Particle and gas emissions from a simulated coal-burning household fire pit

Particle and Gas Emissions from aSimulated Coal-Burning HouseholdFire PitL I N W E I T I A N , † D O N A L D L U C A S , * , ‡

S U S A N L . F I S C H E R , † S . C . L E E , §

S . K A T H A R I N E H A M M O N D , † A N DC A T H E R I N E P . K O S H L A N D † , 4

School of Public Health, Lawrence Berkeley NationalLaboratory, and Energy and Resources Group, University ofCalifornia, Berkeley, Berkeley, California 94720, and TheHong Kong Polytechnic University,Hung Hom, Kowloon, Hong Kong

Received July 5, 2007. Revised manuscript receivedNovember 16, 2007. Accepted January 7, 2008.

An open fire was assembled with firebricks to simulate thehousehold fire pit used in rural China, and 15 different coals fromthis area were burned to measure the gaseous and particulateemissions. Particle size distribution was studied with amicroorifice uniform-deposit impactor (MOUDI). Over 90% ofthe particulate mass was attributed to sub-micrometer particles.The carbon balance method was used to calculate theemission factors. Emission factors for four pollutants (particulatematter, CO2, total hydrocarbons, and NOx) were 2–4 timeshigher forbituminouscoals thanforanthracites. Inpast inventoriesof carbonaceous emissions used for climate modeling, thesetwo types of coal were not treated separately. The dramaticemission factor difference between the two types of coalwarrants attention in the future development of emissioninventories.

IntroductionCoal burning is a major source of the greenhouse gases (GHG)and airborne particulate matter (PM). China leads the worldin coal production and consumption. Coal is the dominantenergy source in China, accounting for 75 and 69% of itsenergy production and consumption (by calorific valuecalculations), respectively, in 2002 (1), and the predominanceof coal in the China energy structure is not expected to changesignificantly within the next few decades. A sizable fractionof the coal in China is consumed in small householdcombustion devices. In 2002, the residential coal consump-tion per capita was 59.4 kg, accounting for 5.6% of the totalcoal consumption (1066.9 kg/capita) (1). Household coal usein urban areas of China has decreased due to policies thatsubstitute gas and other, cleaner burning fuels, but in ruralareas household coal use seems to be increasing to relievethe pressure on biomass resources (2).

Emission factors of greenhouse gases and PM from coalused in household stoves are different from those used in

industrial combustion, since the combustion temperature islower and pollution control measures are generally absentin small stove combustion devices. Zhang et al. (3) system-atically measured emissions from 28 fuel/stove combinationsin China, a large fraction of the combinations in useworldwide, to provide the first database for emissions fromthis type of combustion. In their work, coal stoves includedbrick and metal stoves with and without flues. Not includedwas a “fire pit” (“Huo Tang” in Chinese), a householdcombustion device long used in provincial areas of China,especially the southwest. A primitive fire pit is composed ofthree stones in a pit in the living room. Figure 1 shows a0.6 m × 0.6 m fire pit in the floor of the family residence’smain room in Xuan Wei, China. In the center of the pit, fourto six bricks are arranged to surround the pile of wood orcoal. Between the bricks are open slots that allow air in andash out. A tripod is used to support the cooking pot.

Here we present emissions from burning 15 different fuelsin a laboratory system designed to mimic the fire pits usedin Xuan Wei County, China. In addition, Xuan Wei Countyhas been the focus of many studies linking high lung cancerrates with indoor coal burning in fire pits. The lung cancermortality rate for women in this county is China’s highest,and the men’s is among China’s highest (4). Within thiscounty, the lung cancer rates vary by 2 orders of magnitudeamong different communes that use different fuels. Epide-miological studies revealed a closer association of lung cancerwith the indoor burning of bituminous coal (as opposed toanthracite or wood) than with tobacco use or occupation(5–7). The gas and particle emission results presented in thispaper can be used to improve emission inventories for boththe local air pollution and health studies in Xuan Wei andthe global climate effects of household fire pits.

Experimental ProceduresSimulated Fire Pit. Figure 2 shows the laboratory stoveconstructed to simulate the fire pit shown in Figure 1. Ametal plate at the base contains the ash, and a square firebrickplate (30 cm × 30 cm × 2 cm) provides heat insulation. Fourfirebricks form a base with cross-grooves that allow stokingof the fire and removal of the ash, and six standing firebricksare arranged in a circle that encloses the fire. The assemblyis placed on an insulating fiber mat and an electronic scale(HP-40K) to monitor the fuel mass during the burningprocess. Above the fire is a 2 L stainless steel pot, filled with1.5 L of water. The pot is independently supported so thatthe mass losses due to fuel combustion and water vaporiza-tion can be measured separately.

Fuel Preparation. Figure 3 shows the different types offuels burned. As obtained from local mines in Xuan WeiCounty, the bituminous coals are chunks approximately 10cm in diameter. Before burning, they are broken into 2–5 cmdiameter lumps. The anthracites are mostly powder and aremade into briquettes in a method similar to that used inXuan Wei. The coal powder is mixed with yellow claypurchased from a local art store in a 4:1 by mass ratio. Thecoal and clay are mixed with water and formed into acylindrical pie-shaped mass ∼20 cm in diameter and ∼3 cmin height. The mixture is dried in an oven at 150 °C for about3 h and then broken into 2–5 cm diameter lumps. Besidesthe 15 coal samples, one oven-dried pine wood sample,collected from a household in the research area, was alsoburned. The kindling used to start the coal and wood fireswas Fatwood StarterStix. The composition data for the 15coals (8), categorized by coal rank, are given in the Supporting

* Corresponding author phone: 510-486-7002; fax: 510-486-7303;e-mail: [email protected].

† School of Public Health, University of California, Berkeley.‡ Lawrence Berkeley National Laboratory, University of California,

Berkeley.§ The Hong Kong Polytechnic University.4 Energy and Resources Group, University of California, Berkeley.

Environ. Sci. Technol. 2008, 42, 2503–2508

10.1021/es0716610 CCC: $40.75 2008 American Chemical Society VOL. 42, NO. 7, 2008 / ENVIRONMENTAL SCIENCE & TECHNOLOGY 9 2503

Published on Web 02/21/2008

Information (Table 2). No composition data are available foreither the pine wood sample or the kindling.

Fume Hood. The stove assembly is placed in a chemicalfume hood (Figure 4) with a 26 cm diameter duct fitted witha manual damper. The draft flow, calculated from the cross-sectional area and velocity measured by a hot-wire an-emometer at multiple locations in the duct, is maintainedat 6.4 m3/min. This flow rate was chosen to meet twoconstraints: high enough for the hood to contain all theemissions without visible spillage of the smoke but lowenough to not disturb the fire. The front glass panels of thehood were removed to further reduce the air velocity at thefire. The resulting air velocity at the front surface of the hoodis 0.1 m/s, lower than that of typical air currents (0.25 m/s)in a closed room (9, 10).

Burn Cycle. The coal fire requires about 15 min to light.First, three to five pieces of kindling wood (∼50 g) are lit bya propane torch. After 2 min, when the kindling is burningwell, ∼100 g of additional kindling is added. The kindling ispiled to fill the stove to the highest level possible with alimited amount of wood. Between 5 and 8 min, when somewood embers have built up in the bottom layer of the woodpile, about 500 g of coal are gradually added to the fire. If thecoal is difficult to light, 30 g of kindling are reserved andadded at minute 8. The coal is also divided, with 100 g reservedand added with the reserved kindling. At minute 12, whenmost of the kindling has burned and the bottom bed of thecoal has been lit, the water pot is placed above the fire. Thefire is not disturbed until about minute 20; at this point,more coal is added if the fire does not appear sustainable.If coal lumps become plastic and congeal into a large piece,an event that blocks the air flow and tends to extinguish thefire, the fire is stoked and poked.

The experimental cycle involves heating 1.5 kg of waterfrom ambient temperature to boiling and then simmeringfor 10 min. Water is brought to boil as rapidly as possibleduring the heating phase and is then maintained within5 °C of boiling during the simmering phase. These phasestake 20-60 min, so the entire process takes 30–70 mindepending on the fuel type. After the experiment, theremaining ash is weighed. The burn cycle is similar intime and scale as in the households of Xuan Wei, wherelighting a fire and cooking a meal usually takes 1–1.5 h.The fire can smolder for another hour if left alone. In thefield, the fire is allowed to die or it is maintained until thenext meal is cooked. In the laboratory, the fire isextinguished by a water spray after each experiment.

Because each type of coal burned differently, it wasdifficult at first to obtain repeatable results. Many coal fireswere studied to determine how best to obtain consistentresults; only results from fires that were considered “normal”are presented here.

Gas and Particulate Matter Monitoring. The apparatusfor monitoring gases and particles is illustrated in Figure 4.Total hydrocarbons (THC), oxygen (O2), nitrogen oxides(NOx), carbon monoxide (CO), and carbon dioxide (CO2) aremeasured each minute with a Horiba online gas-analyzersystem (FMA-220, CLA-220, AIA-210, and AIA-220). Thesystem is calibrated and zeroed daily. Particulate matter (PM)is monitored continuously with a MIE personal DataRAM1200 particulate monitor; the instrument is set to measurePM2.5 in the 0.001–400 mg/m3 range (11). The instrument iscalibrated by comparing the recorded time-weighted average(TWA) with the TWA measured gravimetrically on a filterdownstream from the optical chamber.

Size-Segregated Particle Sample Collection. Isokineticair samples are also drawn from an 18 mm diameter thin-wall probe. A microorifice uniform deposit impactor (MOUDIModel 110, MSP Corp.), operating at a flow rate of 30 L/min,is used to collect size-fractionated samples on polycarbonatefilters (FPC4537 filter, Zefon International). The filter samplesare measured gravimetrically and examined with scanningelectron microscopes. The impactor has 50% cut-pointaerodynamic diameters of 10, 5.6, 3.2, 1.8, 1.0, 0.56, 0.32,0.18, 0.1, and 0.056 µm on stages one through ten, respectively.A 37 mm Teflon filter is used as the after-filter. Flow rate andstagnation pressure are continuously monitored. To preventoverloading the filters, particle samples are collected for 30 severy 3 min during the burn cycle.

Airborne Particle Sample Collection. Airborne particlesamples are collected with various filter media for differentanalytic techniques. Teflon membrane filters are used formass weighing and elemental analysis, silver membranefilters for electron microscopy studies, and Pallflex Tefloncoated on glass-fiber filters for free radical and PAHanalysis, and for toxicity tests. The system for collectingairborne particle samples is not isokinetic in that the inletvelocity in the probe is 5 times higher than in the duct (seeFigure 4). Air is drawn from the duct through ¼ in. stainlesssteel sampling probes 0.5 m downstream of the duct inletand 1 m upstream of the damper. Calculations estimatethat up to 20% of particles 3–10 µm in size are lost by theprobe, but the collection efficiency of particles smallerthan 3 µm is not affected (12). The average stack tem-perature at the location of the sampling probes during thecombustion experiments was 30 ( 1 °C, which is 5 °Chigher than that of ambient air. Copper lines are usedbetween sampling probe and the ports for filter samplingand gas monitoring. The mass of particulate matter on the37 mm Teflon filters is measured with a Cahn 29 ElectroBalance (sensitivity, 0.001 mg). Humidity and temperaturein the weighing room were held at 35 ( 5% and 75 ( 5 °F,

FIGURE 1. The fire pit used in Xuan Wei, China

FIGURE 2. Simulated fire pit in the laboratory.

2504 9 ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 42, NO. 7, 2008

respectively. All filters are equilibrated to the roomtemperature and humidity for at least 24 h before weighing.

EC/OC Analysis. The thermal/optical reflectance (TOR)method was applied to determine organic and elementalcarbon (EC/OC) (13). Quartz fiber filter samples wereanalyzed for OC/EC by DRI Model 2001 thermal/opticalcarbon analyzer. The protocol involves heating a 0.526 cm2

punch aliquot of a filter stepwise in a nonoxidizing helium(He) atmosphere to measure organic carbon, and in anoxidizing atmosphere of 2% oxygen in a balance of heliumto measure elemental carbon (14).

Emission Factor Calculation. The carbon balancemethod (3, 15), which does not require the measurementof the duct flow rates, was applied to calculate the emissionfactors (EF) of particulate matter and gases. The resultswere comparable to those obtained by the direct methodof EF calculation (8), which multiplies the average ductconcentration by the air flow rate in the duct and theduration, divided by the fuel consumed. The mass balancefor carbon is described as Cf - Ca ) CCO2 + CCO + CTHC +CPM, with Cf ) carbon mass in the fuel and Ca ) carbonmass in the bottom ash, including the unburned solid fuel

FIGURE 3. Photographs of the fuels burned: (a) kindling used to light the fire; (b) pine wood from Xuan Wei, China; (c) chunks of abituminous coal; (d) powder of an anthracite; (e) yellow clay used as a binder to make coal briquettes; (f) briquettes made from dand e.

FIGURE 4. Schematic of the experiment.

VOL. 42, NO. 7, 2008 / ENVIRONMENTAL SCIENCE & TECHNOLOGY 9 2505

or char. Rearranging and defining K as the ratio of theairborne carbon released as products of incompletecombustion to the carbon released as CO2 leads to

K)CPIC ⁄ CCO2

If the K-factor and the amount of fuel carbon consumed aredetermined for a burn cycle, the emission factor of CO2 permass of fuel consumed (Mf) can be calculated as

EFCO2) 44(Cf -Ca) ⁄ 12Mf(1+K)

Similar expressions are used for other pollutants, includingNOx using its molar ratio to CO2.

Results and DiscussionCharacterization of Coals. Three anthracite and 12 bitu-minous coals from Xian Wei were tested for their moisturecontent, ash, volatile, and fixed carbon content and carbon,hydrogen, nitrogen, sulfur, and oxygen content (Table 2 inthe Supporting Information). The bituminous coals allcontained less than 0.5% sulfur, while the anthracite coalshad 3–5% sulfur. According to the Chinese coal qualitystandards (GB/T 15224-2004), the three anthracites areclassified as high-sulfur (>3%) and high-ash (>30%); the 12bituminous coals contain extremely low levels of sulfur(<0.5%), and their ash content ranged from 13 to 32%.

Real Time Emissions. Figure 5 shows real-time gas andparticle emission monitoring data of single representativeruns for kindling wood, and bituminous and anthracite coals.The two initial peaks for some experiments are a result ofadding reserved kindling and coal for those coals that aredifficult to ignite. About 150 g of kindling (Fatwood StarterStix)were needed to start a coal fire. The kindling was burnedseparately 3 times without adding any coal to determine

how long the wood fire would persist. As shown in the uppergraph in Figure 5, the kindling burned completely in about15 min. Visible flame combustion lasted until the 12th minute,after which the embers smoldered until extinction. By the12th minute, over 90% of the kindling mass was burned andthe CO2 concentration in the duct fell below 0.1%. Each coalburning cycle was divided into two stages; “kindling stage”and “coal stage”, with the 12th minute used as the cut point.Two separate filter samples were collected for the differentstages; the comparison among different coals is based onthe samples collected during the coal stage.

As seen in the lower two graphs in Figure 5, the PMemission profiles are similar during the kindling burningstage, when the concentrations of all the monitored emissionpollutants peak at the fifth minute. Until the 12th minute,combustion is dominated by the kindling, although a smallfraction of the coal begins to devolatilize and burn. An averageof only 12% of the coal mass is burned during the kindlingstage. The anthracite/kindling produces more emissionsduring this period, and the time behavior is repeatable.However, it is difficult to separate the coal emissions fromthe wood emissions during this period. During the sustainedcoal-burning stage the emission profiles are substantiallydifferent between the two types of coals. In the anthracitecoal experiments, the concentration of THC drops continu-ously, whereas it remains relatively constant during thebituminous coal experiments. PM emission from anthracitecoal remains less than 1 mg/m3, and there is no visible smokein the hood; PM emission from bituminous coal is an orderof magnitude higher, and there is always visible smokethroughout the run. This observation is consistent with theChinese names for anthracite and bituminous coal: “smoke-less coal” and “smoky coal”, respectively.

Particle Size Distribution. Figure 6 shows the mass sizedistributions of the particles from anthracite, bituminouscoal, and pine wood. Two of the most studied coal samples

FIGURE 5. Real time monitoring of three representative burncycles of anthracite, bituminous coal, and kindling wood.

FIGURE 6. Particle size distributions of emissions fromanthracite, bituminous coal, and fuel wood.

2506 9 ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 42, NO. 7, 2008

in the previous epidemiology research (called RS and HC)were selected here to represent anthracite and bituminouscoal, respectively (8). Included are data from seven tests:one for anthracite and three tests each for wood and for onetype of bituminous coal. Particles collected on stage 0 of theMOUDI sampler, with a cut point of 18 µm, are not reportedbecause of the low collection efficiency of particles at thisstage. Significant amounts of mass were detected on the after-filters as is discussed below; the lower size limit of particlescollected was arbitrarily selected as 20 nm.

The pine emissions consistently produced a bimodaldistribution with one peak at 0.10–0.18 µm and the other at0.32–0.56 µm. This is similar to the bimodal distributionreported by Rau (16) for residential wood combustionparticles. In the bituminous coal emissions, no bimodaldistribution was observed. The mass at the after-filter stage(particle diameter< 56 nm) ranged from 2 to 8% in the woodand bituminous emissions. It is possible that a fraction ofthe mass at this stage is due to adsorption of volatile orsemivolatile organics on the Teflon membrane after-filter.In the case of anthracite emissions, the mass at the after-filter stage accounted for even a larger fraction (17%) of thetotal mass because of the overall extremely low particleemission during anthracite coal combustion (see Figure 5).

For the PM emission from all three fuels, over 90% of theparticulate mass is attributed to sub-micrometer particles.These results are different from one previous study in XuanWei, which indicated that 51% of the particle mass frombituminous coal combustion was less than 1 µm in diameterand only 6% of particle mass from wood combustion wasless than 1 µm (4). These differences may be partly due tothe different measurement methods: the previous study usedscanning electron microscopy to measure the particle size,and the calculation was based on volume; in the currentstudy the particle sizes are fractionated by aerodynamicdiameter using a MOUDI sampler and were based on massmeasurement. The finer particle size distribution indicatedby the current study is consistent with other reports in boththe older and more recent literatures (16–19).

Correlation between Gaseous THC and PM. Figure 7shows scatter plots of THC vs PM and CO vs PM. The currentstudy, with various fuel types and a single stove, found nosignificant correlation between CO and PM (R ) -0.05, P )0.74). In contrast, some previous studies (18–20) reportedsignificant correlations of various strengths between thesetwo pollutants from various stove/fuel systems. We observedsignificant correlations between emission factors of gaseousTHC and of PM (R) 0.80, P< 0.001). Gaseous THC emissionswere found to be highest for bituminous coals and lower foranthracites and wood (Figure 7); this coincides with the highertar-producing capability of bituminous coals (21). GaseousTHC may serve as an index of the availability of all precursors,gaseous and condensable, in the flame. The correlationbetween gaseous THC and PM suggests that the productionof PM is dominated by the availability of precursors ratherthan by the combustion environment, as the current studyinvestigates several fuel types and a single stove.

Further analysis shows that it is mainly organic carbonemission that correlates with gaseous THC (R ) 0.83, P <0.001), while elemental carbon does not (R ) 0.18, P ) 0.18).The elemental carbon ranges from less than 10% to over50%, with the wood having the highest EC content, and thecoal burning after the 12th minute the lowest. In the scatterplot of gaseous THC vs PM, the observations fall into threeclusters: bituminous coals in the middle to upper right andthe two other fuels, anthracite and wood, in the lower left.

Effect of Fuel Nitrogen on of NOx Emission. Figure 8shows the scatter plot of nitrogen content in coal versusemission factors of NOx. Fuel nitrogen is significantlycorrelated with the NOx emission factor (R) 0.88, P< 0.001).

The combustion temperature of coal in the fire pit had arange of 770–1100 °C (22). As discussed by Sarofim (23), thecontribution of atmospheric nitrogen to coal-generated NOx

becomes significant only at high temperature (>2100 °C)operations such as those encountered in cyclone burners.The mass fraction of nitrogen is below 1.5% in the 15bituminous coals measured in the current study, and thecorresponding emission factor is below 4.0 g/(kg of fuel),while anthracite contains less than 0.8% nitrogen and yieldsless than half as much NOx as the bituminous coals.

Emission Factors. The emission factors (g/kg) for an-thracite and bituminous coals on a fuel mass basis are shownin Table 1.

FIGURE 7. Relationship between the emission factors of PM,THC, and CO.

FIGURE 8. Effect of nitrogen content in coal on the emissionfactor of NOx. The error bar indicates one standard error.

VOL. 42, NO. 7, 2008 / ENVIRONMENTAL SCIENCE & TECHNOLOGY 9 2507

The summary statistics for the anthracite coal group andbituminous coal group were determined from the means forthree anthracite and 12 bituminous coals, respectively, witha minimum of three experiments for each individual coal (62total coal experiments). A detailed table (Table 3) is presentedin the Supporting Information.

For coal fires, emission factors for four pollutants, allexcept for CO, are significantly higher from the bituminouscoals than the anthracites. Bituminous coal PM emissionfactors (11.2 g/kg) are nearly 4 times as high as those ofanthracites (2.9 g/kg). Similarly, Butcher and Ellenbecker(24) reported a PM emission factor for bituminous coal, 10.4g/kg, which was 20 times larger than that for anthracite,0.33–0.62 g/kg. In past inventories of carbonaceous emissionsused for climate modeling, these two types of coal were nottreated separately (25). The dramatic emission factor dif-ference between the two types of coal warrants attention inthe future development of emission inventories.

AcknowledgmentsWe thank Kirk Smith for helpful discussions and RobertChapman of the USEPA for the coals. We also thank ScottMcCormick and his staff for their assistance. This work wassupported by the Environmental Health Sciences SuperfundBasic Research Program (Grant No. P42ESO47050-01) fromthe National Institute of Environmental Health Sciences, theWood Calvert Chair in Engineering (UCB), the NationalCancer Institute, and the University of California ToxicsSubstances Research and Training Program.

Supporting Information AvailableTables on the proximate and ultimate analyses of the 15coals and the emission factors of the pollutants from eachexperiment. This material is available free of charge via theInternet at http://pubs.acs.org.

Literature Cited(1) Lin X. Y. China Energy Statistics Yearbook 2000–2; China Statistics

Publishing House: Beijing, 2004.(2) Smith K. R.; Tian L. W. Indoor coal smoke and cancer. In Air

Pollution and Cancer; Cogliano V, Straif K, Samet J., Eds.; IARC:Lyon, France; in press.

(3) Zhang, J.; Smith, K. R.; Ma, Y.; Ye, S.; Jiang, F.; Qi, W. Greenhousegases and other airborne pollutants from household stoves inChina: A database for emission factors. Atmos. Environ. 2000,34, 4537–4549.

(4) Mumford, J. L.; He, X. Z.; Chapman, R. S.; Cao, S. R.; Harris,D. B.; Li, X. M. Lung cancer and indoor Aair-pollution in Xuan-Wei. China Sci. 1987, 235 (4785), 217–220.

(5) Mumford, J. L.; Harris, D. B.; Williams, K. Indoor air samplingand mutagenicity studies of emissions from unvented coalcombustion. Environ. Sci. Technol. 1987, 21, 308–311.

(6) Chapman, R. S.; Mumford, J. L.; Harris, D. B.; He, Z. Z.; Jiang,W. Z.; Yang, R. D. The epidemiology of lung cancer in Xuan Wei,China: Current progress, issues, and research strategies. Arch.Environ. Health 1988, 43 (2), 180–5.

(7) He X.; Yang R. Lung Cancer and Indoor Air Pollution from CoalBurning; Yunnan Science and Technology Publishing House:Kunming, China, 1994.

(8) Tian L. W. Coal Combustion Emissions and Lung Cancer inXuan Wei, China. Ph.D. dissertation, University of California,Berkeley, CA, 2005.

(9) Ballard-Tremeer G. Emissions of Rural Wood-Burning CookingDevices. Ph.D. dissertation, University of the Witwatersrand,Johannesburg, South Africa, 1997.

(10) Ballard-Tremeer, G; Jawurek, H. H. The ”hood method” ofmeasuring emissions of rural cooking devices. Biomass Bioenergy1999, 16 (5), 341–345.

(11) MIE. Instruction Manual: Personal DataRam Models pDR-1000AN and pDR-1200; MIE: Bedford, MA, 2000.

(12) Hinds W. C. Aerosol Technology: Properties, Behavior, andMeasurement of Airborne Particles, 2nd ed.; Wiley: New York,1999.

(13) Chow, J. C.; Watson, J. G.; Chen, L. W. A.; Arnott, W. P.;Moosmiller, H. Equivalence of elemental carbon by thermal/optical reflectance and transmittance with different temperatureprotocols. Environ. Sci. Technol. 2004, 38, 4414–4422.

(14) Cao, J. J.; Lee, S. C.; Ho, K. F.; Zhang, X. Y.; Zou, S. C.; Fung, K.Characteristics of carbonaceous aerosol in Pearl River DeltaRegion, China during 2001 winter period. Atmos. Environ. 2003,37 (11), 1451–1460.

(15) Smith, K. R.; Khalil, M. A. K.; Rasmussen, R. A.; Thorneloe, S. A.;Manegdeg, F; Apte, M. Greenhouse gases from biomass andfossil-fuel stoves in developing countriessA Manila pilot study.Chemosphere 1993, 26 (1–4), 479–505.

(16) Rau, J. A. Composition and size distribution of residential woodsmoke particles. Aerosol Sci. Technol. 1989, 10 (1), 181–192.

(17) Bond T. C. Covert D. S. Kramlich J. C. Larson T. V. Charlson R. J.Primary particle emissions from residential coal burning: Opticalproperties and size distributions. J. Geophys. Res., [Atmos.] 2002,107, D21.

(18) Venkataraman, C.; Rao, G. U. M. Emission factors of carbonmonoxide and size-resolved aerosols from biofuel combustion.Environ. Sci. Technol. 2001, 35, 2100–2107.

(19) Kleeman, M. J.; Schauer, J. J.; Cass, G. R. Size and compositiondistribution of fine particulate matter emitted from woodburning, meat charbroiling, and cigarettes. Environ. Sci. Technol.1999, 33, 3516–3523.

(20) Gupta, S.; Saksena, S.; Shankar, V. R.; Joshi, V. Emission factorsand thermal efficiencies of cooking biofuels from five countries.Biomass Bioenergy 1998, 14 (5–6), 547–559.

(21) Mitra, A.; Sarofim, A. F.; Barziv, E. The influence of coal typeon the evolution of polycyclic aromatic-hydrocarbons duringcoal devolatilization. Aerosol Sci. Technol. 1987, 6 (3), 261–271.

(22) Xian Y., Qiao J., Wang G. Personal exposure levels of air pollutantsin Xuan Wei. In Lung Cancer and Indoor Air Pollution from CoalBurning; He, X., Yang, R, Eds.; Yunnan Science and TechnologyPublishing House: Kunming, China, 1994; pp 151–157..

(23) Sarofim A. F. Pollutant formation and destruction. In Funda-mentals of the Physical-Chemistry of Pulverized Coal Combus-tion; Lahaye, J., Prado G., Eds.; Kluwer Academic Publishers;:Boston Hingham, MA, 1987; pp 245–268..

(24) Butcher, S. S.; Ellenbecker, M. J. Particulate emission factors forsmall wood and coal stoves. J. Air Pollut. Control Assoc. 1982,32 (4), 380–384.

(25) Cooke, W. F.; Liousse, C.; Cachier, H.; Feichter, J. Constructionof a 1 degrees × 1 degrees fossil fuel emission data set forcarbonaceous aerosol and implementation and radiative impactin the ECHAM4 model. J. Geophys. Res., [Atmos.] 1999, 104 (D18),22137–22162.

ES0716610

TABLE 1. Mean Emission Factors (g/kg) for the Two CoalTypes

mean emission factor (g/(kg of fuel))

pollutant emitted anthracite coals bituminous coals

PM 2.9 ( 0.3 11.2 ( 1.0CO2 1751 ( 105 2269 ( 59CO 92.9 ( 8.7 73.7 ( 3.7THC 15.5 ( 1.3 37.0 ( 3.1NOx 1.10 ( 0.17 2.61 ( 0.16

2508 9 ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 42, NO. 7, 2008