Electric Arc Furnaces for Steel-Making: Hot Spots for Persistent Organic Pollutants

Electric Arc Furnaces forSteel-Making: Hot Spots forPersistent Organic PollutantsM U S T A F A O D A B A S I , * , †

A B D U R R A H M A N B A Y R A M , †

T O L G A E L B I R , † R E M Z I S E Y F I O G L U , †

Y E T K I N D U M A N O G L U , † A Y S E B O Z L A K E R , †

H U L U S I D E M I R C I O G L U , †

H A S A N A L T I O K , † S I N A N Y A T K I N , ‡ A N DB A N U C E T I N §

Department of Environmental Engineering, Faculty ofEngineering, Dokuz Eylul University, Kaynaklar Campus,35160 Buca, Izmir, Turkey, Department of EnvironmentalEngineering, Faculty of Engineering, Namık Kemal University,Corlu, Tekirdag, Turkey, and Department of EnvironmentalEngineering, Faculty of Engineering, Pamukkale University,Denizli, Turkey

Received March 23, 2009. Revised manuscript receivedJune 1, 2009. Accepted June 1, 2009.

Persistent organic pollutant (POP) concentrations weremeasured in stack-gases of ferrous scrap processing steelplants with electric arc furnaces (EAFs) (n ) 5) in Aliaga, Izmir,Turkey and in air (n ) 11) at a site near those plants.Measured stack-gas concentrations for the four plants withoutscrap preheating (611 ( 311, 165 000 ( 285 000, and 33 ( 3ng m-3, average ( SD for Σ41PCBs, Σ16PAHs, and Σ7PBDEs,respectively) indicated that they are significant sources forpolycyclic aromatic hydrocarbons (PAHs), polychlorinatedbiphenyls (PCBs), and polybrominated diphenyl ethers (PBDEs).POP emissions from the plant with scrap preheating weresignificantly higher (13 500, 445 000, and 91 ng m-3 for Σ41PCBs,Σ16PAHs, and Σ7PBDEs, respectively). It was also shown thatthe steel plants emit considerable amounts of fugitive POPs inparticle-phase. Estimated emissions using the emissionfactors generated in this study and the production amountssuggested that the steel plants with EAFs may significantlycontribute to local and global PAH, PCB, and PBDE emissions.Several other compounds (aromatic and aliphatic hydrocarbons,oxygen, sulfur, nitrogen, and chlorine-containing organiccompounds, n ) 49) were identified and determinedsemiquantitatively in the stack-gas and ambient air samples.Ambient air concentrations (62 ( 35, 320 ( 134 ng m-3, 1451 (954 pg m-3, for Σ41PCBs, Σ16PAHs, and Σ7PBDEs, respectively)were significantly higher than those measured previouslyaround the world and in the region, further confirming thatthe steel plants with EAFs are “hot spots” for POPs.

IntroductionA significant portion (32%) of the world’s steel requirement(1244 million tons year-1) is produced from ferrous scrap

metals by iron-steel plants with electric arc furnaces (EAFs)(1). The U.S. steel industry produced about 106 million tonsof raw steel in 2006, and approximately 93 steel plants withEAFs accounted for 57% of the total U.S. steel production.The contribution of EAFs in steel production has increaseddramatically over the past 30 years (from 10% in 1970 to 57%in 2006) (2). Iron-steel production is also an importantindustrial process in Turkey, where the present study wasconducted. In Turkey, ∼23.3 million tons of steel wasproduced in 2006. Plants with EAFs accounted for 71% ofthis production (17.6 million tons year-1). Most of the ferrousscrap (75%) used in EAFs was imported (1).

Emission generating operations during the EAF steel-making are charging the scrap, melting and refining, remov-ing slag, tapping steel, continuous casting, and ladle met-allurgy processes. Emissions from EAFs are generally collectedusing direct shell evacuation supplemented with a canopyhood located above the EAF. In general, the capturedemissions are routed to bag filters for particulate matter (PM)control (2, 3).

The EAF processes produce particle and gas-phasepollutants. The amount and composition of the PM emittedcan vary depending on the scrap composition and types andamounts of furnace additives. Iron and its oxides are theprimary PM components. In addition, zinc, chromium, nickel,lead, cadmium, and other metals (and metal oxides) are alsopresent in the PM. Gaseous pollutants, such as NOx, SO2,CO, HF, and HCl may also be emitted depending on theequipment, operating practices and the additional fuels used.Several organics like volatile organic compounds (VOCs),chlorobenzenes, and polychlorinated dibenzo-p-dioxins/furans (PCDD/Fs) are also emitted. Other than these mostlyregulated pollutants, EAFs emit persistent organic pollutants(POPs) like polycyclic aromatic hydrocarbons (PAHs), poly-chlorinated biphenyls (PCBs), and polybrominated diphenylethers (PBDEs) (2-4). POPs covered in this study are emittedby different mechanisms. PAHs may be present in the scrapand are evaporated during production processes or they mayform as a result of incomplete combustion of scrap organicmatter, fuels, and additives like coal (3). PCBs are also presentin the scrap (3). They may also form by de novo synthesisin thermal processes, similar to PCDD/Fs (3). Ferrous scrapcontains impurities like plastic and foam that could containsignificant amounts of PBDEs (4). PBDEs are emitted duringthe steel production process, during scrap charging (mostlyin the particle-phase), scrap preheating, and at the beginningof the melting cycle (mostly in the gas-phase) (3, 4). All thesePOPs are semivolatile. They may be predominantly in thegas/particle-phases or distributed between two phases,depending on temperature and their physicochemicalproperties.

Fugitive emissions of PM (and POPs contained in particles)may also be significant (5-8). The fraction of the EAF gasesthat could not be collected is emitted from the openings inthe roof (2). Also scrap, slag, filter dust storage, transfer, anddumping operations may emit significant amounts ofparticle-phase POPs (3).

Recent studies based on soil and ambient air samplinghave shown that the ferrous scrap processing steel plantswith EAFs in Aliaga industrial region in Turkey are importantlocal sources for PAHs, PCBs, and PBDEs (5-8). However,there have been a limited number of studies in the literatureon EAF source characterization and emission factor genera-tion for PAHs and PCBs (3, 9, 10) while no studies exist forPBDEs. The objectives of this study were (1) the measurementof emissions and generating emission factors for PAHs, PCBs,

* Corresponding author phone: 90-232-4127122; fax: 90-232-4127080; e-mail: [email protected].

† Dokuz Eylul University.‡ Namık Kemal University.§ Pamukkale University.

Environ. Sci. Technol. 2009, 43, 5205–5211

10.1021/es900863s CCC: $40.75 2009 American Chemical Society VOL. 43, NO. 14, 2009 / ENVIRONMENTAL SCIENCE & TECHNOLOGY 9 5205

Published on Web 06/12/2009

Dow

nloa

ded

by A

NK

OS

CO

NSO

RT

IA T

UR

KE

Y o

n Ju

ly 1

3, 2

009

Publ

ishe

d on

Jun

e 12

, 200

9 on

http

://pu

bs.a

cs.o

rg |

doi:

10.1

021/

es90

0863

s

and PBDEs from ferrous scrap processing steel plants withEAFs and (2) measurement of ambient air levels of POPsnear these industries in Aliaga industrial region in Turkey.

Materials and MethodsAmbient Air and Stack-Gas Sampling. The study area islocated at the Aliaga industrial region, ∼5 km south of theAliaga city center and ∼45 km north of the metropolitan cityof Izmir, Turkey. The area contains several pollutant sourcesincluding a large petroleum refinery and a petrochemicalcomplex, scrap processing iron-steel plants with EAFs, scrapstorage and classification sites, steel rolling mills, a naturalgas-fired power plant, a very dense transportation activity offerrous scrap trucks, heavy road and rail traffic, a shipdismantling area, and busy ports with scrap iron dockyards(Figure 1).

Daily ambient air samples (n) 11) were collected at a sitenear the iron-steel plants between April 26 and May 7, 2007.Meteorological data were obtained from a meteorologicalstation located near the Horozgedigi village (Figure 1). Windspeed ranged between 1.4-4.8 m s-1 during the samplingprogram. The prevailing wind directions in the area are WNWand NW. There were both northerly and southerly windsduring the sampling program. The location of the samplingsite relative to the steel plants and these wind directionsindicate that the sampling site is affected from the steel plantemissions. A sampling site located at SE of the present sitewould be preferable to better capture the emissions of allfive steel-plants. However, an alternative secure samplingsite with power supply was not available. As a result, themeasured air concentrations may have been mainly affectedfrom the three plants located at NW, WNW, and SW of thesampling site. The wind direction having the least possibleinfluence is SE that was encountered only on one samplingday (Figure 1).

Stack-gas samples were collected from the five scrapprocessing iron-steel plants with EAFs located near the airsampling site. Four of the stacks were sampled once whereasone stack was sampled twice (n ) 6). The production

capacities, stack-gas flow rates, and number of the electricarc furnaces of the plants range between 94 and 163 ton h-1,640 000 and 1 227 000 Nm3 h-1 (at 1 atm and 273 K), and 1and 2, respectively.

Air samples were collected using a modified high-volumesampler model GPS-11 (Thermo-Andersen Inc.). Particle-phase POPs were collected on 10.5 cm diameter quartz filtersand the gas-phase compounds were collected in a modifiedcartridge containing XAD-2 resin placed between layers ofpolyurethane foam (PUF). Stack-gas samples were collectedisokinetically using a sampling train consisting of a heatedsampling probe, a filter cartridge, a condenser, a water-cooledresin (XAD-2) cartridge, a vacuum pump with flow controller,and a gas-meter (see Supporting Information (SI) Figure S1).Stack-gas particle-phase POPs were collected on glass-fiberthimble filters and the gas-phase compounds were collectedin the XAD-2 resin cartridge. Average sampling duration was∼24 h for ambient air while it was 2.5 h for stack-gases(covering at least three production cycles). The averagesampling volumes (measured with an uncertainty < 3%) were300 ( 40 m3 and 3.1 ( 0.8 m3 for ambient air and stack-gassamples, respectively. The average temperatures were 17 (3 °C (n ) 11) and 91 ( 7 °C (n ) 5) at ambient air andstack-gas samples, respectively.

Sample Preparation and Analysis. Prior to extraction, allsamples were spiked with PCB, PAH, and PBDE surrogatestandards. Ambient air PUFs were Soxhlet extracted for 24 hwith a mixture of 1:1 acetone:hexane. The remaining samples(ambient air filters, stack-gas thimble filters, and XAD-2 resincartridges) were ultrasonically extracted for 60 min withacetone:hexane (1:1). For the stack samples, the samplingprobe was rinsed with acetone:hexane (1:1) and the solutionwas combined with stack-filter extract (particle-phase). Thesampling line between the filter and resin cartridge was alsorinsed with acetone:hexane. The condensate from thecondenser before the resin cartridge was liquid-liquidextracted with dichloromethane and hexane. The extractsfrom the sampling line and condensate were combined withthe extract from the resin cartridge (gas-phase). The extract

FIGURE 1. Map of the study area. Wind roses show the frequency (%) of prevailing wind directions.

5206 9 ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 43, NO. 14, 2009

Dow

nloa

ded

by A

NK

OS

CO

NSO

RT

IA T

UR

KE

Y o

n Ju

ly 1

3, 2

009

Publ

ishe

d on

Jun

e 12

, 200

9 on

http

://pu

bs.a

cs.o

rg |

doi:

10.1

021/

es90

0863

s

volumes were reduced and were transferred into hexane usinga rotary evaporator and a high purity N2 stream. Afterconcentrating to 2 mL, samples were cleaned up andfractionated on an alumina-silicic acid column containing3 g silicic acid (deactivated with 4.5% DI water) and 2 galumina (deactivated with 6% DI water). The column wasprewashed with 20 mL dichloromethane (DCM) followed by20 mL petroleum ether (PE). Then, the sample in 2 mL hexanewas added to the column and PCBs, PBDEs, and PCNs wereeluted with 35 mL PE (Fraction 1) while PAHs and orga-nochlorine pesticides (OCPs) were eluted with 20 mL DCM(Fraction 2). The final extracts were solvent exchanged intohexane and were concentrated to 1 mL under a stream of N2.

All samples were analyzed with an Agilent 6890N gaschromatograph (GC) equipped with a mass selective detector(Agilent 5973 inert MSD). PAHs and PCBs were analyzedusing electron impact ionization while negative chemicalionization (NCI) mode was used for PBDEs, PCNs, and OCPs.The capillary column used for PAHs and PCBs was HP5-ms(30 m, 0.25 mm, 0.25 µm) while a DB5-ms column (15 m,0.25 mm, 0.1 µm) was used for PBDEs. Helium was the carriergas and high purity methane was the reagent gas for NCI.Fraction 1 was analyzed for PCBs, PBDEs, and PCNs, Fraction2 was analyzed for OCPs in selected ion monitoring mode(SIM). Then, equal volumes of Fraction 1 and 2 werecombined and analyzed for PAHs since lighter PAHs areeluted partly with Fraction 1. In this case, the MSD was runin simultaneous scan and SIM modes. Compounds wereidentified based on their retention times, target and qualifierions, and were quantified using the internal standardcalibration procedure. Further details for sample preparationand instrumental analysis could be found elsewhere (5-8).

Other than the compounds quantified using calibrationstandards, several organics were tentatively identified in thestack-gas and ambient air samples using mass spectral librarysearches. Concentrations of compounds having a matchquality >90% were determined semiquantitatively using theaverage response factors calculated from the responses ofthe calibrated compounds (PAHs). The behavior of thetentatively identified compounds in the analytical system(GC-MS) and as a result their actual response factors maydeviate from those of the calibrated compounds. Therefore,the concentrations determined by this approach are onlysemiquantitative (11).

Average recoveries for the surrogate standards were 85 (14% (acenaphthene-d10), 100 ( 17% (phenanthrene-d10), 94(16% (chrysene-d12), 95(15% (perylene-d12), 93(8% (PCB-14), 95 ( 8% (PCB-65), 92 ( 11% (PCB-166), and 87 ( 23%(PBDE-77). Instrumental detection limits (IDL) were deter-mined from linear extrapolation, based on the loweststandard in calibration curve and using the area of a peakhaving a signal/noise ratio of 3. For 1 µL injection, thequantifiable amounts were 0.15, 0.10, and 0.05-0.35 pg forPAHs, PCBs, and PBDEs, respectively. Blank PUF cartridges,air filters, thimble filters, and resin columns were alsoanalyzed. The limit of detection of the method (LOD, ng)was defined as the mean blank mass plus three standarddeviations (LOD ) mean blank value + 3SD). Instrumentaldetection limit was used for the compounds that were notdetected in blanks. Average analyte amounts in blanks weregenerally <5% of the amounts found in samples. Samplequantities exceeding the LOD were quantified and blank-corrected by subtracting the average blank amounts fromsample amounts.

Results and DiscussionEmissions. Stack-gas PCB, PAH, and PBDE concentrationsmeasured at the steel plants are presented in Table 1 and SITable S1. Recently, new as well as existing EAFs have beenequipped with a system for preheating the scrap, by the off

gas for energy recovery. PCB and PBDE concentrations werenot highly variable for the four plants without preheatingwhile PAH concentrations were variable. This may be due tovariation of organic matter and PAH content of the scrap.Another possibility is that the PAH emissions are mainlyaffected from the variations in production processes sincea significant fraction is formed during the thermal processes(i.e., incomplete combustion). One of the plants was sampledtwice at different days. Stack-gas PCB and PBDE concentra-tions were similar for two samplings (with 16 and 28% RSDs,respectively) while PAH concentrations were not (RSD )128%).

Scrap preheating may result in higher emissions ofaromatic organohalogen compounds such as PCDD/Fs,chlorobenzenes, PCBs as well as PAHs and other incompletecombustion products from scrap contaminated with paints,plastics, lubricants, or other organic compounds (3). It wassuggested that these emissions could be minimized bypostcombustion using additional oxygen burners (3). Thestack-gas concentrations for the plant with preheating weresignificantly (3-22 times) higher (Table 1, see SI Table S1).This is due to desorption of POPs in the scrap when they areheated. Another reason for the significantly higher POPemissions from the EAFs with preheating is the difference inthe stack-gas postcombustion. The postcombustion isachieved through introduction of fresh burning air into thestack gases at very high temperatures. However, since asignificant amount of heat is used for scrap preheating, thetemperature of stack gases from this process is significantlylower, not favoring an efficient postcombustion. When thisis combined with desorption of the organics during preheat-ing, significantly high amounts of POPs are emitted from theEAFs as measured in the present study. Additional burnersshould be used for an efficient postcombustion, howeverthey require additional fuel.

Emission factors for PCBs, PAHs, and PBDEs werecalculated using the stack-gas concentrations, stack-gas flowrates, and steel production amounts (Table 1, see SI TablesS2, S3, and S4). Similar to the stack gas concentrations,emission factors for the plant with preheating were signifi-cantly higher. PCB emission factors were within the rangereported in the literature for EAFs (3). However, the emissionfactor for the plant with preheating was several times higherthan the upper-bound literature value. A wide range ofemission factors has been reported for PAHs. The averageemission factor determined in this study was close to theupper-bound literature value. However, the PAH emissionfactor for the plant with preheating was ∼4 times higher

TABLE 1. Stack-Gas POP Concentrations and Emission Factorsfor Steel Plantsa

stack-gas concentrations (ng Nm3-)b

No Preheating (n ) 4)

AVG SD GMwith

Preheating(n ) 1)

Literaturee

Σ41PCB 611 322 544 13500Σ16PAH 165000 285000 45600 445000Σ7PBDE 33 3 33 91

Emission Factors (mg ton-1)Σ41PCBc 3.8 2.0 3.5 128ΣPCBd 5.4 3.6 4.6 171 1.5-45d

Σ16PAH 992 1618 293 4203 3.5-920Σ7PBDE 0.22 0.07 0.21 0.9

a AVG: average, SD: standard deviation, GM: geometricmean. b At 273 K and 1 atm (T)355-373 K, P ) 0.99-1.01atm for the stack-gas samples) c Sum of 41 PCBs. d ΣPCB )(PCB 28 + 52 + 101 + 153 + 138 + 180)x5, ref (12). e Ref 3.

VOL. 43, NO. 14, 2009 / ENVIRONMENTAL SCIENCE & TECHNOLOGY 9 5207

Dow

nloa

ded

by A

NK

OS

CO

NSO

RT

IA T

UR

KE

Y o

n Ju

ly 1

3, 2

009

Publ

ishe

d on

Jun

e 12

, 200

9 on

http

://pu

bs.a

cs.o

rg |

doi:

10.1

021/

es90

0863

s

than the highest literature value (Table 1). The large variabilityof the PAH emission factors reported previously and deter-mined in the present study supports the hypothesis that thePAH emissions are significantly affected from the variationsin production processes and scrap content.

Annual global and national PCB, PAH, and PBDE emis-sions from EAFs were estimated using the annual productionamounts and the emission factors generated in this study(Table 2). Estimated emissions are also compared to theglobal, regional and national emissions for Turkey reportedin the literature (Table 2). When national emissions for Turkeyare considered, it is obvious that significant amounts of thePOPs are imported into the country along with the ferrousscrap. Estimated emissions have also suggested that the steelplants with EAFs may significantly contribute to local andglobal PAH, PCB, and PBDE emissions (Table 2). However,this is a semiquantitative comparison due to the uncertaintiesassociated with the estimated emissions in the present studyand those reported previously. In the present study, emissionfactors were derived from a limited number of measurements(n ) 5) and there was only one measurement for the processwith preheating. Unlike the conventional pollutants, no highquality emission factors generated from experimental workexist for POPs. Emission estimates are mainly depend oninformation on the production amounts of the POPs (i.e.,PCBs and PBDEs) and the products containing them, theuse and disposal of these products. It is generally assumedthat certain fractions of the POPs are emitted during theseprocesses. Emitted POP amounts have also been estimateddepending on their physicochemical properties (i.e., vaporpressure, octanol-air partition coefficient) or extrapolatedfrom other experimental studies. Therefore, large uncertain-ties are generally associated with the estimated POP emissions(13, 17).

Other than the compounds quantified using calibrationstandards, several organic compounds were identified usingmass spectral library searches and their concentrations weredetermined semiquantitatively in the stack-gas and ambientair samples. These included several aromatic and aliphatichydrocarbons (n ) 22), oxygen (n ) 14), sulfur (n ) 5),nitrogen (n)5), and chlorine-containing (n)3) compounds(see SI Table S5). Squalene, hexadecane, 1,1′-biphenyl,1-methylnaphthalene, dibenzofuran, acetophenone, andbenzo[b]thiophene were the most abundant compounds withconcentrations up to 9372 ng m-3. Similar to the calibratedcompounds, the stack-gas concentrations for the EAF withpreheating were several times higher compared to thosemeasured in stack-gases of other EAFs (see SI Table S6).

Ambient Air Concentrations. Ambient air PCB, PAH, andPBDE concentrations are presented in SI Table S7. Lowmolecular weight compounds (phenanthrene, fluorene,fluoranthene, and pyrene) dominated the Σ16PAH concen-trations. This compound profile is similar to those measuredaround the world and recently in the study area (6, 20). PBDE-209 was the dominant compound for Σ7PBDEs and it isfollowed by PBDE-99 and -47. Recently, a similar ambientPBDE profile was reported in Izmir area (21). Low molecularweight congeners (PCB-18, 28, 31, and 33) dominated theΣ41PCB concentrations (see the SI Table S7).

The ambient air POP concentrations in the Izmir regionhave been recently measured in several studies (5-8, 20-22).These studies included urban and suburban sites in Izmir,and sites close to the air sampling site of the present study(i.e., the Horozgedigi and Aliaga urban sites) (Figure 1).Ambient air concentrations measured at those sites cover arange of spatial (urban, suburban, and industrial sites) anda temporal (winter, summer) variation. Recent comparisonshave indicated that the ambient POP concentrations mea-sured in the area were within the ranges reported in theliterature (7, 20, 21). In general, the concentrations measuredin the present study were significantly higher than thosemeasured recently in the Izmir area (two-tailed t test, p <0.01) (Figure 2). Σ41PCB concentrations reported in theliterature and measured in Izmir area range between a fewhundreds to a few thousands pg m-3 (7) (Figure 2A). Recently,relatively high Σ27PCB concentrations (4230-11350 pg m-3)were measured around an electronic waste dismantlingfacility in China (23). However, considering the previouslyreported concentrations in the study area and around theworld (5, 7), the Σ41PCB concentrations measured in thepresent study (19 000-136 000 pg m-3) are among the highestones measured ever. Similarly, Σ7PBDE concentrationsmeasured in this study (range 356-3968, average ( SD 1451( 954 pg m-3) were significantly higher than those measuredrecently in the area (Figure 2C) and they were close to theupper-bound values reported around the world (21). Li et al.(23) have measured comparably high Σ9PBDE concentrations(range 302-6095, average ( SD 1995 ( 1629 pg m-3) aroundan electronic waste dismantling facility in China. Σ16PAHconcentrations measured in the present study were generallyhigher than those measured recently in the area (Figure 2B).However, there were a few higher concentrations measuredat Izmir and urban Aliaga sites. It was shown that theserelatively higher concentrations measured at urban sites weredue to the increase in PAH emissions from residential heatingin winter (20).

TABLE 2. Global and National PBDE, PCB, and PAH Emissions (kg year-1) from Steel Plants with Electric Arc Furnaces

this study (iron-steel production) literature (All sources)

globala Turkeya European Union/Europe Turkey Global

PBDE 28 5.3–64 0.2–2.6PBDE 47 20–147 0.8–6.1 3.5-750c, 3400-180 000b

PBDE 100 3.3–16 0.1–0.7PBDE 99 14–56 0.6–2.3 1-60c

PBDE 154 2.2–7.4 0.1–0.3PBDE 153 1.6–4.4 0.1–0.2 0.03-1c

Σ6PBDE 47–294 2–12 7505-7774d, 42921e

PBDE 209 41–46 1.7–1.9 138-465d, 25 650f

Σ41PCB 1510–50 790 63–2100 72-1240g 17 600-464 000g

ΣPCB 2150–67 900 89–2800 88-1640g 19 000-603 000g

BbF 5600–21 800 230–900 536 000h 53 000h

BkF 3600–14 200 150–590 218 000h 20 000h

BaP 1500–5800 60–240 493 000h 42 000h

IcdP 200–740 8–31 365 000h 35 000h

a Calculated using global and national steel production Figures (398 and 16.5 million ton year-1), the average andmaximum emission factors generated in this study. b ref 4. (North America). c ref 13. d ref 14. e ref 15. f ref 16. g ref 17, 18.h ref 19.

5208 9 ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 43, NO. 14, 2009

Dow

nloa

ded

by A

NK

OS

CO

NSO

RT

IA T

UR

KE

Y o

n Ju

ly 1

3, 2

009

Publ

ishe

d on

Jun

e 12

, 200

9 on

http

://pu

bs.a

cs.o

rg |

doi:

10.1

021/

es90

0863

s

Ambient air concentrations of semiquantitatively deter-mined compounds are presented in SI Table S8. Similar tothe stack-gas samples, squalene, hexadecane, 1,1′-biphenyl,1-methylnaphthalene, dibenzofuran, acetophenone, andbenzo[b]thiophene were the most abundant compounds withconcentrations ranging between 7.4 and 75.5 ng m-3.

Phase Distribution of POPs in Air and Stack-Gases. PCBs,PAHs, and PBDEs are semivolatile compounds and they aredistributed between gas and particle-phases in the atmo-sphere depending on their physicochemical properties andtemperature. Octanol-air partition coefficient (KOA) has beencommonly used to investigate and to model the gas-particlepartitioning of organic compounds (21, 23-26). Figure 3shows the variation of gas-phase percentages of PCBs, PAHs,and PBDEs in ambient air and in stack-gases as a functionof their temperature-adjusted octanol-air partition coef-ficients. Stack-gas and ambient air gas-phase fractions ofPOPs were significantly different. For the EAFs withoutpreheating, gas-phase fractions of PCBs, PAHs, and PBDEswere generally higher compared to ambient air and theirranges were relatively narrow (55-96, 32-97, 45-74%,respectively). This could be explained by the temperaturedependency of KOA that decreases with temperature (17 and91 °C, in air and in stack, respectively) favoring lesspartitioning for POPs to particle-phase in stack-gases. PMcontrol by bag filters may also alter the gas-particle distribu-

tion. A significant fraction of POPs (especially those withhigh KOA values) might be on fine particles that couldpenetrate through the bag filters. However, this is onlyspeculative. Therefore, it is difficult to assess the effect of PMremoval on gas-particle distribution. For the EAF withpreheating, gas-phase fractions of PCBs, PAHs, and PBDEswere significantly lower compared to other EAFs, indicatingthat the significantly higher POP emissions from the EAFswith preheating are emitted mainly in particle-phase. Thissuggests that fume formation may be a more effectiveemission mechanism compared to evaporation of gas-phasePOPs during scrap preheating.

Ambient air gas-phase fractions of PCBs, PAHs, and PBDEsranged between 20 and 99, 2-98, and 1-88%, respectively(Figure 3A-C). These results are generally consistent withthe previous observations and gas-particle partitioning modelestimations (20, 21, 23, 26). Phase distribution models likethe KOA model is based on equilibrium partitioning. However,using a dynamic uptake model, Cetin and Odabasi (21) haveshown that several months (100-1000 days) are required forcompounds having log KOA > 10 (PCB-128-PCB-209, ben-z[a]anthracene-benzo[g,h,i]perylene, PBDE-47-PBDE-209)to reach equilibrium between the air and atmosphericparticles. Therefore, several POPs are not likely to reachequilibrium between the gas and particle-phases since thereare relatively short distances between the sources and the

FIGURE 2. Comparison of the concentrations measured at the air sampling site of the present study with those measured recently inthe region (7, 8, 20-22).

FIGURE 3. Variation of stack-gas and ambient air gas-phase percentages of PCBs, PAHs, and PBDEs with octanol-air partitioncoefficients (KOA) (24-31).

VOL. 43, NO. 14, 2009 / ENVIRONMENTAL SCIENCE & TECHNOLOGY 9 5209

Dow

nloa

ded

by A

NK

OS

CO

NSO

RT

IA T

UR

KE

Y o

n Ju

ly 1

3, 2

009

Publ

ishe

d on

Jun

e 12

, 200

9 on

http

://pu

bs.a

cs.o

rg |

doi:

10.1

021/

es90

0863

s

sampling site (Figure 1). Gas-phase fractions of severalcompounds were generally decreased significantly in ambientair relative to stack-gas with average (stack gas/ambient air)ratios of 1.4 ( 0.6 (average ( SD), 6.6 ( 7.9, and 12.3 ( 12.5for PCBs, PAHs, and PBDEs, respectively (Figures 3A-C).One of the extreme examples for this observation is PBDE-209 with a log KOA value of 15.3 calculated from itsdimensionless Henry’s law constant (30) and octanol-waterpartition coefficient (31). Gas-phase percentages of PBDE-209 ranged between 19-45 and 0.3-10% in stack-gases andin air, respectively. However, dynamic uptake model esti-mates that several months are required for this much of adecrease in gas-phase fraction of PBDE-209 to take place(21). This suggested that there may be a significant contri-bution to ambient particle-phase POPs by another mech-anism other than gas-particle partitioning. Therefore, varia-tions of average stack-gas/ambient air gas and particle-phaseconcentration ratios with KOA were investigated (see the SIFigure S2). Particle-phase stack-gas/ambient air concentra-tion ratios decreased significantly with KOA (from 1407 to 40,4090 to 13, and 747 to 8 for PCBs, PAHs, and PBDEs,respectively) indicating that ambient concentrations ofcompounds having high KOA values were enriched by sourcesother than stack-gases, while gas-phase concentration ratiosdid not show significant and systematic variations (see theSI Figure S2). During the sampling programs it was observedthat scrap iron, slag, filter dust storage, transfer and dumpingprocesses, and vehicular traffic (especially trucks) on pavedand unpaved roads around the steel plants emit significantamounts of PM. Fugitive PM10 emissions from these sourceswere recently estimated by a detailed emission inventory(32). It was shown that the contributions of paved roads,unpaved roads, transfer/dumping operations, wind erosionfrom storage piles (slag, filter dust, and scrap), and EAFs tothe total PM10 emissions were 77.3, 16.1, 0.7, 0.2, and 5.7%,respectively (32). Also there have been intermittent EAFemissions during fan or bag-filter malfunctioning. Theseresults further suggest that there is a relationship betweenthe fugitive particle-phase emissions and sudden decreasein gas-phase fractions of the compounds with log KOA > 10from source to receptor. The presence of the fugitiveemissions also suggests that the emission factors generatedand the POP emissions calculated in the present study maybe underestimated, especially for compounds with high KOA

values.Recently, it was suggested that BDE-209 is transported in

the environment primarily in particle-phase (33, 34). Also,in recent modeling efforts on the environmental fate of PBDEs(i.e., photochemical removal/conversion), atmospheric BDE-209 was assumed to be particle bound 99.1-99.98 (35) and99.999% (36). Gas-phase fraction of BDE-209 has a crucialimportance since the photolysis rate is significantly (120times) higher in this phase according to Schenker et al. (35),whereas no photolysis takes place on particles according toRaff and Hites (36). Although there have been several studiesreporting atmospheric PBDE concentrations, the number ofstudies that included BDE-209 is relatively small. There arean increasing number of studies reporting appreciable gas-phase BDE-209 fractions. Agrell et al. (37) have measuredthe BDE-209 mainly in gas-phase (>90%) at an urban site,whereas it was 100% in gas-phase at a rural site. The averageproportion of BDE-209 in gas-phase was 30 ( 11% at foursites in Izmir, Turkey (21). Li et al. (23) have reported thaton the average, 5% of BDE-209 was associated with gas-phase at three sites in southeast China. The results of thepresent study are very important since it was shown for thefirst time that significant amounts of heavy congeners likeBDE-209 could be emitted in the gas-phase, contrary tocommon assumption. However, it is not possible to deter-mine the net influence of being emitted in the gas-phase for

the environmental fate of BDE-209. Modeling efforts takinginto account the presence of emissions in the gas-phase,higher photolysis rates in this phase relative to particle-phase,and relatively slow gas-particle partitioning are required tobetter assess the environmental fate of PBDEs.

AcknowledgmentsWe thank the steel plants (HABAS, CEBITAS, EGE CELIK,IZMIR DEMIR-CELIK, and SIDER) for their support duringfield sampling.

Supporting Information AvailableSteel production by electric arc furnaces, stack-gas concen-trations (gas + particle) of PAHs, PCBs, and PBDEs (TableS1), PCB emission factors for steel plants (Table S2), PAHemission factors for steel plants (Table S3), PBDE emissionfactors for steel plants (Table S4), tentatively identifiedcompounds in stack gases of steel plants and in ambient air(Table S5), semiquantitative stack-gas concentrations ofvarious compounds measured in steel plants (Table S6),ambient air concentrations (gas + particle) of PAHs, PCBs,and PBDEs (Table S7), semi-quantitative ambient air con-centrations (gas + particle) of various compounds (TableS8), stack-gas sampling train (Figure S1), variation of stack-gas/ambient air gas and particle-phase concentration ratioswith octanol-air partition coefficient (Figure S2). This materialis available free of charge via the Internet at http://pubs.acs.org.

Literature Cited(1) International Iron and Steel Institute, World Steel in Figures

2007, http://www.worldsteel.org, 2008.(2) U.S. EPA. AP 42, Fifth Edition, Compilation of Air Pollutant

Emission Factors, Volume 1: Stationary Point and Area Sources.2008. Available at: http://www.epa.gov/ttn/chief/ap42/in-dex.html.

(3) IPPC. Best Available Techniques Reference Document on theProduction of Iron and Steel; European Commission, IntegratedPollution Prevention and Control (IPPC): Seville, Spain, 2001.

(4) Alcock, R. E.; Sweetman, A. J.; Prevedouros, K.; Jones, K. C.Understanding levels and trends of BDE-47 in the UK and NorthAmerica: an assessment of principal reservoirs and sourceinputs. Environ. Int. 2003, 29, 691–698.

(5) Bozlaker, A.; Odabasi, M.; Muezzinoglu, A. Dry deposition andsoil-air gas exchange of polychlorinated biphenyls (PCBs) in anindustrial area. Environ. Pollut. 2008, 156, 784–793.

(6) Bozlaker, A.; Muezzinoglu, A.; Odabasi, M. Atmospheric con-centrations, dry deposition and air-soil exchange of polycyclicaromatic hydrocarbons (PAHs) in an industrial region in Turkey.J. Hazard. Mater. 2008, 153, 1093–1102.

(7) Cetin, B.; Yatkin, S.; Bayram, A.; Odabasi, M. Ambient con-centrations and source apportionment of PCBs and traceelements around an industrial area in Izmir, Turkey. Chemo-sphere 2007, 69, 1267–1277.

(8) Cetin, B.; Odabasi, M. Particle-phase dry deposition and air-soil gas exchange of polybrominated diphenyl ethers (PBDEs)in Izmir, Turkey. Environ. Sci. Technol. 2007, 41, 4986–4992.

(9) Gomes, J. F. P. Emissions of polycyclic aromatic hydrocarbonsand polycyclic carbonyl biphenyls from electric arc furnaces.Rev. Metal. (Madrid, Spain) 2008, 44, 280–284.

(10) Almaula, S. Polycyclic aromatic hydrocarbons from steelmaking.Environ. Forensics. 2005, 6, 143–150.

(11) Agilent Technologies Inc., SemiQuant: New GC/MS SoftwareApproaches to Estimating Compound Quantities. TechnicalOverview, No. 5989-4997EN, 2006. Available from: http://www.agilent.com/chem.

(12) Long-range Transboundary Air Pollution, Proposed Annexes toa Draft Protocol on Persistent Organic Pollutants; UN-ECE(United Nations/Economic Commission for Europe): Geneva,Switzerland, 1997.

(13) Prevedouros, K.; Jones, K. C.; Sweetman, A. J. Estimation of theproduction, consumption, and atmospheric emissions of pent-abrominated diphenyl ether in Europe between 1970 and 2000.Environ. Sci. Technol. 2004, 38, 3224–3231.

(14) Source Control of Priority Substances in Europe (SOCOPSE).An Inventory and Assessment of Options for Reducing Emis-

5210 9 ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 43, NO. 14, 2009

Dow

nloa

ded

by A

NK

OS

CO

NSO

RT

IA T

UR

KE

Y o

n Ju

ly 1

3, 2

009

Publ

ishe

d on

Jun

e 12

, 200

9 on

http

://pu

bs.a

cs.o

rg |

doi:

10.1

021/

es90

0863

s

sions: Polybrominated Diphenyl Ethers (PBDEs), SpecificTargeted Research Project (Project contract no. 037038), SecondDraft, June, 2008.

(15) European Union Risk Assessment Report, Diphenyl Ether,Pentabromo Derivative, Final Report; European ChemicalsBureau, European Commission: Ispra, 2001.

(16) European Union Risk Assessment Report, Bis(Pentabromophenyl)Ether, Final Report; European Chemicals Bureau, EuropeanCommission: Ispra, 2002.

(17) Breivik, K.; Sweetman, A.; Pacnya, J. M.; Jones, K. C. Towardsa global historical inventory for selected PCB congeners -a massbalance approach 2. Emissions. Sci. Total Environ. 2002, 290,199–224.

(18) Breivik, K.; Sweetman, A.; Pacyna, J. M.; Jones, K. C. Towardsa global historical emission inventory for selected PCB con-geners-A mass balance approach-3. An update. Sci. TotalEnviron. 2007, 377, 296–307.

(19) EMEP (European Monitoring and Evaluation Programme), 2008.http://www.emep.int/.

(20) Demircioglu E. An investigation on polycyclic aromatic hydro-carbons (PAHs) in Izmir. PhD Thesis. Dokuz Eylul University,Izmir, 2008.

(21) Cetin, B.; Odabasi, M. Atmospheric concentrations and phasepartitioning of polybrominated diphenyl ethers (PBDEs) in Izmir,Turkey. Chemosphere 2008, 71, 1067–1078.

(22) Odabasi, M. Unpublished results. Department of EnvironmentalEngineering, Dokuz Eylul University, Izmir, 2008.

(23) Li, Y.; Jiang, G.; Wang, Y.; Wang, P.; Zhang, Q. Concentrations,profiles and gas-particle partitioning of PCDD/Fs, PCBs andPBDEs in the ambient air of an E-waste dismantling area,southeast China. Chin. Sci. Bull. 2008, 53 (4), 521–528.

(24) Odabasi, M.; Cetin, E.; Sofuoglu, A. Determination of octanol-air partition coefficients and supercooled liquid vapor pressuresof PAHs as a function of temperature: Application to gas/particlepartitioning in an urban atmosphere. Atmos. Environ. 2006, 40,6615–6625.

(25) Odabasi, M.; Cetin, B.; Sofuoglu, A. Henry’s law constant,octanol-air partition coefficient and supercooled liquid vaporpressure of carbazole as a function of temperature: Applicationto gas/particle partitioning in the atmosphere. Chemosphere2006, 62, 1087–1096.

(26) Harner, T.; Shoeib, M. Measurements of octanol-air partitioncoefficients (KOA) for polybrominated diphenyl ethers (PBDEs):Predicting partitioning in the environment. J. Chem. Eng. Data2002, 45, 1069–1074.

(27) Harner, T.; Bidleman, T. F. Measurements of octanol-air partitioncoefficients for polychlorinated biphenyls. J. Chem. Eng. Data1996, 41, 895–899.

(28) Chen, J.; Xue, X.; Schramm, K. W.; Quan, X.; Yang, F.; Kettrup,A. Quantitative structure-property relationship for octanol-airpartition coefficients of polychlorinated biphenyls. Chemosphere2002, 48, 535–544.

(29) Zhang, X.; Schramm, K. W.; Henkelmann, B.; Klimn, C.; Kaune,A.; Kettrup, A.; Lu, P. A method to estimate the octanol-airpartition coefficient of semivolatile organic compounds. Anal.Chem. 1999, 71, 3834–3838.

(30) Cetin, B.; Odabasi, M. Measurement of Henry’s law constantsof seven polybrominated diphenyl ether (PBDE) congenersas a function of temperature. Atmos. Environ. 2005, 39, 5273–5280.

(31) Braekevelt, E.; Titlemier, S. A.; Tommy, G. T. Direct measurementof octanol-water partition coefficients of some environmentallyrelevant brominated diphenyl congeners. Chemosphere 2003,51, 563–567.

(32) Assessment of Current Status of Aliaga Industrial Region for AirPollution, Report prepared for Ministry of Environment andForestry; Department of Environmental Engineering, DokuzEylul University: Izmir, Turkey, 2009; in Turkish.

(33) Gouin, T.; Thomas, G. O.; Chaemfa, C.; Harner, T.; Mackay, D.;Jones, K. C. Concentrations of decabromodiphenyl ether in airfrom Southern Ontario: Implications for particle-bound trans-port. Chemosphere 2006, 64, 256–261.

(34) Zhu, L.; Hites, R. A. Brominated flame retardants in tree barkfrom North America. Environ. Sci. Technol. 2006, 40, 3711–3716.

(35) Schenker, U.; Soltermann, F.; Scheringer, M.; Hungerbuhler,K. Modeling the environmental fate of polybrominateddiphenyl ethers (PBDEs): The importance of photolysis forthe formation of lighter PBDEs. Environ. Sci. Technol. 2008,42, 9244–9249.

(36) Raff, J. D.; Hites, R. A. Deposition versus photochemical removalof PBDEs from Lake Superior air. Environ. Sci. Technol. 2007,41, 6725–6731.

(37) Agrell, C.; ter Schure, A. F. H.; Sveder, J.; Bokenstrand, A.; Larsson,P.; Zegers, B. N. Polybrominated diphenyl ethers (PBDEs) at asolid waste incineration plant I: Atmospheric concentrations.Atmos. Environ. 2004, 38, 5139–5148.

ES900863S

VOL. 43, NO. 14, 2009 / ENVIRONMENTAL SCIENCE & TECHNOLOGY 9 5211

Dow

nloa

ded

by A

NK

OS

CO

NSO

RT

IA T

UR

KE

Y o

n Ju

ly 1

3, 2

009

Publ

ishe

d on

Jun

e 12

, 200

9 on

http

://pu

bs.a

cs.o

rg |

doi:

10.1

021/

es90

0863

s