ARTICLE pubs.acs.org/EF
Homo- and Heterogeneous Mercury Oxidation in a Bench-Scale Flame-Based Flow Reactor Clara A. Smith,*,† Balaji Krishnakumar,‡ and Joseph J. Helble† † ‡
Thayer School of Engineering, Dartmouth College, Hanover, New Hampshire 03755, United States Department of Chemical, Materials, and Biomolecular Engineering, University of Connecticut, Storrs, Connecticut 06269, United States ABSTRACT: An experimental study of hetero- and homogeneous mercury oxidation chemistry was conducted in a bench-scale flame-based flow reactor with a residence time of 2.9 s. Homogeneous mercury oxidation levels increased with an increasing HCl concentration, with oxidation increasing to 29% at HCl concentrations of 555 parts per million by volume (ppmv). The presence of SO2 alone also led to Hg oxidation, with approximately 20% oxidation observed at SO2 concentrations ranging from 100 to 900 ppmv. When both HCl and SO2 were present, mercury oxidation was enhanced in the presence of SO2 when the concentration of HCl was 200 ppmv and inhibited when concentration of HCl was 555 ppmv. To examine heterogeneous effects, iron oxide or montmorillonite particles were injected into the post-flame gases of the system. Mercury oxidation by HCl was enhanced at high iron oxide particle concentrations (>100 m2α-Fe2O3/m3 of flue gas) compared to the homogeneous system. When iron oxide particles were injected in the presence of 100400 ppmv SO2 without HCl, mercury oxidation was enhanced at particle loadings less than or equal to 1 m2α-Fe2O3/m3 of flue gas and decreased with an increasing particle loading. At a SO2 concentration of 500 ppmv, mercury oxidation increased with an increasing particle concentration. Without HCl or SO2 in the system, mercury oxidation in the presence of iron oxide particles alone was found to be negligible. To determine if the increased oxidation by HCl or SO2 occurred solely because of an increased surface area, montmorillonite particles were injected, and under these conditions, no increase in the extent of mercury oxidation was observed, suggesting that iron oxide particle surfaces are an important contributor to the promotion of mercury oxidation in coal combustion systems.
’ INTRODUCTION Current efforts to remove gaseous mercury from flue gas streams focus on the use of activated carbon as an injected sorbent. Activated carbons that are brominated or sulfur-impregnated preferentially capture elemental mercury, while undoped activated carbons preferentially capture oxidized forms of mercury, such as mercuric chloride (HgCl2).1 Oxidized mercury is also more likely to be removed by wet flue gas desulfurization units than elemental mercury because of its higher water solubility.2 For these reasons, the development of a better understanding of the factors determining the extent of mercury oxidation in combustion systems is an important consideration in the development of approaches to reduce mercury emissions. Although mercury oxidation in coal combustion systems likely occurs through both homo- and heterogeneous pathways simultaneously, most studies have examined homo- and heterogeneous pathways separately.310 For example, the effect of concentrations of vapor-phase chlorine species, such as HCl, on homogeneous mercury oxidation was examined in several studies using bottled gases in different systems and at temperatures ranging from 20 to 900 °C. Oxidation was consistently found to increase with increasing chlorine concentrations.35 In an isothermal system maintained at 520 °C, mercury oxidation was found to increase from 20% oxidation at a HCl concentration of 300 parts per million by volume (ppmv) to 60% oxidation at a HCl concentration of 3000 ppmv.3 Other studies examined the role of sulfur dioxide (SO2) alone in mercury oxidation. No effect on mercury oxidation was found r 2011 American Chemical Society
when 115 ppm SO2 was added to bottled gases in an isothermal continuous flow reactor system in the absence of HCl at temperatures ranging from 20 to 900 °C.6 In contrast, another study using bottled gases and higher concentrations of SO2 in the absence of chlorine species found that mercury oxidation increased from 20% in the absence of SO2 to 37% in the presence of SO2 at two concentrations (400 and 1200 ppm) at isothermal temperatures from 200 to 900 °C.5 No significant effect of the SO2 concentration was observed on mercury oxidation. Later studies considered SO2 in the presence of chlorine species and found that, under these conditions, SO2 inhibited mercury oxidation. This behavior was observed in an isothermal system using a simulated flue gas at 400 °C, with mercury oxidation decreasing from a level of 72% at a concentration of 60 ppmv HCl to 46% in the presence of 60 ppmv HCl and 1200 ppmv SO2.5 Another isothermal study using bottled gases at 175 °C found that 84% mercury oxidation occurred at a Cl2 concentration of 10 ppmv, which decreased to 0.5% upon the addition of SO2 at a concentration of 1500 ppmv.7 In the same system, the presence of 50 ppmv HCl resulted in mercury oxidation of 0.3%, which decreased to 0.1% in the presence of SO2 at a concentration of 1500 ppmv and HCl at a concentration of 50 ppmv. A more recent study by Cauch et al. indicated that, in experiments using 10% potassium chloride (KCl) impingers in the measurement Received: May 4, 2011 Revised: August 31, 2011 Published: September 07, 2011 4367
dx.doi.org/10.1021/ef200676f | Energy Fuels 2011, 25, 4367–4376
Energy & Fuels of mercury oxidation, measured mercury oxidation in the presence of Cl2 may be overestimated by nearly 90%, possibly because of the formation of the hypochlorite ion (OCl) in the KCl solution, unless SO2 is present in sufficient quantities in the flue gas or sodium thiosulfate (Na2S2O3) is added to the KCl solution to counteract the hypochlorite ion formation.11 The presence of water vapor was found to increase the inhibitory effect of SO2 on mercury oxidation in studies using bottled gas mixtures. Ghorishi et al.8 found that the addition of 500 ppmv SO2 at 754 °C to a simulated flue gas stream containing HCl at concentrations of 50200 ppmv decreased mercury oxidation from 16% at 100 ppmv HCl to 9% and the addition of 1.7% water further decreased Hg oxidation to 7%. Agarwal et al.9 drew similar conclusions in a study of mercury oxidation using bottled gases that were cooled in a flow system from 550 to 200 °C. In that study, 72% mercury oxidation occurred at a Cl2 concentration of 2 ppmv. When SO2 was added, so that the gas contained 370 ppmv SO2 and 2 ppmv Cl2, the extent of mercury oxidation decreased to 52%. Mercury oxidation was completely suppressed in the presence of water vapor at a concentration of 13%, SO2 at a concentration of 370 ppmv, and Cl2 at a concentration of 2 ppmv. Similarly, Zhao et al.10 found that, in the presence of 2000 ppmv SO2 in a bottled gas mixture containing 13 ppmv Cl2, mercury oxidation decreased from 41 to 38% and, when the gas mixture was further modified to contain 8% H2O, the extent of oxidation decreased further to 4%. A few mechanisms have been suggested to explain the effects of chlorine species and SO2 on mercury oxidation. Kinetic modeling has suggested that chlorine radicals are responsible for the majority of homogeneous gas-phase mercury oxidation observed at temperatures between 400 and 800 °C.12 In systems containing Cl2 and SO2 it has been suggested that the formation of SO2Cl2 inhibits mercury oxidation by removing Cl2 necessary for oxidation.9 In modeling efforts, SO2 was also suggested to scavenge OH radicals, producing the sulfite radical (HOSO2) or sulfur trioxide (SO3).13 Zhao et al. attributed the inhibited mercury oxidation in the presence of both SO2 and H2O to the transformation of Cl2 and Cl to HCl, which is less effective at oxidizing mercury. These studies show that, when added alone, Cl2 strongly oxidizes elemental mercury in homogeneous systems, HCl oxidizes mercury, and SO2 either slightly enhances or has no effect on mercury oxidation. When these gases are added in combination, SO2 strongly inhibits mercury oxidation by Cl2, SO2 weakly inhibits oxidation by HCl, and the addition of H2O further inhibits mercury oxidation by chlorine species in the presence of SO2. In nearly all cases reported in the literature, however, the extent of mercury oxidation is considerably less than that observed in operating coal combustion systems, suggesting that heterogeneous pathways play an important role in mercury oxidation. Heterogeneous mercury oxidation has been studied in the presence of solids without the addition of chlorine or sulfur species. Mercury oxidation has been correlated with the level of unburned carbon in fly ash in a coal combustion system,14 the gas oxygen concentration in the presence of activated carbon or fly ash in an isothermal reactor using bottled gases,15 and fly ash in both a 500 MW lignite-fired power plant and a laboratory-scale fixed-bed reactor using bottled gases.16 Other compounds found to catalyze mercury oxidation in a low-temperature bench-scale study using bottled gases include iron compounds, palladium, and unburned carbon.4 Copper chloride was also found to catalyze mercury oxidation in a bench-scale fixed-bed reactor study using bottled gases at a temperature of 250 °C.17
ARTICLE
Other studies with HCl added to simulated flue gas also suggested the importance of heterogeneous pathways. Copper oxide at a temperature of 250 °C17 and iron-containing and ironfree fly ash at a temperature of 180 °C18 were all found to promote mercury oxidation in a study using simulated flue gas containing 50 ppmv HCl. Gold, platinum, and palladium were also found to promote mercury oxidation in the presence of 50 ppmv HCl at a temperature of 149 °C in experiments using simulated flue gas.19 In contrast, mercury oxidation was found to decrease in the presence of 50 ppmv HCl at a temperature of 250 °C when calcium oxide was added to a model fly ash containing silica, alumina, and iron oxide or copper oxide.17 Using simulated flue gas, another study examined heterogeneous mercury oxidation pathways using both HCl and SO2. Oxides of vanadium, chromium, manganese, iron, nickel, copper, and molybdenum were all found to enhance mercury oxidation in experiments using simulated flue gas in the presence of 10 ppmv HCl and 160 ppmv SO2 at reaction temperatures between 250 and 400 °C.20 Measurements conducted at the baghouse outlet of a pilotscale coal-fired combustion system using varying blends of subbituminous/bituminous coals examined mercury oxidation as a function of the SO2 and HCl concentrations, with a constant HCl concentration ranging from 1 to 33 ppmv. In this system, the temperature remained between 127 and 138 °C and mercury oxidation was found to increase with increasing concentrations of SO2 over the range of 200700 ppmv. This increasing extent of mercury oxidation was also weakly correlated with the increasing unburned carbon content in the fly ash and with the increasing iron oxide/calcium oxide ratio in fly ash.21 Mercury oxidation by HCl was also observed to be inhibited by SO2 in moist flue gas in a laboratory study when exposed to fly ash at temperatures of 120 and 180 °C.18 Similarly, mercury oxidation was inhibited when the ratio of concentrations of SO2/HCl was much greater than 1 in the presence of CuO and at a temperature of 250 °C.17 When SO2/HCl ratios of 1:1 or 1:4 were examined, however, heterogeneous mercury oxidation over the CuO-containing model fly ash was enhanced from the corresponding case where HCl was present alone.17 These heterogeneous oxidation studies have shown that copper chloride (CuCl) and different fly ashes oxidize mercury without the addition of HCl or SO2, while in the presence of iron oxide or copper oxide, the presence of HCl is needed for mercury oxidation to occur. Calcium oxide appears to inhibit mercury oxidation by HCl, whereas the presence of SO2 in flue gas is found to have both enhancing and inhibiting effects under different conditions. While these studies provide general guidance on compounds that may catalyze mercury oxidation, a clearer understanding of hetero- and homogeneous mercury oxidation in a well-controlled flame-based system, where radicals deemed necessary for mercury oxidation are present, is needed.
’ EXPERIMENTAL SECTION A methane flame-based experimental system described in detail elsewhere22,23 and depicted in Figure 1 was used to conduct homo- and heterogeneous mercury oxidation experiments in this study. The flamebased system was chosen because it provides a radical pool at high temperatures that may be relevant to mercury oxidation in coal combustion systems. The importance of HCl and SO2 concentrations, hematite (α-Fe2O3) particle concentrations, and montmorillonite particle 4368
dx.doi.org/10.1021/ef200676f |Energy Fuels 2011, 25, 4367–4376
Energy & Fuels
ARTICLE
Figure 1. Homo- and heterogeneous mercury oxidation experimental system. The figure was modified from the work by Sterling et al.23 concentrations was examined under entrained flow conditions. The following is a brief description of the system. Simulated Flue Gas. Commercial-grade CH4 was burned with O2 at an equivalence ratio of 0.9 in a multi-element micro-diffusion flatflame burner to generate the flue gas used in this study. The peak flame temperature in this study measured using a K-type thermocouple was 1174 °C. Particles, HCl (1% in N2), and SO2 (1.5% in N2) were added by injection into a stainless-steel mixing chamber [16 cm (L) 17 cm (W) 21 cm (H)] located directly above the flat-flame burner. The chamber surfaces in contact with the gases were coated with furnace cement to reduce heat loss. Elemental mercury, HCl, SO2, and particles were injected using four separate ports on the mixing chamber to investigate Hg oxidation under well-controlled conditions. A quartz reactor inserted into a 13 cm orifice in a vertical wall of the mixing chamber provided the necessary temperature zone for examining mercury oxidation. The junction between the reactor and the mixing chamber was sealed with furnace cement prior to the start of experiments each day. Quartz Reactor. A total of 5 cm of the cylindrical quartz reactor [59 cm (L) and 13 cm (outer diameter)] was inserted into the mixing chamber and arranged with the outlet lower than the inlet at a downward angle of 2.5° to prevent accumulation of condensation at the outlet. The first 20 cm of the quartz reactor length were covered with heating tape and insulation to reduce heat loss and produce the desired temperature profile. At the inlet of the quartz reactor, the combustion gases entered the mixing chamber orthogonal to the flow through the quartz reactor. There the combustion flue gas mixed with the injected HCl, SO2, and Hg gases before flowing horizontally through the quartz reactor and exiting into the fume hood. Three sampling ports were located at distances of 6.5, 25, and 45 cm from the mixing chamber. Throughout this work, mercury measurements were conducted by extracting flue gas from the sample port 2 because the temperature profile of this configuration is similar to the thermal history of the economizer zone of coal-fired power plants, where mercury oxidation is expected to occur. During this study, it was observed that, after several experiments, Hg concentrations would not return to baseline levels when injection of reactive species was terminated. When this occurred, the experiments were stopped and the reactor was cleaned with 30% nitric acid to remove any accumulated deposits prior to continuation of the measurements. Experiments conducted after the acid wash showed no degradation and produced results similar to those observed in the literature.23
Table 1. Measured System Temperatures under Homo- and Heterogeneous Experimental Conditions homogeneous
heterogeneous
port
temperatures (°C)
temperatures (°C)
P1
620
580
P2
560
530
P3
550
520
P4
560
540
SP1
470
470
SP2
450
390
Previous studies have reported that the surface of a quartz reactor can affect mercury oxidation, with the extent increasing with an increasing reactor surface area/volume (SA/V) ratio. One report indicated that mercury oxidation increased with an increasing surface area at 297 K and a high Cl2 concentration of 5000 ppmv, over a SA/V range from 0.005 to 0.01 m2/m3.24 In contrast and counterintuitively, a separate study found mercury oxidation to be inhibited by increasing the surface area in experiments with Cl2 at concentrations from 100 to 600 ppmv in a methane combustor sampled at a temperature of 573 K when comparing two configurations: 85 and 400 m2/m3.25 The SA/V of the quartz reactor used in this study was 0.001 m2/m3, and the range of HCl concentrations was 100-555 ppmv. Consequently, little interaction with mercury was expected in these experiments. Temperature Profile. The temperature at different points in the experimental system was measured using a K-type thermocouple. The peak post-flame temperature under homogeneous conditions measured within the mixing chamber was 1017 °C. Temperatures in the mixing chamber injection ports (P1P4) and in the quartz reactor sample ports (SP1 and SP2) measured about 2.5 cm from the inside wall and under homogeneous conditions are presented in Table 1. Entraining particles for heterogeneous oxidation experiments required an additional 5 standard liters per minute (SLPM) of N2. This altered the temperatures within the system, as seen in Table 1. The additional 5 SLPM N2 gas stream was included for all heterogeneous experiments, regardless of particle loading, to avoid any temperature change related to varying flow rates of N2. This nitrogen was added postcombustion only in the heterogeneous experiments, slightly decreasing the oxygen concentration in the flue gas from 2.2% in the homogeneous experiments to 1.9% in all heterogeneous experiments. 4369
dx.doi.org/10.1021/ef200676f |Energy Fuels 2011, 25, 4367–4376
Energy & Fuels Mercury Vapor Generation System. Elemental mercury was injected into the experimental system by entraining mercury vapor from a heated permeation tube under a nitrogen stream flowing through the tube at 0.75 SLPM. To begin an experiment, the mercury permeation tube was inserted into a glass U-tube placed in a 20% commercial antifreeze water solution and then brought to a temperature of 93 °C. The manufacturer-specified mercury vaporization rate for the permeation tube was 3100 ng of Hg/min at 100 °C. The temperature was controlled using an immersion heater and stirrer inserted into the antifreeze bath. The Hg/N2 gas stream traveled from the U-tube to port 2 of the mixing chamber through a stainless-steel line maintained at temperatures above 95 °C using heating tapes. Under these conditions, the measured elemental mercury concentration ranged between 35 and 40 μg/Nm3. Particle Injection System. Particles to be injected into the experimental system were loaded into a perfluoroalkoxy polymer vessel. Nitrogen, at a flow rate of 5 SLPM, entered the vessel through tubing inserted to a distance 1 cm from the bottom of the vessel and was used to entrain the particles. The feed line leading from the vessel to the mixing chamber was vibrated using a standard handheld engraver operating at its lowest setting to prevent accumulation of particles within the line. Hematite (α-Fe2O3) and montmorillonite were both considered in the examination of heterogeneous mercury oxidation chemistry, chosen because they are both found in combustion fly ash. Hematite is known to catalyze the oxidation of mercury,17 while montmorillonite is believed to have little catalytic effect on mercury oxidation.26 Hematite powder (Alfa Aesar) had a manufacturer-specified particle size range of 2050 nm. The size range of the hematite particles was verified by visual inspection using scanning electron microscopy, and the surface area of the iron oxide powder was measured using N2-adsorption Brunauer EmmettTeller (BET) analysis and found to be 29.9 m2/g. Montmorillonite (WARD’S Natural Science) was size-segregated with a cyclone followed by a 0.2 μm Isopore filter to remove larger particles. The size distribution of the montmorillonite sample was determined by cascade impaction to have 50% aerodynamic cutoff diameters (d50) between 0.6 and 2.7 μm. The N2-adsorption BET surface area of the montmorillonite sample was 28.9 m2/g. Although the montmorillonite particles were much larger in diameter than the hematite particles, the measured specific surface area was similar because nitrogen molecules partially penetrate the silicate layers.27 To calculate the average particle feed rate during an experiment, the weight loss of the vessel was divided by time elapsed during particle injection. Flue Gas Sampling System. Samples were obtained by withdrawing 4 SLPM of flue gas from the quartz reactor via sample port 2. The sampled gas traveled through a heated Teflon line to a heated quartz filter, with the Teflon line controlled at a temperature of 127 °C to prevent condensation on the inner surface of the line. The quartz filter was maintained at 150 °C in a temperature-controlled oven and used to remove particulate from the sampled stream. Downstream of the filter, the gas stream passed through a second heated line, transferring the gas to a cryotherm, a heat exchanger using impingers where circulating cold water (