Environ. Sci. Technol. 1991, 25, 744-759
(2) Kramer, J. R.; Herbes, S. E.; Allen, H. E. Phosphorus: Analysis of Water, Biomass, and Sediment. In Nutrients in Natural Waters;Wiley Interscience: New York, 1972;
pp 51-100. (3) Loder, T. C.; Liss, P. S. Lirnnol. Oceanogr. 1985,30,418-421. (4) Ridgeway, H. F.; Olson, B. E. Appl. Environ. Microbiol. 1981, 41, 274-287. ( 5 ) Kirk, J. 0. Light Processes in Aquatic Ecosystems; Cambridge University Press: New York, 1983. (6) Johnson, D. L. Scanning Electron Microsc. 1983, 3, 1211-1228. ( 7 ) Bernard, P. C.; VanGrieken, R. E.; Eisma, D. Enuiron. Sei. Technol. 1986, 20, 267-273. (8) Carder, K. L.; Steward, R. G.; Johnson, D. L.; Prospero, J. M. J . Geophys. Res. 1986, 91D, 1055-1066. (9) Weideman, A. D.; Bannister, T. T.; Effler, S. W.; Johnson, D. L. Limnol. Oceanogr. 1985, 30, 1078-1083. (10) Effler, S. W.; Johnson, D. L. Water Res. Bull. 1987, 23, 73-79. (11) Yin, C. Q.; Johnson, D. L. Limnol. Oceanogr. 1984, 29, 1193-1201. (12) Effler, S. W.; Johnson, D. L. Heavy metal distribution in particles from the flocculent sediments of Onondaga Lake, N.Y. In X I V Trace Contaminants Conference; Hemphill, D. D., Ed.; University of Missouri: Columbia, MO, 1980; Vol. 14, pp 489-499. (13) Kullenberg, G. Observed and computed scattering functions. In Optical Aspects of Oceanography;Jerlov, N. G., Nielson, E. S., Eds.; Academic Press: New York, 1974; pp 25-49. (14) Devan, S. P.; Effler, S. W. J . Environ. Eng. 1983, 110, 93-109.
(15) Yin, C. Q. A research of sources and budget of mineral sediments in Onondaga Lake, Central New York. M.S.
Thesis, SUNY-CESF, 1982; p 135. (16) Effler, S. W. Water, Air, Soil Pollut. 1987, 33, 85-115. (17) Sze, P.; Kingsbury, J. M. J. Phycol. 1972, 8, 25-37. (18) Field, S. D.; Effler, S. W. Arch. Hydrobiol. 1983,98,409-421. (19) Effler, S. W.; Perkins, M. G.; Brooks, C. M. Water, Air, Soil Pollut. 1986, 29, 117-134. (20) Effler, S. W.; Driscoll, C. T. Environ. Sei. Technol. 1985, 19, 716-720. (21) Dean, W. E.; Eggleston, J. R. Sediment. Geol. 1984, 40, 217-232. (22) Preisendorfer, R. W. Limnol. Oceanogr. 1986,31,909-926. (23) Effler, S. W. J . Enuiron. Eng. 1988, 114, 1436-1447. (24) Wetzel, R. G. Limnology; Saunders College Publishing: Philadelphia, PA, 1983. (25) Sze, P. Phytoplankton and zooplankton. In Onondaga County Monitoring Program Annual Report 1987; Onondaga County Department of Drainage and Sanitation, 1989. (26) Effler, S. W.; Johnson, D. L.; Jiao, J. F.; Perkins, M. G. Optical impacts and sources of tripton in Onondaga Creek USA. Water Resour. Bull., submitted. (27) Aas, E. The refractive index of phytoplankton. In Institute for Geofysikk; Institute Report Series 46; Universitetet I Oslo, 1981. (28) Stumm, W.; Morgan, J. J. Aquatic Chemistry;J. Wiley and Sons: New York, 1981. (29) Reynolds, R. C., Jr. Limnonol. Oceanogr. 1978,23,583-597. Received for review November 29, 1989. Revised manuscript received July 24, 1990. Accepted November 16, 1990.
Chemical Composition of Emissions from Urban Sources of Fine Organic Aerosol Lynn M. Hildemann,t Gregory R. Markowski, and Glen R. C a s *
Environmental Engineering Science Department and Environmental Quality Laboratory, California Institute of Technology, Pasadena, California 91 125
A dilution source sampling system was used to collect primary fine aerosol emissions from important sources of urban organic aerosol, including a boiler burning No. 2 fuel oil, a home fireplace, a fleet of catalyst-equipped and noncatalyst automobiles, heavy-duty diesel trucks, natural gas home appliances, and meat cooking operations. Alternative dilution sampling techniques were used to collect emissions from cigarette smoking and a roofing tar pot, and grab sample techniques were employed to characterize paved road dust, brake lining wear, tire wear, and vegetative detritus. Organic aerosol constituted the majority of the fine aerosol mass emitted from many of the sources tested. Fine primary organic aerosol emissions within the heavily urbanized western portion of the Los Angeles Basin were determined to total 29.8 metric tons/day. Over 40% of these organic aerosol emissions are from anthropogenic pollution sources that are expected to emit contemporary (nonfossil) aerosol carbon, in good agreement with the available ambient monitoring data. Introduction Chemical compounds that will exist as organic aerosols in the atmosphere are found as a mixture of aerosol plus vapor-phase organics when sampled directly from a hot stack exhaust. Present methods of measuring particulate emissions from sources vary greatly, with large differences Present address: Department of Civil Engineering, Stanford Stanford, CA 94305-4020.
University, 744
Environ. Sci. Technol., Vol. 25, No. 4, 1991
in the way that condensible organic vapors are collected. Motor vehicle emission tests typically are conducted by dilution sampling (1,2),where condensible organic vapors are scavenged by condensation onto preexisting aerosol as the exhaust is cooled and diluted with filtered air in a large laboratory facility. In contrast, in commonly used stationary source sampling methods like EPA method 5 (3, 4) and the Source Assessment Sampling System (SASS) (5),organic vapors that will form aerosol upon cooling and dilution of the stack exhaust in the atmosphere pass through hot particulate filters without being collected, and then in some systems the condensible vapors are captured in impinger baths or cold traps. However, these vapor traps may also collect other gas-phase organic material that would not form aerosol under ambient conditions. Most of the equipment commonly used a t present for collecting aerosol source samples is constructed from components that contain plastics, rubber compounds, and/or greases that can release organics into the system, contaminating the aerosol sample. While slight contamination of the sample with artifact organics may not pose a major problem if only total carbon is to be measured, even slight contamination introduced during sampling can pose overwhelming problems if the ultimate goal is identification and quantification of single organic species in the aerosol by gas chromatography/mass spectrometry (GC/MS). Thus, variations in test methods, contamination control procedures, and analytical procedures will frustrate most attempts to compare the relative character
0013-936X/91/0925-0744$02.50/0
0 1991 American Chemical Society
Table I. Preliminary Estimate of the Major Sources of Fine Aerosol Organic Carbon (OC) Emissions within an 80 Heavily Urbanized Area Surrounding Los Angeles for 1982 ( 7 , 8 )
oc
source type (1)
(2) (3) (4) (5) (6) (7)
(8) (9) (10) (11)
(12) (13) (14) (15) (16) (17) (18) (19) (20) (21) (22) (23) (24) (25) (26) (27) (28) (29) (30)
emitted," kglday
70 of oc emissions
source to be tested
5114 4389 3710 2100 1480 1433 1336 1270 877 752 692 692 671 393 335 278 228
18.6 15.9 13.5 7.6 5.4 5.2 4.9 4.6 3.2 2.7 2.5 2.5 2.4 1.4 1.2 1.0 0.8 0.8 0.7 0.7 0.6 0.6 0.4 0.3 0.3 0.3 0.3 0.2 0.2 0.2 0.8
X X X X X
paved road dust charcoal broilers noncatalyst gasoline vehicles diesel vehicles brake lining surface coating cigarettes fireplaces forest fires roofing tar pots natural gas combustion organic chemical processes tire wear misc industrial point sources catalyst-equipped gasoline vehicles misc petroleum industry processes primary metallurgical processes railroad (diesel oil) residual oil stationary sources refinery gas combustion secondary metallurgical processes mineral industrial processes other organic solvent use jet aircraft asphalt roofing manufacturing coal burning wood processing residual oil-fired ships structural fires distillate oil stationary sources other sources
total
211
206 195 167 158 106 92 81 76 74 66 63 61 226 27532
X X
X
80 km
notes
b C C C
b d C C
e
X X
C C
X
d b d
X
C
X
C
"Annual average emissions stated a t a daily rate. *Tested by grab sampling and then resuspension of the collected particulate matter. cTested via dilution sampling. dThis category represents a collection of many small dissimilar sources for which a small number of tests cannot be used to represent the emissions from the group as a whole. 'Forest fires are an occasional emission source that does not affect ambient samoles on a routine basis.
and importance of the different organic aerosol sources based on the current technical literature. Recently, a new source sampling system has been designed specifically to collect organic particulate matter including the vapor-phase material that would have condensed into the aerosol phase under ambient conditions, while minimizing contamination and particle loss problems (6). Briefly, the sampler operates as a portable dilution tunnel and can be used to test both mobile and stationary sources. The organic vapors present in hot exhaust gases are cooled to ambient temperature and pressure by dilution with cool, filtered air, causing vapor-phase organics to condense onto preexisting aerosol in much the same manner as will occur in the plume downwind of the source. Aerosol samples are collected a t ambient temperature and pressure on filter substrates identical with those used for the sampling of ambient aerosol, which facilitates later detailed chemical analysis and comparison of source and ambient aerosol samples. In the present study, this dilution source sampling system is used to explore the relative importance of the various sources of fine (dp < 2 pm) primary organic aerosol emissions to an urban atmosphere, using the Los Angeles Basin as a test case. Emission source tests are conducted for both mobile and stationary sources by a single method, so that comparisons between sources reflect differences between the sources rather than differences in sampling and analysis methods. Based on existing engineering estimates of Los Angeles Basin organic aerosol emission rates (estimated by literature review; see refs 7 and 8), source
types are selected for testing that are predicted to contribute close to 80% of the fine organic aerosol emitted to the Los Angeles area atmosphere (see Table I). The present paper documents the source test procedures and presents the bulk chemical characteristics of the sources sampled. In most cases, both the mass emission rate and chemical composition are determined. This is particularly useful in support of quantitative organic aerosol emission inventory studies, since many studies of aerosol source chemical composition have focused either on mass emission rate determination or on relatiue chemical abundances (9-11), but not both. On the basis of these tests, the existing organic aerosol emission inventory for Los Angeles is revised. In future work, the source samples documented here will be characterized by high-resolution gas chromatography (GC) and GC/MS in order to describe the emission rates of key organic compounds that can be used as markers for the presence of these sources in the ambient aerosol.
Experimental Section 1. Source Sampler Design. The source sampler used in this work is shown in Figure la. In its stationary source test configuration, it employs an in-stack stainless steel cyclone separator intended to collect particles with aerodynamic diameter greater than 10 pm. Exhaust gases and particles smaller than 10 Fm in diameter are drawn through a heated Teflon sampling line into a stainless steel dilution tunnel. Following rapid and thorough mixing with cooled, purified air, approximately 15% of the source Environ. Sci. Technol., Vol. 25, No. 4, 1991
745
RESIDEKC CHAMBER
BOILER STACK
DILUTION TUNNEL
DILUTION AIR SAMPLER
I
FILTER HOLDERS
I
2pmCYCLOHES DILUTION
3 .k,
7
'10
/
I
SAMPLE FILTER
1 TAR POT
um CYCLONE
Figure 1. Source sampler design. (a) Dilution stack sampler for boiler experiments; (b) vehicle test configuration (top view): (c) cigarette sampling; (d) tar pot sampling.
sample is drawn into a stainless steel chamber where 4C-60 s of residence time is provided to allow condensation processes to go to completion. The sample next flows from the residence time chamber through several parallel cyclone separators in which particles larger than 2-pm aerodynamic diameter are collected. The fine organic aerosol that remains in the sample stream then is collected by filtration a t ambient temperature and pressure by using equipment identical with that commonly used for lowvolume ambient aerosol sampling. All flows are metered and controlled so that data on quantitative emission rate per unit of source activity are obtained. Details of the sampler's design, sampling efficiency, and performance characteristics (including a comparison with EPA method 5 ) have been published previously (6). 2. Source Sampler Preparation. The sampler was cleaned before use a t each source type sampled. Large parts first were vapor-degreased by using tetrachloroethylene, with all open ends wrapped with clean aluminum foil immediately after cleaning to prevent recontamination of the interior surfaces. These sampler parts were then assembled in the laboratory and baked for a t least 4 h at temperatures of greater than 70 "C by wrapping electrical heating tape around the stainless steel shell of the dilution sampler while circulating purified air through the system. The smaller sampler parts were sonicated first for 5 min in glass-distilled methanol and then for another 5 min in glass-distilled hexane. Again, all open ends were wrapped with clean aluminum foil after the cleaning process to prevent recontamination during storage and transit. After the sampler was assembled a t the site of the combustion source, the system was checked for leaks of unfiltered ambient air into the system. Under a typical operating vacuum of 2500 Pa, this leakage rate was re746
Environ. Sci. Technol., Vol. 25, No. 4, 1991
quired to be less than 0.1% of the total flow rate through the system. System blank samples then were taken at the site prior to sampling the source in order to detect any potential contaminants, such as ambient organics passing through the dilution air purification system, artifacts resulting from the handling and storage of the filters, or residues remaining after cleaning the sampler. Collection of the system blank samples was achieved by taking filter samples of the dilution air over a period of several hours at the normal stack aerosol sample extraction points (see Figure la). Immediately before each experiment, the stack gas flow rate and temperature were measured at various points, using a pitot probe and thermocouple. The flow rate measurements were used to select an appropriate nozzle size for the in-stack cyclone to achieve isokinetic sampling. The temperature measurements were used to determine the temperature setting for the heated sample inlet line in order to avoid premature condensation of aerosol in the inlet line upstream of the dilution tunnel. Before and after each sampling experiment, the flow rate through each of the filters was measured with a rotameter calibrated to an accuracy of f l % . During each experiment, the flow rates of the incoming dilution air, and of the mixture exiting the tunnel, were measured a t regular intervals by using magnehelic gauges that determined the pressure both upstream and across calibrated orifice plates. The exhaust gas flow through the stack sampler inlet line was measured in the same manner with a venturi meter. All flow rates were corrected to standard temperature and pressure (STP). System temperature and relative humidity also were recorded frequently, and parameters such as excess oxygen, temperature, and concentrations of both CO and C 0 2were measured with a Kane-May Model 9103
combustion analyzer using a probe inserted into a sampling port in the stack. 3. Source Sampling Procedures. For tests conducted on ducted stationary sources, the dilution stack sampler was erected on a wheeled platform adjacent to the stack, and the entire sampler was moved during sampling to traverse the stack. The hot exhaust gases withdrawn from the stack passed through a straight, 1.2-m-long heated Teflon tube a t a typical rate of 22-32 L/min (at STP) before being mixed with dilution air in the main part of the dilution sampling system. The dilution air was cooled slightly to achieve an ambient temperature of 25-35 "C upon mixing with the hot source effluent, and the dilution air was purified before use by passing it through an activated carbon bed and a HEPA filter. The source effluent was typically diluted by a factor of between 35:l and 50:l. Fine aerosol samples were collected from the residence time chamber a t the downstream end of the sampling system. Aerosol samples were withdrawn through six parallel AIHL cyclone separators (12), each operated a t 27.9 f 0.3 L/min, which removed particles with aerodynamic diameters greater than or equal to 2.0 pm. Downstream of each of these cyclone separators, three 47-mmdiameter filter holders were used to collect the aerosol at a sample flow rate of 9.0-9.6 L/min through each filter holder, yielding 18 separate fine particle samples. Three of the 18 filter holders contained filter substrates for determination of bulk fine aerosol chemical properties. Two of these three filter holders contained Teflon filters (Gelman Teflo, 2.0-pm pore size) for determination of (1) aerosol mass by gravimetric analysis of both filters, (2) trace-metals content by X-ray fluorescence, and (3) ionic species by ion chromatography and atomic absorption spectrophotometry. The third filter holder contained a quartz fiber filter (Pallflex 2500 QAO) of the type customarily used for collection of ambient organic aerosol samples. All quartz fiber filters were prebaked a t 750 "C for a t least 2 h to lower their carbon blank. These filter types were chosen in order to match the sampling characteristics of an existing set of Los Angeles ambient aerosol samples (8, 13, 14). From chemical analysis of the two Teflon and one quartz fiber filters, a nearly complete material balance on the bulk chemical content of the aerosol can be obtained (15, 16). The remaining 15 filter holders contained prebaked quartz fiber filters destined for later detailed examination of the organic species by high-resolution GC and GC/MS. All filter handling was performed with clean, Teflon-coated tweezers. The glass jars used for storing organic aerosol source samples were annealed overnight a t 500 "C, and had solvent-washed Teflon liners in the lids. Samples for trace elements, ionic species, and gravimetric analysis (collected on Teflon filters) were stored in Petri dishes sealed with Teflon tape. All samples were stored a t -25 "C within 2 h of the end of an experiment. The aerosol mass emission rate from each combustion source was determined gravimetrically by weighing the Teflon filters before and after sampling on a Mettler balance (Model M-55-A) in a temperature- and humidity-controlled room (20-24 "C, 50-5570 RH). Thirty-four trace elements were determined by X-ray fluorescence. Sulfate, nitrate, and chloride emission rates were measured by ion chromatography (Model 2020i, Dionex Corp.), and sodium and magnesium were analyzed by flame atomic absorption (Model AA-6, Varian Techtron). The emission rate of aerosol ammonium ion was measured by a modified indophenol colorimetric method (17) with a rapid-flow analyzer (Model RFA-300, Alpkem Corp.). Organic and
elemental carbon mass emission rates were determined from a quartz fiber filter by a thermal evolution and combustion technique (18, 19). The mass emission rate of organic compounds was estimated as 1.2 times the organic carbon measured (20). The mass emission rate of each chemical species found in the exhaust aerosol was determined after subtraction of any trace pollutant concentrations measured in the dilution air supply. 4. Sources Tested by Dilution Sampling. The sources selected for sampling, identified in Table i, were chosen to represent source types responsible for close to 80% of the fine primary organic aerosol emissions to the Los Angeles atmosphere based on existing engineering estimates drawn from a literature review (7, 8 ) . Those sources where dilution sampling was necessary are indicated by footnote c in Table I. 4.1. Stationary Industrial Sources. A Babcock & Wilcox dual fuel FM-type industrial-scale watertube boiler (steam production capacity 53 x lo6 kJ/h) was tested equipped with a steam atomizer burner for No. 2 fuel oil combustion. For each test, the boiler being sampled was operated in steady-state mode a t -60% of capacity (using a second boiler to absorb the demand fluctuations), and the actual rate of fuel consumption was recorded continuously by a plotter connected to a flowmeter in the fuel line. Sampling was conducted on the roof of the steam plant facility, using access ports built into the exhaust stacks. These access ports allowed samples to be taken by traversing only along one axis and prevented samples from being taken closer than 8 cm from the wall. The 1.1-m-diameter stack was traversed twice during each experiment, and samples were taken from five equidistant points within the stack. The cross-sectional area of the stack was divided into five zones surrounding each sampling point, and the duration of sample extraction at each point was directly proportional to the volumetric flux of exhaust gas through the area of the stack assigned to that point. Hence, samples were taken for long periods near the stack wall, while the center of the stack was sampled for a shorter time. The volumetric emission rate of the stack exhaust gases (as measured by traversing the stack with a pitot probe equipped with a thermocouple) was adjusted to 3% oxygen, dry, a t standard temperature and pressure, based on fuel consumption and excess oxygen measurements. 4.2. Mobile Sources. A fleet of automobiles and heavy-duty diesel trucks was tested a t the California Air Resources Board's Haagen-Smit Laboratory dynometer facilities. Vehicles tested are described in Table 11. For the automobiles, the cold-start Federal Test Procedure Urban Driving Cycle (see Figure 2a) was used to simulate city driving conditions. Cars were tested in as-received condition, and an effort was made to include both domestic and foreign model vehicles. Each catalyst-equipped automobile was tested while burning the brand of unleaded gasoline normally purchased by its owner. For the noncatalyst Mercury and Buick, leaded premium with a lead content of 0.22 g/gal was used. For the other four noncatalyst cars, leaded regular gasoline with a lead content of 0.34 g/gal was used. The trucks, which were provided by the City of Pasadena, were both late-model, low-mileage vehicles and were tested while burning the diesel fuel normally purchased by that city. The heavy-duty dynomometer used for the trucks could not simulate a complicated driving cycle. Instead, the cycle illustrated in Figure 2b was used. T o sample an amount of vehicle exhaust proportional to the total amount being emitted at any time, an addiEnviron. Sci. Technol., Vol. 25, No. 4, 1991
747
TIME rnin.
COLD STAR1 ENGINE
0
5
10
20
15
SHUT OFF ENOINE
2.5
30
MOT START ENGINE
35
ENGINE OFF
40
45
50
TIME (rnin)
Figure 2. Driving cycles. (a) Federal Test Procedure urban driving cycle; (b) driving cycle for heavy-duty diesel trucks.
Table 11. Characteristics of Vehicles Tested no. of engine odometer cylin- displac., reading, ders cu in. mi
FTP fuel econ, mi/gal
Noncatalyst Automobiles 1965 Mercury Monterey 8 390 1969 Ford Mustang 8 302 1970 Buick Skylark 8 350 1972 Chevrolet Caprice 8 400 1974 Ford Pinto 4 140 1976 Volkswagen Beetle 4 97
139 192 141 843 88 726 147 249 107 138 80 876
10.7 14.7 13.0 12.3 16.7 26.6
Catalyst Automobiles 1977 Chevrolet Vega 4 140 1980 Honda Civic 150 4 91 4 91 1980 Honda 1500 1980 Toyota Corolla 4 108 4 119 1981 Datsun 200SX 1983 Chevrolet Malibu CL 6 231 1983 Dodge Omni 4 135
88598 83862 149424 66832 81541 38826 27280
22.0 27.5 28.0 20.6 24.6 17.7 22.7
Heavy-Duty Diesel Trucks 1987 GMC Truck 8 636 (2-axle) 1987 Ford Dump Truck 8 636 (3-axle)
2920
7.9"
5581
7.2'
"Fuel economy is for the truck driving cycle shown in Figure 2b.
tional dilution tunnel segment was added upstream of the normal intake to the dilution sampler (see Figure lb). A combination of vehicle exhaust plus filtered dilution air was drawn through this tunnel segment at a constant flow rate, with the dilution air flow rate varying to maintain a constant total flow rate in the face of changing engine 748
Environ. Sci. Technol., Vol. 25, No. 4, 1991
speed. With this approach, the concentration of the diluted emissions in this tunnel was proportional to the flux of exhaust from the tailpipe at any time. A fraction of this prediluted exhaust was withdrawn into the stack sampler at a constant flow rate and then was diluted a second time by the standard dilution air supply to the sampler, effectively sampling an exhaust aerosol concentration proportional to the total emission rate a t any time. 4.3. Dispersed Area Sources. An undampered, traditional brick fireplace in an older single-family home was utilized for the sequence of wood combustion experiments, each involving a different type of wood: seasoned pine, seasoned oak, seasoned almond, and a synthetic log (Pine Mountain brand, 5 lb). For the wood, pieces ranging from 1000 to 6000 g were preweighed to the nearest 5 g and labeled, as was the kindling. T o mimic the course of a traditional undampered fire in a home fireplace, wood was added at intervals through the course of the fire, and the fire was periodically stirred. For the synthetic log, the fire was left undisturbed, in accordance with the manufacturer's directions. Emissions were withdrawn from the flue through a hole bored into the side of the chimney, using four different sampling points along the long axis of the chimney. For each test, emissions were sampled from the time that the fire was lit to the late-ember stage, a 3-4 h period. Pieces of newspaper were used to ignite the fire, but otherwise the material being combusted (including the kindling) was strictly limited to the type of wood being used for the particular experiment. For the meat-cooking experiments, a local commercial-scale kitchen was utilized. Two types of hamburger meat, regular (approximately 21 70 fat) and extra-lean (approximately 10% fat), were cooked by two methods:
Table 111. Market Share Data for Cigarettes"
cigarette type regular filtertips (nonmenthol) lights/ultralights (nonmenthols) menthols
1981 market share, 70
brand sampled
1981 market share, 70
22.2
Camel Merit
2.3 3.2
16.5
Winston
4.1
27.5
Benson & hedges
2.1
33.7
Calculated from data in ref 21.
by charbroiling over a natural gas flame, and by frying. In each of four experiments, 80 quarter-pound (113 g) hamburger patties were cooked until they were medium- to well-done, eight at a time, over a period of 70-80 min. The aerosol generated during the cooking was withdrawn through a 14000 cfm (6.6 m3 s-l) overhead exhaust hood equipped with a baffle-type grease extractor. The source sampler, located on the roof of the kitchen, withdrew samples a t the exit to the exhaust duct. The air flow rate through the exhaust vent was verified through pitot probe measurements. A natural gas fired space heater (Western Gravity Heat, Model 8G100) and a water heater (American Standard, Model G-531-H) were used to generate emissions from natural gas combustion a t a single-family home. The two devices were connected to a common exhaust duct, and emissions were withdrawn from five different points in the 23 X 23 cm exhaust duct. For each experiment, the heaters were turned off for 1-3 min approximately twice per hour in an attempt to mimic the cyclic nature of their operations; otherwise, both heaters were left on continuously for the 8-12-h duration of each experiment. For collection of cigarette smoke, a vertically oriented dilution tunnel made of stainless steel, aluminum, and Teflon was erected (see Figure IC). This tunnel, which was solvent cleaned prior to use, was constructed to collect all the aerosol released to the atmosphere while a cigarette was smoked. As the cigarette was smoked, the smoker would exhale into the triangular opening at the bottom of the tunnel. During the periods between puffs, the cigarette was placed in an ash tray at the bottom of the tunnel, and the sidestream smoke from the smoldering cigarette was drawn into the tunnel. As the smoke was drawn up into the tunnel, it underwent dilution and then passed through 2-pm-size cut cyclone separators; the fine aerosol finally was captured on a set of filters identical with those used with the dilution sampling system. Consumption of regular, filtertip, light, and menthol cigarettes was estimated based on market-share data from 1981 (2),and one of the five most popular brands of each category of cigarette was used for testing, as shown in Table 111. The roofing tar pot tested contained petroleum-based built-up roofing asphalt (GAF brand) maintained a t 250-300 "C. A smaller scale dilution sampler constructed of stainless steel, aluminum, and Teflon (see Figure Id), solvent-cleaned prior to use, was employed for sampling tar pot emissions. A Variac was used t o regulate the dilution air flow rate such that the sample was diluted by 18-fold before collection of fine aerosol on filter sets downstream of two 2-~m-sizecut cyclones. Samples were taken from a vent opening a t the top of the tar pot for 10-15 min. 5. Sources Tested by Resuspension. For collection of paved road dust, a grab sampling technique was used. A small vacuum sweeper truck (Tennant 255,36HP) was
driven up and down several blocks of Pasadena-area streets that had accumulated road dust during a 2-week dry period in May 1988. The material collected was resuspended in a clean Teflon bag, using purified air to maintain a slightly positive pressure in the bag. The contents of the Teflon bag were agitated and samples were withdrawn through 2-pm-size cut cyclone separators for subsequent collection on filter substrates. A bulk sample of organometallic brake dust from the rear drum brakes of a late-model light-duty truck was obtained by brushing the dust from the inside of the drums after the wheels were removed. The sample was resuspended and the fine particles present were collected in the same manner as for the road dust. A radial tire (195/60R15 Toyo) that had already been driven for 7200 miles was run on a rolling resistance test a t a tire-testing laboratory over a period of several days. Wear particles that accumulated on the horizontal surfaces of the machine were collected as a grab sample. When the tire particles were resuspended, they became electrically charged and adhered to the walls of the sampling equipment. The attempt to obtain a fine aerosol sample was abandoned. Instead, a total particle sample was collected directly onto filters, giving a sample that was mainly composed of coarse particles. Organic aerosol produced by the abrasion of surface waxes from the leaves of plants is readily identifiable as a component of ambient particulate matter samples (14, 22,23). T o collect a representative sample of vegetative abrasion products, the predominant plant species in the Los Angeles area, both cultivated and natural, were identified. On the basis of a study of vegetation in the Los Angeles Basin (24,251,51 cultivated species representing six types of urban vegetation (broad-leaf trees, conifers, palms, shrubs, grasses, and ground cover), 10 species of natural vegetation and 1 example of agricultural vegetation were selected for study, as shown in Table IV. Separate samples of green and dead leaves were acquired for each of these 62 species over a 3-day period from botanical collections a t the Los Angeles County Arboretum in Arcadia, CA, Rancho Santa Ana in Claremont, CA, and Eaton Canyon Park in Pasadena, CA. Leaves from the species collected were composited to form a mixed sample in which the mass of leaves from each species was proportional to the fraction of the estimated total leaf mass in the Los Angeles region represented by that species (see Table IV) Two composite samples were formed-one with dead leaf material, and one with green leaf material. Each composite sample was then agitated in a clean Teflon bag for -2 h to release the plant waxes and other vegetative detritus, collecting the sample by the same approach as for the road dust.
Results and Discussion The results of individual experiments, including the standard deviation of the analytical method, are available as a supplementary microfiche appendix to this paper. In the text that follows, the average of the results from each source type will be discussed. These average source profiles are shown in Table V. Values shown in Table V that are accompanied by the symbol t were found a t a concentration a t least twice the standard deviation of the measurement method in a t least half of the source tests conducted on sources of the type indicated. Entries marked with the symbol * are present a t concentrations that are twice the standard deviation of the mean emission rate for the group of sources tested. 1. Oil-Fired Boiler Emissions. The operating parameters for the five tests conducted on a boiler burning Environ. Sci. Technol., Vol. 25, No. 4, 1991
749
Table IV. Vegetation Sample Composite leaf mass per area of land covered, g/m2
broad-leaf trees ribbon gum ash CA live oak elm maple C A sycamore Peruvian pepper jacaranda Victorian box black locust broad-leaf trees crape myrtle avocado camphor magnolia Brazilian pepper olive willow silver dollar gum citrus (orange) others shrubs CA sage brush golden wattle juniper glossy privet bottlebrush Chinese juniper camellia oleander hibiscus rose shiny xylosma coyote bush Japanese pittosp. toyon cape honeysuckle cotoneaster heavenly bamboo India hawthorn holly podocarpus yucca others conifers Monterey pine other pine Italian cypress others palms cocos palm fan palm Canary Is. palm others mixed grasses dichondra ground cover ivy African daisy ice plant others
750
area covered by veget in Los Angeles, m2/km2
scale factor"
fraction of mass in composite Los Angeles leaf sample,b 70
Urban Vegetation E u c a l y p t u s uiminalis Fraxinus s p . Quereus agrifolia Ulmus sp. Acer s p . Platanus racemosa S c h i n u s molle Jacaranda mimosafolia Pittosporum u n d u l a t u m Robinia psudoacacia
490 756 1230 140' 300 342 362 283 2500 50
7300 3100 1600 3000 1200 1000 970 900 630 570
1.20 1.20 1.20 1.20 1.20 1.20 1.20 1.20 1.20 1.20
2.10 1.36 1.20 0.25 0.20 0.22 0.21 0.15 0.91 0.017
Lagerstroemia indica Persea americana C i n n a m o n u m camphora Magnolia grandiflora S c h i n u s terebinthifoliu Olea europaea Salix s p . E u c a l y p t u s polyanthemos Citrus s p .
2140 215 153 1110 109 1190 176 503 490
520 490 440 400 280 230 500 340 380 4900
1.20 1.20 1.20 1.20 1.20 1.20 1.20 1.20 1.20
0.66 0.061 0.039 0.26 0.018 0.16 0.051 0.10 0.11
A r f e m e s i a californica Acacia longifolia J u n i p e r u s sp. Ligustrum i u c i d u m Callistenon sp. J u n i p e r u s chinensis Camellia sp. N e r i u m oleander Hibiscus sp. Rosa sp. X y l o s m a congestum Baccharis pilularis Pittosporum tobira Heteromeles arbutifolia Tecomaria capensis Cotoneaster sp. N a n d i n a domestica Raphiolepis indica Ilex s p . Podocarpus sp. Yucca sp.
70 647 3430 482 773 2360 2080
1500 1300 1000 820 820 660 580 540 530 520 420 400 380 340 97 250 160 240 150 300 290 3200
1.28 1.28 1.28 1.28 1.28 1.28 1.28 1.28 1.28 1.28 1.28 1.28 1.28 1.28 1.28 1.28 1.28 1.28 1.28 1.28 1.28
0.065 0.51 2.17 0.26 0.41 0.98 0.75 0.14 0.26 0.11 0.16 0.25 0.52 0.32 0.028 0.11 0.053 0.075 0.056 0.047 0.31
P i n u s radiata Pinus s p . Cupressus sempervirens
941 1000 8500
2500 820 450 450
1.12 1.12 1.12
1.29 0.45 2.07
Arecastrum r o m a n z o f f i a n u m Washingtonia sp. Phoenix canariensis
722 1570 2050
720 620 320 38 92000 2200
1.02
1.02 1.02
0.26 0.48 0.33
1.02
17.42
1.10
2.55 1.05 1.02
417
789 329 589 lOO0d
2160 l500d 478 717 547 496 600d 255 1700d
(various unidentified) Dichondra repens
380d
Nedera sp. Osteospermum fruticosum L a m p r a n t h u s spectabulus
580d
Environ. Sci. Technol., Vol. 25, No, 4, 1991
800d 1400d
8200 2500 1300 1900
1.10 1.10
Table IV (Continued) leaf mass per area of land covered, g/m2 black sage buckthorn buckwheat CA sage brush ceanothus chamise manzanita scrub oak sugar bush grass others
Salvia sp. R h a m n u s sp. Eriogonum sp. Artemesia californica Ceanothus sp. Adenostoma s p . Arctostaphylos s p . Quercus sp. Rhus sp. Gramineae
citrus lemon
Citrus limon b u r m
Natural Vegetation 305 225 205 18 770 92 5050 64 305 400d
Agricultural Vegetation 429
scale factorn
fraction of mass in composite Los Angeles leaf sample,b %
34000 3200 11000 33000 31000 52000 3600 13000 11000 58000 41000
1.23 1.23 1.23 1.23 1.23 1.23 1.23 1.23 1.23 1.23
6.25 0.44 1.40 0.36 14.27 2.86 10.84 0.53 2.08 13.85
23000
1.00
4.81
area covered by veget in Los Angeles, m2/km2
Other Land 66000 430000 18000
nonvegetated urban barren natural barren
"Scale factor accounts for species not sampled, which appear as the last item in each vegetation category. Amounts of the species actually sampled were scaled up by this factor within each category. *Calculated as (100 X column 1 X column 2 X column 3)/C(column 1 X column 2 X column 3). 'Average of values for Chinese and American elm. dLeaf mass per area estimated for this study. Other values calculated from data in refs 24 and 25.
.
1