 1.42, indicating that chloride accounts for approximately two-thirds of the chlorine in the PM. PAHs. PAH emission rates were measured primarily to help create source profiles, although PAHs are also air toxics. The PAHs were measured on composited samples. A total of 24 samples were analyzed, 12 tier 0, 8 tier 1 and 4 highemitter samples. Each sample, except those for the high emitter, consisted of two vehicles with similar emission rates. Only one high emitter had a compete set of PAH samples for analysis. VOL. 35, NO. 1, 2001 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
31
Table 5 gives the average emission rates of selected PAHs, including the oxy-PAH for the oxygenated fuel tests. More details are available in ref 8, or the complete data set can be requested from the Coordinating Research Council. The total PAH emission rate is dominated by the relatively large naphthalene and methlynaphthalene emissions. The FTP PAH emission rates were 7.5, 5.4, and 2.3 times higher than those from the REP05 for the tier 0, tier 1, and high-emitter samples, respectively. PAH emission rates were determined for the nonoxygenated fuel tests as well. The nonoxygenated total PAH emission rates were consistently higher than those from the oxygenated fuel, by an average factor of 1.55. An examination of individual PAH emission rates showed that the distribution of PAH compounds was not significantly different for the two fuels. The effect of driving cycle on the PAH emission rates of individual compounds was examined by averaging together all the samples in each of the three emission categories: tier 0, tier 1, and high emitter. Both the tier 0 and tier 1 vehicles were found to have a larger proportion of naphthalene, methlynaphthalene, and the di- and trisubstituted methlylnaphthalenes on the FTP than on the REP05. Steranes and Hopanes. Hopanes and steranes are of interest as possible marker compounds for mobile source PM emissions. These compounds were measured in the same sample extracts as the PAHs. Most of the sterane and hopanes were below detection limits. There was a significant background correction due the presence of these species in the media blank extracts. Given the relatively few number of samples with measurable quantities of these compounds, the main conclusion is that, at current analytical sensitivities, these compounds cannot be used as source profile fitting species for low emitting vehicles.
Acknowledgments We gratefully acknowledge the support of the Coordinating Research Council, who co-sponsored this project as CRC Project E-46. We also gratefully acknowledge the assistance of Bev Lynn from the CDPHE Information Technology Section for her work with Colorado’s Motor Vehicle Registration database, Jim Sidebottom from the CDPHE Mobile Sources Section for providing daily updates on high emitters identified from the enhanced I/M database, and the following CDPHE staff: Dawn Mirabile for recruiting vehicles; Steve Sargent, Thad Pyzdrowski, and Michael Waida for their continued conscientious efforts in conducting the vehicle emission tests; and Jerry Lyons for preparing the vehicles and de-prepping them for return to owners. We also thank Dennis Creamer and Conoco for supplying the test fuel.
32
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 35, NO. 1, 2001
Literature Cited (1) Watson, J. G.; Fujita, E.; Chow, J. C.; Zielinska, B.; Richards, L. W.; Neff, W.; Dietrich, D. Northern Front Range Air Quality Study Final Report; Desert Research Institute Document 6580-6858750.1F2; June, 1998. (2) Cadle, S. H.; Mulawa, P. A.; Hunsanger, E. C.; Nelson, K.; Ragazzi, R. A.; Barrett, R.; Gallagher, G.; Lawson, D. R.; Knapp, K. T.; Snow, R. J. Air Waste Manage. Assoc. 1999, 49, PM164PM174. (3) Cadle, S. H.; Mulawa, P. A.; Hunsanger, E. C.; Nelson, K.; Ragazzi, R. A.; Barrett, R.; Gallagher, G.; Lawson, D. R.; Knapp, K. T.; Snow, R. Environ. Sci. Technol. 1999, 33, 2328-2339. (4) Fujita, E.; Watson J. G.; Chow, J. C.; Robinson, N. F.; Richards, L. W.; Kumar N. Northern Front Range Air Quality Study Final Report Volume C: Source Apportionment and Simulation Methods and Evaluation; Desert Research Institute: 1998. (5) Mulawa, P. A.; Cadle, S. H.; Knapp, K.; Zweidinger, R.; Snow, R.; Lucas, R.; Goldbach, J. Environ. Sci. Technol. 1997, 31, 13021307. (6) Ragazzi, R.; Nelson, K. The Impact of a 10% Ethanol Blended Fuel on the Exhasut Emissions of Tier 0 and Tier 1 Light duty Gasoline Vehicles at 35°F; Colorado Department of Public Health and Environment Report; May 1999. (7) Burns V. R.; Benson, J. D.; Hochhauser, A. M.; Koehl, W. J.; Kreucher, W. M.; Reuter, R. M. Description of Auto/Oil Air Quality Improvement Research Program; Society of Automotive Engineers Publication SAE 912320; SAE: Warrendale, PA, 1991. (8) Cadle, S. H.; Mulawa, P. A.; Groblicki, P. J.; Laroo, C.; Ragazzi, R. A.; Nelson, K.; Gallagher, G.; Zielinska B. In-Use Light-Duty Gasoline Vehicle Particulate Matter Emissions on the FTP, REP05, and UC Cycles; CRC Project E-46 Final Report; June 1999. (9) Chow, J. C.; Watson, J. G.; Pritchett, L. C.; Pierson, W. R.; Frazier, C. A.; Purcell, R. G. Atmos. Environ. 1993, 27A, 1185-1201. (10) Chow, J. C.; Watson, J. G. Guidelines for PM-10 Sampling and Analysis Applicable to Receptor Modeling; U.S. Environmental Protection Agency, Office of Air Quality Planning and Standards; Washington, DC, 1994; EPA-452/R-94-009. (11) Zielinska, B.; Fujita E.; Sagebiel, J.; Harshfield, G.; Uberna, E.; Hayes T.; Keene, F. J. Air Waste Manage. Assoc. 1998, 48, 10381050. (12) Wang, Z.; Fingas, M. LC-GC 1995, 13, 950-958. (13) Durbin, T. D.; Norbeck, J. M.; Smith, M. R.; Truex,; T. J. Environ. Sci. Technol. 1999, 33, 4401-4406. (14) Cadle, S. H.; Mulawa, P. A.; Ragazzi, R. A.; Knapp, K. T.; Norbeck, J. M.; Durbin, T. D.; T. J. Truex, T. J.; Whitney, K. A. Exhaust Particulate Matter Emissions From In-Use Passenger Vehicles Recruited in Three Locations: CRC Project E-24; Society of Automotive Engineers Publication 1999-01-1545; SAE: Warrendale, PA, May 1999.
Received for review February 28, 2000. Revised manuscript received September 26, 2000. Accepted October 12, 2000. ES0010554
