Policy Analysis. Estimate of Mercury Emissions to the Atmosphere from Petroleum. EPA Estimates of Mercury in Crude Oil.

Transcription

1 Policy Anlysis Estimte of Mercury Emissions to the Atmosphere from Petroleum S. MARK WILHELM* Mercury Technology Services, Luthern Church Rod, Tombll, Texs An estimte of the contribution of mercury to the tmospheric environment from petroleum processed in the United Sttes is constructed from recent dt. The estimte is bsed on mss blnce pproch for mercury in crude oil, in ined products, nd in wste strems (ir, wter, solid wste) from ineries. Although there re insufficient dt t present to hve high degree of confidence in the men mount nd rnge of mercury concentrtions in crude oil or in ined products, the frmework of the estimte should ssist direction for the cquisition of dditionl dt. On the bsis of selected dt tht put the estimted men concentrtion of totl mercury in crude oil close to 10 ppb, it is clculted tht the totl mount of mercury in U.S. petroleum processed yerly is slightly over 8000 kg/yr. Of this mount, pproximtely 6000 kg/yr is estimted to be emitted to the tmosphere from combustion of liquid hydrocrbon fuels, which represents bout 10% of the U.S. yerly emission rte of tmospheric mercury from col combustion. The mteril blnce predicts tht the mount of mercury in ir emissions from ll U.S. ineries is on the order of 1500 kg/yr bsed on the ssumption tht fugitive mercury emissions from ineries re negligible. Atmospheric emissions of mercury from fuel oil burned in the United Sttes re estimted in the U.S. EPA Mercury Report to Congress to be pproximtely kg/yr, nd this estimte my be in error on the high side by fctor of If the men mounts of mercury in U.S. distillte nd residul fuel oils re in the rnge of 5-15 ppb, s suggested by more recent dt, then U.S. fuel oil combustion should contribute no more tht bout kg/yr (emission rtio ) 1) of mercury to the tmospheric burden. Introduction The U.S. EPA recently nnounced (December 2000) tht regultion of mercury nd other hzrdous ir emissions from oil- nd col-fired electric utility stem-generting units is necessry nd pproprite. The EPA decision derived from the mndte given to the EPA by Congress under the Clen Air Act nd followed numerous studies of the hzrds to the public helth nticipted to occur s result of hzrdous ir emissions by electric utility stem-generting units. The studies tht were the bsis of the EPA decision re summrized in the Report to Congress (1) nd included exmintion of mercury emissions from both col nd oil combustion. While the concentrtion of mercury in U.S. cols is * Corresponding uthor phone: (281) ; fx: (281) ; e-mil: smw@hgtech.com. TABLE 1. THg in Crude Oil by INAA (7) source mount SD notes Cliforni detection limit ) 4 ppb Cliforni Cliforni Cliforni Cymric Cliforni Liby Liby Liby Louisin Wyoming men rnge ppb known with some certinty, the EPA dmitted tht the mounts of mercury in crude oil nd fuel oils were much less certin. Under the Emergency Plnning nd Community Right- To-Know Act (EPCRA), the U.S. EPA requires certin fcilities processing or using listed toxic chemicls to report the nnul quntity of such chemicls entering the environment (2). The reporting requirements re termed the toxic relese inventory or TRI. In 2000, the reporting threshold for mercury ws revised from lb to 10 lb. The current (June 2001) U.S. EPA estimte (for TRI guidnce purposes) proposes tht the verge concentrtion of mercury in crude oil is 1.5 ppm, bsed on dt on mercury concentrtions in crude oil compiled by Wilhelm nd Bloom (3). The newly proposed guidnce estimte replces prior 6 ppm estimte proposed in n erlier (November 2000) drft of the guidnce document. The estimte of 6 ppm proposed erlier by the EPA ws bsed on dt compiled by Brooks (4). The U.S. EPA estimtes (4-6) of the men concentrtion of mercury in crude oil re likely in error for resons to be discussed, but they stem minly from relince on dt from nomlous sources tht skew the smple ensembles used in previous compiltions. The dt nd nlysis provided herein lso cst some doubt on the EPA estimtes for mercury concentrtions in fuel oil nd the resulting verge U.S. emissions estimtes. Using more recent dt, n ttempt is mde to construct better frmework for estimting tmospheric mercury emissions from petroleum. EPA Estimtes of Mercury in Crude Oil The U.S. EPA estimtes of the men concentrtion of totl mercury (THg) in crude oil, both historic (4-6) nd present (1), re bsed primrily on dt compiled by Brooks (4). Brooks compiltion consisted mostly of dt published by Shh, Filby, nd vrious collegues (7, 8) who were the first to pply instrumentl neutron ctivtion nlysis (INAA) to U.S. crude oils. Shh et l. (7) reported concentrtions for 10 crude oils s shown in Tble 1 in The dt of Shh et l., in its severl reported forms, re the bsis for U.S. EPA estimtes (4-6) of prts per million levels for the men mount of mercury in crude oil. The exercise (by EPA) to rrive t men mount involved verging the men or medin of the rnge of concentrtions from the erly studies of Shh nd Filby (7, 8). This method to rrive t men concentrtion is questionble in the sense tht the verge ws constructed from severl ppers nd reports tht contin the sme dt but were considered s seprte crude oil source studies (4) ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 35, NO. 24, /es001804h CCC: $ Americn Chemicl Society Published on Web 10/31/2001

2 TABLE 2. THg in Albert Crude Oils (9) strtigrphic er smples no. bove DL high low men medin Upper Cretceous DL Lower Cretceous DL Jurssic DL Trissic DL Crboniferous DL Devonin DL totl DL Clculted ssuming <DL ) 1 ppb. SD DL TABLE 3. THg in Crude Oil nd Gs Condenstes smples rnge THg men SD notes (12) crude oil Libyn (13) crude oil (14) crude oil 76 NR ll (14) crude oil 39 NR lowest 39 smples (14) crude oil 37 NR highest 37 smples (15) crude oil 1 <1 Asi (16) condenste 4 NR 15 (14) condenste 18 NR (14) condenste 8 NR lowest 8 smples (14) condenste 10 NR highest 10 smples (17) condenste SE Asi (15) condenste Asin NR, not reported. One of the crude oils exmined by Shh nd Filby (Cliforni Cymric) ws unusul in hving hd totl mercury concentrtion (THg) bove 20 ppm. This crude ws extensively nlyzed during the 1970s becuse the high mercury concentrtion cused processing difficulties for ineries, nd hence it ws included in most of the erly investigtions. In ddition, the high concentrtion lent itself to nlyticl method development nd thus becme populr mong nlysts. The repetitive inclusion of Cymric THg dt in the erly dt compiltions used by the EPA to derive emissions estimtes provided disproportionte emphsis on this source s representtive of crude oils. For resons unknown, the EPA compiltion of dt for mercury in crude oil (4), upon which its estimtes re bsed, did not include or consider dt published by Filby nd Shh nd Hitchon et l. (9, 10), who mesured totl mercury in 86 crude oils (nd two tr snds) from Albert, Cnd. Thirtynine smples hd THg below the detection limit (DL) of 2 ppb. Forty-nine smples were bove DL with men of 50.0 ppb (mximum concentrtion of 399 ppb). The dt re summrized in Tble 2. The verge of 86 crude oils (22 ppb) ws clculted by ssigning vlue of hlf DL to those exhibiting THg below the detection limit. The discrepncy in the men mount of mercury in crude oil s clculted from the dt of Shh nd Filby (3-6 ppm) nd from Hitchon nd Filby (22 ppb) using the identicl nlyticl method begs the question s to which, if either, is closer to ctul men. Neither smple ensemble is sttisticlly representtive of U.S. processed crude oils, but Hitchon s dt re more rndom ensemble for prticulr region (Albert, Cnd) nd spn ll the geologic strt. The U.S. EPA clculted men concentrtions of mercury in crude oil processed domesticlly likely re in error. The principl reson is tht the older dt contin n inordinte contribution of mesured concentrtions from the Cymric field in Cliforni. The Cymric field contributes pproximtely 17 of the roughly 5000 million brrels of crude oil processed nnully in the United Sttes (foreign nd domestic) or bout TABLE 4. Recent THg in Crude Oil Dt smples rnge men THg SD notes (18) crude oil 2 <10 Middle Est (18) crude oil 4 <10 Afric (18) crude oil 11 ND North Americ 1560 Cymric (18) crude oil 4 <10 Asi (18) crude oil 4 <10 South Americ (18) crude oil 1 <10 North Se (19) crude oil Afric (19) crude oil Middle Est (19) crude oil Cnd (19) crude oil North Se (19) crude oil Mexico (19) crude oil South Americ (19) crude oil mixed (20) crude oil 11 <15 U.S. (20) crude oil 2 <15 Cnd (20) crude oil 2 <15 Mexico (20) crude oil 3 <15 South Americ (20) crude oil 2 <15 North Se (21) crude oil 8 < Cndin imports 0.34% (11). Concentrtions of mercury in Cymric crude, however, ccount for more thn 10% of the dt used by the EPA to clculte n verge concentrtion of mercury in crude processed in the United Sttes. Given tht Cymric hs close to the highest reported concentrtion of mercury in the world, the EPA estimtes (1) re likely mny times higher thn they would be if clcultions of the verge were bsed on volumeweighted contributions. Recent Dt on Mercury in Crude Oil nd Condenste More recent dt for totl mercury concentrtions in crude oil nd gs condenstes re compiled in Tbles 3 nd 4. Tble 3 summrizes THg in crude oils tht re not clerly identified in terms of origin or import sttus. Mus et l. (12), using neutron ctivtion nlyticl methods, reported THg VOL. 35, NO. 24, 2001 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

3 TABLE 5. THg in Refined Products smples rnge men THg SD notes (22) gsoline NR U.S. (22) gsoline NR foreign (22) diesel NR U.S. (22) diesel NR foreign (22) kerosene NR U.S. (22) heting oil NR U.S. (14) light distilltes 14 NR* U.S. (14) utility fuel oil 32 NR U.S. (14) sphlt 10 NR U.S. (16) nphth NR (15) nphth NR Asin (27) petroleum coke U.S. NR, not reported. in Libyn crude oils to be in the rnge of ppb. Ling et l. (13) reported the men concentrtion of mercury in 11 crude oils s 4 ppb (rnge of 1-7 ppb). Bloom (14) found totl Hg in unfiltered crude oils rnging between sub-prts per billion levels to oversturtion (severl prts per million). Also included in Tble 3 re dt for mercury in condenstes (15-17) tht re hydrocrbon liquids originting from gs fields but processed in ineries s opposed to gs plnts. Condenstes cn be ined or used s feeds to petrochemicl mnufcture. The dt tht pper in the literture lect the fct tht mercury is problemtic to petrochemicl processes when concentrtions exceed few prts per billion. Thus it is not likely tht the dt in Tble 3 re representtive of gs field condenstes s whole. One might surmise, however, tht condenstes my hve higher verge THg thn crude oils. Recent dt for mercury in crude oil tht include informtion on origin, nd theore processed volumes re summrized in Tble 4. Mgw et l. (18) reported THg for 26 crude oil s tht re purchsed by U.S. (West Cost) ineries nd tht spn the mjor U.S. crude strems, both domestic nd imported. Mesured vlues re ll less thn 10 ppb (the detection limit of the CVAA instrument) excepting one Cliforni crude oil (Cymric) reported contining 1.5 ppm THg. The single prts per million vlue of the Cymric crude rises the men clculted from Mgw s dt from 5 to 146 ppb. The New Jersey Deprtment of Environmentl Protection Mercury Tsk Force performed study of oil processed in New Jersey ineries. The dt, reported by Morris (19), indicted mercury concentrtions in U.S Est Cost crude oil imports between 1 nd 10 ppb. Morris lso identified crude oil origin, but the number of smples nlyzed for ech source nd stndrd devitions were not reported. The men mount of mercury in crude oil imported to the U.S. Est Cost ineries, ccording to Morris, is close to 5 ppb. Environment Cnd (20) compiled dtbse on oil properties tht included multi-element metls nlysis using inductively coupled plsm-tomic emission spectrometry (ICP-AES). The nlyticl method is not highly sensitive or selective for mercury. The dt, orgnized by region of origin in Tble 4, re below 15 ppb, the mercury detection limit of the instrument (ICP-AES). Duo et l. (21) reported nlyticl dt for 8 crude oils described s representtive of 50% of ll crude oil processed in Cnd. The exct origins of the crude oils were not divulged, but mny of these sme oils re likely processed in the United Sttes s well s Cnd. The method used ws vrition of digestion/cvaa. The method hd minimum detection limit for mercury of 2 ppb. Most of the dt re below this detection limit s shown in Tble 4. TABLE 6. Dt for THg in Fuel Oils Used by the U.S. EPA To Clculte Emission Fctors smples rnge men SD notes (23) RFO 6 3 <100 Brooks (4) (24) RFO 6 20 Brooks (4) (25) RFO Brooks (4) (26) DFO Brooks (4) (5) RFO (medin) (5) DFO 2 <120 (6) RFO Totl mercury concentrtions in crude oil cnnot be nlyzed sttisticlly t present, in prt becuse of the uncertinties in the nlyticl methods nd lso due to the fct tht much of the dt reported in the literture re not well-documented s to origin. While the mjority of recently cquired dt re reported less thn 10 ppb totl mercury, there re notble exceptions in the prts per million rnge, principlly Bloom (14). The men concentrtions for totl mercury in condenstes nd in crude oil (1.5 ppm for crude oil, 4.0 ppm for condenste) tht Bloom reported re much higher thn other reported dt. This is due to the fct tht in both cses the dt sets contin lrge number of smples from somewht nomlous sources tht present processing difficulties nd hence re extensively nlyzed. Bloom s reported men mounts were clculted from the number of smples nlyzed nd not bsed on crude oil sources. Inclusion of Bloom s dt when clculting men concentrtion for mercury in U.S.-processed crude oil leds to n obvious error becuse concentrtions re not lective of source distribution. The upper hlf of Bloom s dt (39 smples) ll lect smples from one field (South Americ) tht produces less thn brrel/dy (10 million brrel/ yr) or bout 0.2% of crude processed in U.S. ineries nnully (ssuming this crude is processed in the United Sttes, which is not ctully known). The condenstes in the upper hlf of Bloom s dt come mostly from the Gulf of Thilnd. These Asin condenstes re not processed in the United Sttes but re prevlent in reported dt (14, 15, 17) becuse they re problemtic to petrochemicl mnufcture. The men of the lower hlf of condenste smples nlyzed in Bloom s lbortory ws reported s pproximtely 20 ppb. Mercury in Refined Products Compiled in Tble 5 re reported dt for mercury in ined products. Bloom (14) reported mercury in U.S. light distilltes nd fuel oil close to 1 ppb (46 smples). Ling et l. (22) reported mercury in U.S. gsoline nd diesel less thn 1 ppb. Tble 6 summrizes the older dt for mercury concentrtions in fuel oil used by the U.S. EPA to clculte emissions fctors (23-26) compiled by Brooks (4). These dt were obtined prior to 1982 nd do not disclose fuel oil origin by region or inery. Thus, there ppers to be some disgreement between THg in fuel oil dt reported in EPA studies from the 1970s nd the more recent dt. A focused study using sttisticl smpling nd modern nlyticl methods would be required to rrive t better verge. A sttisticl ensemble for mercury in inery products does exist in one cse. Totl mercury in petroleum coke hs been reported s prt of the U.S. EPA reporting requirements on fuel feeds to utility boilers (27), nd the men is pproximtely 50 ppb (1000 dt points, 2 million t). The distribution of mercury concentrtions in petroleum coke is shown in Figure 1. Estimtion Method for Mercury Emissions from Petroleum One method to estimte mercury emissions from petroleum is to consider the entire ining process (see Figure 2) s ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 35, NO. 24, 2001

4 FIGURE 1. Distribution of totl mercury concentrtions in petroleum coke. FIGURE 2. Schemtic of typicl crude oil ining process. well s the vrious combustion processes for the fuel products derived from crude oil. Centrl to the estimtion process is the men mount of mercury in crude oil. A sttisticl ensemble of mercury concentrtions in crude oils does not exist s it does for col (27). Some of the existing dt for mercury in crude oil were obtined to identify problemtic feeds to processing, nd thus men concentrtions clculted from these dt re likely higher thn ctul. Mercury concentrtion must be resonbly sptilly homogeneous in petroleum reservoir (becuse of its fluid nture), but most reported mercury concentrtions from nlysis of totl mercury in crude oil from prticulr reservoir cnnot be extrpolted to geologiclly specific concentrtion. This is due to the fct tht the process smple loction for crude oil strongly ffects the mount tht is mesured in totl mercury nlysis. Crude oil contins both dissolved nd suspended mercury compounds. A typicl nlysis for totl mercury in crude oil yields the sum of both forms, but the suspended mount is not necessrily lective of the reservoir concentrtion. The smpling nd nlyticl protocols for differentition of the vrious forms of mercury in hydrocrbon liquids hve been only recently pplied, but the vritions seen in THg from prticulr fields re due more to vrition in the suspended frction s opposed to the truly dissolved frction. For exmple, order of mgnitude differences in totl mercury concentrtion cn be obtined for the sme crude oil between smples tken upstrem nd downstrem of seprtors nd deslters. Crude oil smpled from tnkers typiclly comes from the top of the tnk where suspended forms cn be lower in concentrtion due to settling in trnsit. Uncertinty in reported nlyticl dt lso comes from the fct tht smple continers my or my not be gitted prior to nlysis to resuspend the solid forms tht settle to the bottom of continers. Similrly, some crude oil smples re filtered prior to nlysis, nd thus the reported mercury concentrtion is ctully dissolved mercury only. Oxidtion of elementl mercury by oxygen from exposure of smples to ir or by sulfur in blended oils is lso postulted to disrupt the nturl distribution of species nd thus to obfuscte not only species distribution but lso totl mercury mesurements s well (3). The ftes of dissolved nd suspended forms of mercury in the ining process (see Figure 2) re different. The suspended forms (mostly HgS) re nonvoltile nd tend to VOL. 35, NO. 24, 2001 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

5 TABLE 7. Estimtes of Mercury in Crude Oil nd Refined Products (1999) (11) brrel/yr (11) (10 9 ) kg/yr (10 11 ) THg estd totl (kg/yr) crude oil domestic (40%) Alsk (18%) GOM (20%) other (62%) ? 2000? imported (60%) Cnd (15%) Mexico (15%) Middle Est (20%) other (50%) ? 2400? totl (in) ined products d ) 0.75 motor fuels (60%) ? 900? d ) 0.80 nphths (5%) ? 200? d ) 0.85 distilled fuel oil (21%) ? 900? d ) 0.85 residul fuel oil (5%) ? 400? d ) 1.10 petroleum coke (3%) d ) 0.90 hevy oils (3%) d ) 0.55 still gs (3%) ? 900? wstewter (29) solid wste (30) ? 1200? ir fugitive 250? totl (out) 8000 d, density.? denotes mjor uncertinty. remin in the residuum in the primry nd vcuum distilltions. Dissolved forms (mostly Hg 0 ) re voltile nd re more likely to trnsfer to lighter products. Orgnic forms, if present, my be converted to Hg 0 by hydrotreting. The degree to which Hg 0 is converted to HgS in the ining process is unknown but very importnt to the process of estimting tmospheric mercury emissions from ineries. Chemicl considertions suggest tht some of the elementl mercury in middle distilltes should be converted to sulfide in crcking nd orming units, especilly for crudes hving higher sulfur content. Ionic forms of mercury, if present, would be expected to concentrte in wter extrctions (seprtors, deslting, custic treting). To obtin rough estimte of the mount of mercury in crude oil ined in the United Sttes, one must exmine the origin nd of oil tht is processed blnced ginst the fte of mercury in the ining process. The concentrtions for mercury in rw crudes of known origins must be normlized to the volumes they represent. No ttempt hs been mde here to estimte mercury in worldwide crude production s dt re lcking for some of the mjor crude strems. For U.S. crude consumption, some generl estimtes re possible. Becuse more thn hlf of the crude oil tht is processed in the United Sttes is imported, the concentrtion rnge for totl mercury in crude oil tht is estimted for the United Sttes is likely close to tht for the world in generl. Estimte of Mercury Emissions from Petroleum An estimte of mercury emissions from petroleum in the United Sttes for the yer 1999 is constructed in Tble 7. The mcro-nlysis considers the mount of mercury in crude oil, the distribution of mercury in the ining process, nd the combustion processes for the fuel products derived from crude oil. Anlyticl nd source uncertinties obfuscte ssignment of cler men concentrtion of mercury to severl crude oil feedstocks nd to some ined product strems. The estimtion model relies on more recently reported THg vlues for identified crude sources nd is constructed to provide frmework to identify mjor strems tht require better sttisticl definition. Estimtes of the totl yerly mount of mercury in mjor crude oil strems were clculted by multiplying the source crude feedstock mounts for yer 1999 (11) by the rounded (upwrd) concentrtion of mercury reported for regionl crude oil sources s compiled in Tbles 3 nd 4. Dt for mercury in crude oil originting in the field reported by Bloom (14) nd dt for the Cymric field (7, 8, 18) were excluded. It is redily cknowledged tht the ctul men concentrtions for crude oil from severl sources re uncertin nd could be n order of mgnitude higher (or lower) thn those used in Tble 7. If the mount of mercury in crude oil (including condenstes) processed in the United Sttes is pproximtely 10 ppb on verge, s suggested by the recent dt, then the totl mount of mercury in crude oil is close to 8100 kg. Of this mount, pproximtely 6300 kg resides in inery products. Approximtely 10% of inery products (sphlts, lube oils, solvents) is not burned, leving pproximtely 5700 kg emitted to the tmosphere minly by combustion (includes inery fuel combustion nd ssumes n emission rtio of 1). The men mount of mercury in U.S. trnsporttion fuels (gsoline + diesel + jet fuel) hs mjor impct on the estimte (due to the fct tht hlf of ined products fll into this ctegory) nd ws considered to be no greter thn 2 ppb bsed on the dt of Ling et l. (22) nd dt for other distilltes (14). THg in distilltes likely lects the mount of voltile elementl mercury in crude oil, nd THg in coke nd sphlt lects the mount of nonvoltile suspended forms (mostly HgS nd HgSe). The known mount in coke (50 ppb) (27) seems resonble reltive to the mount in lighter frctions bsed on wht is presently known bout the species distribution of mercury in crude oil (3, 14). Elementl mercury in crude oil my distribute to still gs in the tmospheric tower perentilly. In fct, some ineries hve reported observtions of condensed mercury in distilltion tower overheds. It would seem resonble, theore, to obtin better dt on the mount of mercury in inery still gs so s to hve more direct estimte of the mount in inery ir emissions, given tht most still gs is burned in ineries to fire process heters nd boilers. Upwrd perturbtion of the model cuses imblnces tht re difficult to reconcile unless the concentrtions of mercury in trnsporttion fuels nd nphths re higher thn ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 35, NO. 24, 2001

6 TABLE 8. Fuels from Crude Oil Used by Refineries fuel bpy (11) (10 6 ) kg/yr (10 9 ) THg mt kg (Hg)/yr LPG <10 4 still gs <30? 600 RFO/DFO <10 8 hevy oils coke totl 1500 Residul fuel oil/distillte fuel oil. estimted (2-5 ppb). For exmple, if the men concentrtion of (voltile) mercury in processed crude oils is 50 ppb, the concentrtion of mercury in trnsporttion fuels would need to be on the order of t lest 50 ppb lso. Although dmittedly not sttisticlly verified, there re few (if ny) known reports of THg in trnsporttion fuels tht re bove few prts per billion, with most close to 1 ppb. Perturbtion of the model to lower (thn 10 ppb) men concentrtion of mercury in crude oil would be more resonble to preserve the overll blnce if mercury concentrtions in light fuels re s low s suggested by the dt now vilble (14, 22) nd if nonvoltile forms re not dominnt in their contribution to THg of crude oils. Atmospheric emissions of mercury from ineries must mss blnce with the other venues for mercury egress from ineries. The mss blnce is Hg (crude oil) ) Hg (ir) + Hg (products) + Hg (wstewter) + Hg (solid wste) For ineries, voltile nd prticulte mercury emissions to the tmosphere re postulted to come mostly from the fuels tht re used to fire inery process heters nd possibly some (unknown) mount from fugitive emissions (mostly process vents). Energy usge t U.S. ineries is rigorously documented (11) nd includes purchsed nturl gs (THg less thn 0.2 µg/m 3 ; 28), electricity, nd stem. The fuels used in ineries tht derive from processed petroleum re compiled in Tble 8 nd include mostly still gs nd coke. Applying the totl mercury estimtes used in Tble 7, the estimted mount of mercury in ir emissions from fuel burning t ll U.S. ineries is estimted to be pproximtely 1500 kg/yr or bout 25% of the totl contined in combusted (ined) fuel products (6000 kg/yr). The higher percentge mount ssigned to inery combustion ppers to be due to the fct tht the mjor fuels utilized t ineries (coke, still gs) hve higher mercury concentrtions thn fuel products (motor fuels, fuel oils) tht re sold. While the concentrtion of mercury in coke is known, the verge concentrtion of mercury vpor in still gs is much less certin. It is rtionlized tht fugitive emissions of mercury from ineries re negligible to the present estimte, but this my not be the cse. Vent gses from distilltion nd other unit processes re typiclly sent to flre nd could be component of tmospheric emissions. The quntities of distilltion nd other vented gses re smll, however, s compred to inery fuel combustion. Vent gses from ctlyst regenertion tht dischrge to the tmosphere in some ineries re suspect s possible sources of tmospheric mercury emissions, but here lso the mounts of such dischrges re typiclly smll nd should mke negligible contribution to the totl point source mount when compred to inery combustion gses. Admittedly, dt re needed to support this hypothesis. Using the stted rtionle for the distribution of mercury nd its compounds in the ining process nd the rounded mens for mercury in crude nd ined products, the estimte for the mss of mercury leving ineries roughly blnces with the mount of mercury estimted to be in crude oil, s it must. On the bsis of the stted ssumptions, if the men mercury concentrtion in crude oil tht is processed in the United Sttes is pproximtely 10 ppb, then, on yerly bsis, bout 20% (1500 kg/yr) is lost to the tmosphere in the ining process (ll U.S. ineries), roughly 25% is in solid nd liquid wste strems from ineries, nd bout 55% is in sold products tht re mostly used s fuels. The tmospheric mercury burden from U.S. petroleum is then pproximtely 6000 kg/yr or pproximtely n order of mgnitude less thn tht which comes from burning col ( kg/yr). Severl mjor uncertinties exist in the cited nlysis including in prticulr the estimtes for inery wstewter (29) nd solid wste (30). The estimte for mercury in wstewters from ineries comes from 1982 U.S. EPA study (29) nd my not be relible given the dvnces mde in mercury nlysis in wter in the pst two decdes. Refinery solid wstes tht pss leching criteri for disposl my contin some mounts of HgS nd HgSe tht re insoluble in queous lechnts. Given tht sttisticlly derived dt for mercury in wstewter nd solid wste strems re lcking t present, cre should be exercised in drwing conclusions from the presented estimtes. The gretest uncertinty reltive to ssigning low men THg to crude oil derives from the smpling nd nlyticl procedures tht were employed to cquire the cited dt. Significnt mounts of mercury (in some cses s much s 75% of the totl) cn dhere to the wlls of smple continers, thus evding mesurement. Loss of voltile mercury from smples is notorious for producing low results. Mny routine crude oil ssys plce crude oil smples in metl continers where mercury cn rect with the corrosion products on the continer wlls irreversibly. Suspended mercury tends to settle out in crude oils stored for ny length of time. All of the bove suggest tht the currently vilble dt my be bised low. A more complete sttisticl understnding of the mounts of mercury in crude oil nd in ined products is needed to be ble to provide defensible estimte of mercury emissions to the tmospheric environment. Likewise, mercury emission fctors for liquid fuels require better definition. Attempts to cquire dditionl dt on mercury in worldwide petroleum should use nlyticl protocols tht specite t lest the voltile nd nonvoltile chemicl forms, thus testing the ssumptions of the distribution of mercury species in the ining process. Accurte mesurements of THg in crude oil nd ined products will require more rigorous smpling protocol tht ccounts for the forms of mercury tht typiclly evde mesurement. Acknowledgments I pprecite helpful discussions with the U.S. EPA (Dvid Kirchgessner, EPA/ORD) nd the Americn Petroleum Institute (Krin Ritter nd Roger Clff). The opinions nd interprettions in the mnuscript, however, re solely those of the uthor nd do not lect the opinions or positions of either the U.S. EPA or the Americn Petroleum Institute. Literture Cited (1) Mercury Study Report to Congress, Volume II: An Inventory of Anthropogenic Mercury Emissions in the United Sttes; U.S. Environmentl Protection Agency, Office of Air Qulity Plnning & Stndrds nd Office of Reserch nd Development: Wshington, DC, nd Reserch Tringle Prk, NC, 1997; EPA-452/ R (2) Emergency Plnning nd Community Right-To-Know Act- Section 313: Guidnce for Reporting Toxic Chemicls: Mercury nd Mercury Compounds Ctegory, DRAFT; U.S. Environmentl VOL. 35, NO. 24, 2001 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

7 Protection Agency, Office of Informtion Anlysis & Access: Wshington, DC, November 2000; EPA-745/B-00-00X. (3) Wilhelm, S.; Bloom, N. Fuel Proc. Technol. 2000, 63, 1. (4) Brooks, G. Estimting Air Toxic Emissions from Col nd Oil Combustion Sources; Rdin Corportion Report to U.S. Environmentl Protection Agency, Office of Air Qulity, Environmentl Protection Plnning nd Stndrds: Reserch Tringle Prk, NC, 1989; EPA-450/ (5) Locting And Estimting Air Emissions from Sources of Mercury nd Mercury Compounds; U.S. Environmentl Protection Agency, Office of Air Qulity, Environmentl Protection Plnning nd Stndrds Agency: Reserch Tringle Prk, NC, 1997; EPA-454/ R (6) Study of Hzrdous Air Pollutnt Emissions from Electric Utility Stem Generting UnitssFinl Report to Congress; U.S. Environmentl Protection Agency, Office of Air Qulity, Environmentl Protection Plnning nd Stndrds Agency: Reserch Tringle Prk, NC, 1998; EPA-453/R (7) Shh, K. R.; Filby, R. H.; Hller, W. A. J. Rdionl. Chem. 1970, 6, 413. (8) Filby, R. H.; Shh, K. R. In The Role of Trce Metls in Petroleum; Yen, T. F., Ed.; Ann Arbor Science Publishers: Ann Arbor, MI, (9) Hitchon, B.; Filby, R. Geochemicl StudiessTrce Elements in Albert Crude Oils; Open File Report ; Albert Reserch Council for Albert Energy nd Utilities Bord nd Albert Geologicl Survey: Albert, Cnd, (10) Hitchon, B.; Filby, R. H.; Shh, K. R. In The Role of Trce Metls in Petroleum; Yen, T. F., Ed.; Ann Arbor Science Publishers: Ann Arbor, MI, (11) Energy Sttistics for 1998 nd 1999; U.S. Deprtment of Energy, Energy Informtion Administrtion, Ntionl Energy Informtion Center: Wshington, DC, (12) Mus, M.; Mrkus, W.; Elghondi, A.; Etwir, R.; Hnnn, A.; Arf, E. J. Rdionl. Nucl. Chem. 1995, 198 (1), 17. (13) Ling, L.; Lzoff, S.; Horvt, M.; Swin, E.; Gilkeson, J. Fresenius J. Anl. Chem. 2000, 367, 8. (14) Bloom, N. S. Fresenius J. Anl. Chem. 2000, 366, 438. (15) To, H.; Murkmi, T.; Toming, M.; Miyzki, A. J. Anl. At. Spectrom. 1998, 13, 1085H. (16) Olsen, S.; Westerlund, S.; Visser, R. Anlyst 1997, 122, 12. (17) Shfwi, A.; Ebdon, L.; Foulkes, M.; Stockwell, P.; Corns, W. Anlyst 1999, 124, 185. (18) Mgw, R. I.; McMillen, S. J.; Gl, W. R.; Try, J. H.; Trocine, R. P. Risk Evlution of Metls in Crude Oils; SPE/EPA Explortion & Production Environmentl Conference; SPE Pper 52725; Society of Petroleum Engineers: Dlls, TX, (19) Morris, R. New TRI Reporting Rules on Mercury. Presented t Ntionl Petroleum Refiners Assocition Meeting, Sn Antonio, TX, September (20) Environment Cnd. Properties of Crude Oils nd Oil Productss 12/99; Emergencies Science Division: Ontrio, Cnd, (21) Duo, W.; Guthrie, T.; Edwrds, W. C. Mercury Emissions from the Petroleum Refining Sector in Cnd; Report to Environment Cnd, Trns-Boundry Air Issues Brnch, Hzrdous Air Pollutnts Progrm: Ottw, Cnd, 2000; Contrct K (22) Ling, L.; Horvt, M.; Dnilchik, P. Sci. Totl Environ. 1996, 187, 57. (23) Crter, W.; Buening, H.; Hunter, S. Emission Reduction on Two Industril Boilers with Mjor Combustion Modifiction; U.S. Environmentl Protection Agency: Reserch Tringle Prk, NC, 1978; EPA-600/ (24) Slter, S.; Hll, R. Electricity Genertion by Utilities: 1974 Ntionwide Emissions Estimtes; AIChE Symposium Series 165; Americn Institute of Chemicl Engineers: New York, 1977; pp (25) Shih, C.; Ackermn, D.; Scinto, L.; Moon, E.; Fishmn, E. POM Emissions from Sttionry Conventionl Combustion Processes; U.S. Environmentl Protection Agency, Industril Reserch Lbortory: Reserch Tringle Prk, NC, 1980; EPA Contrct (26) Cstldini, C.; Wterlnd, L.; Mson, H. Combustion Modifictions Controls for Residentil nd Commercil Heting Systems, Vol. IIsOil-Fired Residentil Fuel Test; U.S. Environmentl Protection Agency, Industril Reserch Lbortory: Reserch Tringle Prk, NC, 1981; EPA-600/ b. (27) U.S. Environmentl Protection Agency. Unified Air Toxics Progrm: Electric Utility Stem Generting Units Hzrdous Air Pollutnt Emission Study, utiltox/utoxpg.html, (28) Cho, S.; Attri, A. Chrcteriztion nd Mesurement of Nturl Gs Trce Constituents, Prt 1: Nturl Gs Survey; Institute of Gs Technology Report to Gs Reserch Institute; Contrct ; Gs Reserch Institute: Chicgo, IL, (29) Ruddy, D. Finl Development Document for Effluent Limittions Guidelines nd New Source Performnce Stndrds nd Pretretment for the Petroleum Refining Point Source Ctegory; U.S. Environmentl Protection Agency: Wshington, DC, 1982; EPA/ 440/1-82/014. (30) Wste Minimiztion for Selected Petroleum Refining Residuls; U.S. Environmentl Protection Agency, Office of Solid Wste: Wshington, DC, 1996; EPA/530/R-96/009. Received for review October 24, Revised mnuscript received August 15, Accepted September 18, ES001804H ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 35, NO. 24, 2001