Environ. Sci. Technol. 2002, 36, 567-575
Measurement of Emissions from Air Pollution Sources. 4. C1-C27 Organic Compounds from Cooking with Seed Oils J A M E S J . S C H A U E R , * ,†,| M I C H A E L J . K L E E M A N , †,‡ G L E N R . C A S S , †,§ A N D BERND R. T. SIMONEIT# Environmental Chemistry and Technology Program, University of WisconsinsMadison, Madison, Wisconsin 53706, Environmental Engineering Science, California Institute of Technology, Pasadena, California 91125, and College of Oceanic and Atmospheric Sciences, Oregon State University, Corvallis, Oregon 97331
The emission rates of gas-phase, semivolatile, and particle-phase organic compounds ranging in carbon number from C1 to C27 were measured from institutionalscale food cooking operations that employ seed oils. Two cooking methods and three types of seed oils were examined: vegetables stir-fried in soybean oil, vegetables stirfried in canola oil, and potatoes deep fried in hydrogenated soybean oil. The emission rates of 99 organic compounds were quantified, and these include n-alkanes, branched alkanes, alkenes, n-alkanoic acids, n-alkenoic acids, carbonyls, aromatics, polycyclic aromatic hydrocarbons (PAH), and lactones. Carbonyls and fatty acids (n-alkanoic and n-alkenoic acids) make up a significant portion of the organic compounds emitted from all three seed oil cooking procedures. The compositional differences in the organic compound emissions between the different cooking operations are consistent with the differences in the organic composition of the various cooking oils used. The distribution of the n-alkanoic acids between the gas and particle phases was found to be in good agreement with gas/particle partitioning theory. The relative importance of emissions from commercial deep frying operations to the total emissions of C16 and C18 n-alkanoic acids in the Los Angeles urban area was estimated using the available information and is estimated to account for approximately 7% of the total primary emissions of these acids. Additional emissions of these n-alkanoic acids from stir-frying and grill frying operations are expected. Estimates also indicate that seed oil cooking may make up a significant fraction of the emissions of lighter n-alkanoic acids such as nonanoic acid.
Introduction The C16 and C18 fatty acids (both n-alkanoic acids and n-alkenoic acids) are among the most prominent single * Corresponding author phone: (608)262-4495; fax: (608)262-0454; e-mail:
[email protected]. | University of WisconsinsMadison. † California Institute of Technology. ‡ Present address: Civil and Environmental Engineering Department, University of California at Davis, Davis, CA 95616. § Deceased. # Oregon State University. 10.1021/es002053m CCC: $22.00 Published on Web 12/28/2001
2002 American Chemical Society
organic compounds found in the urban atmospheric fine particulate mixture (1). These particle-phase acids are known to be emitted from many sources such as meat cooking operations, wood combustion, motor vehicle exhaust, and road dust (2), but air pollution modeling results for the Los Angeles Basin indicate that there must be additional as yet unquantified sources of these compounds (3, 4). Seed oils are comprised largely of esters of n-alkanoic acids (5), and fatty acids have been identified in the exhaust from heated seed oils (6). Thus, food-cooking operations that employ seed oils are a likely source of the missing fatty acids emissions. In addition, the effect of the gas-phase, semivolatile, and particle-phase organic compounds emitted from food frying operations on photochemical smog and secondary aerosol formation has not been evaluated largely due to the lack of emissions data for sources of this type. To this end, the emissions from cooking with seed oils are investigated in the present study. This study demonstrates the importance of emissions of n-alkanoic acids and carbonyls from selected food cooking operations. Although these emissions cannot necessarily be applied to food cooking operations at other conditions, the study demonstrates the importance of emissions from food cooking operations. Likewise, the study demonstrates the need for additional measurements of this nature, which are not currently available for the study of air pollution. The results reported in this manuscript, that cover the emissions of gas-phase and particle-phase organic compounds from cooking with seed oils, are part of a series of papers that address the emissions of organic compounds from a broad range of air pollution source. Previous presented results report the emissions from meat charbroiling operations (7), diesel powered motor vehicles (8), and fireplace combustion of wood (9).
Experimental Methods Comprehensive Source Sampling. The seed oil cooking emissions tests reported here were conducted using a large institutional-scale deep fryer and a large industrial-scale electric grill operated by professional chefs as they prepared commercially distributed food products. Emissions were sampled downstream from the filters and grease extractors located in the ventilation system above the appliances. The overall exhaust system was operated at an air flowrate of 400 m3 min-1 and provided sufficient dilution with ambient air to bring the food cooking effluents to ambient temperature prior to collection of source samples. The diluted source effluent was withdrawn from the kitchen vent stack through AIHL-design cyclone separators (10) which were operated at flowrates such that coarse particles with aerodynamic diameters greater than 1.8 µm were trapped, while fine particles and gases passed through the cyclone separator. Fine particle emissions data are emphasized in the present study, because such data are needed for use in the development of urban and regional emissions control strategies for fine particles that will be required under the newly adopted National Ambient Air Quality Standard for fine particulate matter in the United States (11). Semivolatile and fine particle-phase organic compounds were collected directly from the exhaust vent using both a denuder/filter/PUF sampling train and a filter/PUF sampling train. The denuder/filter/PUF sampling train was comprised of an AIHL-cyclone (10), an XAD-coated annular denuder (URG, Inc.; 400 mm long, four channel denuder), three prebaked quartz fiber filters (47 mm diameter, Pallflex Tissuequartz 2500 QAO), and three sets of polyurethane foam VOL. 36, NO. 4, 2002 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
567
plug (PUF) cartridges (Atlas Foam; density ) 0.022 g cm-3, ILD ) 30; 5.7 cm diameter by 7.6 long). The filter/PUF sampling train was comprised of three sets of filter/PUF units operated in parallel using the same sampling media as used for the denuder/filter/PUF sampling train. Details of these sampling trains are described by Schauer et al. (7). The filter/ PUF samples collected during the seed oil cooking tests were not analyzed as part of this study and have been stored for use in future research projects. Fine particulate matter, carbonyls, organic acids, and gasphase hydrocarbons were also collected during these tests using the same cyclone-based unit as described by Schauer et al. (7). Briefly, the sampling train was comprised of an AIHL-cyclone (10) operated at 30 lpm. Downstream of the cyclone, the flow was split among four sample lines: a Teflon membrane filter (Gelman, 47 mm diameter Teflo) operated at 10 lpm, a Teflon membrane filter (Gelman, 47 mm diamater Teflo) followed by two KOH impregnated quartz fiber filters operated at 10 lpm, two quartz fiber filters (47 mm diameter, Pallflex Tissuequartz 2500 QAO) operated at 10 lpm, and two DNPH impregnated C18-cartridges operated in series at 0.2 lpm. A small volume of diluted air, collected downstream of the single Teflon membrane filter, was used to fill a 6 L polished stainless steel SUMA canister for the collection of volatile hydrocarbons during the test of stir-fried vegetables in soybean oil. The ambient air entrained into the exhaust hood above the appliances contained low concentrations of background contaminants that must be subtracted from the source effluent. Identical samples were collected in the kitchen as described above for the collection of samples in the exhaust hood, except that only a filter/PUF sample was used for the collection of semivolatile and particle-phase organic compound concentrations. A denuder/filter/PUF sample was not collected in the kitchen and was only collected in the vent. Analysis of the filter/PUF samples that were collected over the nominal 1-h tests, showed very low levels of semivolatile and particle-phase organic compounds, which were in the range of field blanks. To this end, these samples were used for blank correction and a denuder field blank was used for blank correction of the denuder sample. The concentrations of fine particle mass, fine particle elements, carbonyls, organic acids, and gas-phase hydrocarbons that were present in the background air were above detection and were used to correct the emissions for the concentrations in the dilution air. Source Testing Procedure. The vegetable stir-frying source tests were conducted on an industrial-size electric grill using a commercially distributed mixture of precut broccoli, red and green peppers, celery, and onions (Ingardia Brothers Produce Inc., Costa Mesa, CA). The grill temperature was not directly measured during the test but was estimated to be in the range of 110-125 °C by the chef. Soybean oil (Vons Company Inc., Los Angeles, CA) was used for one test, and canola oil (Cargill Foods, Minneapolis, MN) was used for the other vegetable stir-fry source test. Both the soybean oil and the canola oil tests were conducted while stir-frying 22.6 kg of vegetables using approximately 1.5 L of seed oil and 3 kg of stir-fry sauce (Chef Mate, Nestles; main ingredients: water, soy sauce, high fructose corn syrup, sherry, vegetable oil, and modified food starch) over a period of 1 h. The denuder/filter/PUF sampling train was operated throughout the stir-fry source tests except for the approximately 1 min required to replace the denuder in the middle of the test. The full suite of chemical measurements described above was made during the soybean oil source test, while all measurements except for gas-phase volatile hydrocarbons and fine particle trace elements and ionic species were made during the canola oil stir-fry source test. Table 1 provides a summary of the vegetable stir-fry source tests. 568
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 36, NO. 4, 2002
TABLE 1. Source Tests of Cooking Vegetables with Seed Oil source test vegetable stir-fry with soybean oil vegetable stir-fry with canola oil deep fried potatoes
mass of vegetables cooked (kg)
cooking time (min)
22.6
60
22.6
60
68
75
The deep fried potatoes source test was conducted using an industrial scale deep fryer at 175 °C which was filled with commercial hydrogenated soybean oil (Creamy Liquid Shortening, Kraft Food Service, Preferred). The hydrogenated soybean oil in the deep fryer during the source test had been in use for several days of cooking prior to the tests and was considered relatively fresh by the cooking staff. Sixty-eight kilograms of deep frying potatoes (Kraft 3/8 in. Grade A fancy, Frozen) were cooked in 1.13 kg batches in two parallel fryers operating simultaneously over a period of 75 min by placing the frozen potatoes directly in the hot cooking oil. The denuder/filter/PUF sampling train was operated for two 30minute time periods during the cooking test, while all other sampling equipment was operated continuously throughout the test. Gas-phase volatile hydrocarbons and fine particle trace elements and ionic species measurements were not made during the deep fried potatoes cooking source test. Table 1 also includes a summary of the deep-frying potatoes source tests. Prior to the start of the sources tests, a blank test was conducted to determine the emissions from the ventilation system. In this test the ventilation system was in full operation, but no cooking was being done in the kitchen. The results of these tests indicated that the background air in the kitchen dominated the particulate matter and associated organic compound concentrations in the exhaust of the ventilation system when there was no cooking in the kitchen. Chemical Analysis. The extraction procedure employed for semivolatile organic compounds collected on XAD-coated annular denuders and PUF cartridges as well as particlephase organic compounds collected on quartz fiber filters are discussed by Schauer et al. (7). The extracts collected on these substrates were reduced in volume to approximately 250 µL and were then split into two fractions. One fraction was derivatized with diazomethane to convert organic acids to their methyl ester analogues. Filter, PUF, and denuder field blanks were analyzed with each set of source samples. The field blanks were prepared, stored, and handled by exactly the same procedures as used for the source samples. Field blanks were very similar to lab blanks, which contained only small levels of contaminants that were present at concentrations that were significantly lower than the source samples. Both the derivatized and underivatized sample fractions are analyzed by GC/MS on a Hewlett-Packard GC/MSD (GC Model 5890, MSD Model 5972) using a 30 m × 0.25 mm diameter HP-1701 capillary column (Hewlett-Packard). 1-Phenyldodecane was used as a coinjection standard for all sample extracts and standard runs. Deuterated internal standards were used to determine extraction recovery for the compounds quantified in the underivatized samples, and deuterated acids were used to verify that the diazomethane reactions were driven to completion. In addition, deuterated n-alkanoic acid recoveries were used in conjunction with the recovery of deuterated tetracosane to determine the recovery of the compounds quantified in the derivatized fraction. The recovery of the nonvolatile deuterated ntetradecane internal standard was 81 ( 6% (average ( standard deviation) for the filter analyses and 88 ( 6% for
FIGURE 1. Material balance on the gas-phase, semivolatile, and particle-phase organic compounds emitted from stir-frying vegetables in soybean oil. the denuder and PUF analyses. The recovery of the semivolatile deuterated n-pentadecane internals standard was 65 ( 4% for the filter analyses and 57 ( 6% for the denuder and PUF analyses. The recovery for the semivolatile deuterated n-alkanoic acid internal standards (n-hexanoic acid and n-decanoic acid) were 69 ( 15% for the filter analyses and 62 ( 7% for the denuder and PUF analyses. Hundreds of authentic standards have been prepared for the positive identification and quantification of the organic compounds found in the current source test program. When quantitative standards could not be obtained for a given compound or compound class, significant effort was made to obtain a nonquantitative secondary standard that could be used for unique identification of the organic compounds. Quantification of compounds identified using secondary standards has been estimated based on the response factors for compounds having similar retention times and chemical structures. The uncertainties of quantification of organic compounds by GCMS using the procedures of this study have been determined to be (20% ((1σ). Total non-methane organic gases (NMOG, EPA method TO12) and individual organic vapor-phase hydrocarbons ranging from C1 to C10 were analyzed from samples collected in internally eletropolished stainless steel canisters by gas chromatography/flame ionization detection (GC/FID). Carbonyls collected using DNPH-impregnated C18 cartridges were analyzed by liquid chromatography/UV detection (12). The Teflon filters were weighed before and after the source tests in a humidity and temperature controlled room to determine the mass emissions rates. One Teflon filter from the vegetable stir-fry test that used soybean oil was used for XRF analysis (13), and the other Teflon filter was extracted in water and analyzed by ion chromatography (IC) for watersoluble ions (14). The water extract of the Teflon filter was also analyzed for ammonium ion as described by Solorzano
(15). Quartz filters were analyzed for organic and elemental carbon (OC/EC) as described by Birch and Cary (16).
Results and Discussion A material balance for the gas-phase and fine particle-phase organic compounds emitted from stir-frying vegetables in soybean oil is shown in Figure 1. The total mass of organic carbon in the particle phase was measured from a quartz fiber filter located downstream of an XAD coated denuder. Organic compound mass was estimated by assuming a ratio of 1.4 between organic compound mass and organic carbon, which is based on the quantified fraction of the organic compounds in the particulate matter and the expectation of that the unelutable fraction of the organic compounds are comprised of more polar organic compounds. Of the 15.7 mg of fine particle-phase organic compound mass emitted per kg of vegetables cooked during the soybean oil-based vegetable stir-fry test, 82% of the mass is composed of free fatty acids (n-alkanoic acids and n-alkenoic acids). n-Alkanes and olefinic n-aldehydes make up smaller but noticeable portions of the mass emitted. It should be noted that virtually the entire fine particle mass can be identified at the single compound level. A similar mass balance can be constructed for the fine particulate matter emitted during the canola oil based stir-fry and the deep fried potatoes source tests, and one finds that the free fatty acids also make up a major portion of the particle mass. The total fraction of the fine particle organic compound mass that can be identified on a single compound level for the canola oil based stir-fry test and the deep frying of potatoes test are 83% and 73%, respectively. The gas-phase organic compound emissions from stirfrying vegetables in soybean oil are dominated by carbonyls, as shown in Figure 1. A total of 242 mg of carbonyls are emitted per kg of vegetables cooked, which accounts for 84% of the mass of quantified gas-phase organic compound VOL. 36, NO. 4, 2002 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
569
TABLE 2. Average Fine Particle Mass Emission Rate and Fine Particle Chemical Composition Emitted from Cooking Vegetables in Seed Oils vegetables stir-fried in soybean oil canola oil mass emission rate (mg kg-1 of vegetables cooked) organic carbon (wt %)b elemental carbon (wt %) Na+ (wt % by IC) NH4+ (wt % by IC) Cl- (wt % by IC) N03- (wt % by IC) S04) (wt % by IC) potassium (wt % by XRF) sulfur (wt % by XRF) a
Not measured.
b
21.5 ( 1.2
deep frying of potatoes
29.5 ( 1.3
13.1 ( 1.2
Particle Composition 69.6 ( 5.50 58.3 ( 4.7
62.7 ( 5.1
