Elemental mercury oxidation in an electrostatic precipitator enhanced with in situ soft X-ray irradiation

Abstract

Corona discharge based techniques are promising approaches for oxidizing elemental mercury (Hg⁰) in flue gas from coal combustion. In this study, in-situ soft X-rays were coupled to a DC (direct current) corona-based electrostatic precipitator (ESP). The soft X-rays significantly enhanced Hg⁰ oxidation, due to generation of electrons from photoionization of gas molecules and the ESP electrodes. This coupling technique worked better in the positive corona discharge mode because more electrons were in the high energy region near the electrode. Detailed mechanisms of Hg⁰ oxidation are proposed and discussed based on ozone generation measurements and Hg⁰ oxidation behavior observations in single gas environments (O₂, N₂, and CO₂). The effect of O₂ concentration in flue gas, as well as the effects of particles (SiO₂, TiO₂, and KI) was also evaluated. In addition, the performance of a soft X-rays coupled ESP in Hg⁰ oxidations was investigated in a lab-scale coal combustion system. With the ESP voltage at +10 kV, soft X-ray enhancement, and KI addition, mercury oxidation was maximized.

Implications: Mercury is a significant-impact atmospheric pollutant due to its toxicity. Coal-fired power plants are the primary emission sources of anthropogenic releases of mercury; hence, mercury emission control from coal-fired power plant is important. This study provides an alternative mercury control technology for coal-fired power plants. The proposed electrostatic precipitator with in situ soft X-rays has high efficiency on elemental mercury conversion. Effects of flue gas conditions (gas compositions, particles, etc.) on performance of this technology were also evaluated, which provided guidance on the application of the technology for coal-fired power plant mercury control.

Introduction

Mercury (Hg) pollution is arousing increasing attention due to its toxicity. Coal-fired power plants are the primary emission sources of anthropogenic releases of mercury and accounted for 53% of the total mercury emission in United States in 2005 (U.S. Environmental Protection Agency [EPA], 2011). In 2011, the EPA finalized the first U.S. national standard to regulate mercury emission from coal-fired power plants.

In flue gas from coal-fired power plants, mercury is usually present in ionic, particulate, and elemental states (Korpiel and Vidic, 1997; Fujiwara et al., 2002). Due to its high solubility in water, the ionic state of mercury can be removed by treatments such as wet scrubbers. Particulate mercury can be removed by particle treatment facilities such as electrostatic precipitators (ESPs) or a fabric filter baghouse. However, elemental mercury (Hg⁰) that is present as vapor is hard to capture.

Several types of approaches have been proposed to treat Hg⁰ in flue gas, including activated carbon injection (ACI) (Sjostrom et al., 2010), noncarbon sorbent injection (Worathanakul et al., 2008; Suriyawong et al., 2009), addition of oxidizing chemicals, such as chlorine (Richardson et al., 2003), and selective catalytic reaction (SCR) (Gao et al., 2013). Among them, ACI and SCR are already commercially available. ACI is the most commonly used technique, but it has limitations, such as high cost and a low applicable temperature range. The activated carbon also affects the reuse of the fly ash. The oxidation effect of SCR was quite variable and appears to be coal specific and catalyst specific (Feeley et al., 2003).

Corona discharge has been proven to oxidize Hg⁰ (Urabe et al., 1988), and its operation cost is much lower than that of ACI. There are three types of corona discharge: alternating current (AC), direct current (DC), and pulsed. Dielectric barrier discharge (DBD), a type of AC corona discharge, has been well studied as a Hg⁰ oxidation technique (Chen et al., 2006; Jeong and Jurng, 2007; Byun et al., 2008, 2011a, 2011b; Ko et al., 2008c; Lin et al., 2010; Wang et al., 2010; Jiang et al., 2012). Similarly, DC corona discharge (Liang et al., 2002; Wang et al., 2009, 2011a, 2013a) and pulsed corona discharge (Liang et al., 1998; Ko et al., 2008a, 2008b; Xu et al., 2009) have also been used to oxidize Hg⁰. Notably, corona discharge has additional beneficial effects other than Hg⁰ oxidation. DBD, pulsed corona discharge, and the system based on DC corona discharge by Wang et al. (2013a) can also oxidize nitrogen oxides (NO_x) and sulfur oxides (SO_x). The corona-based ESP developed by Liang et al. (2002) can remove both Hg⁰ and zinc-rich fly ash particles.

Although the process of Hg⁰ oxidation by corona discharge has been studied for years, its role in Hg⁰ oxidation by DC-energized ESPs in practical flue gas treatment is not clear. Hg⁰ could be captured and partially oxidized in ESPs at coal-fired power plants. In pilot-scale and full-scale ESPs, 8–58% Hg⁰ concentration reductions have been reported (Xie et al., 2002; Wang et al., 2008; Hu et al., 2009; Yang et al., 2013). However, the Hg⁰ oxidation in such processes was attributed to oxidizing compounds in the flue gas, such as hydrogen chloride (HCl). The effect of corona discharge was not investigated.

Soft X-ray radiation has been coupled with ESP to enhance its ability to collect particulate matter (Kulkarni et al., 2002; Hogan et al., 2004; Kettleson et al., 2009, 2013; Li et al., 2009b). Collection efficiency is enhanced when X-rays interact with particles, electrodes, and gas molecules to emit electrons, thereby generating positively charged particles and ions, which then increase particle charging. Considering this, soft X-ray radiation may also be able to enhance Hg⁰ oxidation in ESPs, based on the higher ionization environment in ESPs and also the potential interactions between soft X-rays and Hg atoms.

This study first evaluated the effect of a DC corona discharge–based ESP with in situ soft X-ray irradiation on mercury oxidation, considering both DC corona discharge alone and the combined effect of corona discharge and soft X-rays. In addition, the effects of flue gas composition and representative fly ash and sorbent particles in flue gas were evaluated in synthetic flue gas. Finally, mercury oxidation by ESP and soft X-rays was tested in a flue gas from coal combustion in a drop-tube furnace.

Experimental

Experimental setup

A cylinder-wire ESP was used in this study. The configuration of the ESP and the coupling of the ESP and the photoionizer are shown in Figure S1 of Supplemental Material. A DC power supply (Glassman High Voltage Inc., series EL; High Bridge, NJ) provided the voltage. The current through the ESP was monitored with a microampmeter (Nishizawa, U-120, class 1.5; Sakaki Sakakimachi, Japan). The soft X-rays were generated by a photoionizer (Hamamatsu Photonics Ltd., Iwata City, Japan; model C7114-01; 3.5–9.5 keV).

Two experimental systems were used. In the simulated flue gas system (Figure 1a), flue gas was obtained by mixing different types of gases (O₂, CO₂, and N₂). Their flow rates were regulated by mass flow controllers (Omega, Stamford, CT; FMA 5510, FMA 5512). The total flow rate was kept at 1 liters per minute (lpm). Mercury was generated at 12 ng/min by a mercury permeation tube (Dynacal; Valco Instruments Co. Inc. and VICI AG, Houston, TX, USA). The effects of silicon dioxide (SiO₂; Snowtex-o; Nissan Chemical, Houston, TX), titanium dioxide (TiO₂; Aerodisp-W740X; Evonic Industries, Essen, Germany), and potassium iodide (KI; Mallinckrodt Chemicals, St. Louis, MO) particles were studied. Particles were generated by an atomizer (TSI model 3076; Shoreview, MN). Gases and particles were mixed in a clean quartz tube (61.0 cm length, 1.9 cm inner diameter). The ESP was set downstream of the tube. Hg⁰ concentration and submicron particle size distribution were measured by a mercury analyzer (Nippon Instruments Corporation, Tokyo, Japan; WA-4) and scanning mobility particle sizer (SMPS; TSI, model 3080; Shoreview, MN), respectively. The ozone concentration was measured by an ozone analyzer (Thermo, model 49i; Waltham, MA). An EPA method using a sampling train was used to measure both Hg⁰ and Hg²⁺ concentrations (Hedrick et al., 2001).

Figure 1. (a) Simulated flue gas experimental system (for experiment sets I to III). (b) Drop-tube furnace experimental system (for experiment set IV).

In the second system (Figure 1b), coal was combusted in a drop-tube furnace (Lindberg/Blue; ThermoElectronCorp., Madison, WI, USA) (Zhuang and Biswas, 2001; Suriyawong et al., 2006; Wang et al., 2013b). Pulverized PRB (Powder River Basin) sub-bituminous coal (supplied by Ameren UE, St. Louis, MO) was used in the experiment without KI addition, whereas Chinese S03 coal was used in the KI addition case. The composition of the coals used is given in Table S1. They were fed by a coal feeder (Quann et al., 1982) at 1 g/hr. The coal was carried by a 0.5 lpm airflow, and another 2.5 lpm airflow was directly injected into the ceramic tube. The ESP was again set downstream of the furnace, and the same measurements were taken as in the simulated flue gas system. A cascade impactor (Mark III; Pollution Control Systems Corp., Seattle, WA) was set before the SMPS to remove particles larger than 1 μm.

Experimental design

The experimental plan is summarized in Table 1. The effects of corona discharge and X-rays on elemental mercury oxidation were first evaluated (in set I). Then, the effects of two factors on this process were evaluated, namely, (1) the flue gas composition effect (pure O₂, CO₂, and N₂, in set II) and (2) the particle effect (in set III). Finally, in set IV the effects of corona discharge and X-rays on elemental mercury oxidation were tested on practical flue gas from coal combustion.

Table 1. Test plan

Set	Purpose	Particle	ESP Operation	Gas Composition
I	Test effects of corona discharge and soft X-rays	No particles	Corona −10, −8, +8, and +10 kV, soft X-rays off;	4% O2–12% CO2–84% N2
			Corona −10, −8, 0, +8, and +10 kV, soft X-rays on
II	Test effects of O2, CO2, and N2 in the flue gas	No particles	Corona +10 kV, soft X-rays off and on	100% CO2, 100% N2, and 100% O2, respectively
III	Test effect of particles in flue gas	SiO2 particles; TiO2 particles; KI particles	Corona +10 kV, soft X-rays off and on	4% O2–12% CO2–84% N2
IV	Test mercury oxidation by ESP and soft X-rays in flue gas from coal combustion	Fly ash particles from drop-tube furnace; particles removed by HEPA filter; fly ash particles from drop-tube furnace with KI addition	Corona +10 kV, soft X-rays off and on	Gas from drop-tube furnace

Results and Discussion

Effects of corona discharge and soft X-rays

Based on previous studies (Kulkarni et al., 2002; Namiki et al., 2005; Suriyawong et al., 2008) and industrial ESP operation (Parker, 2003), −10, −8, 0, +8, and +10 kV were selected as the test conditions. Hg⁰ conversion was used as the indicator of mercury oxidation. The conversion efficiency was calculated using the following equation:

(1)

Figure 2 shows that at +10 kV, conversion efficiencies of 49.0% without soft X-rays and 65.6% with soft X-rays were reached, whereas at −10 kV, the conversion efficiencies were lower, 24.1% and 31.5% for ESP and ESP with soft X-rays, respectively. In contrast to the +10 kV and −10 kV cases, the conversion was not significant (below 4.8%) under other applied voltages, due to low electron concentrations in the ESP, which can be inferred from current-voltage characteristics (Figure S2). The phenomenon of higher Hg⁰ conversion at +10 kV than that at −10 kV could be explained by the electron density distribution, which will be modeled in a following section.

Figure 2. Elemental mercury conversion efficiencies under various ESP voltages, with soft X-ray corona (SXC) off and on (gas combination: 0.04 lpm O₂, 0.12 lpm CO₂, 0.84 lpm N₂).

Possible mercury oxidation mechanisms

In the soft X-ray–coupled ESP, a possible Hg⁰ oxidation mechanism can be summarized in four steps. In step 1, energetic photons from corona discharges and soft X-rays (if applied) ionize gas molecules or electrodes in the ESP and form positive ion-electron pairs (Goldman et al., 1985; Chen, 2002; Kulkarni et al., 2002). In step 2, the newly generated high-energy electrons collide with gas molecules and the electrodes, which may result in four outcomes: (a) a gas molecule absorbs energy from the collision and dissociates into a positive ion and a new electron; (b) a new electron is released from the electrode after the collision between an electron and the electrode; (c) the gas molecule absorbs energy from the collision and dissociates into atoms; and (d) the gas molecule absorbs energy from the collision and transforms to the excited state. As a result of processes a and b, a great number of electrons are generated, and they initiate more collisions. The processes occurring in steps 1 and 2 are summarized in Figure 3. The positive-voltage case (Figure 3a) and negative-voltage case (Figure 3b) have two major differences: the drifting directions of positive ions are opposite between the two cases, so are the drifting direction of electrons; and the process 2b just happens in the negative-voltage case. In step 3, nonelectron species generated in step 2 take part in further chemical reactions. The detailed reactions and produced species depend on the gas composition in the ESP will be discussed in later sections. In step 4, the oxidizing radicals generated in steps 2 and 3 react with and oxidize Hg⁰. For example, oxygen atoms (O) and ozone molecules (O₃) oxidize Hg⁰ in the following pathways (Van Veldhuizen, 2000; Wang et al., 2009):

(2)

(3)

Figure 3. Schematic of ion and electron generation and movement in ESP coupled with soft X-rays (the serial numbers correspond to the steps discussed in the text). (a) Positive-voltage case; (b) negative-voltage case.

Production of Hg²⁺ was confirmed with mercury speciation measurement (Figure S3).

Applying soft X-rays resulted in more electron production in step 1, which promoted the electron impact reactions and chemical reactions in steps 2 and 3. Consequently, more radicals were formed and resulted in a high Hg⁰ oxidation rate. However, under soft X-rays alone, only an insignificant amount of Hg⁰ was oxidized, as shown in Figure 2. There are two possible reasons. First, without an electric field in the ESP, the energy of electrons generated by photoionization due to X-rays was too low to initiate the chain reactions and finally could not oxidize Hg⁰. Second, when the electric field was off, there was no force driving electrons to move. Therefore, the possibility of collisions between electrons and gas molecules was much lower than when the ESP was on, which resulted in fewer reactions. For these reasons, soft X-rays alone were not used as a test condition in the experiments discussed in the following sections.

The O₃ generated by an ESP is an active chemical species that oxidizes mercury. O₃ generations under various voltages with soft X-rays on or off are shown in Figure 4. In the voltage ranging from −8.8 to +10 kV, under the same voltage, the concentration of O₃ was slightly higher when the soft X-rays were on than when it was off. However, when the voltage was between −10 and −8.8 kV, X-rays reduced O₃ generation. Comparing the O₃ generation with the Hg⁰ conversion result in Figure 2, a higher O₃ concentration was obtained under −10 kV than under +10 kV; however, Hg⁰ conversion was higher under +10 kV. Therefore, O₃ might not be the most important species in oxidizing Hg⁰. This conclusion matches the theoretical calculation. The reduction of Hg⁰ concentration due to reactions with O₃ in the ESP was calculated using

(4)

Figure 4. Ozone generation from ESP under base condition (0.04 lpm O₂, 0.12 lpm CO₂, 0.84 lpm N₂).

Kinetic parameters for this reaction have been reported in the literature (k is 2.1 × 10⁻¹⁸ cm³ molecule⁻¹ sec⁻¹, a is 1.0, and b is 0.8 [Hall, 1995]; k is 7.5 × 10⁻¹⁹ cm³ molecule⁻¹ sec⁻¹, a is 1.0, and b is 1.0 [Pal and Ariya, 2004]). With these kinetic parameters, the estimated Hg⁰ conversion efficiency due to reactions with O₃ is at least 1 order of magnitude lower than the Hg⁰ conversion efficiencies for the experimental cases. Thus, it can be inferred that the reaction between Hg⁰ and O₃ might not be the key mechanism of Hg⁰ conversion, and that O radicals contribute most in Hg⁰ oxidation in the system.

The above-mentioned results showed that a positive ESP voltage created better Hg⁰ oxidation than a negative one. Hence, a voltage of +10 kV was applied to the ESP in subsequent experiments.

Effect of flue gas composition

To further elucidate the mechanism, especially how gas composition determines radical formation (step 3 in the mechanism), Hg⁰ conversion was measured under pure O₂, CO₂, and N₂, respectively. The Hg⁰ conversions and current measurements are shown in Figure 5, with simulated flue gas scenarios as reference, and O₃ generations in those cases are shown in Figure S4.

Figure 5. Mercury conversion efficiencies by ESP and soft X-rays under single-gas environments (total flow rate: 1 lpm; ESP currents for various cases are above the bars; gas combination: 4% O₂, 12% CO₂, 84% N₂).

Under pure O₂, O and O₃ radicals were formed in the following reactions (Chen and Davidson, 2002):

(5)

(6)

where M is a gas molecule. ESP currents and Hg⁰ conversions were both high under ESP alone and ESP with soft X-rays. One noteworthy phenomenon is that the current was lower when the soft X-rays were on. The reason might be that electrons produced by corona discharge reached an extremely high concentration in the ionization area due to high concentration of O₂ and the drifting effect shown in Figure 3a, and the bipolar charging due to soft X-ray photoionization actually neutralized the electrons. However, the Hg⁰ conversions were at the same level in both cases. Perhaps the radical concentrations were high enough that they were no longer limiting factors of Hg⁰ oxidation reactions in both cases.

When N₂ was used, soft X-ray and ultraviolet (UV) radiations as well as high-energy electrons were present in the system, but there was no source for O radicals. In this case, Hg⁰ conversion was below 5%. These results verify that radicals generated from oxygen are the main cause of Hg⁰ oxidation rather than direct reactions between Hg⁰ molecule and high-energy radiations or electrons. The slight Hg⁰ conversion may be due to ionization of excited Hg atoms in the presence of N₂ (Sibata et al., 1979).

Under pure CO₂, no Hg⁰ was oxidized by ESP alone, and no current was detected. This result agrees with the previous understanding that CO₂ plays a role in stabilizing the discharge process (Wang et al., 2009). However, when soft X-rays were added, a current of 41.5 μA and an Hg⁰ conversion of 49.1% were reached. The main reaction involved is as follows (Skalny et al., 2008a, 2008b):

(7)

This reaction leads to generation of O radicals. However, the extent of reaction was extremely low under corona discharge alone, because the electron impact dissociation rate of CO₂ is 4–5 orders of magnitude lower than that of O₂ (Li et al., 1995; Wang et al., 2011a). On the other hand, soft X-ray ionization can generate electrons, which promote the chain reactions and eventually result in a high concentration of radicals. ESP current measurements support this mechanism. No current was observed under ESP alone. However, when soft X-rays were on, a current of 41.5 μA was observed. The Hg⁰ conversion result indicates that soft X-ray enhancement has great potential in Hg⁰ oxidation by ESP when high concentration of CO₂ is present in flue gas, such as in the oxy-combustion case.

The effect of O₂ concentration on Hg⁰ oxidation by ESP and soft X-rays was also investigated in simulated flue gas (Figure S5). When ESP alone was used, the mercury conversion efficiency significantly increased as the oxygen concentration increased, because O₂ was the major source for generation of chemically active species. However, when ESP and soft X-rays were both applied, the Hg⁰ conversion was relatively stable, because CO₂ was also an important source for generation of chemically active species in this condition. The O₃ concentrations under various O₂ concentrations are shown in Figure S6.

Modeling of electron density in an ESP

Based on the above-discussed mechanisms and reaction pathways, a model for predicting electron density distribution was built. The simulation results can explain why the ESP coupled with soft X-rays oxidized Hg⁰ better under positive voltages than negative ones. Considering the configuration of the ESP, a one-dimensional axisymmetric model was used to represent a cross-section of the ESP. Its schematic diagram is shown in Figure S7.

Shown in eqs 8–10, the drift-diffusion equation, multicomponent diffusion equations, and Gauss Law are the governing equations for electron density, nonelectron species densities, and the electric field.

(8)

(9)

(10)

The generation rates of species were calculated based on 12 electron impact reactions and 11 chemical reactions (Table S2). The ionization rate due to soft X-ray radiation was estimated based on data from Kulkarni et al. (2002).

Electron density distributions under +10 and −10 kV with and without soft X-rays are shown in Figure 6. The number of electrons that were close to the electrode, where the electrons had higher energies, was higher for the positive-voltage cases than negative-voltage cases. Electrons having higher energy were more likely to trigger reactions, such as in eqs 5 and 7, through which O radicals were produced. This effect can explain our experimental observation that the Hg⁰ oxidation rate was higher under positive voltage. Meanwhile, with soft X-rays, the number of electrons was higher, which resulted in a higher Hg⁰ oxidation rate.

Figure 6. Electron density distributions in ESP at +10 and −10 kV, with and without soft X-rays.

Effect of particles in flue gas

Flue gas usually contains high concentrations of particulate matter. The particles may participate in and influence the process of Hg⁰ oxidation by ESP and soft X-rays in various ways. The effects of three kinds of particles were studied in this section, including SiO₂, TiO₂, and KI particles. SiO₂, a chemically inert substance, is one of the main components of fly ash particles. TiO₂ is a well-known and widely applied photocatalyst that can be activated under UV light and catalyze many reactions (Wold 1993; Wang et al., 2011b, 2012). It has been reported that oxidation of Hg⁰ in flue gas by oxygen takes place in the presence of nanostructured TiO₂ particles and UV light (Wu et al., 1998). KI particles were reported to have the ability to oxidize Hg⁰ in flue gas (Li et al., 2009a).

Table 2 summarizes Hg⁰ conversions in the ESP with particle injection. The size distributions of the particles are shown in Figure S8. The SiO₂ particles had no significant influence on Hg⁰ conversion. The slight decrease might be due to a small amount of electrons attaching to the particles instead of participating in the chain reactions.

Table 2. Elemental mercury conversion efficiencies under SiO₂, TiO₂, and KI particle injection (gas combination: 0.04 lpm O₂, 0.12 lpm CO₂, 0.84 lpm N₂)

Particle	Conversion When ESP and Soft X-rays Were Both Off	Conversion When Just ESP Was On	Conversion When ESP and Soft X-rays Were Both On
SiO2	0	53.7 ± 1.3%	60.3 ± 1.5%
TiO2	0	14.0 ± 0.8%	17.3 ± 0.1%
KI (high concentration)	92.5 ± 2.2%	87.1 ± 2.8%	90.4 ± 1.1%
KI (low concentration)	19.4 ± 1.3%	66.1 ± 3.2%	70.0 ± 2.5%

The TiO₂ particles significantly reduced Hg⁰ conversion. UV photons from corona discharge and soft X-rays had energies higher than 3.0 eV, which is the band gap of TiO₂. In this case, UV photons and soft X-rays were absorbed by TiO₂ particles, and they excited electrons from TiO₂ particles. The generated electron-hole pairs reacted with electron donors or acceptors adsorbed on the surface of TiO₂ particles (Chen and Mao, 2007; Yates, 2009). Thus, the decrease of Hg⁰ conversion might be because both soft X-rays and UV light from corona discharge were partially absorbed by TiO₂ particles. As a result, the extent of gas molecule ionization was reduced, and eventually fewer radicals were produced. On the other hand, the amount of Hg⁰ that absorbed on the surface of TiO₂ particles was low, so the contribution of the TiO₂ photocatalytic pathway on Hg⁰ oxidation was insignificant. Overall, TiO₂ particle injection in this test reduced the Hg⁰ conversion. However, it has been reported that TiO₂ particles fixed in a DBD reactor can react with corona discharge in the reactor and enhance mercury oxidation (Yang et al., 2012a, 2012b), which is contrary to the results of this study. The explanation is that in the specific experimental conditions of Yang et al. (2012a, 2012b), active sites containing OH functional groups were formed on the surface of TiO₂ under UV radiation, which resulted in Hg⁰ oxidation.

KI particles increased the Hg⁰ conversion by introducing a new pathway for Hg⁰ oxidation (Li et al., 2009a):

(11)

(12)

Considering the corona discharge and soft X-ray radiation environment, the following reactions were likely to form I₂ radicals:

(13)

(14)

It should be noted that most of the KI particles were quickly deposited on the collection electrode of the ESP once they flew in, and the chance that they could participate in the reaction was much lower. This is evidenced by the comparison of Hg⁰ conversions under different operating conditions with a high concentration of KI particles. When the ESP and soft X-rays were off, Hg⁰ conversion of 92.5% was achieved; however, when the ESP was turned on or they were both turned on, the conversion decreased by 5.4% and 2.1%. Another issue is that reaction between KI and Hg⁰ at room temperature was observed in this study. However, in a previous study (Li et al., 2009a), no conversion was observed. The difference may be due to the longer residence time of the flue gas in the system in this study (30.0 sec compared with 5.8 sec).

Mercury oxidation in flue gas from coal combustion

Mercury oxidation by ESP and soft X-rays was tested in flue gas from coal combustion in a drop-tube furnace (Figure 1b). The particle size distribution in the flue gas from PRB coal combustion is shown in Figure 7. The size distribution is typical for coal combustion (Zhuang and Biswas, 2001; Suriyawong et al., 2006; Li et al., 2009b; Wang et al., 2013b). Figure 7 also shows that the ESP can remove at least 99.7% of the fly ash particles. For flue gas with particles, no clear Hg⁰ conversions were observed (conversion efficiencies were 1.1 ± 0.1% with ESP alone, and 1.5 ± 0.2% with both ESP and soft X-rays).

Figure 7. Particle size distributions of PRB coal combustion downstream of the ESP.

Comparing the flue gas from drop-tube furnace with the simulated gas used in experiment set I, the composition of the former is more complicated and contains water vapor, SO₂, HCl, and NO_x. Water vapor can enhance Hg⁰ oxidation due to ∙OH radical generation under corona discharge (Ko et al., 2008a; Wang et al., 2009; Yang et al., 2012b). SO₂ and HCl also can enhance Hg⁰ conversion (Figures S10 and S11). NO interferes Hg⁰ oxidation because it competes with Hg⁰ on reacting with oxidizing radicals (Jeong and Jurng, 2007; Ko et al., 2008a, 2008c; Wang et al., 2009).

Based on the above discussion and the results presented in the last section, two factors, NO and particles in the flue gas, are possible to interfere Hg⁰ oxidation. However, Hg⁰ conversions in particle-free flue gas (particles were filtered out by a HEPA [high-efficiency particulate arrestance] filter before flying into the ESP) were still extremely low (conversion efficiencies were 0 with ESP alone, and 3.9 ± 0.4% with both ESP and soft X-rays). Therefore, particles were not the main reason for the conversion reduction. NO in the flue gas was the cause of the conversion reduction.

Figure 8 shows the conversions in the cases where 200 ppm KI was mixed into the coal before combustion. Higher conversions were observed when the ESP and soft X-rays were turned on, which evidences that Hg⁰ can be oxidized by ESP and soft X-rays in this condition. If the Hg⁰ concentration in the flue gas with ESP and soft X-rays off was used as baseline, the conversions due to ESP and to ESP with soft X-rays were 21.5% and 27.7%, respectively, which are 43.9% and 42.3% of the values for synthetic flue gas. One explanation might be that with I₂ in flue gas working as an oxidizer in the ESP, the radicals generated by the ESP and soft X-rays were in excess after the preferred reaction with NO, and they oxidized the Hg⁰.

Figure 8. Elemental mercury conversion efficiencies by ESP and soft X-rays in practical flue gas from coal combustion, with KI addition (200 ppm KI in Chinese S03 coal).

Conclusion

The process of Hg⁰ oxidation by ESP and soft X-rays was investigated. Studies were first conducted in synthetic flue gas. Hg⁰ was oxidized under +10 kV and −10 kV voltages of the ESP, and +10 kV had the better effect. Soft X-rays enhanced the oxidation. A mechanism for Hg⁰ oxidation was proposed: electrons were emitted due to photoionization caused by UV radiation from corona discharge and by soft X-ray radiation in the ESP, and then they initiated chain reactions that produced electrons, intermediate species, and other products. Radicals, such as O and O₃, were produced in the chain reactions, and they oxidized Hg⁰. A model simulating the plasma processes was built for predicting the electron distribution in the ESP.

Oxidation mechanisms were further demonstrated by experiments conducted under single-gas environments. Especially, in pure CO₂, soft X-ray photoionization showed significant enhancement on Hg⁰ oxidation. SiO₂ particles did not have a significant effect on Hg⁰ oxidation process; TiO₂ particles absorbed photons from corona discharge and soft X-rays and inhibited the Hg⁰ oxidation; KI particles enhanced Hg⁰ oxidation, because I⁻ could be oxidized into I₂ radicals.

In practical flue gas from coal combustion, the effects of ESP and soft X-rays on Hg⁰ oxidation were not significant, which might be due to interference from NO. With KI addition to the coal, beneficial effects of ESP and soft X-rays on Hg⁰ oxidation could be observed. With the ESP voltage at +10 kV, soft X-ray enhancement, and KI addition, mercury oxidation was maximized.

Nomenclature

α =	=	order of reaction between elemental mercury and ozone with respect to ozone
b =	=	order of reaction between elemental mercury and ozone with respect to elemental mercury
c =	=	concentration
=	=	average concentration in the ESP
∆c =	=	change of concentration
=	=	diffusivity
=	=	intensity of electric field
=	=	diffusive flux vector
k =	=	reaction rate constant
n =	=	number concentration
R =	=	generation rate
s =	=	space charge density
t =	=	time
=	=	mass averaged fluid velocity vector
w =	=	mass fraction
=	=	permittivity
=	=	electric mobility
ρ =	=	density
Subscripts	=
e =	=	electron
Hg0–O3 =	=	reaction between elemental mercury and ozone
i =	=	species i
off =	=	ESP is off
on =	=	ESP is on
0 =	=	reference index (represent vacuum in eq 10)

Acknowledgment

This work was supported by the Consortium for Clean Coal Utilization. The authors thank Dr. Brent Williams and Dhruv Mitroo for helping measure the ozone concentration, and Dr. James Ballard for careful reading and editing of the manuscript.

Supplemental Material

Supplemental data for this article can be accessed at http://dx.doi.org/10.1080/10962247.2014.998352.

Additional information

Notes on contributors

He Jing

He Jing and Xiaofei Wang are doctoral students, Wei-Ning Wang is a research assistant professor, and Pratim Biswas is the department chair and Lucy & Stanley Lopata Professor at the Department of Energy, Environmental and Chemical Engineering, Washington University in St. Louis, St. Louis, MO, USA.