Gas-phase elemental mercury removal in a simulated combustion flue gas using TiO₂ with fluorescent light

Abstract

A previously proposed technology incorporating TiO₂ into common household fluorescent lighting was further tested for its Hg⁰ removal capability in a simulated flue-gas system. The flue gas is simulated by the addition of O₂, SO₂, HCl, NO, H₂O, and Hg⁰, which are frequently found in combustion facilities such as waste incinerators and coal-fired power plants. In the O₂ + N₂ + Hg⁰ environment, a Hg⁰ removal efficiency (η_Hg) greater than 95% was achieved. Despite the tendency for η_Hg to decrease with increasing SO₂ and HCl, no significant drop was observed at the tested level (SO₂: 5–300 ppm_v, HCl: 30–120 ppm_v). In terms of NO and moisture, a significant negative effect on η_Hg was observed for both factors. NO eliminated the OH radical on the TiO₂ surface, whereas water vapor caused either the occupation of active sites available to Hg⁰ or the reduction of Hg₀ by free electron. However, the negative effect of NO was minimized (η_Hg > 90%) by increasing the residence time in the photochemical reactor. The moisture effect can be avoided by installing a water trap before the flue gas enters the Hg⁰ removal system.

Implications:

This paper reports a novel technology for a removal of gas-phase elemental mercury (Hg⁰) from a simulated flue gas using TiO₂-coated glass beads under a low-cost, easily maintainable household fluorescent light instead of ultraviolet (UV) light. In this study, the effects of individual chemical species (O₂, SO₂, HCl, NO, and water vapor) on the performance of the proposed technology for Hg⁰ removal are investigated. The result suggests that the proposed technology can be highly effective, even in real combustion environments such as waste incinerators and coal-fired power plants.

Introduction

Mercury (Hg) is a toxic heavy metal that is released into the environment at a global annual emission rate of 1930 tons from anthropogenic sources such as waste incinerators (municipal, medical) and coal-fired power plants (United Nations Environmental Programme, 2008a). Hg emitted from these sources is deposited in seas, rivers, and soil, mostly by rain or snow. Hg contamination in aquatic species and food leads to the accumulation of Hg and causes biomagnification through the food chain. This accumulation has a significant effect on human health (Munthe et al., 2007; Sundseth et al., 2010; Arctic Monitoring and Assessment Programme, 2005). Hg pollution has been discussed by the United Nations Environmental Programme (UNEP) Governing Council since 2002. The UNEP mercury program established an ad hoc open-ended working group to discuss appropriate measures and international treaties for the prevention of mercury pollution (United Nations Environmental Programme, 2008b). As a result, the development of a legally binding global treaty on mercury is under way, and the treaty is expected to be signed in 2013 in Minamata, Japan.

Hg in a flue gas exists in three forms: elemental mercury (Hg⁰), oxidized mercury (Hg²⁺), and particle-bound mercury (Hg^p). Each form of Hg has different characteristics, and numerous studies have reported the removal of individual Hg species by conventional air pollution control devices (APCDs) installed at combustion facilities. Hg^p can be removed by a fabric filter (FF) or an electrostatic precipitator (ESP) with an efficiency of 27–58% depending on the operating conditions (Choi et al., 2009). Hg²⁺ is soluble in aqueous solution and has good reactivity with particulate matter. Thus, wet flue-gas desulfurization (WFGD) can remove Hg²⁺ with efficiencies in the range of 53–87% (U.S. Department of Energy, 2006). Meanwhile, Hg⁰ accounts for more than 35% of the total Hg in flue gas in most combustion facilities and is very difficult to remove. APCDs are generally ineffective in removing Hg⁰ because Hg⁰ is insoluble in water and has a poor reactivity with particulate matter.

Activated-carbon injection (ACI) technology is currently the most widely used technology and can achieve a Hg removal efficiency greater 90%. Activated carbons (virgin or impregnated with one of the following: sulfur, iodine, chlorine, or bromine) are injected into the flue gas immediately before the gas enters the ESPs or baghouse filters. However, the application and effectiveness of ACI technology is limited by the types of APCDs installed in exhaust facilities. In addition, the disposal procedure used for adsorbents containing Hg is complicated and expensive.

In a previous study, we reported the removal of Hg⁰ in a system containing only Hg and air using a TiO₂ catalyst and a common household fluorescent light (Lee et al., 2004):

(1)

During the process, HgO was formed on the activated TiO₂ surface through the photo-oxidization of Hg⁰ by the hydroxyl (OH) radical in an O₂ atmosphere (Lee et al., 2001; Li and Wu, 2006; Li and Wu, 2007; Pitoniak et al., 2003, 2005; Snider and Ariya, 2010; Wu et al., 1998).

In this study, the previously proposed technology was further tested for its Hg⁰ capture capability in a simulated flue-gas system. The flue gas was simulated by the addition of the chemical constituents (O₂, SO₂, HCl, NO, H₂O, and Hg⁰) frequently found in combustion facilities at the concentrations based on both the Korean Emission Standards for Hazardous Air Pollutants (The Korean Ministry of Environment, 2011) and actual reported data from flue gas analyzed in real time by the TMS (Tele Metering System) installed at waste incinerators in the Republic of Korea.

Experimental

TiO₂-coated glass beads

The photocatalytic adsorbents were prepared by coating the glass beads with anatase-TiO₂ nanoparticles (NanoPac Ltd., Suwon, Korea). Clear glass beads were used to allow fluorescent light to penetrate further toward the center of the tubular photochemical-reaction cell reactor. First, glass beads 3 mm in diameter were washed clean and placed on a sieve. The glass beads on the sieve were then dipped into a TiO₂ solution while being sonicated. The beads were then dried for 1 hr at 20 °C followed by 1 hr at 80 °C in an oven. This process was repeated three times. The amount of TiO₂ coated on the beads was calculated by measuring the total weight of the beads before and after the coating. Approximately 0.4 mg of TiO₂ was coated per 1 g of photocatalyst (TiO₂-coated glass beads).

The crystallinity of the anatase TiO₂ used in this work was confirmed by powder X-ray diffraction (XRD; D/max 2200V/PC; Rigaku Co., Tokyo, Japan). The measured crystallite size was 5–10 nm. The average particle size measured using a transmission electron microscope equipped with an energy dispersive X-ray analyzer (TEM-EDX; JEM-1210; JEOL Ltd., Tokyo, Japan) was 60 nm, whereas the specific surface area measured by the Brunauer-Emmett-Teller (BET) method (ASAP2020; Micromeritics Instrument Corp., Norcross, MN, USA) was approximately 275 m²/g.

Photocatalytic reactor

A schematic diagram of the experimental setup is shown in Figure 1. A photochemical-reaction cell made of quartz was placed inside a stainless steel housing equipped with four common household fluorescent lamps (4 × 20 W, 220V, model E20EX-D; Dooyoung Lighting Co., Seoul, Korea). The wavelength spectrum was also measured (ILT900; International Lighting Technologies Inc., Peabody, MA, USA). A major peak was observed between 400 and 800 nm and a small but noticeable peak appeared between 360 and 370 nm. All of the household fluorescent lamps commercially available in Korea exhibited a minor peak in the ultraviolet A (UVA) range (360–370 nm). The cell was 65 cm in height and 2.1 cm in diameter, and was located at an equal distance of 7 cm from each of the four fluorescent lamps. A cooling fan was installed to prevent the temperature from rising inside the housing. A gas mixer was used to blend the individual input gases, and all of the gas lines were covered with heating jackets capable of controlling the temperature from 30 to 300 °C.

Figure 1. A schematic diagram of the experimental system.

Simulated flue gas

In this work, a combustion flue gas was simulated by a particle-free, dry gas containing O₂, SO₂, HCl, NO, Hg⁰, and water vapor with a balance gas, N₂. Except for the water vapor, all of the gases were introduced into the system at a precisely controlled flow rate using mass-flow controllers. The concentrations of the introduced gases were determined based on both the Korean Emission Standards for Hazardous Air Pollutants (The Korean Ministry of Environment, 2011) and reported analysis data from waste incinerator flue gas using TMS in the Republic of Korea. Elemental mercury vapor was introduced into the system by passing N₂ gas at a precisely controlled flow rate above liquid mercury contained in a temperature-controlled gas-washing bottle (10 °C). The amount of water vapor in the inlet gas stream was controlled by adjusting the amount of water injection from the syringe pump. The gas lines and photocatalytic reactor were maintained at a temperature of 50 °C to prevent water vapor from condensing and saturating the TiO₂ surface. Finally, a water trap was installed ahead of the on-line Hg⁰ analyzer. The volume of the water vapor corrected to standard conditions, V _w(std), was calculated using the following equation (ASTM International, 2008):

(2)

where M _w is the molecular weight of water, 18.0 g/g·mol; P _std is the standard absolute pressure, 101.3 kPa; T _std is the standard absolute temperature, 293 K; R is the ideal gas constant, 0.008314 kPa·m³/K·g·mol; V _w(std) is the volume (m³) of water vapor in the gas sample corrected to standard conditions; W _ws is the total weight (g) of water supplied by the syringe pump; and K ₂ is 0.001336 m³/mL.

Then, the moisture content (volume ratio of water vapor to the stack gas), B _ws, was calculated using eq 3, which was in the range of 0.5–5% in this study:

(3)

where V _m(std) is the volume (m³) of the total dry gas corrected to standard conditions.

Hg analysis

The concentrations of gaseous Hg⁰ at the inlet and outlet of the reactor were measured by an on-line Hg⁰ analyzer (VM 3000; Mercury Instruments, Karlsfeld, Germany). This analyzer uses a cold vapor atomic absorption spectrometry (CVAAS) method based on the resonance absorption wavelength of Hg atoms at 253.7 nm (Norton et al., 2002). It provides continuous measurements of Hg⁰ in ambient air and in flue gas. It was periodically calibrated with an Hg⁰ calibrator (Series 3310; Tekran Instruments Corp., Toronto, Canada). The Hg⁰ concentration in the gas stream passing through the reactor with no light source was set as the baseline. The Hg⁰ removal efficiency, η_Hg, was calculated using the following equation:

(4)

where Hg⁰ _in is the Hg⁰ concentration when the fluorescent light was off and Hg⁰ _out is the Hg⁰ concentration when the photocatalytic reaction was allowed to occur in the presence of fluorescent light.

Results and Discussion

To test the influence of SO₂, HCl, NO, and water vapor on the Hg⁰ removal by TiO₂-coated beads with fluorescent light, comparative experiments were performed with each chemical constituent at various concentrations. Twenty-five grams of the TiO₂-coated beads were placed in a photocatalytic reactor with a total input-gas flow rate of 3 L/min, resulting in a residence time of 0.16 sec. The base gas was composed of O₂, Hg⁰, and N₂ (for Hg carrying and input-gas balancing) producing an inlet Hg concentration of 50–60 μg/m³.

Effect of O₂

First, the effect of O₂ on Hg⁰ removal by TiO₂-coated beads with fluorescent light was investigated. The gas tested was composed of Hg⁰, N₂, and O₂ (14%). As shown in Figure 2, for the first 30 min without any light source, no Hg⁰ removal was observed. Then, the fluorescent light was turned on for 300 hr, and the output Hg⁰ concentration was measured. A η_Hg value greater than 90% was achieved instantly once the light was switched on. After 100 min, η_Hg reached 99%. A η_Hg value greater than 95% was maintained for over 300 hr. The presence of O₂ indicated no negative effect on the Hg⁰ removal by TiO₂-coated beads with fluorescent light.

Figure 2. Hg⁰ removal efficiency in the presence of O₂, Hg⁰, and N₂.

Effect of SO₂

Next, SO₂ was added to the base gas while maintaining the total flow rate at 3 L/min. The concentration of SO₂ was precisely controlled at 50, 100, 200, and 300 ppm_v. It should be noted that SO₂ interferes with the Hg analyzer measurement by absorbing light with a wavelength in the range of 250–370 nm (Vandaele et al., 1994), which overlaps with the light absorption wavelength of Hg (253.7 nm). To account for the overrated on-line Hg measurement caused by SO₂, a blank test using a gas sample with SO₂ (50, 100, 200, and 300 ppm_v) balanced with N₂ (no Hg⁰) was carried out first. The blank test results were taken into account when calculating η_Hg in the presence of SO₂.

After the inlet Hg⁰ concentration reading (50–60 μg/m³) was stabilized for 30 min, the fluorescent light was turned on for 2 hr. Figure 3 shows the Hg⁰ removed by TiO₂-coated beads with fluorescent light in the presence of SO₂. For all of the tested SO₂ concentrations, η_Hg greater than 90% was observed. At SO₂ concentrations below 100 ppm_v, η_Hg greater than 95% was achieved. Despite the tendency for η_Hg to decrease with increasing SO₂, no significant drop was observed at the tested level of SO₂ (50–300 ppm_v).

Figure 3. Effect of SO₂ on the Hg⁰ removal efficiency.

Effect of HCl

Figure 4 shows the Hg⁰ removed by TiO₂-coated beads with fluorescent light in the presence of HCl. For all of the tested HCl concentrations (C _HCl = 30, 60, 90, and 120 ppm_v), η_Hg greater than 90% was observed. A η_Hg value greater than 99% was observed for C _HCl < 60 ppm_v. As in the case of SO₂, the technology proposed in this work shows highly effective Hg⁰ removal despite the presence of HCl.

Figure 4. Effect of HCl on the Hg⁰ removal efficiency.

Effect of NO

Figure 5 shows the effects of NO (20, 30, 40, and 60 ppm_v) on Hg⁰ removal. At a NO concentration (C _NO) of 20 ppm_v, η_Hg reached 90% after 40 min. At C _NO = 30 ppm_v, the initial Hg⁰ removal rate was very slow compared with that at C _NO = 20 ppm_v, taking more than 2 hr to achieve 90% η_Hg. No Hg⁰ removal was observed at C _NO > 40 ppm_v, even after 2 hr. Unlike SO₂ or HCl, the addition of NO had a significant negative effect on the Hg⁰ removed by TiO₂-coated beads with fluorescent light. As the NO concentration increased, the degree of interference increased until the amount of Hg⁰ removed became essentially zero. Niksa et al. reported that NO was the most effective inhibitor during Hg⁰ oxidation and that this inhibition was caused by the elimination of the OH radical on the TiO₂ surface by the following reaction (Niksa et al., 2001):

Figure 5. Effect of NO on the Hg⁰ removal efficiency.

(5)

M represents any other molecule available to carry off excess energy. The influence of NO was further examined by evaluating η_Hg at input-gas flow rates of 1, 2, and 3 L/min, exhibiting residence times of 0.48, 0.24, and 0.16 sec, respectively. The NO concentration was fixed at 40 ppm_v, a concentration at which no Hg⁰ removal was previously observed for the total input-gas flow rate of 3.0 L/min. The amount of photocatalyst was 25 g, and the inlet Hg⁰ concentration was 50–60 μg/m³. As shown in Figure 6, η_Hg was below 10% for a residence time of 0.16 sec, whereas it exceeded 90% with residence times of 0.24 and 0.48 sec. This result suggests that the negative effect of NO in flue gas might be overcome by increasing the residence time in the photocatalytic reactor.

Figure 6. Hg⁰ removal efficiency at various residence times (C _NO = 40 ppm_v).

Effect of water vapor

Experiments analogous to those discussed in the preceding sections were carried out at various moisture contents. Gas streams with 0.5–5% water vapor were introduced into the system.

The values of η_Hg observed at different moisture contents are shown in Figure 7. At 5% moisture, η_Hg was well below 5%. This was most likely due to the occupation of TiO₂ active sites by the water-vapor molecules. The amount of Hg⁰ removed increased as the water-vapor content was decreased. At 0.5% H₂O, η_Hg instantly reached 40% and gradually increased to 90%.

Figure 7. Effect of moisture on the Hg⁰ removal efficiency.

Another set of experiments was performed to further elucidate the moisture effect. In step I, a gas containing Hg⁰, O₂, N₂, and no water vapor was injected into the system. The light was turned off during step I. In step II, the light was turned on and η_Hg instantly reached close to 100% and stabilized. The photocatalyst (TiO₂) was allowed to uptake Hg⁰ for 4 hr during step II until the injection of Hg⁰ and O₂ was stopped and the light was turned off. Then, in step III, N₂ gas was introduced into the system to remove any Hg⁰ physically attached to the surface of the reactor and TiO₂ (step III). Finally, in step IV, a gas containing 5% water vapor and 14% O₂ balanced with N₂ was passed through the Hg⁰-adsorbed TiO₂ photocatalysts, now with the fluorescent light turned on. The outlet Hg⁰ concentration was continuously monitored at all times throughout the experiments; the results are shown in Figure 8. A significantly high Hg⁰ release was observed as soon as the water vapor was introduced into the system (step IV), gradually decreasing until no Hg⁰ release was detected after 200 min. A similar phenomenon was observed and discussed by Li and co-workers for the case of UV (Li and Wu, 2006; Li et al., 2008). They reported that water vapor in the presence of UV generated free electrons on the surface of TiO₂, causing the reduction of HgO by the following reaction:

Figure 8. Desorption of adsorbed Hg⁰ from TiO₂ surface by water vapor: (a) step I, (b) step II, (c) step III, and (d) step IV.

(6)

The effects of moisture on the Hg⁰ removal in the flue gas can be summarized as follows: first, the water vapor molecules compete with Hg⁰ gas molecules for the available empty active sites on the TiO₂ surface. At the same time, they go after the active sites occupied by HgO and detach the adsorbed Hg⁰ back into the gas stream.

Conclusions

A previously proposed novel gaseous Hg⁰ removal technology was further tested under simulated flue-gas conditions. The effects of the selected combustion flue gas constituents (SO₂, HCl, NO, and moisture) on Hg⁰ removal by TiO₂-coated glass beads under common household fluorescent light were investigated. First, despite the tendency for the Hg⁰ capture efficiency to decrease with increasing SO₂ and/or HCl, the proposed technology was proved to be highly effective under the tested conditions (SO₂: 50–300 ppm_v, HCl: 30–120 ppm_v), simulating those in conventional combustion facilities, specifically, waste incinerators. Both NO and moisture exhibited negative effects on Hg⁰ removal. NO was the most effective inhibitor during Hg⁰ oxidation, eliminating the OH radical from the TiO₂ surface. However, we were able to minimize the negative effects caused by NO by increasing the residence time. The water vapor caused either the occupation of the active sites available for Hg⁰ removal or the reduction of HgO by free electron in the presence of fluorescent light and subsequent detachment from the TiO₂ surface in the form of Hg⁰. In real combustion facilities such as waste incinerators (typically with 10–20% H₂O in the flue gas), the moisture effect can be avoided by installing a water trap for the flue gas before it enters the Hg⁰ removal system. Clearly, the design parameters to maximize the Hg⁰ removal efficiency in real combustion environments deserve further exploration. However, the TiO₂ + fluorescent light system is readily applicable for Hg recovery from used fluorescent light bulbs, which typically involves a moisture-free and acid-free environment.

Acknowledgments

This work was supported by the National Research Foundation of Korea (NRF) grant funded by the Korea government (MEST) (no. 2009-0079977).