Fuel Processing Technology 90 (2009) 643–651

Contents lists available at ScienceDirect

Fuel Processing Technology

j ourna l homepage: www.e lsev ie r.com/ locate / fuproc

Experimental study on mercury transformation and removal in coal-fired boilerflue gases

Yunjun Wang a, Yufeng Duan a,⁎, Liguo Yang a, Changsui Zhao a, Xianglin Shen a, Mingyao Zhang a,Yuqun Zhuo b, Changhe Chen b

a School of Energy and Environment, Southeast University, Nanjing 210096, PR Chinab Department of Thermal Engineering, Tsinghua University, Beijing 100084, PR China

⁎ Corresponding author. Tel.: +86 25 83795652; fax:E-mail address: yfduan@seu.edu.cn (Y. Duan).

0378-3820/$ – see front matter © 2008 Elsevier B.V. Adoi:10.1016/j.fuproc.2008.10.013

a b s t r a c t

a r t i c l e i n f o

Article history:

This paper reported mercu Received 26 July 2008Received in revised form 20 October 2008Accepted 24 October 2008

Keywords:Coal-fired power plantMercury transformationMercury removalFlue gas

ry speciation and emissions from five coal-fired power stations in China. Thestandard Ontario Hydro Method (OHM) was used into the flue gas mercury sampling before and after fabricfilter (FF)/electrostatic precipitator (ESP) locations in these coal-fired power stations, and then variousmercury speciation such asHg0, Hg2+ andHgP influe gas,was analyzed byusing EPAmethod. The solid samplessuch as coal, bottom ash and ESP ash, were analyzed by DMA 80 based on EPA Method 7473. Through analysisthe mercury speciation varied greatly when flue gas went through FF/ESP. Of the total mercury in flue gas, theconcentration of Hg2+ is in the range of 0.11–14.76 μg/Nm3 before FF/ESP and 0.02–21.20 μg/Nm3 after FF/ESP;the concentration of Hg0 ranges in 1.18–33.63 μg/N m3 before FF/ESP and 0.77–13.57 μg/N m3 after FF/ESP, andthat of HgP is in the scope of 0–12.11 μg/Nm3 before FF/ESP and 0–0.54 μg/Nm3 after FF/ESP. The proportion ofHg2+ ranges from 4.87%–50.93% before FF/ESP and 2.02%–75.55% after FF/ESP, while that of Hg0 is between13.81% – 94.79% before FF/ESP and 15.69%–98% after FF/ESP, with that of HgP is in the range of 0%–45.13% beforeFF/ESP and 0%–11.03% after FF/ESP. The mercury in flue gas mainly existed in the forms of Hg0 and Hg2+. Theconcentrations of chlorine and sulfur in coal and flue gas influence the species of Hg that are formed in the fluegas entering air pollution control devices. The concentrations of chlorine, sulfur and mercury in coal and thecompositions of fly ash had significant effects on mercury emissions.

© 2008 Elsevier B.V. All rights reserved.

1. Introduction

Various kinds of air pollutants emitted from coal-fired powerstations, and heavy metals were included as hazardous air pollutants.As one of heavy metals, worldwide mercury emissions from anthro-pogenic sources were currently estimated to about 4000 tons/annum.It emitted to the air in the process of combustion.

According to the global mercury mass balancing model [1], 34% ofthe total emissions originated from coal burning. The largest amountof mercury emitted from coal-fired power stations. US EPA submitteda report to USA Congress in 1997, and also pointed out 33% of mercuryemissions, induced by human beings, was from coal-fired powerstations [2]. The average mercury content of Chinese coals was about0.22 mg/kg. From 1978 to 1995 the total mercury emissions in Chinareached 2493.8 tons with an average increasing rate of 4.8% per yearduring the process of coal combustion [3]. Mercury emissions fromcoal combustion increased from 202 tons in 1995 to 257 tons in 2003at an average annual rate of 3.0%. Among all of the coal consumptionsectors, the mercury emissions growth of the power sectors was the

+86 25 83795652.

ll rights reserved.

largest up by 5.9% annually [4]. Mercury emissions of coal-fired powerplants, occupying 33.6% in China, were the second largest from coalcombustion in China [5].

Mercury, emitted from coal-fired power stations, exists in threeprimary forms, which are namely elemental mercury (Hg0), gaseousoxidized mercury (Hg2+) and particle-bound mercury (HgP). All ofthem will do direct or potential harm to human health. Therefore,researching mercury speciation and emissions and removing themfrom coal-fired power stations are especially important. Differentspeciation of mercury has different physical and chemical properties.The gaseous oxidized mercury is soluble and has the tendency toassociate with particulate matter. Thus, emissions of the gaseousoxidized mercury may be efficiently controlled by air pollutionparticulate-controlling devices, such as a flue gas wet desulphuriza-tion scrubber system, electrostatic precipitator (ESP) or fabric filter(FF). Also, particle-bound mercury can be easily collected by dustremoval devices, whose resident time is shorter in the atmosphere. Bycontrast, elemental mercury is extremely volatile and insoluble.Elemental mercury has a high pressure at the typical operatingtemperatures of air pollution control devices. Therefore, it is difficultto capture elemental mercury by wet flue-gas desulphurization(WFGD), flue-gas desulphurization (FGD) or particle collection device

Table 1Information of five coal-fired power plants.

Item MW The system of power plant Fuel Load (%)

Plant 1 50 PC+FF Zhunger bituminous 100Plant 2 200 PC+FF Zhunger bituminous 100Plant 3 220 PC+ESP Shenhua bituminous 100Plant 4 600 PC+ESP Zhunger bituminous 90Plant 5 600 PC+ESP+WFGD Shenhua bituminous 90

644 Y. Wang et al. / Fuel Processing Technology 90 (2009) 643–651

(PCD) system. Nearly, elemental mercury completely releases into theatmosphere. As a result, it is very important for its removal to analyzethe distributions and the factors of mercury speciation in flue gas.

In order to effectively control the mercury emitted from coal-firedpower stations, many countries started to research the distributionsand the factors influencing on the mercury speciation and itstransformation. So some researchers focused on understanding themechanisms of the mercury oxidization by injection of elementalmercury into solid or gas fuel flame or simulated flue gas. Otherresearchers worked on gas-phase mercury speciation using chemicalkinetics or the fate of mercury in the flue gas during combustionprocesses. Bench-scale experiments by Yan et al. [6] indicated thatHg0 could be oxidized by Cl2 in simulated flue gas and flue gasconstituents (SO2, NOx, and CO) and fly ash impacted on the reactionbetween Hg0 and Cl2. In the presence of fly ash, NO2, HCl, and SO2

resulted in greater levels of Hg oxidation, while NO inhibited Hgoxidation [7]. The effect of coal chlorine can also been discussed in theplants that use selective catalytic reduction (SCR) units for NOx

control. In general, the oxidation was influenced by the presence ofchlorine. In the bench and field tests, significant oxidation (65–91%)has been observed for bituminous coal-fired units. However, for thelow-chlorine sub-bituminous coal-burning units, little catalyticoxidation has been observed [8,9]. Galbreath et al. [10] studied themercury transformations of coal combustion flue gas in 42-MJ/hcombustion system. The results showed that mercury chlorination,catalysis of mercury oxidation by Al2O3(s) and/or TiO2(s), andcalcium-rich (25.0 wt.% CaO) fly ash from subbituminous coal had agreat effect on mercury speciation. Helble [11] conducted the researchto observe the harmful microelement distribution in gasificationprocess. Eswaran and Stenger [12] studied mercury conversion inselective catalytic reduction catalysts, whichwas carried out in a pilot-scale system under the following conditions: a temperature of 371 °C,space velocities of 4000 h−1, and a flue gas consisting of N2, CO2, SO2,and NO. A maximum of 70% Hg0 oxidation was observed with HCl inthe flue gas, at the maximum tested concentrations of 35 ppm. Inrecent years, China have commenced on the related researches on thetrace element characteristics and distribution in the coal-fired fluegas. Zhou et al. [13] investigated mercury speciation and transforma-tion in the post combustion condition and the gas-phase interactionsbetween flue-gas constituents and mercury in the bench-scaleexperiment. Yang et al. [14] researched the mercury speciation anddistribution in a selected 220MWpulverized-coal boiler system. Chenet al. [15] studied that mercury could be oxidized while the flue gaspassed through ESP/FF. This paper reported that the field measure-ment results of mercury based on the US-EPA recommended OntarioHydro Method (OHM), which was carried out in five pulverized-coalboiler systems, and the characteristics of mercury transformation andremoval by ESP/FF were obtained.

2. Experimental section

2.1. Sampling

The Ontario Hydro Method (OHM) was used to take the flue gassamples, which was the standard method of measuring and speciatingmercury influe gas. OHMhas twopossible configurations based on EPAMethods 5 and 17 out-of-stack filtration and in-stack filtration,respectively. The EPA Method 17 configuration was used at thesampling point after the electrostatic precipitator. Due to the highamount of fly ash immediately before the ESP region, a modifiedsampling train (EPA Method 5) with both in-stack and out-of-stackfiltration was used. In order to collect the representative samples, theisokinetic sampling was conducted to collect sufficient particulatematters and flue gas. Sampling gas went through a probe/filter systemmaintained at 120 °C or the flue gas temperature. Then the sampleflowed through a series of impingers,which immerged into an ice bath.

Particle-bound mercury was collected in the front tip of the samplingprobe. The first three impingers containing 1 N potassium chloride(KCl) solutionwere connected to absorb oxidizedmercury (Hg2+). Thefourth impinger containing acidified hydrogen peroxide (H2O2) wasused to absorb elemental mercury, and elemental mercury wasmainlycaptured in the fifth, sixth and seventh impingers which contained thesolutions of acidified potassium permanganate (KMnO4). In addition,the eighth impinger containing silica gel was provided to ensure thatthe flue gas was thoroughly dried-up before it left the impinger train.

2.2. Analysis

The samples, taken from flue gas, were immediately recovered anddigested using Ontario Hydro Method. The OHM solutions wereanalyzed using a Leeman Labs Hydra AA. The Hydra AA is a cold vaporatomic absorption (CVAA) instrument dedicated to mercury analysis.It has a detection limit of 1 ppt. According to EPA Method 7473, thesolid samples were analyzed by Milestone DMA80.

2.3. Testing locations

The results discussed in this paper were obtained from fiveselected power stations with pulverized coal-fired boiler systems inChina. These boilers in commercial plants were operated normallyduring testing. Table 1 listed some information of five coal-fired powerplants. The boiler loads ranged from 50 MW to 600 MW, and it couldbasically represent the overall structure of the current coal-firedpower plants. Five coal-fired power stations were installed with dust-removing equipment (FF/ESP), and were tested at full capacity or closeto full-load operating condition. A diagram of the boiler was shown inFig. 1. The mercury concentration in the flue gas was measured at twolocations, namely before the fabric filter (FF) or electrostaticprecipitator (ESP), and after the FF or ESP in the duct leading to thestack. OHM was used to measure the mercury concentration at bothlocations. In order to get the samples accurately, the coal sampleswere collected in the air-coal powder pneumatic conveying ductimmediately before burners. The bottom ash was sampled on belt ofthe slag-discharging machine, and the flue gas samples used the OHMsystem before and after FF/ESP respectively at the same time. The flyashes in the FF/ESP were drawn by a vacuum pump.

3. Results and discussion

3.1. Data analysis

Several studies suggested that many different factors mayinfluence mercury speciation and emissions. Mercury speciation andemissions in coal-fired flue gas from the power stations are stronglydependent on the coal types (e.g., bituminous, sub-bituminous orlignite), the operating conditions of the combustion system (in termsof unburned carbon in fly ash), and the temperature and residencetime in particulate control device et al. This study researched severalfactors influencing on mercury speciation and distribution in coal-fired flue gas. Table 2 showed the ultimate and proximate analyses andtrace element contents in coal. The content of mercury in coal iswithin 0.01–0.25 mg/kg, which is in the range of 0.02–1.0 mg/kg [16].

Fig. 1. Sampling locations in the PC boiler with FF/ESP system.

645Y. Wang et al. / Fuel Processing Technology 90 (2009) 643–651

Fig. 2 showed the mercury speciation concentrations, which weremeasured in thefluegasbeforeandafter FF/ESP. ThecontentofHg0/Hg2+

was far larger than that of HgP in pulverized-coal boiler flue gas. Thecontent of divalent mercury was more than elemental mercury in the FFoutlet flue gas of Plants 1 and 2. However, the content of elementalmercury was higher than divalent mercury in the ESP outlet flue gas ofPlants 3, 4 and 5. The share of elemental mercury in flue gas ranged from13% to 98%, and the divalent mercury was in the range of 2%–80%.Prestbo [17]measuredmercuryconcentrations in coal combustiongasof14 coal-fired plants, and it indicated that the shares of elementalmercury and divalent mercury in flue gas were in the range of 6%–60%and 40%–94%, respectively. Various forms of mercury concentration inthe flue gas from five power stations were accorded with Prestbo'sconclusions. From Fig. 2, it was seen clearly that HgP in FF inlet of Powerplant 1was exceptionally higher; the reasonmight be that the unburnedcarbon in the FF fly ash of power plant 1 was the highest, and it couldadsorb themercury in flue gas effectively. However, HgPwas very low inother places, and therewas noHgP detected before and afterWFGD. Thereasonwas that therewas little fly ash in the filter cylinder. So thanks toOHM sampling, there was error in the measurement.

3.2. Mercury speciation

3.2.1. Effect of chlorine in coalChlorine in coal plays a primary role in mercury oxidation, which

emits into flue gas at high temperatures. With the temperature andchlorine in flue gas increasing, furthermore the oxidized mercury willbe growing up. The reason may be that chlorine is formed atomicchlorine at high temperatures, which has quick reactionwithmercury.And the gas–solid (heterogeneous) oxidations involve surface inmercury oxidation and subsequent bindingwith a chlorinated surface.Mercury chlorination, HgCl2(g), was generally considered to be one of

Table 2The ultimate and proximate analyses in coal.

Item Cad Had Nad Oad Sad Clad Hgad Wad Aad Vad FCad

wt.% wt.% wt.% wt.% wt.% mg/kg mg/kg wt.% wt.% wt.% wt.%

Plant 1 47.58 3.7 1.24 10.97 0.74 305 0.244 1.88 33.89 26.11 38.13Plant 2 51.55 3.62 1.31 10.28 0.81 288 0.188 3.54 28.9 28.2 39.37Plant 3 68.25 4.54 1.36 9.84 0.37 154 0.011 8.05 7.59 30.82 53.55Plant 4 46.01 3.09 1.21 9.58 0.52 198 0.209 3.18 36.41 24.47 35.95Plant 5 52.66 3.83 1.91 8.43 0.51 510 0.05 1.83 30.83 28.7 38.64

the dominant mercury transformation mechanisms in coal combus-tion flue gas, even though HgCl2(g) has never been directly measured.The ratio of oxidized mercury to total mercury in flue gas was affectedby chlorine in coals (Fig. 3). Liu et al. [18] provided some insight intomercury emissions from high-chlorine coal-fired FBC systems, andfound out that bituminous coals had the highest chlorine concentra-tions in the range of 500–1300mg/kg. At the first air pollution controldevice (APCD) inlet, the amount of Hg2+ valuing from 70% to 88% inbituminous coals was the highest in the flue gas from these coals.However, lignite had some of the lowest chlorine concentrations,below detection limit to 60 mg/kg. Flue gas from the combustion oflignite coal contained a corresponding low percentage of Hg2+.

3.2.2. Effects of HCl and Cl2 in flue gasFigs. 4 and 5 were shown that chlorine was mainly in the form of

HCl in the coal-fired flue gases. HCl was much more than Cl2 in fluegas. HCl played an important role in mercury oxidation, as shown inFig. 4. With HCl increasing, the oxidized Hg increased. It was the mostimportant species affecting mercury oxidation since the majoroxidized mercury species in coal-fired flue gas was HgCl2. HCl had apositive impact on increasing Hg2+ in the bituminous coal-burning fluegas in this study. The results were consistent with the formerresearches [19]. Cl2 also had an important effect on mercury oxidationin Fig. 5, although it had a very small proportion. There was a positive

Fig. 2. The content of mercury before and after FF/ESP in five coal-fired power plants.

Fig. 3. Hg2+ vs. chlorine content in coal.

646 Y. Wang et al. / Fuel Processing Technology 90 (2009) 643–651

effect on oxidized mercury due to its concentration. Experiments andkinetic calculations also indicate that homogeneous chlorinationreactions may be greatly affected by the rapid quench ratesexperienced in a coal-fired boiler and by related changes in chlorinespecies, including Cl2 and super-equilibrium concentrations of Cl[19,20]. Niksa et al. [20] identified the predominant reactions to be thefirst, which was, the oxidation of Hg0 by atomic Cl yielded labile HgCl,followed by the oxidation of HgCl by Cl2 to produce HgCl2 withassociated regeneration of atomic Cl. In this chlorine recycle process,the concentrations of both atomic Cl and Cl2 were important.Equilibrium calculations conducted for mercury at stack gas conditionsindicate that the oxidized form is thermodynamically stable. Chlorinein flue gas is a very important constituent for mercury oxidation, andmercury chloride (HgCl2) is reported as the most prominent form ofoxidized mercury. Hall [21] researched various compositions of coal-fired flue gas at 20–900 °C, and found out that Hg0 (g) could react withHCl or Cl2 rapidly. Kellie et al. [22] concluded that the divalent mercuryincreased with the increasing of HCl in flue gas. Agarwal and Stenger[23] investigated effects of H2O, SO2, and NO on homogeneous Hgoxidation by Cl2, and found that chlorine gas (Cl2) oxidized Hgeffectively. In all these experiments, Cl2 was intentionally added into aflue gas stream and the percent of Hg0 oxidationwas observed. In otherexperiments, Agarwal et al. [24] investigated Hg0 oxidation by chlorinespecies in the temperature range of approximately 120–550 °C. Twotests showed that Cl2 became less effective in oxidizing Hg0 at highertemperatures, but more than 50% Hg0 was oxidized, and mercuryoxidation increased with the temperature decreasing.

The reaction mechanism may be as follows [25]:

Hg0ðgÞ þ HClðgÞ→HgClðgÞ þ H

HClðgÞ→ClðgÞ þ H

Hg0ðgÞ þ ClðgÞ→HgClðgÞ

HgClðgÞ þ HClðgÞ→HgCl2ðgÞ þ H

Hg0ðgÞ þ Cl2ðgÞ→HgCl2ðg; sÞ

Hg0ðgÞ þ Cl2ðgÞ→HgClðgÞ þ ClðgÞ

HgClðgÞ þ ClðgÞ→HgCl2ðgÞ

3.2.3. Effect of NOx in flue gasIn flue gas, NOx also influenced the mercury speciation distribu-

tion, as shown in Fig. 6. Having some catalytic effect and making theflue gas oxidized, NOx might cause partially elemental mercury to beoxidized. Thus, the higher the content of NOx, the higher the ratio of

oxidized mercury. NO was predicted to either promote or inhibitoxidation, depending on its concentration. Higher quench ratesincreased mercury oxidation in the presence of NO, whereas fasterquenching decreased mercury oxidation without NO [20]. In thepresence of NO2, Hg0 was catalytically oxidized on the surface to formthe nonvolatile nitrate Hg(NO3)2 possibly, which was bound to basicsites on the carbon. Capture continued until the binding sites wereused up and breakthrough occurred [26]. Hall et al. [27] studied thehomogeneous reaction between NO2 and Hg0. A small but significant(16.7%) oxidation of Hg0 was observed at 340 °C when the NO2

concentrationwas 400 ppm. The oxidation of mercury increased up to30%with increasing the NO2 concentration to 1000 ppm. Eswaran alsoinvestigated the removal of mercury by using zeolite [28]. In theexamination, flue gas was composed of N2, O2 and CO2 and the contentof elemental mercury was 15 μg/m3. When there was NO in thesimulativefluegas, the contentof elementalmercury reduced to3 μg/m3

sharply. It indicated that NO could boost the mercury oxidation. Basedon thebench-scale tests, therewas a significant interaction betweenNOx

and the fly ash generated from an eastern bituminous coal burning,which greatly impacted mercury speciation. The concentration of theadded Hg0 measured as Hg2+ was greater than 25% when the NOx waspart of simulated flue gas passing through the fly ash [29]. Carey et al.[30] investigated the adsorption by several different sorbents using bothelemental mercury and mercuric chloride under simulated flue gasconditions in the bench-scale and fixed-bed experiment. The gasconcentration of NOx was much lower at Site 1 than that at Site 2. Inaddition, Site 2 was believed to have a higher ratio NO2/NO in the fluegas. This ratio affected the measured adsorption capacity. The totalmercury concentration at the two sites varied from10–30 μg/Nm3,with47% of the mercury being oxidized at Site 1 versus 65% at Site 2.

The reaction mechanism may be as follows [13]:

NOðgÞ þ O2→NO2ðgÞ þ O

Hg0ðgÞ þ O→HgOðgÞ

Hg0ðgÞ þ NO2→HgOðg; sÞ þ NOðgÞ

3.2.4. Effect of sulfur in coalMeasurements infive Chinese power plants showed that Hg2+/HgT

increased with coal S increasing (Fig. 7). Frandsen put forward that Sand Cl could oxide the elemental mercury [31]. The concentration ofSO2 was an important factor in mercury oxidation; furthermore, itconcealed the effect of Cl. The model of Frandsen was the mostcomprehensive, based on the number of elements and mercuryspecies considered. The modeling results predicted that, at lowtemperatures in conventional coal combustion systems, in the

Fig. 4. Hg2+ vs. HCl in flue gas.

647Y. Wang et al. / Fuel Processing Technology 90 (2009) 643–651

absence of chlorine HgSO4(s) was the stable form of mercury. Themodel also predicted that when the temperature of flue gas wasmorethan 700 K, the divalent mercury was mainly HgCl2. As the tem-perature of flue gas was less than 590 K, the divalent mercury wasmainly HgSO4. Kellie et al. [22] provided sight into the factors, whichaffected mercury speciation in a 100-MW coal-fired boiler with low-NOx burners. Hg0 decreased as SO2 in the flue gas increased in themeasurement. In contrast, Hg0 was more responsive to SO2 levels inthe flue gas than to the coal sulfur content.

Equilibrium reactions proposed by Frandsen et al. were shown asfollows:

Hg0ðgÞ þ 1=2O2ðgÞ→HgOðgÞ

HgOðgÞ þ 2HClðgÞ→HgCl2ðgÞ þ H2OðgÞ

HgOðgÞ þ SO2ðgÞ þ 1=2O2ðgÞ→HgSO4ðsÞ

HgOðgÞ→HgOðsÞ

HgCl2ðgÞ þ SO2ðgÞ þ O2ðgÞ→HgSO4ðsÞ þ Cl2ðgÞ

3.2.5. Effect of SO2 in flue gasAs shown in Fig. 8, there was the relationship between oxidized

mercury and SO2. The results showed that both the SO2 concentrationin flue gas and sulfur content in coal had a positive activity in oxidizingmercury [32]. It was well established that adsorption of SO2 occurred

Fig. 5. Hg2+ vs. C

on the activated carbon surface and subsequent oxidation with O2 ledto the formation of adsorbed-SO3. The reaction of adsorbed-SO3 andwater vapor led to the formation of adsorbed-H2SO4 on the activatedcarbon surface [33,34]. Hg0 could react with O2 to form HgO whichcould then react with the adsorbed-SO2 to produce HgSO4 [35].

However, some other researchers held that some of the catalyticsites were converted to a sulfate form in the presence of SO2 where Hg(NO3)2 was no longer formed. Mercury was oxidized on the surfacewith NO2 acting as the oxidizing agent, but the product formed was alabile sulfur compound. The bisulfate in turn reacted with NO3

− toform a stable but volatile acidic form of the mercuric nitrate [32].

3.3. Mercury removal by FF/ESP

3.3.1. Mercury in coalMercury emissions factor (EF) represents the total amount of

mercury emission from coal combustion. According to the mercuryconcentration tests innine coal-firedplants in1996byUSDOE, the resultsshowed that mercury emissions factor ranged from 1.9–22 lb/1012 Btu(0.82–9.46 g/1012 J) [36]. The mercury emission factor in this paperreferred to University of North Dakota's report, but international unitswere adopted and defined as follows:

Mercury emissions factor g= Jð Þ¼ Mercury emissions to atmosphere g= hð ÞCoal in boiler g= hð Þ × Coal lowheat value J= gð Þ

Due to the mercury control technology, burning low-S coal is apopular method of reducing S emissions. Likewise, burning low-Hg

l2 in flue gas.

Fig. 6. Hg2+ vs. NOx in flue gas.Fig. 8. Hg2+ vs. SO2 in flue gas.

648 Y. Wang et al. / Fuel Processing Technology 90 (2009) 643–651

coal is a likelymethod of reducing Hg emissions. Fig. 9 showed that Hgemissions from five Chinese coal-fired power plants increased withthe Hg content of the coal increasing, and the results showed thatabout 75% of mercury in coal was emitted to the air and 25% ofmercury in coal was captured by FF/ESP. Kolker et al. [37] studied thatHg emissions from many American coal-fired power plants variedsystematically according to the Hg content of the coal, and the best-fitline indicated that for conventional coal-fired plants, 64% of coal Hgwas emitted to the atmosphere and 36% was captured.

3.3.2. ChlorineThe halogen content in coal affects mercury oxidation in flue gas by

homogeneous, heterogeneous and catalytic pathways. Chlorine is themost crucial halogen in combustion systems as far as mercurychemistry is concerned. Mercury removal increased with the contentof chlorine increasing (Fig. 10). It is generally acknowledged that theprimary product of Hg oxidation in flue gas is believed to be HgCl2.Thermodynamic equilibrium calculations performed by Frandsen etal. predicted that all gas-phase mercury in a coal combustion flue gasshould exist as mercuric chloride (HgCl2) at temperatures below450 °C [31]. The gas-phase transformation pathway involved theoxidation of Hg0 by atomic Cl. The concentration of Cl depended oncomponents of flue gas such as O2, H2O, hydrocarbons, Cl compounds,and S compounds. Higher concentration of chlorinewithin the flue gasstream resulted in higher mercury oxidation, where Hg2+ was thedominant species form. Senior et al. researched that the cooling rate in

Fig. 7. Hg2+ vs. sulfur content in coal.

the flue gas strongly influenced homogeneous oxidation of Hg0 [18].High cooling rates of flue gas between the air preheater inlet and airpollution control device inlet limited reaction rates associated withhomogeneous oxidation reactions.

3.3.3. SulfurEmpirical equations, derived from measurements in five China

power plants, showed mercury removal increased with coal Sincreasing (Fig. 11). Morimoto et al. reported that adsorption of SO2

occurred on the activated carbon surface and subsequent oxidationwith O2 led to the formation of adsorbed-SO3 [38]. Norton et al. [7]studied heterogeneous oxidation of mercury in simulated postcombustion conditions. In the presence of fly ash mercury oxidationmechanisms were very complex. Although fly ash played a vital role inmercury oxidation, the flue gas compositions were more importantthan the ash compositions and the sources of the fly ash. The SO2

promoted mercury oxidation, with NO2 being the most importantfactor. Hutson et al. [39] have recently used X-ray absorptionspectroscopy (XAS) and X-ray photoelectron spectroscopy (XPS) toevaluate mercury binding mechanism on conventional (non-haloge-nated) and chlorinated activated carbon sorbents. In this work theappearance of XAS and XPS for both the conventional and the pre-chlorinated Cl sorbents suggested that some mercury may have beenbound to sulfate species that were incorporated onto the carbon fromSO2 in the flue gas mixture.

Fig. 9. Mercury emissions vs. Hg content in coal.

Fig. 10. Mercury removal vs. chlorine content in coal.Fig. 12. Mercury in fly ash vs. unburned carbon.

649Y. Wang et al. / Fuel Processing Technology 90 (2009) 643–651

3.3.4. Compositions of fly ashThe carbon content of fly ash in plant 1 was larger than in plant 2,

and studies showed that activated carbon was the most effectiveadsorbent for mercury. Therefore, the residual carbon in fly ash on theremoval of mercury was very favorable. The residual carbon content inFF/ESP fly ash is higher, and the removal efficiency will be higher(Fig. 12) [40]. Unburned carbon in fly ash has been shown to be animportant factor inmercury capture by FF/ESP. Unburned carbon in flyash had higher Hg adsorption capacity. The surface area of sorbentshad a positive correlation with Hg adsorption capacity. Carbon–oxygen radicles CfO on unburnt carbon surface were beneficial tooxidation and chemical adsorption of Hg [41]. It was obtained that theporous structure and huge surface area of the unburned carbon couldbenefit the Hg adsorption in flue gas. Unburned carbon in ESP fly ashof plant 3 was higher; so much Hg2+ was adsorbed by ESP fly ash.Senior et al. [42] carried out the experiments on the relationshipbetween loss-on-ignition and mercury content in FF/ESP fly ash andfound that the removal efficiency also increased with unburnedcarbon increasing. Gale et al. [43] investigated the correlation betweenflue-gas parameters andmercury oxidation and capture across a pilot-scale FF/ESP in full-scale boilers. The mercury oxidation increasedwith the increase of UBC in the fly ash. UBC was the dominantparameter affecting the flue-gas Hg-oxidation state. In other tests,hydrated-lime injected into the baghouse increased Hg-removal from25% to nearly 80%. They proposed an empirical quadratic model,derived from response-surface concepts, which was used to record

Fig. 11. Mercury removal vs. sulfur content in coal.

and describe the data correlation of mercury removal with CaO and Cconcentrations in the flue gas:

HgRemoval kð Þ = b0 + b1CC + b2CCao + b3C2C + b4C

2Cao + b5CCCCao

where CC and CCaO represent the concentration of C and CaOrespectively in the flue gas (lb/MMacf).

Several researchers have investigated the possibility of using theunburned carbon of fly ash to instead of activated carbon. In bench-scale experiments, fly ashes containing between 2% and 35% UBCcould adsorb significant amounts of mercury. The amount of mercuryadsorbed generally increasedwith surface area of the ash, but was alsoinfluenced by the particle size and porosity [44].

Researches indicated that other components of fly ash (such as CaO[45], Fe2O3 [46] etc.) could also adsorb mercury. The adsorptionperformance of calcium-based sorbents was worse in the absence ofSO2 including lime, hydrated lime, and a mixture of fly ash andhydrated lime, but the Hg capture efficiency enhanced from about 10%to 40% in the presence of SO2, and the adsorption capacity wasincreased in 30 min of exposure of sorbents to flue gas, and it wasadvantageous at higher temperatures. The observed higher capturesat higher temperatures supported the chemisorptions theory of Hg0

capture [47]. Calcium-based sorbents have been shown to havemercury binding capacity in the presence of SO2. It was thought thatthe calcium-based sorbents first capture SO2, and then form stronglyacidic sites on the surface of the sorbent. These acidic sites werebenefit for mercury oxidation [48]. Laboratory studies were conductedto develop an elemental mercury (Hg0) removal process using iron-based sorbents for coal-derived flue gas. The Hg0 removal experimentswere carried out in a laboratory-scale fixed-bed reactor at 80 °C using

Table 3Summary of fly ash composition collected in flue gas before ESP/FF.

Oxides share Plant 1 Plant 2 Plant 3 Plant 4 Plant 5

SiO2 45.39 49.53 41.76 48.46 44.12Al2O3 42.38 42.51 14.49 39.97 24.93TiO2 1.34 1.95 0.79 2.03 1.4Fe2O3 2.35 2.29 11.08 2.76 6.89CaO 2.89 2.17 22.85 3.59 13.26MgO 0.31 0.28 1.97 0.72 1.35Na2O ND 0.12 1.18 0.16 0.95K2O 0.42 0.33 1.14 0.64 1.15UBC 2.87 1.13 2.59 0.9 1.69A/B 14.93 18.11 1.49 11.49 2.99Hg/(A/B) 0.016 0.010 0.007 0.018 0.017

UBC—Unburned carbon, A=SiO2+Al2O3+TiO2, B=Fe2O3+CaO+MgO+Na2O+K2O,ND: not detectable.

650 Y. Wang et al. / Fuel Processing Technology 90 (2009) 643–651

simulated flue gas with a composition of Hg0 (4.8 ppb), H2S(400 ppm), CO (30%), H2 (20%), H2O (8%), and N2 (balance gas) at80 °C. More than 50% Hg0 could be removed by iron oxide (Fe2O3) inthe absence of HCl within 4 hours [49].

The ratio of acid to alkali also had an effect on mercury removal inTable 3. The ratio of mercury to A/B was defined as the evaluationcriteria. Because Hg/(A/B) of Plant 1wasmore than that of Plant 2, theremoval efficiency of plant 1 was higher. In addition, when the ratio ofHg to (A/B) was larger, in other words, alkaline oxide contents in flyash were larger, fly ash was able to adsorb more gaseous mercury asspecified above in plant 4 and 5. However, residual carbon in fly ash ofplant 4 was the least compared with others. Therefore, it can concludethat the adsorption behavior andmechanism of fly ash formercury arevery complex, and multiple factors have different influences onmercury adsorption by fly ash.

4. Conclusions

The mercury contents of consumed coal in the five power stationsexamined was measured, and the range of its contents in coal was0.01–0.25 mg/kg.

The mercury speciation varies greatly when flue gas goes throughFF/ESP. Of the total mercury in flue gas, the proportion of Hg2+ rangesfrom 4.87%–50.93% before ESP and 2.02%–75.55% after ESP, while thatof the Hg0 is in the range of 13.81%–94.79% before ESP and 15.69%–98%after ESP, and that of the HgP is in 0%–45.13% before ESP and 0%–11.03%after ESP. Coal quality, notable Cl and S contents, and the combustioncharacteristics of the coal all influence Hg speciation in flue gasentering air pollution control devices. Elemental Hg is difficult toremove from the gas phase, but particulate boundHg forms after post-combustion can be more readily captured in FF/ESP. The contents ofchlorine and sulfur in coal have a positive correlation with theformation of oxidized mercury. The contents of chlorine, sulfur andmercury in coal have effects on mercury emissions. The flue gascomponents have an important influence on the mercury speciation.With the NOx, SO2, HCl and Cl2 contents increasing, the proportion ofoxidized mercury increases.

The same dust removal equipment has higher efficiency ofmercury removal, which works on different coal-fired power plants.The reason may be that the contents of unburned carbon and alkalinemetal oxides et al are different in the removed ashes from the dustremoval equipment of different power plants. The mercury removalefficiency by FF/ESP is related not only with the adsorption of mercuryby fly ash particles, but also with many other factors, such as thecarbon content of fly ash and flue gas compositions. The chlorine andsulfur contents in coal have an important impact on mercury removalby FF or ESP.

Acknowledgements

This sub-project was joint-funded by the State Basic ResearchDevelopment Program (973 Plan) of China (No. 2002CB211604 &2006CB200300) and the Developing Plan of the Ministry of Educationof China (985-I). The authors are grateful to acknowledge Suli Mengand Zhijun Huang for assistance with revision of the paper.

References

[1] J.M. Pacyna, B. Ottar, Control and Fate of Atmospheric Trace Metals, NATO ASISeries, Kluwer Academic Publishers, The Netherlands, 1989.

[2] E. Grant, U.S. Environmental Protection Agency, Mercury Study Report to CongressVolume I: Executive Summary, Office of Air Quality Planning and Standards andOffice of Research and Development, Washington, DC, December 1997.

[3] Q.C. Wang, W.G. Shen, Z.W. Ma, The estimation of mercury emission from coalcombustion in China, China Environ. Sci. 19 (1999) 318–321.

[4] Y. Wu, S.X. Wang, D.G. Streets, Trends in anthropogenic mercury emissions inChina from 1995 to 2003, Environ. Sci. Technol. 40 (2006) 5312–5318.

[5] D.G. Streets, J.M. Hao, Y. Wu, J.K. Jiang, M. Chan, H.Z. Tian, et al., Anthropogenicmercury emissions in China, Atmos. Environ. 39 (2005) 7789–7806.

[6] N.Q. Yan, S.H. Liu, S.J. Chang, Method for the study of gaseous oxidants for theoxidation of mercury gas, Ind. Eng. Chem. Res. 44 (2005) 5567–5574.

[7] G.A.Norton,H.Q. Yang,R.C. Brown,D.L. Laudal,G.E. Dunham, J. Erjavec,Heterogeneousoxidation of mercury in simulated post combustion conditions, Fuel 82 (2003)107–116.

[8] C.W. Lee, R.K. Srivastava, S.B. Ghorishi, T.W. Hastings, F.M. Stevens, Investigation ofselective catalytic reduction impact on mercury speciation under simulated NOx

emission control conditions, J. Air Waste Manage. Assoc. 54 (2004) 1560–1566.[9] C.W. Lee, R.K. Srivastava, S.B. Ghorishi, J. Karwowski, T.W. Hastings, J.C. Hirschi,

Pilot-scale study of the effect of selective catalytic reduction catalyst on mercuryspeciation in Illinois and Powder River Basin coal combustion flue gases, J. AirWaste Manage. Assoc. 56 (2006) 643–649.

[10] K.C. Galbreath, C.J. Zygarlicke, Mercury transformations in coal combustion fluegas, Fuel Process. Technol. 65–66 (2000) 289–310.

[11] J.J. Helble, Trace element partitioning during coal gasification, Fuel 75 (1996)931–939.

[12] S. Eswaran, H.G. Stenger, Understanding mercury conversion in selective catalyticreduction (SCR) catalysts, Energy Fuels 19 (2005) 2328–2334.

[13] J.S. Zhou, Z.Y. Luo, C.X. Hu, K.F. Cen, Factors impacting gaseous mercury speciationin post-combustion, Energy Fuels 21 (2007) 491–495.

[14] X.H. Yang, Y.Q. Zhuo, Y.F. Duan, L. Chen, L.G. Yang, L. Zhang, Mercury speciation andits emissions from a 220 MW pulverized coal-fired boiler power plant in flue gas,Korean J. Chem. Eng. 24 (2007) 711–715.

[15] L. Chen, Y.F. Duan, Y.Q. Zhuo, L.G. Yang, L. Zhang, X.H. Yang, et al., Mercurytransformation across particulate control devices in six power plants of China: theco-effect of chlorine and ash composition, Fuel 86 (2007) 603–610.

[16] W.H. Huang, Y.C. Yang, Mercury in coal of China, Coal Geol. China 14 (2002) 37–40.[17] M. Prestbo, S. Bloom, Mercury speciation adsorption (MESA) method for

combustion flue gas: methodology, artifacts, inters comparison, and atmosphericimplications, Water Air Soil Pollut. 80 (1995) 145–148.

[18] K.L. Liu, Y. Gao, J.T. Riley, W.P. Pan, An investigation of mercury emission from FBCsystems fired with high-chlorine coals, Energy Fuels 15 (2001) 1173–1180.

[19] C.L. Senior, A.F. Sarofim, T.F. Zeng, J.J. Helble, R. Mamani-Paco, Gas-phase transforma-tions ofmercury in coal-fired power plants, Fuel Process. Technol. 63 (2000) 197–213.

[20] S. Niksa, J.J. Helble, N. Fujiwara, Kinetic modeling of homogeneous mercuryoxidation: the importance of NO and H2O in predicting oxidation in coal-derivedsystems, Environ. Sci. Technol. 35 (2001) 3701–3706.

[21] B. Hall, P. Schager, O. Lindqvist, Chemical reactions of mercury in combustion fluegases, Water Air Soil Pollut. 56 (1991) 3–14.

[22] S. Kellie, Y. Cao, Y.F. Duan, L.C. Li, P. Chu, A. Mehta, et al., Factors affecting mercuryspeciation in a 100-MW coal-fired boiler with low-NOx burners, Energy Fuels 19(2005) 800–806.

[23] H. Agarwal, H.G. Stenger, Effects of H2O, SO2, and NO on homogeneous Hgoxidation by Cl2, Energy Fuels 20 (2006) 1068–1075.

[24] H. Agarwal, C.E. Romero, H.G. Stenger, Comparing and interpreting laboratory resultsof Hg oxidation by a chlorine species, Fuel Process. Technol. 88 (2007) 723–730.

[25] R.N. Sliger, J.C. Kramlich, N.M. Marinov, Towards the development of a chemicalkinetic model for the homogeneous oxidation of mercury by chlorine species, FuelProcess. Technol. 65–66 (2000) 423–438.

[26] S.J. Miller, G.E. Dunham, E.S. Olson, T.D. Brown, Flue gas effects on a carbon-basedmercury sorbent, Fuel Process. Technol. 65–66 (2000) 343–363.

[27] B. Hall, P. Schager, O. Lindqvist, Chemical-reactions of mercury in combustion flue-gases, Water Air Soil Pollut. 56 (1991) 3–14.

[28] S. Eswaran, H.J. Stenger, Gas-phase mercury adsorption rate studies, Energy Fuels21 (2007) 852–857.

[29] D.L. Laudal, T.D. Brown, B.R. Nott, Effects of flue gas constituents on mercuryspeciation, Fuel Process. Technol. 65–66 (2000) 157–165.

[30] T.R. Carey, C.F. Richardson, R. Chang, F.B. Meserole, M. Rostam-Abadi, S. Chen,Assessing sorbent injection mercury control effectiveness in flue gas streams, Fall19 (2000) 167–174.

[31] F. Frandsen, K. Dam-Johansen, P. Rasmussen, Trace elements from combustion andgasification of coal—an equilibrium approach, Prog. Energy Combust. Sci. 20(1994) 115–138.

[32] J.H. Pavlish, E.A. Sondreal, M.D. Mann, E.S. Olson, K.C. Galbreath, D.L. Laudal, et al.,Status review of mercury control options for coal-fired power plants, Fuel Process.Technol. 82 (2003) 89–165.

[33] I. Mochida, K. Kuroda, S. Kawano, Y. Matsumura, M. Yoshikawa, Kinetic study of thecontinuous removal of SOx on polyacrylonitrile-based activated carbon fibres: 1.Catalytic activity of PAN-ACF heat-treated at 800 °C, Fuel 76 (1997) 533–536.

[34] S. Kisamori, I. Mochida, H. Fujitsu, Roles of surface oxygen groups on poly(acrylonitrile)-based active carbon fibers in SO2 adsorption, Langmuir 10 (1994)1241–1245.

[35] R.B. Heslop, P.L. Robinson, Inorganic Chemistry: A Guide to Advanced Study,Elsevier, Amsterdam, 1960.

[36] S.J. Miller, S.R. Ness, G.F. Weber, T.A. Erickson, D.J. Hassett, S.B. Hawthorne, et al., Acomprehensive assessment of toxic emissions from coal-fired power plants: phaseI results from the U.S. Department of Energy study, Final Report on Contract DE-FC21-93MC30097, Energy and Environmental Research Center, Grand Forks, ND,1996.

[37] A. Kolker, C.L. Senior, J.C. Quick, Mercury in coal and the impact of coal quality onmercury emissions from combustion systems, Appl. Geochem. 21 (2006) 1821–1836.

[38] T. Morimoto, S.J. Wu, U.M. Azhar, E. Sasaoka, Characteristics of the mercury vaporremoval from coal combustion flue gas by activated carbon using H2S, Fuel 84(2005) 1968–1974.

651Y. Wang et al. / Fuel Processing Technology 90 (2009) 643–651

[39] N.D. Hutson, B.C. Attwood, K.G. Scheckel, XAS and XPS characterization of mercurybinding on brominated activated carbon, Environ. Eng. Sci. 41 (2007) 1747–1752.

[40] C.L. Senior, S.A. Johnson, Impact of carbon-in-ash on mercury removal acrossparticulate control devices in coal-fired power plants, Energy Fuels 19 (2005)859–863.

[41] J. Jurng, T.G. Lee, G.W. Lee, S.J. Lee, B.H. Kim, J. Seier, Mercury removal fromincineration flue gas by organic and adsorbents, Chemosphere 47 (2002) 907–913.

[42] C. Senior, C.J. Bustard, M. Durham, K. Baldrey, D. Michaud, Characterization of flyash from full-scale demonstration of sorbent injection formercury control on coal-fired power plants, Fuel Process. Technol. 85 (2004) 601–612.

[43] T.K. Gale, B.W. Lani, G.R. Offen, Mechanisms governing the fate of mercury in coal-fired power systems, Fuel Process. Technol. 89 (2008) 139–151.

[44] S.D. Serre, G.D. Silcox, Adsorption of elemental mercury on the residual carbon incoal fly ash, Ind. Eng. Chem. Res. 39 (2000) 1723–1730.

[45] F. Goodarzi, Characteristics and composition of fly ash from Canadian coal-firedpower plants, Fuel 85 (2006) 1418–1427.

[46] S.J. Wu, M.A. Uddi, E. Sasaoka, Characteristics of the removal of mercury vapor incoal derived fuel gas over iron oxide sorbents, Fuel 85 (2006) 213–218.

[47] S.B. Ghorishi, C.B. Sedman, Low concentration mercury sorption mechanisms andcontrol by calcium-based sorbents: application in coal-fired processes, J. Air WasteManage. Assoc. 48 (1995) 1191–1198.

[48] S.B. Ghorishi, C.F. Singer, W.S. Jozewicz, C.B. Sedman, R.K. Srivastava, Simultaneouscontrol of Hg0, SO2, and NOx by novel oxidized calcium-based sorbents, J. AirWasteManage. Assoc. 52 (2002) 273–278.

[49] S.J. Wu, M. Ozaki, M.A. Uddin, E. Sasaoka, Development of iron-based sorbents forHg0 removal from coal derived fuel gas: effect of hydrogen chloride, Fuel 87(2008) 467–474.