Separation of Fine Particulate Matter Emitted from Gasoline and

Department of Civil and Environmental Engineering, Rice University, Houston, ... With regard to the impact of particle emissions on human health, dies...
0 downloads 0 Views 89KB Size
Environ. Sci. Technol. 2003, 37, 3904-3909

Separation of Fine Particulate Matter Emitted from Gasoline and Diesel Vehicles Using Chemical Mass Balancing Techniques M . P . F R A S E R , * ,† B . B U Z C U , † Z . W . Y U E , † G. R. MCGAUGHEY,‡ N. R. DESAI,‡ D. T. ALLEN,‡ R. L. SEILA,§ W. A. LONNEMAN,§ AND R. A. HARLEY| Department of Civil and Environmental Engineering, Rice University, Houston, Texas 77005, Center for Energy and Environmental Resources, University of Texas, Austin, Texas 78758, U.S. Environmental Protection Agency, Research Triangle Park, North Carolina 27711, and Department of Civil and Environmental Engineering, University of California, Berkeley, California 94720

Samples of fine particulate matter were collected in a roadway tunnel near Houston, TX over a period of 4 days during two separate sampling periods: one sampling period from 1200 to 1400 local time and another sampling period from 1600 to 1800 local time. During the two sampling periods, the tunnel traffic contained roughly equivalent numbers of heavy-duty diesel trucks. However, during the late afternoon sampling period, the tunnel contained twice as many light-duty gasoline-powered vehicles. The effect of this shift in the vehicle fleet affects the overall emission index (grams pollutant emitted per kilogram carbon in fuel) for fine particles and fine particulate elemental carbon. Additionally, this shift in the fraction of diesel vehicles in the tunnel is used to determine if the chemical mass balancing techniques used to track emissions from gasoline-powered and diesel-powered emissions accurately separates these two emission categories. The results show that the chemical mass balancing calculations apportion roughly equal amounts of the particulate matter measured to diesel vehicles between the two periods and attribute almost twice as much particulate matter in the late afternoon sampling period to gasoline vehicles. Both of these results are consistent with the traffic volume of gasoline and diesel vehicles in the tunnel in the two separate periods and validate the ability for chemical mass balancing techniques to separate these two primary sources of fine particles.

Introduction With mounting evidence that atmospheric fine particle exposure results in serious detriment to human health, increased attention has been focused on the sources of ambient fine particles (1, 2). With the possibility that emissions from a specific source category may be responsible * Corresponding author phone: (713)348-5883; fax: (713)3485203; e-mail: [email protected]. † Rice University. ‡ University of Texas. § U.S. Environmental Protection Agency. | University of California. 3904

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 37, NO. 17, 2003

for the observed association between elevated fine particle concentrations and human morbidity and excess mortality, the attribution of ambient particulate matter to the emission sources from which they originate is increasingly important (3). One method used to attribute ambient particulate matter to an original source is the chemical mass balance (CMB) model (4). CMB models calculate source contributions by determining the combination of known source profiles required to reconstruct ambient concentrations of chemical constituents. CMB calculations have been applied to both particulate matter and also volatile organic compounds where the ambient mixture of pollutants has been used to determine the relative contribution from different emission sources that have distinct composition (5-7). With regard to the impact of particle emissions on human health, diesel particulate matter has been the focus of numerous studies focusing on carcinogenicity, respiratory morbidity, and mortality (8-12). For this reason, there is a need to be able to accurately quantify the contribution of emissions from diesel sources to ambient pollutant levels. Previous research has focused on using organic compounds as tracers for diesel exhaust (6, 13, 14). The application of the CMB technique is complicated by the fact that there are numerous other sources of the most common compounds emitted in diesel exhaust (15, 16). Organic compounds present in diesel particulate matter, such as the petroleum biomarkers (including the hopanes and stearanes), n-alkanes, and polycyclic aromatic hydrocarbons are also present in the emissions from gasoline-powered vehicles (15, 17). One significant difference between emissions of gasoline and diesel engines is the quantity of elemental carbon (or black carbon) (15, 16). As a result, elemental carbon has been used in combination with organic molecular markers in several studies to separate ambient levels of diesel particulate matter from gasoline vehicle emissions (6, 13, 14). However, there are discrepancies between the results of different applications of chemical mass balancing techniques (18). Studies that use organic molecular markers to trace emissions from both gasoline and diesel powered vehicles suggest that the ambient particle mass attributable to emissions from diesel vehicles is greater that the ambient particle mass attributable to emissions from gasoline vehicles (13, 14). Other studies which rely on bulk compositional measurements and incorporate source profiles from poorly maintained gasoline vehicles and the cold-start of gasoline vehicles estimate that the combined emissions from gasoline vehicles (including poorly maintained vehicles and coldstart emissions) are a greater contribution to ambient particle mass than those of diesel powered vehicles (7). While these discrepancies rise largely from differences in the composition of the source profiles used as inputs to the chemical mass balance model (18, 19), the ability to accurately separate diesel particulate matter is vitally important as it may constitute an important determinant in the overall health effect of ambient particulate matter. For this reason, sampling was conducted in a roadway tunnel in Houston, TX to characterize the emissions of fine particulate matter from the on-road vehicle fleet. Samples were collected during two different times of the day: afternoon rush-hour periods when traffic passing through the tunnel was dominated by light-duty gasoline powered vehicles and mid-day periods where diesel vehicles represented a larger fraction of the tunnel traffic. Samples were analyzed for fine PM mass and composition, and source apportionment calculations were performed on the data 10.1021/es034167e CCC: $25.00

 2003 American Chemical Society Published on Web 07/26/2003

collected inside the tunnel to determine whether the shift in the vehicle fleet could be accurately represented by the CMB calculations. The results of the CMB calculations were then used to calculate separate emission factors for the light-duty gasoline vehicles and the heavy-duty diesel vehicles.

Experimental Methods Tunnel Characteristics. Air samples were collected in the Washburn Tunnel which runs between the cities of Galena Park and Pasadena, TX, underneath the Houston Ship Channel (20). The tunnel has two lanes of traffic (one in each direction) in an undivided single bore that is 895 m long with roadway grades of up to 6% near both ends of the tunnel. Ventilation air is forced into the tunnel at roadway level throughout the length by two large blower fans atop the north entrance to the tunnel. The height of vehicles entering the Washburn tunnel is restricted to 5.1 m prohibiting oversized vehicles from entering the tunnel. Traffic flow is controlled at each end of the tunnel by a round-about intersection, and vehicle traffic typically passes through the tunnel at speeds of 55-70 km h-1 (35-45 mph) with little impediment. Video cameras were used to record the traffic flow through the tunnel during each sampling period. The images from one camera were used to determine the number and type of vehicles traveling in the tunnel. Samples were collected over a period of 4 days between August 29, 2000, and September 1, 2000, during two different sampling times: a mid-day sample from 1200 to 1400 CDT (local time) and an afternoon sample from 1600 to 1800 CDT. While the number of heavy-duty vehicles passing through the tunnel was roughly constant between the two sampling periods, an increase in the light-duty vehicle traffic in the afternoon period decreased the relative fraction of heavyduty vehicles in that sampling period. Pollutant Measurements. Samples collected for the current work included gaseous samples of carbonaceous species over an integration time of 1 h and fine particulate matter integrated over a collection period of 2 h. Two consecutive gas phase samples were collected during the particulate matter sample collection period. Samples were collected from inside the tunnel at a roadside access point accessible from the tunnel control room approximately 50 m from the north exit of the tunnel. Samples of gaseous pollutants from the background ventilation air pumped into the tunnel were collected from the ventilation ducting but space requirements necessitated that fine particulate matter sampling be conducted in the fan room where the air is pulled into the blower fans. Gas-phase pollutants were collected in 6-L Summa stainless steel canisters with a passive canister sampler calibrated to fill from high vacuum in 1 h. Carbon monoxide was measured by GC-FID that employed Ni coated diatomaceous earth catalyst to convert CO to methane via reaction with hydrogen at a temperature of 425 °C. The system was calibrated with a 100 ppmV CO in nitrogen standard (NIST SRM-1679a). Concentrations of CO2 were measured from the canister samples using a nondispersive LI-COR model 6262 gas analyzer which had been zeroed with the headspace vapor from a liquid nitrogen tank and spanned with a 350 ppmV CO2 standard (NIST SRM-1672). Specific chromatographic conditions for these analyses are given elsewhere (20). The two sets of fine particulate matter sampling equipment deployed included a low-volume sampler collecting PM2.5 samples on 47 mm diameter quartz fiber (QF) and Teflon membrane (TM) filters plus a high-volume sampler collecting PM2.5 on a 20 cm × 25 cm QF filter. The lowvolume sampler pulled air through an AIHL Teflon-coated cyclone separator to remove particles larger than 2.5 µm diameter then collected fine particles on three parallel

sampling lines. Flows over the three parallel filters were controlled by calibrated orifice plates with 10 L/min flowing over the two TM filters and 5 L/min flowing over the QF filter for a total flow through the AIHL cyclone of 25 L/min. The high-volume sampler employed an Anderson TSP HighVolume Sampler operating at a flow rate of 1.1 m3 min-1 equipped with a MSP High-Volume Virtual Impactor inlet. Prior to sampling, QF filters were fired at a temperature of 550 °C for at least 2 h to remove sorbed organic material and TM filters were equilibrated in a climate-controlled room at 22 °C and relative humidity between 40% and 60% and weighed on an electronic microbalance. After sampling, filters were sealed in Petri dishes (47 mm diameter filters) and prefired glass jars (20 cm × 25 cm filters) and returned to the laboratory where they were frozen until analysis. TM filters were reweighed after 24 h equilibration at 22 °C and RH between 40% and 60% to determine PM2.5 mass. After reweighing, the TM filters were used to determine concentrations of inorganic metals by inductively coupled plasmamass spectroscopy (21). Punches from QF filters from both the low-volume sampler and the high-volume sampler were analyzed for elemental and organic carbon by the thermal-optical transmission method (22). The concentration of elemental and organic carbon PM2.5 measured from both the low-volume sampler and high-volume sampler were equal to within (15%. Since carbonaceous aerosols constitute a significant portion of total PM2.5, this agreement in the carbon measurements implies that the two separate samplers collected comparable samples. Field blanks of both the TM and QF filters were analyzed and showed no significant biases in the mass or elemental carbon measurements. QF field blanks showed a low blank concentration of organic carbon which was subtracted from all measurements of PM2.5 organic carbon. Samples collected by the high-volume sampler were analyzed for individual organic compounds by following an established analysis procedure (23), which will only be summarized here. Samples were spiked with a known quantity of eight nonpolar perdeuterated recovery standards (n-decane-d10, n-pentadecane-d32, n-tetracosane-d50, nhexatraicontane-d74, acenaphthalene-d10, phenanthrene-d10, chrysene-d12, dibenz[a,h]anthracene-d14) and one polar perdeuterated recovery standard (hexadecanoic acid-d31) prior to extraction and allowed to air-dry. The filters were then cut into approximately 5 cm × 5 cm squares and returned to the storage jar for extraction which was performed with two 35 mL aliquots of hexane and three 35 mL aliquots of 2:1 benzene:2-propanol under mild ultrasonication for 10 min. Prior to use, the benzene was redistilled on site in a glass distillation apparatus. The extracts were combined, reduced by rotary evaporation under mild vacuum to a volume of approximately 5 mL, transferred to a conical vial, and further reduced under a stream of prepurified nitrogen. When the extract volume was approximately 500 µL, the volume of the extract was determined, and the extract split into two storage vials. One vial was derivitized with diazomethane to convert organic acids to their methylester derivatives. Compounds were identified and quantified by gas chromatography-mass spectrometry (GC-MS) using authentic standards for instrument calibration. The authentic standards used for instrument calibration include n-hexacosane, n-octacosane, n-tricontane, 17R(H),21β(H)-hopane, pyrene, chrysene, benzo[b]fluoranthene, benzo[k]fluoranthene, hexadecanoic acid, and octadecanoic acid. The precision of the quantification of individual compounds by GC-MS was estimated by propagation of uncertainties associated with the measurement (including measurement uncertainties in volumetric air flow, extract volume, and GCMS instrument response) to be (20% (24). Additional details VOL. 37, NO. 17, 2003 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

3905

TABLE 1. Summary of Measurements of Carbon Monoxide, Carbon Dioxide, and PM2.5 and PM2.5 Elemental Carbon Inside the Tunnel and in the Tunnel Ventilation Air Plus the Emission Index of PM2.5 and PM2.5 Elemental Carbona

tunnel measurements date

time (CDT)

CO (ppm)

CO2 (ppm)

8/29/00 1200-1400 8.14 684.7 8/30/00 1200-1400 7.32 658.8 1600-1800 17.10 855.5 8/31/00 1200-1400 5.82 654.6 1600-1800 19.51 924.4 9/1/00 1200-1400 8.32 675.4

ventilation air measurements

PM2.5 PM2.5 elemental CO (µg m-3) carbon (µgC m-3) (ppm) 127.0 83.5 97.6 103.8 107.9 94.3

a In grams of pollutant per kg carbon in fuel. 0.2 µgC m-3.

22.0 NDb 13.3 23.8 18.0 16.5 b

0.53 0.48 0.88 0.46 1.18 0.56

∆[P] ∆[CO2] + ∆[CO]

(1)

where the background-subtracted pollutant concentration ∆[P] has units of µg m-3 and the background subtracted concentrations of CO2 and CO have units of µgC m-3 giving the emission index units of mass of pollutant emitted per mass of carbon in fuel (27). This is related to the emission factor describing mass per volume of fuel by the carbon content of the fuel (27). For background subtraction of CO2 and CO, samples from the forced-air ventilation of the tunnel are used. Chemical Mass Balance Model. To apportion emissions to specific sources, a chemical mass balance model (CMB version 8.2) was used. For the chemical mass balance (CMB) model, the optimum solution to eq 2 is sought n

C(ij) )

∑R

(jk)s(ik)

+ e(ij)

(2)

k)1

where the concentration of chemical species i in the fine particles at receptor site j (Cij) is reconstructed as a linear combination of contributions to total fine particle mass concentrations at receptor site j from source k (Rjk) with known composition profiles of s(ik) (4). Using the uncertainty of pollutant concentrations and source profiles to weight influence of each measurement, the optimum solution is found by minimizing the error term (e(ij)) which represents the difference between the ambient concentration of each component in each sample and the reconstructed ambient concentration of marker compounds. For sources of fine particulate matter inside the tunnel, the CMB model is limited to three source categories that were expected to dominate fine particle levels inside the 3906

9

395.0 391.9 399.0 395.3 401.0 397.5

PM2.5 elemental PM2.5 PM2.5 elemental carbon (µg m-3) carbon (µgC m-3) PM2.5 52.2 17.5 27.5 30.9 42.1 43.9

0.4 NDb 2.1 0.5