Doctor of Philosophy (Ph.D.)

Transcription

1

2 Mercury Oxidation and Adsorption over Cupric Chloride-Based Catalysts and Sorbents for Mercury Emissions Control A dissertation submitted to the Division of Graduate Studies and Research of the University of Cincinnati in partial fulfillment of the requirements of the degree of Doctor of Philosophy (Ph.D.) In the School of Energy, Environmental, Biological and Medical Engineering of the College of Engineering & Applied Science 2012 By Xin Li M.S. Chemical Engineering, Nanjing Forestry University, 1998 Committee: Dr. Joo-Youp Lee (Chair) Dr. Junhang Dong Dr. Soon-Jai Khang Dr. Timothy C. Keener

3 Abstract Mercury emissions control is of great importance in environment protection as well as public health. Current mercury emissions control technologies are not well designed nor optimized, mainly due to the lack of fundamental understanding of adsorption and/or catalytic mechanisms and necessary kinetic modeling and reliable simulation data. This work aims to advance the fundamental mechanistic understanding of heterogeneous catalytic oxidation reaction and adsorption by using the reaction between Hg(0) vapor and CuCl 2 and the subsequent adsorption of resultant oxidized mercury onto sorbents. XANES and EXAFS were used to determine mercury compounds formed on AC sorbents. The XANES study on raw and CuCl 2 -impregnated AC sorbents suggests that little or no elemental mercury is formed onto any spent sorbents and the chemisorption of Hg(0) vapor is very likely to be the dominant mechanism. HgCl 2 is found to be a major oxidation reaction product when CuCl 2 and HCl were impregnated onto raw AC regardless of the type of the carrier gas (i.e. N 2 or O 2 ). The adsorption isotherms of HgCl 2 on DARCO-HG and CuCl 2 -impregnated AC were found to be of the Langmuir type. The kinetic adsorption constants were estimated by fitting the model simulation with experimental data. The breakthrough data from experiments are in good agreement with the calculation results from the modified kinetic model. The simulation results indicate that pore diffusion resistance significantly increases with an increase in sorbent particle size. HgCl 2 adsorption removal performance was also predicted in an entrained flow system using a modified model. i

4 The CuCl 2 /α-al 2 O 3 catalyst possesses high activity for the oxidation of Hg(0) to Hg 2+, with an excellent stability under the environment similar to the flue gas from coal-fired power plants. The CuCl 2 crystallites formed onto α-al 2 O 3 were very stable up to 300 o C, and undergo the thermal reduction process from Cu(II) to Cu(0) via Cu(I). In the absence of HCl and O 2 gases, CuCl 2 was found to follow a Mars-Maessen mechanism by consuming lattice chlorine of CuCl 2 for Hg(0) oxidation and to be reduced to CuCl. In the presence of 10 ppmv HCl, 2,000 ppmv SO 2, and 6% O 2 gases, the CuCl 2 /α-al 2 O 3 sample works as an Hg(0) oxidation catalyst exhibiting >90% conversion with good resistance to SO 2 at 140 o C. The reduced CuCl was able to be re-chlorinated to CuCl 2 under HCl and O 2 gases by following the Deacon reaction. Multiple copper species were found to be formed when γ-al 2 O 3 is used as a substrate as opposed to one Cu(II) species on α-al 2 O 3. The CuCl 2 /γ-al 2 O 3 catalysts with low CuCl 2 loading (<3.5 wt%) showed low catalytic performances in mercury oxidation. In contrast, the high loading (i.e. 10 wt%) CuCl 2 /γ-al 2 O 3 catalyst showed almost complete Hg(0) oxidization in the presence of 10 ppmv HCl and 6%(v) O 2 balanced with N 2, regardless of the presence of 2,000 ppmv SO 2 gas over 140 hrs of the performance evaluations. CuCl 2 is expected to be used as a catalyst and a sorbent by impregnating onto non-carbonaceous and carbonaceous substrates in a temperature window after the air preheater. ii

5 iii

6 Acknowledgements Great thanks are first extended to Professor Joo-Youp Lee for his encouragement and guidance during my PhD study. The accomplishments presented in this thesis would have been impossible without his consistent advice and assistance. I am also grateful to Prof. Timothy C. Keener, Prof. Soon-Jai Khang, and Prof. Junhang Dong for their insightful reviews and helpful suggestions. I would like to thank Dr. Sang-Sup Lee for his great help at the beginning of this research work, and Mr. Jin-Soo Kim, Ms. Bala Lingaraju, Mr. Zhouyang Liu for their help during my study. Special thanks are given to Ms. Emily Smith for her correcting any errors in my draft. I greatly appreciate Dr. Steve M. Heald, and Dr. Dale L. Brewe of Argonne National Laboratory for their assistance in the XAFS measurements, data processing, and interpretation. Use of the Advanced Photon Source at Argonne National Laboratory is also supported by the U.S. Department of Energy, Office of Science, Office of Basic Energy Sciences, under Contract DE-AC02-06CH Thanks, as well, to Dr. Lisa Hommel of Surface Analysis & Mass Spec Lab, Ohio State University, for her help in XPS experiments. This study was funded by the College of Engineering and Applied Science at the University of Cincinnati through faculty start-up funds, and the National Science Foundation, NSF CAREER Grant # Finally, I remain greatly indebted to my family my parents, my husband Zhong Tang, and my son Kailin Tang, for their physical and mental support. iv

7 Table of Contents Nomenclatures... V List of Acronyms and Abbreviations... VI List of Figures... VIII List of Tables... X Chapter 1 Introduction Mercury Emission from Coal-fired Power Plants Hazard and Regulations Mercury in Coal-fired Power Plant Fate and Control Technologies Objective of this Research Reference... 8 Chapter 2 Literature Review Mercury in Different Coal Sources Behavior of Mercury in Coal-fired Utility Boilers Mercury Removal by Existing Air Control Devices Selective Catalytic Reduction of NOx Mercury Capture in Particulate Matter Control System Flue Gas Desulfurization (FGD) Summary of Mercury Control by Enhancing Multi-pollutant Control Emerging Mercury Control Technologies Sorbent Injection Catalyst-enhanced Elemental Mercury Oxidation Electro-Catalytic Oxidation (ECO) Photochemical oxidation technology (PCO) Mechanism of Mercury Oxidation Kinetics Modeling of Mercury Oxidation I

8 2.7 Summary Reference Chapter 3 Preparation and Characterization of CuCl 2 -Impregnated Activated Carbon for Mercury Emissions Control Introduction Experimental Section Sorbent Preparation, Characterization, and Mercury Loading onto Sorbents TGA/MS Characterization XAFS Analysis XANES Data Analysis EXAFS Data Analysis XPS and XRD Characterization Results and Discussion TGA-MS Characterization XANES Analysis Results EXAFS Analysis and Modeling Results XPS and XRD Results Conclusions Acknowledgements Reference Chapter 4 Modeling and Simulation of Mercuric Chloride Adsorption onto CuCl 2 - Impregnated Activated Carbon Sorbents Introduction Experimental Section Kinetic Model of HgCl 2 Vapor Adsorption on Activated Carbon Model Assumptions Model Equations Parameters in PDEs Results and Discussion II

9 4.4.1 Breakthrough Curves and Langmuir Isotherms Simulation results Effects of Pore Diffusion Effects of Axial Diffusion and Film Resistance Comparison of pore diffusion and surface adsorption Model and Simulation for Entrained-flow System Conclusions Reference Chapter 5 Oxidation of Elemental Mercury Vapor over CuCl 2 /α-al 2 O 3 Catalysts for Mercury Emissions Control Introduction Experimental Section Catalyst Preparation Evaluation of CuCl 2 /α-al 2 O 3 Catalyst Characterization Results and Discussion Hg(0) Oxidation over CuCl 2 /α-al 2 O 3 Catalyst Characterization Results of CuCl 2 /α-al 2 O 3 Catalyst Catalytic Oxidation of Hg(0) over CuCl 2 Using the Deacon Reaction Conclusion Reference Chapter 6 Oxidation of Elemental Mercury Vapor over γ-al 2 O 3 Supported CuCl 2 Catalysts for Mercury Emission Control Introduction Experimental Section Catalyst Preparation Performance Evaluation of CuCl 2 /γ-al 2 O 3 Catalyst Characterization Results and Discussion III

10 6.3.1 Performance Evaluation of CuCl 2 /γ-al 2 O 3 Catalyst on Hg(0) Oxidation Characterization of CuCl 2 /γ-al 2 O 3 Catalyst Catalytic Oxidation of Hg(0) over CuCl Conclusion Reference Chapter 7 Summary Summary Perspective Bibliography of XIN LI IV

11 Nomenclatures C HgCl 2 concentration in gas phase, g/m 3 C B Bulk HgCl 2 concentration in gas phase, g/m 3 D Diffusion coefficient, m 2.s -1 D e Effective pore diffusion coefficient, m 2 /s D m Molecular diffusivity, m 2.s -1 d k Kinetic diameter, m D Kn Knudsen diffusivity, m 2.s -1 d p Pore diameter, m Ea i Activation energy of diffusion, kj.mol -1 F Gas flow rate, ml.min -1 K Equilibrium constant, K=k 1 /k 2 k Boltzman constant -1 k 1 Adsorption rate constant, m 3 g -1 s k 2 Desorption rate constant, s -1 K G Mass transfer coefficient, m/s K L Adsorption equilibrium constant M w Molecular weight q HgCl 2 adsorbing on the sorbent, g HgCl 2 /g sorbent R Gas constant, J.mol -1.K -1 ra Adsorption rate, g/(m 3 sorbent particle s) t Time, s T Temperature, o C u Superficial velocity, m/s z Diffusion coordination number Greek letters ε Fraction of pores accessible to non-adsorbing molecules ε b Bed porosity ε p Particle porosity. ρ p Sorbent particle density, g/m 3 σ Molecular diameter, m σ HgCl2,N2 τ φ Ω HgCl2,N2 Average kinetic diameter of HgCl 2 and N 2 molecules; Pore tortuosity Macroporosity Dimensionless collision integral V

12 List of Acronyms and Abbreviations AC ACI AEA APCD B.C. B/H CAA CAIR CAMR CVAA ECO EPRI ESP FF FGD HAP I.C. ID ICR IPD LSFO MACT MEC MEL NAS NTI PAC PCO PDEs PHR PJFF PM PRB RSF RTIL SCR Activated Carbon Activated Carbon Injection Air Entraining Admixture Air Pollution Control Devices Boundary Condition Bag House Clean Air Act Clean Air Interstate Rule Clean Air Mercury Rule Cold Vapor Atomic Absorption Electro-Catalytic Oxidation Electric Power Research Institute Electrostatic Precipitation Fabric Filter Flue Gas Desulfurization Hazardous air pollutants Initial Condition Inner Diameter Information Collection Request Inflection point difference Limestone with Forced Oxidation Maximum Achievable Control Technology Methylpolyoxyethylene octadecanammonium chloride Magnesium-Enhanced Lime National Academy of Sciences National Toxics Inventory Powder Activated Carbon Photochemical Oxidation Partial Differential Equations Peak height ratio Pulse-Jet Fabric Filter Particulate Matter Powder River Basin Radial structure function Room Temperature Ionic Liquid Selective Catalytic Reduction VI

13 SDA SEA SEM-EDS STS TGA-MS TPR XAFS XANES XPS XRD Spray Dryer Absorbers Sorbent Enhancement Additive Scanning Electric Microscopy-Energy Dispersive Spectroscopy Sodium Tetrasulfide Thermal Gravity Analyzer-Mass Spectroscopy Temperature Program Reduction X-ray Absorption Fine-structure Spectroscopy X-ray Absorption Near-Edge Structure X-ray Photoelectron Spectroscopy X-ray Diffraction VII

14 List of Figures Figure 2.1 Process flow diagram of a typical pulverized-coal-fired power plant Figure 2.2 Percent oxidized mercury into ESP[18] Figure 2.3 Mercury removal, various coals and air pollution control configurations [18] Figure 2.4 Mercury adsorption/oxidation mechanisms proposed by basic site model [84].. 38 Figure 3.1 SEM image of DARCO HG activated carbon Figure 3.2 TGA results of CuCl 2 and 10%CuCl 2 impregnated sorbents Figure 3.3 The XANES portion of Hg L III -edge XAFS spectra (left) and the first derivative of the spectra (right) Figure 3.4 k-weighted EXAFS spectra, k 2 χ(k), and the Fourier transformed EXAFS spectra χ(r) (shown in amplitude χ(r) ) Figure 3.5 Experimental Hg L III -edge EXAFS spectra and their fitting results for (a) Raw AC-O 2 and (b) 10% CuCl 2 -AC-N Figure 3.6 Cu 2p XPS of CuCl 2, CuCl, fresh and spent 10%CuCl2-AC samples Figure 3.7 X-ray diffraction patterns of CuCl 2 -impregnated sorbents Figure 4.1 Schematic of HgCl 2 adsorption fixed-bed reactor system Figure 4.2 Mass transfer over a single spherical model Figure 4.3 Setting up the mass balance in a fixed-bed system Figure 4.4 Breakthrough curves of sorbents with different inlet mercury concentrations, (a) Raw-AC, (b) 4%(wt) CuCl 2 -AC, and (c) 10%(wt) CuCl 2 -AC Figure 4.5 Langmuir adsorption isotherms of HgCl 2 adsorption on AC Figure 4.6 Adsorption constant (k 1 ) estimation by fitting simulation results with experimental breakthrough data (Raw AC, C B _ in =10 ppbv, d p =15 μm) Figure 4.7 Simulation results for (a) Raw-AC, (b) 4%CuCl 2 -AC, and (c) 10%CuCl 2 -AC (d p =15 μm) Figure 4.8 HgCl 2 concentration (C) inside raw AC particle along the particle radius, (a) at the time of 50 hr and (b) at the position of z/l=0.2 (Raw AC, C B _ in =5 ppbv, d p =15 μm) Figure 4.9 HgCl 2 uptake (q) on raw AC along the particle radius, (a) at the time of 50 hrs and (b) at the position of z/l=0.2 (Raw AC, C B _ in =5 ppbv, d p =15 μm) Figure 4.10 (a) HgCl 2 concentration in the bulk gas phase and (b) HgCl 2 uptake on the sorbents along fixed-bed length (Raw AC, C B _ in =5 ppbv, d p =15 μm) Figure 4.11 Breakthrough curves at different particle sizes (Raw AC, C B _ in =10 ppbv) Figure 4.12 HgCl 2 concentration profile inside particle with different particle sizes after 50 hrs (Raw AC, C B _ in =10 ppbv) Figure 4.13 HgCl 2 concentration profiles inside sorbent particles with different sizes along the fixed-bed length after 50 hrs, (a) 10 µm, (b) 20 µm, (c) 30 µm, and (d) 40 µm Figure 4.14 Four models describing adsorption kinetics with different assumptions Figure 4.15 A comparison of simulation results obtained from four models taking into account the effects of axial diffusion and film resistance Figure 4.16 HgCl 2 concentration profiles inside particle with particle sizes of (a) d p =10 μm and (b) d p =30 μm (Raw AC, C B _ in =5 ppbv, z/l=0.2, Time=40 hr) Figure 4.17 Schematic of HgCl 2 adsorption in an entrained-flow system Figure 4.18 Bulk-phase HgCl 2 concentration (C B ) as a function of sorbent residence time in VIII

15 the ductwork for different particle sizes, (Raw AC, C B _ in =10 ppbv, m AC =10 g/m 3 ) Figure 4.19 HgCl 2 concentration profile inside particle for different particle sizes after 1 s (Raw AC, C B _ in =10 ppbv, m AC =10 g/m 3 ) Figure 4.20 HgCl 2 removal with respect to sorbent loading in entrained flow system for different particle sizes (Raw AC, C B _ in =10 ppbv, time=10 s) Figure 4.21 HgCl 2 removal with respect to sorbent loading in entrained flow system for different sorbents (C B _ in =10 ppbv, dp =20 µm, time=10 s) Figure 5.1 Breakthrough curves of Hg(0) by 10%(wt) CuCl 2 /α-al 2 O 3 catalyst under different gas conditions at 140 C Figure 5.2. SEM images of α-al 2 O 3 pellet before and after CuCl 2 impregnation Figure 5.3. TGA-MS results for CuCl 2 2H 2 O and 10%(wt) CuCl 2 /α-al 2 O 3 catalyst Figure 5.4. H 2 -TPR profiles for 10%(wt) CuCl 2 /α-al 2 O 3, CuCl 2 2H 2 O, and CuCl Figure 5.5. X-ray diffraction patterns of CuCl 2 /α-al 2 O 3 obtained under different gases Figure 5.6. Cu 2p 3/2 high resolution XPS results for 10%(wt) CuCl 2 /α-al 2 O 3 obtained before and after Hg(0) oxidation under different gases Figure 5.7. Survey XPS of 10%CuCl 2 -alumina before and after mercury oxidation Figure 5.8. Schematic of Hg(0) oxidation reaction over CuCl 2 /α-al 2 O Figure 6.1 Catalytic performances of CuCl 2 /γ-al 2 O 3 catalysts with different CuCl 2 loadings under N 2 gas at 140 C Figure 6.2 Catalytic performance of 10%(wt) CuCl 2 /γ-al 2 O 3 catalyst under different gas conditions at 140 C (30 ppbv inlet Hg(0) concentration and 25 mg catalyst) Figure 6.3 SEM images of CuCl 2 /γ-al 2 O 3 and CuCl 2 /α-al 2 O 3 with the same 10%(wt) loading Figure 6.4 XRD patterns of CuCl 2 /γ-al 2 O 3 catalysts with different CuCl 2 loadings Figure 6.5 XRD patterns of 10%(wt) CuCl 2 /γ-al 2 O 3, fresh and spent samples obtained under different gas conditions Figure 6.6 Cu 2p 3/2 high resolution XPS results of various CuCl 2 /γ-al 2 O 3 samples Figure 6.7 Cu coordination and bonding (a) in Cu-aluminate, and (b) in paratacamite Figure 6.8 TGA results for CuCl 2 2H 2 O and 10%(wt) CuCl 2 /γ-al 2 O 3 catalyst Figure 6.9 H 2 -TPR profiles of 3.5%, 10%(wt) CuCl 2 /γ-al 2 O 3, 10%(wt) CuCl 2 /α-al 2 O Figure 6.10 Schematic of Hg(0) oxidation reaction over CuCl 2 /γ-al 2 O IX

16 List of Tables Table 2.1. Mercury content of coals [3] Table 2.2. Some properties of coals burned in U.S. power plants [18] Table 2.3. Average mercury capture by existing post-combustion control configurations used for pulverized-coal-fired boilers [35] Table 2.4. Speciated mercury results from AC Injection field tests[18] Table 3.1. Element analysis for DARCO HG activated carbon using SEM/EDX (this work), and XPS (Reference [11]) Table 3.2. Particle size, surface area, and pore volume of sorbents used in this study Table 3.3 XANES features for reference mercury compounds and spent samples Table 3.4. EXAFS results for reference mercury compounds Table 3.5. EXAFS results for sorbent samples Table 3.6. Thermodynamic values for possible mercury oxidation reactions on sorbents used in this study Table 4.1. Summary of adsorption isotherms Table 4.2. Summary of fixed-bed testing conditions for HgCl 2 adsorption Table 4.3. Properties of the selected sorbents Table 4.4 Molecular diffusivity of mercury species in the gas phase Table 4.5 Langmuir parameters determined from three sorbents Table 4.6 Kinetic parameters for the adsorbing materials Table 5.1. Mercury speciation results from 10%(wt) CuCl 2 / -Al 2 O 3 fixed-bed tests Table 5.2 Results of quantitative XPS analysis for 10%(wt) CuCl 2 /α-al 2 O X

17 Chapter 1 Introduction 1.1 Mercury Emission from Coal-fired Power Plants Hazard and Regulations Mercury (Hg) and most of its compounds have been widely recognized as an extremely toxic substance and must be handled with great care. The 1990 amendments to the Clean Air Act (CAA) present a list of 188 Hazardous Air Pollutants (HAPs) of which mercury is one. The CAA requires the US Environmental Protection Agency (EPA) to identify HAP sources, quantify the emissions by source category, develop regulations for each source category, and assess the public health and environmental impacts after regulations are implemented (CAA section 112(b)). EPA issued the Mercury Study in December 1997.[1] According to the Mercury Study, Hg cycles in the environment as a result of natural and human (anthropogenic) activities.[1] Most of the Hg in the atmosphere is elemental Hg vapor, which circulates in the atmosphere for up to one year, and hence, could be widely dispersed and transported thousands of miles from likely sources of emission. The Mercury Study also found that most of the Hg in water, soil, sediments, or plants and animals is in the forms of inorganic Hg salts and organic forms of Hg (e.g. MeHg). The inorganic form of Hg, when either bound to airborne particles or in a gaseous form, is readily removed from the atmosphere by precipitation and is also dry deposited. Wet deposition is the primary mechanism for transporting Hg from the atmosphere to surface waters and land. Even after it deposits, Hg commonly is emitted back to the atmosphere either as a gas or associated with particles, to be re-deposited elsewhere. As a toxic pollutant, mercury occurs naturally in the environment, but is released into the atmosphere in significant quantities as the result of the burning of fossil fuels. Coal-fired power 1

18 plants are responsible for about one-third of mercury emissions from human activity. In the United States, it was estimated that average annual mercury emission from the coal-fired power plants from the year 2002 to 2006 was 44 metric tons, which was considered as the major anthropogenic source of mercury emissions.[2] Over the time, mercury has been identified as one of the most toxic HAPs released by coal-fired power plants, primarily because of its ability to impair functioning of the central nervous system. Mercury released into the air settles in rivers and lakes, where it moves through the food chain to the fish that people eat. Mercury exposure can harm the brain development of infants and children. A Study conducted by National Academy of Sciences (NAS) noted that mercury deposited on land and water can be metabolized by microorganisms into MeHg, which is a highly toxic and more bioavailable form, that biomagnifies in the aquatic food chain.[3] Bolger et al. also confirmed that mercury can complex with microorganisms to form the organic complex methyl mercury, which is a neurotoxin that can kill nerve cells, cause a lack of coordination, slurred speech, and even death.[4] This neurotoxin can concentrate in the bodies of fish through the food chain, and can ultimately concentrate in human bodies and cause serious health problems. Study showed that the brain and developing neurological system of the fetus are particularly sensitive to mercury and can be damaged by even a tiny exposure.[5] According to the EPA, roughly half of the nation's lakes and reservoirs have levels of mercury that exceed safe levels.[6] Each year, more than 300,000 newborn babies may have an increased risk of learning disabilities due to in utero exposure to mercury, which is a neurotoxin that causes impaired neurological development in fetuses, infants and children.[1] Numerous studies have shown that Hg poses hazards to public health and the environment and that its emission must be controlled.[3, 4, 6-8] 2

19 In 1993, EPA began developing the National Toxics Inventory (NTI), a national repository of emission inventory data for HAPs. The 1996 NTI is an emissions inventory for use in dispersion and exposure modeling that can be used not only to predict ambient air concentrations and the resultant risk to the American population, but also to measure progress under the CAA in reducing HAP emissions. In 2005, EPA promulgated the Clean Air Mercury Rule (CAMR) to cap and reduce emissions of mercury from coal-fired power plants. It was later vacated by the D.C. Circuit Court of Appeals in early [9] EPA has initiated work to develop emissions standards for power plants under CAA section 112, consistent with the D.C. Circuit s opinion regarding the CAMR.[10] In July 2010, EPA issued a new proposed rule, the Transport Rule, which replaces the 2005 Clean Air Interstate Rule (CAIR) and will start to regulate sulfur dioxide (SO 2 ) and nitrogen oxide (NO x ) emissions from power plants in 28 states from 2012.[9, 11] The U.S. EPA also estimates that a total gigawatts (GW) of Flue Gas Desulfurization (FGD) and GW of Selective Catalytic Reduction (SCR) units of a total 373 GW to be generated from coal combustion would be operative by 2020 in order to meet the Transport Rule requirements (according to the TR SB Limited Trading model).[11] On December 21, 2011 the U.S. Environmental Protection Agency (EPA) announced standards to limit mercury, acid gases and other toxic pollution from power plants (MATS). The standards, as the part of new rule of National Emission Standards for Hazardous Air Pollutants published as the Federal Register publication (Vol 17, No. 32), regulate mercury, other toxic heavy metals, and acid gases from coal- and oil-fired utility, industrial, commercial, and institutional power plants.[5] The new rule, which will be effective on April 16, 2012, sets the standard to reduce mercury emissions by >90% starting from 2016 by adopting a Maximum Achievable Control Technology (MACT) approach, instead of a Cap & Trade approach adopted 3

20 by the rescinded Clean Air Mercury Rule (CAMR). The new rule makes it necessary to more strictly control elemental mercury emissions. EPA estimates that the installation of activated carbon injection systems will increase from current 48 GW to 99 GW by 2020 (according to the ToxR policy case).[12] 1.2 Mercury in Coal-fired Power Plant Fate and Control Technologies Mercury is present in coal in trace amounts (average content of μg g 1 ). It exists in three forms in coal-derived flue gas: elemental (Hg(0)), oxidized (Hg 2+ ), and particle-bound (Hg(p)). Hg 2+ and Hg(p) are relatively easy to remove from flue gas using typical air pollution control devices (APCD). Hg(0), however, is difficult to capture. It is insoluble in water and therefore cannot be removed by wet flue gas desulfurization equipment. At the high temperatures in combustion zone of boilers, combustion releases the Hg in coal into the exhaust gas as elemental mercury (Hg(0)). The Hg(0) vapor may then be oxidized by HCl, SO 2, and fly ash in flue gas due to thermo-chemical processes.[13, 14] Oxidized mercury (Hg 2+ ) is soluble and has a tendency to associate with the particles in flue gas to form particulate-bound mercury (Hg(p)). Therefore, emissions of Hg 2+, may be efficiently controlled by typical APCD, such as electrostatic precipitators (ESP), fabric filter (FF), and flue gas desulfurization (FGD) systems.[15, 16] However, the relative proportions of Hg 2+, Hg(p) and Hg(0) can vary widely, the corresponding reductions in total mercury achieved by APCD vary.[17-19] For example, the removal efficiency of Hg from the flue gas by a combination of cold side ESP and wet FGD range from 24% to 70%. Emission speciation is an important source of uncertainty when assessing the atmospheric fate of mercury because Hg 2+, Hg(p) and Hg(0) have very different physico-chemical characteristics, and resulting in different atmospheric lifetimes. 4

21 Under the new rule,[5] the following control options are considered feasible: (1) chemically-promoted carbonaceous sorbent injection followed by its removal at a particulate matter control device; and (2) oxidation of Hg(0) vapor using oxidation catalysts or oxidants followed by the capture of oxidized mercury in wet FGD systems.[15, 20] Sorbent injection is currently the most mature technology, and does not require a high capital cost, because its cost primarily depends on sorbent and disposal costs. Adsorption in solid materials is a process that offers great potential for achieving high quality air emissions with respect to mercury. A suitable adsorbent may have the ability to capture both elemental and oxidized mercury. To date, the most widely tested and promising adsorbent is found to be raw and chemically promoted activated carbon (AC).[14, 15, 17] As a near-term mercury control technology, injecting AC as a sorbent to capture flue gas mercury has shown its efficiency, costeffectiveness, and easy-operation. However the process applied to coal-fired boilers is still in the early stages and its effectiveness under varied conditions (e.g., fuel properties, flue gas temperatures, and trace-gas constituents) need to be further investigated. It is known that Hg(0) can be oxidized to mercuric oxide (HgO), mercuric sulfate (HgSO 4 ), mercuric chloride (HgCl 2 ), mercuric nitrate Hg(NO 3 ) 2, or other Hg compounds. Most mercury compounds that can be formed in typical coal combustion flue gas are known to be weakly bonded, but only HgCl 2 and HgBr 2 are estimated to have significant bond strengths.[21] Studies showed that, among these oxidized mercury species, HgCl 2 has high solubility in water (i.e g/100 g water at 25 C) and other oxidized forms have very low solubility.[22] Therefore, HgCl 2 is the most desirable oxidized form for capture in wet FGD systems. In this context, heterogeneous Hg(0) oxidation using catalysts or oxidants is highly expected to play a critical role in future mercury emissions control in the U.S.[23] 5

22 The transformation of Hg(0) to Hg 2+ primarily occurs in coal-fired power plants utilizing coal with a high chlorine content, such as Eastern bituminous coal.[24, 25] A large portion of this transformation occurs across the SCR unit. Meanwhile, plants burning Western subbituminous or lignitic coals observe significantly less Hg(0) oxidation across the SCR system.[18, 21] While it is evident that a certain amount of chlorine species in the flue gas is necessary for these reactions to occur, the exact mechanisms for Hg(0) oxidation and their dependence on flue gas properties are currently unknown. In recent years, the use of Western subbituminous coal (such as Powder River Basin (PRB), which generates higher percentages of Hg(0) vapor) in coal-fired power plants is increasing,[26] and the proposed Transport Rule is very likely to increase the installation of wet FGD systems (>95% for SO 2 control on a basis of total electricity generation) and SCR units for large coal-fired power plants. In this sense, an efficient mercury oxidation technology upstream of the wet FGD system would provide both SO 2 and mercury emissions control. We have been conducting research on CuCl 2 -impregnated carbonaceous [27-29] and noncarbonaceous sorbents,[30, 31] which have demonstrated high efficiency in mercury capture. In our previous work, various combinations of dopants and substrates have been explored in order to develop efficient and cost-effective non-carbonaceous sorbents, which do not give adverse impacts on by-product utilization. It was found that metal halides (e.g., CuCl, CuCl 2, CuBr 2, ZnCl 2, ZnBr 2, NiCl 2, etc.) have potential for Hg(0) oxidation that can be dispersed over both carbonaceous and non-carbonaceous substrates. An entrained-flow reactor system has been developed in our lab to evaluate the novel mercury sorbents for elemental mercury control.[32-35] It has also been shown that the Hg(0) removal by CuCl 2 -impregnated activated carbon sorbent is a very complicated process, involving Hg(0) oxidation and the resultant oxidized 6

23 mercury adsorption. Our previous results indicated that Hg(0) vapor reacts with CuCl 2 and the desorbed resultant oxidized mercury is re-adsorbed onto substrate surfaces, suggesting different sites for Hg(0) oxidation and re-adsorption of the resultant oxidized mercury.[27, 28] To date, the specific Hg-sorbent reaction and binding mechanisms have not been well understood because of the difficulties in the identification of surface functional groups and mercury species, binding of functional groups onto the carbon surface, reaction between functional groups and Hg(0) vapor, and active adsorption sites of reacted mercury species on the carbon surface. Fundamental understanding of the mechanisms and kinetics is definitely needed in order to offer a useful predictability that will aid in scaling up laboratory experiments to pilot or larger scale. Therefore, further studies on probing the mechanisms and kinetics of mercury adsorption and oxidation are of great interest. 1.3 Objective of this Research This research aims to advance the fundamental mechanistic understanding in heterogeneous catalytic oxidation reaction and adsorption by using the reaction between Hg(0) vapor and CuCl 2, and subsequent adsorption of resultant oxidized mercury (Hg 2+ ) onto sorbents or catalysts. A literature review is conducted to provide a comprehensive review of the research topics in this work, as presented in Chapter 2. This dissertation is arranged immediately following the specific research objectives as shown below, (1) Identify the resultant oxidized mercury species formed from the reaction between Hg(0) vapor and CuCl 2 -impregnated activated carbon (CuCl 2 -AC) by means of several characterization techniques, including X-ray Absorption Near-Edge Structure (XANES), and X- ray absorption fine-structure spectroscopy (XAFS);[29] 7

24 (2) Conduct kinetic study of oxidized mercury adsorption and oxidation on different ACbased sorbents on a fixed-bed system. Perform the simulation of mercuric chloride adsorption on AC at various conditions through the adsorption kinetics, mass balance, and diffusion theory considering different rate control steps. (3) Investigate the reaction (oxidation) mechanisms of elemental mercury vapor with cupric chloride through the characterization of several techniques, including Scanning electronic microscopy-energy dispersive spectroscopy (SEM-EDS), Thermo-gravimetric analysis-mass spectroscopy (TGA-MS), Temperature programming reduction (TPR), X-ray diffraction (XRD), and X-ray photoelectron spectroscopy (XPS);[36] (4) Prepare novel Cu-based catalysts on various supports (i.e. porous alumina). Investigate the mercury adsorption/oxidation mechanism by performing extensive characterization through BET, TGA-MS, TPR, XRD, and XPS. The research work in this study will help us to fundamentally understand the speciation of mercury and copper from CuCl 2 -impregnated sorbents and catalysts to elucidate the adsorption and oxidation pathway. The outcome of this research is expected to provide systematic and scientific information on reaction and adsorption of mercury in flue gas for its control and speciation studies. 1.4 Reference 1. US EPA: Mercury Study Report to Congress, vol. VIII, EPA-452/R , US Environmental Protection Agency, Washington, DC, USA,

25 2. Wiedinmyer, C.; Friedli, H., Mercury Emission Estimates from Fires: An Initial Inventory for the United States. Environmental Science & Technology 2007, 41, (23), "Toxicological Effects of Methylmercury", Committee on the Toxicological Effects of Methylmercury, Board on Environmental Studies and Toxicology, National Research Council (NAS), Bolger, P. T.; Szlag, D. C., An Electrochemical System for Removing and Recovering Elemental Mercury from a Gas Stream. Environmental Science & Technology 2002, 36, (20), National Emission Standards for Hazardous Air Pollutants From Coal- and Oil-Fired Electric Utility Steam Generating Units and Standards of Performance for Fossil-Fuel-Fired Electric Utility, Industrial-Commercial-Institutional, and Small Industrial-Commercial- Institutional Steam Generating Units. 40 CFR Parts 60 and 63. Federal Register 77:32 (Feb 16, 2012) p US EPA, online available at: 7. An Environmental Health and Engineering Report 17505, Emissions of Hazardous Air Pollutants from Coal-fired Power Plants, 2011, 8. Grandjean, P.; Satoh, H.; Murata, K.; Eto, K., Adverse Effects of Methylmercury: Environmental Health Research Implications. Environ Health Perspect 2010, 118, (8). 9. Clean Air Mercury Rule. Office of Air Quality Planning and Standards and Office of Research and Development, U.S. Environmental Protection Agency: Washington, DC. 2005, "Federal Implementation Plans To Reduce Interstate Transport of Fine Particulate Matter and Ozone." 40 CFR Parts 51, 52, 72, 78, and 97. Federal Register 75:147 (August 2, 2010) p "Federal Implementation Plans To Reduce Interstate Transport of Fine Particulate Matter and Ozone; Correction." 40 CFR Parts 51, 52, 72, 78, and 97. Federal Register 75:177 (September 14, 2010) p EPA, Integrated Planning Model (IPM) v.4.10 Model Runs. 9

26 13. Meij, R.; Vredenbregt, L. H. J.; Winkel, H. t., The Fate and Behavior of Mercury in Coal- Fired Power Plants. Journal of the Air & Waste Management Association 2002, 52, (8), Park, K. S.; Seo, Y. C.; Lee, S. J.; Lee, J. H., Emission and speciation of mercury from various combustion sources. Powder Technology 2008, 180, (1 2), US EPA: Control of Mercury Emissions from Coal-fired Electric Utility Boilers, EPA- 600/R , US Environmental Protection Agency, Washington, DC, USA, US Environmental Protection Agency: ICR data, online available at: Srivastava, R. K.; Hutson, N.; Martin, B.; Princiotta, F.; Staudt, J., Control of mercury emissions from coal-fired in electric utility boilers. Environmental Science & Technology 2006, 40, (5), Pavlish, J. H.; Sondreal, E. A.; Mann, M. D.; Olson, E. S.; Galbreath, K. C.; Laudal, D. L.; Benson, S. A., Status review of mercury control options for coal-fired power plants. Fuel Processing Technology 2003, 82, (2 3), Lee, C. W.; Serre, S. D.; Zhao, Y.; Lee, S. J., Mercury oxidation promoted by a selective catalytic reduction catalyst under simulated Powder River Basin coal combustion conditions. Journal of the Air & Waste Management Association 2008, 58, (4), Jones, A. P.; Hoffmann, J. W.; Smith, D. N.; Feeley, T. J.; Murphy, J. T., DOE/NETL's Phase II Mercury Control Technology Field Testing Program: Preliminary Economic Analysis of Activated Carbon Injection. Environmental Science & Technology 2007, 41, (4), Schofield, K., Fuel-Mercury Combustion Emissions: An Important Heterogeneous Mechanism and an Overall Review of its Implications. Environmental Science & Technology 2008, 42, (24), Lide, D. R., CRC Handbook Chemistry and Physics, 85th Edition CRC Press: Boca Raton, FL: Presto, A. A.; Granite, E. J., Survey of Catalysts for Oxidation of Mercury in Flue Gas. Environmental Science & Technology 2006, 40, (18), Bock, J., Hocquel, M.J.T., Unterberger, S., Hein, K.R.G., In Mercury Oxidation Across SCR Catalysts Of Flue Gas With Varying HCl Concentration, Mega Symposium and Air & Waste Management Association, Specialty Conference, Washington, DC, May 19-20, 2003, 2003; Washington, DC, May 19-20, 2003,

27 25. Lee, C. W., Srivastava, R. K., Ghorishi, S. B., Hastings, T. W., Stevens, F. M. In Study of Speciation of Mercury under Simulated SCR NOx Emission Control Conditions,Specialty Conference,, Mega Symposium and Air & Waste Management Association,, Washington, DC, May 19-20, 2003, 2003; Washington, DC, May 19-20, 2003, DOE, U. S. Annual Energy Outlook 2007 with Projections to 2030; Report No. DOE/EIA-0383(2007); Energy Information Administration.: In. 27. Lee, S.-S.; Lee, J.-Y.; Keener, T. C., Mercury oxidation and adsorption characteristics of chemically promoted activated carbon sorbents. Fuel Processing Technology 2009, 90, (10), Lee, S.-S.; Lee, J.-Y.; Keener, T. C., The effect of methods of preparation on the performance of cupric chloride-impregnated sorbents for the removal of mercury from flue gases. Fuel 2009, 88, (10), Li, X.; Lee, J.-Y.; Heald, S., XAFS characterization of mercury captured on cupric chloride-impregnated sorbents. Fuel 2012, 93, Lee, J. Y.; Ju, Y. H.; Keener, T. C.; Varma, R. S., Development of cost-effective noncarbon sorbents for Hg-0 removal from coal-fired power plants. Environmental Science & Technology 2006, 40, (8), Lee, J.-Y.; Ju, Y.; Lee, S.-S.; Keener, T.; Varma, R., Novel Mercury Oxidant and Sorbent for Mercury Emissions Control from Coal-fired Power Plants. Water, Air, Soil Pollution: Focus 2008, 8, (3), Lee, S.-S.; Lee, J.-Y.; Keener, T., Elemental Mercury Control by Novel Oxidant and Sorbent in an Entrained-Flow System. Water, Air, Soil Pollution: Focus 2008, 8, (3), Lee, S.-S.; Lee, J.-Y.; Keener, T. C., Novel sorbents for mercury emissions control from coal-fired power plants. Journal of the Chinese Institute of Chemical Engineers 2008, 39, (2), Lee, S. S.; Lee, J. Y.; Keener, T. C., Performance of Copper Chloride-Impregnated Sorbents on Mercury Vapor Control in an Entrained-Flow Reactor System. Journal of the Air & Waste Management Association 2008, 58, (11), Lee, S. S.; Lee, J. Y.; Keener, T. C., Bench-Scale Studies of In-Duct Mercury Capture Using Cupric Chloride-impregnated Carbons. Environmental Science & Technology 2009, 43, (8),

28 36. Li, X.; Lee, J.-Y.; Kim, J., Oxidation of Elemental Mercury Vapor over Cupric Chloride- Based Catalyst for Mercury Emissions Control. Environmental Science & Technology, 2012, in review. 12

29 Chapter 2 Literature Review 2.1 Mercury in Different Coal Sources The mercury content of coal varies across U.S. coal basins. Mercury removal rates at coal-fired power plants are largely based on the type of coal burned and the plant equipment installed. According to U.S. Geological Survey's COALQUAL database, the average mercury content in coal is approximately 0.2 μg/g.[1] The values for conterminous U.S. coal areas are estimated to range from 0.08 μg/g for coal in the San Juan and Uinta regions to 0.22 μg/g for the Gulf Coast lignite.[2] Table 2.1 lists the mercury content in several typical coals from different sources.[3] It shows that the content of mercury in the coal varies wildly depending on different geography sources without apparent relationship between the types of the coal. The U.S. coalfired power plants typically burn one of three types of fuel: (1) bituminous coal (also referred to as high rank coal), (2) subbituminous coal, and (3) and lignite (subbituminous coal and lignite are referred to as low rank coals). Table 2.1. Mercury content of coals [3] Number of Average Hg content Range Coal type analyses (ppm, dry) (ppm, dry) Anthracite Bituminous coal Bituminous, High S Bituminous, Low S Subbituminous coal Lignite

30 Studies have shown that coal-fired power plants are the largest anthropogenic emitters of mercury in the U.S.[4] The current capacity of U.S. coal-fired power plants is over 300 GW. These coal-fired utility boilers emit over 40 tons of mercury each year. Although a number of power plants have been adopting various measures to control hazardous air pollutants (HAPs), including a wide range of combinations of installed air pollution control configurations, the specific control of mercury emission has not been put in force so far. When the new rule comes into force,[5] EPA estimates that the installation of activated carbon injection systems will increase from current 48 GW to 99 GW by 2020 (according to the ToxR policy case) in order to effectively control the mercury emission.[6] 2.2 Behavior of Mercury in Coal-fired Utility Boilers In raw coal, mercury may exist in various forms, including elemental mercury (Hg(0)), ionic mercury (Hg 2+ ) compounds. During combustion, all the mercury in coal is volatilized at the combustion temperatures, and converted to Hg(0) vapor in coal-fired boilers. As the flue gas is cooled, however, a series of complex reactions occur converting Hg(0) to mercury compounds and/or particle-associated mercury (Hg(p)) that are in a solid-phase at flue gas cleaning temperatures, or Hg that is adsorbed onto the surface of other particles.[7-9] The vapor phase Hg(0) might be oxidized by HCl, SO 2, and fly ash in flue gas through thermo-chemical processes.[10, 11] Since the Hg(0) oxidation reactions are kinetically limited, Hg actually enters the flue gas cleaning device(s) as a mixture of Hg(0), Hg 2+, and Hg(p). This partitioning of Hg into Hg(0), Hg 2+, and Hg(p) is known as mercury speciation, which is one of the critical factors in selecting mercury control approaches. The fate of mercury in a typical pulverized-coal-fired power plant can be schematically depicted in Figure 2.1. Emission speciation is an important 14

31 source of uncertainty when assessing the atmospheric fate of mercury because Hg 2+, Hg(p), and Hg(0) have very different physico-chemical characteristics and, consequently, different atmospheric lifetimes. Figure 2.1 Process flow diagram of a typical pulverized-coal-fired power plant Research on the mercury speciation in the coal-fired flue gas stream has been carried out for decades. Both laboratory and field studies indicate that the mercury speciation is affected by a number of factors, including coal properties, operating conditions, fly ash properties, and flue gas control processes.[12] Among these factors, chlorine content in coal is believed to be one of the most important factors.[13] For example, Cao, et al, conducted field tests in an 100-MW utility boiler with cold-side electrostatic precipitators (CS-ESP).[14] Their tests have shown that 15

32 chorine in coal promotes the mercury oxidation process significantly. This conclusion has been further verified by other observations. In coal-fired power plants utilizing coal with a high chlorine content, such as Eastern bituminous coal, it was found that a significant transformation of Hg(0) to Hg 2+ occurs mainly across the selective catalytic reduction (SCR) unit.[15, 16] On the other hand, power plants burning Western subbituminous or lignitic coals with low chlorine content showed much less Hg(0) oxidation across the SCR system.[15, 17] As shown in Table 2-2, the chlorine concentration of coal is dependent on coal type. [18] In general, bituminous coals have higher concentrations of chlorine than subbituminous coal and lignite. Hence, relatively higher concentrations of Hg 2+ are typically found in bituminous coalfired flue gas streams. Study by Pavlish et al. has shown that for the combustion of bituminous coals, about 20% of the total mercury is in the elemental form, 35% is oxidized, and 45% is particulate-bound.[19] For subbituminous coals, about 65% of the total mercury is in the elemental form, 20% is oxidized, and 15% is particulate-bound. For lignite coals, about 85% of the total mercury is in the elemental form, 10% is oxidized, and 5% is particulate-bound. In recent years, the use of Western, such as Powder River Basin (PRB) subbituminous coal in coal-fired power plants is increasing.[20] Less transformation of Hg(0) to Hg 2+ will be expected even if the installation of wet FGD systems, and SCR units for large coal-fired power plants. Therefore, as required by the new rule, [21] a mercury oxidation technology upstream of the wet FGD system would be necessary to provide both SO 2 and mercury emissions control. 16

33 Table 2.2. Some properties of coals burned in U.S. power plants [18] Coal Mercury, ppm Chlorine, ppm Sulfur, % Range Average Range Average Range Average Bituminous Subbituminous Lignite Among elemental, oxidized, and particulate-bound mercury species present in flue gas, Hg(0) vapor is most difficult to control because of its low concentrations at a ppb level, high volatility, low reactivity in the gas phase (due to its closed electron structure of 5d 10 6s 2 ), and low solubility in water. Although thermodynamics predicts that oxidized mercury is a predominant mercury species in flue gas below ~450 C, the transformation of Hg(0) to Hg 2+ is kinetically limited and Hg 2+ fraction ranges from almost 0 to 100%, primarily depending on coal chlorine and unburned carbon content. Compared to Hg(0), oxidized mercury species have low volatility, high surface affinity (particularly with carbon), and high solubility (particularly HgCl 2 ). Thus they can be readily controlled by AC injection or wet FGD systems. In contrast, once Hg(0) vapor is emitted, it is a long residence time in the atmosphere, and contributes to global-scale deposition. Therefore, it is critical to utilize the heterogeneous oxidation of Hg(0) vapor for its removal in downstream air pollution control devices such as ESP, FF, and wet FGD system depending on the availability of a utility plant. It is known that Hg(0) can be oxidized to mercuric oxide (HgO), mercuric sulfate (HgSO 4 ), mercuric chloride (HgCl 2 ), mercuric nitrate Hg(NO 3 ) 2, or other Hg compounds. [22] Among these oxidized mercury species, HgCl 2 has high solubility in water (i.e g/100 g 17

34 water at 25 C) and other oxidized forms have very low solubility.[23] Therefore, HgCl 2 is the most desirable oxidized form for capture in wet FGD systems. In this context, transformation of Hg(0) to Hg 2+ (mainly HgCl 2 ) through heterogeneous oxidation using catalysts or oxidants is highly expected to play a critical role in future mercury emissions control in the U.S.[24] 2.3 Mercury Removal by Existing Air Control Devices Before the implementation of the new rule, which will be effective on April 16, 2012, the control of mercury emissions from coal-fired boilers is mainly achieved via existing air control devices from removing particulate matter (PM), sulfur dioxide (SO 2 ), and nitrogen oxides (NO x ). This includes capture of Hg(p) in PM control equipment, e.g. electrostatic precipitator (ESP), or fabric filter (FF), and capture of soluble Hg 2+ compounds in wet FGD systems. Recent data also reflect that use of SCR NO x control can enhance oxidation of Hg(0) in flue gas and results in increased mercury removal in wet FGD. [25, 26] Currently, available mercury oxidation or mercury capture processes are reviewed below in the order of air control device sequence in existing coal-fired power plants Selective Catalytic Reduction of NOx Selective catalytic reduction (SCR) is a means of converting nitrogen oxides (NOx) into nitrogen and water catalyzed by selective catalysts at an elevated temperature. During the combustion process, the nitrogen that is present naturally in the coal (e.g. nitrogen compounds), and the nitrogen and oxygen present in the combustion air combine to form NOx. In order to control the emission of NOx, the exhaust gas from the combustion is passed through a large catalyst bed where the NOx reacts with the catalyst and anhydrous ammonia and is converted to nitrogen and water. SCR technology is a proven and effective method to control NOx emissions 18

35 from coal-fired power plants, typically removing between 80-90% of the NOx in the exhaust gas of a coal-fired power plant. Because Hg 2+ can be captured much more effectively than Hg(0) in wet FGD systems, converting Hg(0) to Hg 2+ in the upstream of the wet FGD should improve mercury capture in the wet FGD system. SCR catalysts have been shown to promote the oxidation of Hg(0) to Hg 2+, particularly for bituminous coal. Since 2003, the impact of SCR on mercury oxidation has been investigated in two series of field tests: (1) EPRI-EPA-DOE sponsored field tests, [27] and (2) DOE sponsored tests being conducted by CONSOL.[28] The results of field test programs suggest that oxidation of elemental mercury by SCR catalyst may be affected by the following conditions:[27] 1) The coal characteristics, especially the chlorine content; 2) The type and the amount of catalyst used to treat the gas stream; 3) The temperature of the reaction; 4) The concentration of ammonia; 5) The age of the catalyst. It has been shown that the use of SCR unit can result in 85-90% mercury oxidation when firing bituminous coals. Figure 2.2 shows data from the EPRI-EPA-DOE field tests, [27] the DOE-CONSOL field tests [28] and from field tests conducted at Dominion Resources Mount Storm Unit 2.[29] In particular, the figure reflects the percent of Hg 2+ measured at the inlet of the CS-ESP for boilers equipped with SCR. In most cases, the percent of Hg 2+ increased when burning bituminous coals, indicating a positive impact on transformation of Hg(0) to Hg 2+ by SCR process. 19

36 Obviously, it would be desirable to increase the oxidation of mercury by SCR when firing subbituminous or even lignite coals as in the case of bituminous coals. Senior and Linjewile compared the results of thermochemical equilibrium calculations of mercury species concentration to full-scale and pilot test results.[30] Results from both calculations and tests suggested that, when firing subbituminous coal, the oxidation of Hg(0) to Hg 2+ with SCR is limited by equilibrium rather than by kinetics. They also found that when ammonia was injected, oxidation of Hg(0) to Hg 2+ tended to drop somewhat.[30] This suggests that the presence of ammonia may interfere with mercury oxidation on the catalyst. Additionally, as SCR catalyst ages, the oxidation of Hg(0) tends to decline due to the loss of catalyst activity. Figure 2.2 Percent oxidized mercury into ESP[18] ESP = Electrostatic Precipitator; LSFO = Limestone with Forced Oxidation; SDA = Spray Dryer Absorber system; FF = Fabric Filter; SCR = Selective Catalytic Reduction. PRB = Powder River Basin; Bit = Bituminous coal; Mt. Storm = Mt. Storm Coal 20

37 Nevertheless, conversion of Hg(0) to Hg 2+ in the SCR unit is a promising approach especially when burning bituminous coals. Modified SCR catalysts or addition of special designed oxidation catalysts to the existing SCR unit may further benefit Hg oxidation Mercury Capture in Particulate Matter Control System During combustion, the noncombustible mineral materials in coal remain after the coal is burned. Most of this material is carried by the exhaust gas and finally exit the boiler in the form of fly ash. In power plants, these remaining unburned materials are collected in the bottom of the boiler. Currently, there are two technologies used to capture fly ash, i.e. ESP, and FF (also named bag house ). An ESP works like a big fly ash magnet. The exhaust gas enters the precipitator where a negative electric charge is imposed on the fly-ash particles. The negatively charged particles are then attracted to a series of positively charged metal plates. An FF works like a household vacuum cleaner. It consists of a series of cloth bags that filter fly ash from the exhaust gas. Both ESP and FF can typically remove more than 99% of the fly ash contained in the exhaust gas. It has been reported that fly ash has the capability to capture oxidized mercury from flue gas in the ductwork upstream and inside of a PM control system.[31, 32] The EPA s Information Collection Request (ICR) data showed that a FF is more effective for mercury capture than an ESP due to more efficient contact between mercury and fly ash in FF [13] [ as shown in Figure 2.3]. 21

38 Figure 2.3 Mercury removal, various coals and air pollution control configurations [18] CS-ESP = cold-side electrostatic precipitator; HS-ESP = hot-side electrostatic precipitator; FF = fabric filter; PS = particle scrubber; SDA = spray dryer absorber system Flue Gas Desulfurization (FGD) Wet and dry FGD systems are typical control technologies for SO 2 emissions from coalfired utilities and installed downstream of the PM control device. The wet FGD scrubbers use caustic slurry composed of water and limestone or water and lime. Commonly referred to as a scrubber, the wet FGD is a proven and effective method for removing SO 2 emissions from the exhaust of coal-fired power plants. Scrubbers are typically designed to remove more than 95% of the SO 2 contained in the exhaust gas. The dry FGD uses a fine mist of lime slurry (spray dryer absorbers) for SO2 removal. Hg 2+ can be absorbed in the slurry of both wet and dry FGD, but Hg(0) is still hardly captured. Mercury Capture in Wet FGD Systems Because oxidized mercury (e.g. HgC1 2 ) is highly water soluble. It can be readily captured in wet FGD systems. Field tests showed that over 90% Hg 2+ can be captured in most 22

39 calcium-based wet FGD systems. Niksa, et al studied the scrubber equilibrium chemistry and discussed the capability for total mercury capture in wet FGD systems using a thermochemical equilibrium model.[33] It is found that the reduction of Hg 2+ to Hg(0) may occur and result in subsequent re-emission. This undesirable process can be controlled with the help of a sulfidedonating liquid reagent.[34] They reported that limiting FGD scrubber chemistry and reemission of mercury may negatively affect Hg 2+ capture resulting in Hg 2+ removal of less than 90%. Mercury Capture in Dry FGD Systems A document released by EPA reports that dry FGD system combined with FF exhibited an efficient mercury removal an average of 95%, especially for bituminous coal.[18] Mercury mostly in the form of Hg 2+ at the inlet of the SDA with bituminous coals is captured in the filter cake of the FF. However, mercury capture in FF/SDA systems tends to be much less in low-rank coals. For low-rank coals, (e.g. subbituminous, and lignite) the low capture of mercury by FF/SDA systems is believed to be a result of the scrubbing of HCl in the SDA, which makes oxidation and capture of mercury (mostly in the form of Hg(0) for these coals) in the downstream FF less effective. In fact, Figure 2.3 shows higher mercury capture by FF when firing subbituminous coal than mercury capture by FF/SDA. This is believed to be a result of the SDA scrubbing effect in removing HCl that could otherwise be available to react on the FF.[18] Summary of Mercury Control by Enhancing Multi-pollutant Control The average mercury capture results by existing post-combustion control configurations used for pulverized-coal-fired boilers are shown in Table 2.3. It can be summarized from the Table 2.3 that, the amount of Hg captured by a given control technology is greater for bituminous coal than for either subbituminous coal or lignite. For example, the average capture 23

40 of Hg in plants equipped with a CS-ESP is 36% for bituminous coal, 3% for subbituminous coal, and 0% for lignite. Table 2.3. Average mercury capture by existing post-combustion control configurations used for pulverized-coal-fired boilers [35] Post-combustion control strategy Post-combustion emission control device configuration Average percentage mercury capture, % Bituminous Subbituminous Lignite PM control only CS-ESP HS-ESP 9 6 N/A FF N/A PS N/A 9 N/A PM SDA+CS-ESP N/A 35 N/A control+spray SDA+FF dryer adsorber SDA+FF+SCR 98 N/A N/A PM control+wet PS+FGD FGD system CS-ESP+FGD HS-ESP+FGD N/A FF+FGD 98 N/A N/A CS-ESP = cold-side electrostatic precipitator HS-ESP = hot-side electrostatic precipitator FF = fabric filter PS = particle scrubber SDA = spray dryer absorber system 24

41 While both industry and academia are still in the process of understanding how much mercury will be removed by these control devices on a consistent basis, it is generally estimated that the reductions in mercury from these devices are between 60-80%. Plants that employ only PM controls experienced average Hg emission reductions ranging from 0-90%. Units with FFs obtained the highest average levels of control. Decreasing average levels of control were generally observed for units equipped with a CS-ESP, HS-ESP, and PS. For units equipped with dry scrubbers, the average Hg emission reductions ranged from 0-98%. The estimated average reductions for FGD scrubbers were similar and ranged from 0-98%. 2.4 Emerging Mercury Control Technologies Coal-fired power plant mercury control technologies described in 2.3 are sometimes called as native capture, which means mercury capture without add-on mercury-specific control technology.[18] Unlike mercury removal by existing APCDs, where mercury capture was achieved as a co-benefit with removal of other pollutants, mercury control via add-on specific control devices are under development Sorbent Injection Dry sorbent may be injected into the ductwork upstream of a PM control device normally either an ESP or FF. Alternatively, powdered sorbent injection approaches may also be employed in combination with existing SO 2 control devices. For example, powdered sorbent may be injected prior to the SO 2 control device or after the SO 2 control device, subject to the availability of a means to collect the powdered sorbent downstream of the injection point. There are many factors that can impact the performance of the sorbents, such as properties of sorbents, 25

42 amount and method of injecting sorbent concentration, flue gas conditions, APCD configuration. These factors are briefly discussed in the following sub-sections. Mercury Control by Conventional AC Injection The most widely tested sorbent for mercury control at utility boilers is powdered activated carbon (AC). AC has numerous applications in removing pollutants from air or water streams both in the field and in industrial processes. When it comes to the materials for spill cleanup, water treatment, air purification, volatile organic compounds capture from painting, dry cleaning and gasoline dispensing operations, AC is often one of the efficient commercial goods to be chosen. Other powdered sorbent materials such as enhanced AC and silica-based sorbents, which are specifically formulated for controlling mercury emissions from coal-fired power plant are also proposed and tested. Commercial AC manufactures such as Norit (Texas, USA), Rheinbraun Brennstoff GmbH (Cologne, Germany) can provide tailored products for specific mercury adsorption. For example, Norit DARCO FGD is a lignite coal-based AC manufactured specifically for the removal of heavy metals and other contaminants typically found in incinerator flue gas emission streams.[36] Activated Lignite HOK is produced on the basis of Rhenish lignite in the so-called rotary hearth furnace process. The special properties of Rhenish lignite, coupled with the activation conditions in the rotary hearth furnace, yield Activated Lignite HOK which has been used as low-cost adsorption in waste gas and effluent treatment.[37] The Energy & Environmental Research Center (EERC) at University of North Dakota has pursued a research program (sponsored by US DOE/NETL) for producing AC from North Dakota lignite that can be competitive with commercial grade AC. It was reported that the AC produced from North Dakota lignite was superior to commercial grade DARCO FGD and 26

43 Rheinbraun s HOK activated coke product with respect to iodine number. The iodine number of North Dakota lignite-derived AC was between 600 and 800 mg I 2 /g, whereas the iodine number of DARCO FGD was between 500 and 600 mg I 2 /g, and the iodine number of Rheinbraun s HOK activated coke product was around 275 mg I 2 /g.[38] Norit later developed DARCO HG which is also a lignite coal-based AC but specifically for the removal of mercury in coal fired utility flue gas emission streams.[39] It's open pore structure and fine particle size allow fast adsorption, which is critical for high removal efficiency in flue gas streams where contact times are short. According to Norit, DARCO HG has been proven in numerous full scale operating facilities to be highly effective for the removal of mercury. The U.S. Department of Energy/National Energy Technology Laboratory (DOE/NETL), the Electric Power Research Institute (EPRI), and many utility companies have sponsored a number of field tests to evaluate the use of powdered sorbent, especially powdered AC, on capture of mercury from power plants. Table 2.4 shows the test programs that have evaluated standard AC injection for mercury control, i.e. the Hg removal via AC injection is measured between the inlet and outlet of the PM control device (ESP). 27

44 Table 2.4. Speciated mercury results from AC Injection field tests[18] Particulate Elemental Oxidized Total (μg/dnm3) (μg/dnm3) (μg/dnm3) (μg/dnm3) Pleasant Prairie (Baseline) ESP Outlet ESP Outlet Removal Efficiency (%) * 5.27 Pleasant Prairie (ACI, 11 lbs/mmacf) ESP Inlet ESP Outlet Removal Efficiency (%) Salem Harbor (Baseline) ESP Inlet < < ESP Outlet < 0.34 < < 1.25 Removal Efficiency (%) > 96 * * ~ 88 Salem Harbor (ACI, 10 lbs/mmacf) ESP Inlet 4.9 < < 5.24 ESP Outlet < 0.09 < < 0.62 Removal Efficiency (%) > 98 * ~ 88 Gaston (Baseline) COHPAC Inlet COHPAC Outlet Removal Efficiency (%) * 6.25 Gaston (ACI injection, 1.5 lbs/mmacf) COHPAC Inlet COHPAC Outlet 0.1 < Removal Efficiency (%) 50 > * Efficiency calculation not appropriate because outlet value is greater than inlet value 28

45 At Pleasant Prairie (flue gas with more Hg(0)), ESP can only remove ~5.27% total mercury if there is no AC injection. With AC injection, a mercury removal efficiency of 73.0% was achieved. At Salem Harbor (with more Hg(p) in flue gas), ESP alone can achieve ~88% mercury removal efficiency. Addition of AC injection resulted in no significant improvement in total mercury removal efficiency. Worse, the Hg(0) concentration increased from 0.27 µg/nm 3 at ESP inlet to µg/nm 3 at ESP outlet, with/without AC injection. However, in another case, at Gaston field (with more Hg 2+ in flue gas) a total mercury removal efficiency of 90.3% was observed with even higher Hg(0) removal efficiency up to 97%. In addition, Hg 2+ concentrations at the outlet of PM controls were below 1 μg/nm 3 under sorbent injection conditions at all plants. Therefore, AC injection was quite efficient in decreasing Hg 2+ concentrations in flue gas at PM control outlets as seen in results from Pleasant Prairie and Gaston. It is fair to say that raw AC is a great sorbent to remove Hg(p) and Hg 2+, but is not efficient when the flue gas contains more Hg(0). Mercury Control by Halogenated AC Injection Nelson et al. [40] reported that halogenated AC sorbents can overcome some of the limitations associated with AC injection for mercury control in power plant applications. Halogenated ACs offer several potential benefits. It may expand the usefulness of sorbent injection to many situations where standard AC may not be as effective. Halogenated ACs can be injected at lower injection rates, which potentially lead to fewer plant impacts and a lower carbon content in the captured fly ash. Brominated-AC may result in better performance with low-sulfur (including low-rank) coals because of less competition from SO 3. 29

46 Devolatilized Char Sorbents Other lower cost sorbents and additives are also under development. The devolatilized char from the furnace has been tested as a low-cost sorbent or oxidizing catalyst.[41] This carbon, known as the Thief Process, was tested in a TOXECON TM slipstream at the Pleasant Prairie Unit 1. The mercury reduction provided by the Thief carbon was not as high as for conventional AC injected at the same rate, however, if the cost is significantly less it may justify the higher injection rate. Several other low-cost sorbents such as corn char or treated fly ash, which may provide cost-effective benefits in some cases, have also been investigated. [42] The low-cost sorbents can also be used in combination with an additive to enhance sorbent capacity. Since chlorine plays an important role in facilitating the capture of Hg(0) on sorbent, lowrank coals may not have adequate chlorine to achieve high Hg(0) capture efficiency with AC. One of the solutions to this issue is to inject a chemical into the fuel (or into the gas stream), to provide the halogens for high sorbent capacity. Nanostructured Chelating Sorbents Previously, Abu Daabes, et al. developed a nanostructured chelating adsorbent for the removal of gaseous oxidized mercury from flue gas.[43] This cysteine-based chelating adsorbent showed high oxidized mercury uptake, capturing as much as 30 mg Hg/g adsorbent in a fixed-bed experiments under simulated flue gas conditions. It is reported that multiple bonds (covalent bonds and electrostatic bonds) formed among mercury and other atoms such as nitrogen and sulfur, resulting in a very strong binding. Ji, et al. [44] proposed that an ionic liquid nano-coating can ionize the captured vapor phase mercury, hence the mercury ion can be captured by the chelating ligands. They proved that the ionization of the vapor phase mercury is the first and key step for the mercury capture 30

47 process. It was found that the room temperature ionic liquid (RTIL), e.g. methyl polyoxyethylene octadecan ammonium chloride (MEC), with a vapor pressure of <0.01 kpa at 20ºC, is a good solvent for HgCl2. There is a disadvantage, however, for chelating sorbents serving as mercury capture materials in a practical scale the cost of such chelating sorbents. Given the difficulty of recycle, it would be hard for this sorbent to find its way into commercial applications Catalyst-enhanced Elemental Mercury Oxidation As discussed previously, conversion of Hg(0) to Hg 2+ is essential for mercury capture because FGD cannot remove Hg(0), but easily removes Hg 2+ due to its solubility in water. Therefore, Hg(0) oxidation and oxidation catalysts during the pollutant control processes have been one of the critical topics in the R&D of mercury control technologies. A variety of potential mercury oxidation catalysts have been tested under experimental conditions ranging from laboratory-scale packed beds using simulated flue gas to full scale tests. Generally, the research on mercury oxidation to date has focused on three broad catalyst areas: (1) SCR catalysts; (2) Carbon-based materials; (3) Metals and metal oxides. Each of these groups of materials has its relative merits and shortcomings, and none has emerged as a clear favorite in terms of either mercury conversion efficiency or economic viability. Thus, research into each of the three catalyst groups remains active. Selective catalytic reduction (SCR) catalysts The SCR catalysts are originally used to reduce the NOx concentration in flue gas. SCR catalysts usually operate at temperatures above 300ºC. Studies showed that SCR catalysts oxidize Hg(0) to Hg 2+, particularly in the presence of HCl.[45] Eswaran et al. also observed 31

48 mercury oxidation in the presence of H2SO4, presumably to HgSO4. [46] The effect of SCR catalysts on mercury oxidation in full-scale power plants have been investigated by several authors.[47, 48] For example, Benson et al. tested the performance of SCR catalysts for mercury oxidation at power plants burning subbituminous and lignite coals.[47] They observed that alkali and alkaline species can reduce the effectiveness of SCR catalysts for mercury oxidation, likely because these species deposited on the catalyst and reacted with acid sites on the catalyst surface. They also pointed out that fly ash might block the pores of the catalyst and reduce both NO and Hg(0) conversion. Carbon-Based Catalysts As the sorbents, ACs can remove a number of different species from flue gas including NO,[49, 50] SO 2,[51, 52] and HCl.[53, 54] However, activated carbon (AC) injection is also reported to be functional for removing both Hg(0) and Hg 2+ from flue gas. [55, 56] Early studies showed that the unburned carbon (UBC) in fly ash can adsorb Hg(0). [32, 57-59] While raw AC is believed to have little capability to absorb Hg(0), further investigation found that Hg(0) in the flue gas can be somehow converted into Hg 2+, and the extent of mercury oxidation in flue gas depends upon the amount of UBC present in fly ash.[58, 60] Fly ash has been considered as a convenient, low-cost material to be employed as the mercury oxidation catalyst. According to the research conducted by several groups, mercury oxidation on fly ash particles is believed to take place at carbon sites in the ash with various impurities.[57-61] Although the surface reaction to form HgCl 2 is widely accepted to be the mercury oxidation pathway, there are many competing factors in play. For example, Kellie et al. [62] reported that higher coal-cl, which generates higher HCl concentrations in flue gas, favors 32

49 greater formation of Hg 2+. Galbreath, et al. observed that in the presence of fly ash increasing the concentration of HCl in simulated flue gas caused an increase in Hg 2+.[63] Metal and Metal Oxide Catalysts Since SCR reactors operate at high temperatures, and with the mercury species in the flow stream containing more Hg(0), high conversion of Hg(0) to Hg2+ poses a challenge, especially when a number of other contaminants (e.g. SO 2, NH 3, fly ash, etc. ) are present in the flow stream. In addition to SCR catalysts (usually V 2 O 5 ), other metals and metal oxides have been investigated as potential mercury oxidation catalysts. As with the SCR and carbon-based catalysts discussed above, the reaction mechanisms for the metal and metal oxide catalysts are uncertain and still being investigated. Langmuir-Hinshelwood, Eley-Rideal (with either Hg(0) or HCl as the adsorbed species), Mars-Maessen, and the Deacon process have all been proposed to explain the possible reaction mechanisms. It is understandable that there is no single mechanism can be universally applied to mercury capture under all conditions. Those successful mechanisms are usually limited to the mercury oxidation under specific conditions, such as catalytic active components, substrates used to support the metal or metal oxide, specific composition in flue gas, and operation conditions. Noble metals, including copper, gold, silver, and palladium, have been tested as potential mercury oxidation catalysts. Ghorishi et al. [64] prepared a model fly ash containing CuO that oxidized >90% of the Hg(0) present in simulated flue gas containing HCl at 250 o C. However, only 10% oxidation was achieved when the CuO-free fly ash was tested. The same study found that CuCl, present in a model fly ash, was reactive enough to oxidize Hg(0) even in the absence of gas-phase HCl. This may suggest a Mars-Maessen reaction. 33

50 Zhao et al. reported 40-60% Hg(0) oxidation across a gold catalyst in the presence of Cl 2 at o C.[65] Interestingly, they observed that the presence of HCl reduced Hg(0) oxidation relative to Cl 2 alone. Palladium catalysts were also tested for Hg(0) oxidation by several groups. Initial studies of palladium catalysts showed less than 30% mercury oxidation at ~150 o C.[66] Several tests of palladium catalysts at sites burning lignite, [67] subbituminous, [68] and bituminous [69] coals have shown >90% conversion of Hg(0) to Hg 2+ for short (3-9 day) tests. However, the tests also showed a sharp decline in catalyst activity due to ash buildup on the surface. Some researchers investigated iron and its oxides as the Hg(0) oxidation catalysts. [17, 64] Dunham, et al. reported that the extent of mercury oxidation in the presence of fly ash increased with the magnetite (Fe 3 O 4 ) content of the ash at 120 and 180 o C.[70] It was suggested that simply injecting Fe 2 O 3 into flue gas may not play the same role of Fe 2 O 3 present in fly ash Electro-Catalytic Oxidation (ECO) Electro-Catalytic Oxidation (ECO) technology was first proposed and tested by Powerspan Corp.[71] It is an integrated air pollution control technology that can be installed in conventional coal-fired utility plants for the control of oxides of sulfur and nitrogen. The ECO process involves a high-energy oxidation reactor followed by an ammonia-based scrubber and a wet ESP. Once the flue gas enters the ECO oxidation reactor, hydroxyl radicals and atomic oxygen are formed that oxidize pollutants in the gas stream to produce water-soluble compounds and aerosols. The flue gas then enters an ammonia-based scrubber where the gas is saturated and cooled while being scrubbed of SO2 and NO2. The absorber vessel is similar to a wet SO2 34

51 scrubber except that it operates at a high ph and utilizes a smaller tower. The flue gas enters a wet ESP after exiting the absorber vessel; the aerosols and fine PM are captured in the wet ESP. The technology can be installed downstream of a facility s existing APCD such as dry ESP or FF. Pilot tests have shown that this technology can reduce SO2 emissions by 98%, NOx emissions by 90%, and mercury emissions by 80-90%. However, high energy use (up to 5% of electricity generated at the site) poses an economics issue of this technology in field applications Photochemical oxidation technology (PCO) Photochemical oxidation technology (PCO) technology, is a mercury control process developed by the DOE/NETL and was licensed to Powerspan Corp in PCO technology uses ultraviolet (UV) light to oxidize and remove mercury. The flue gas is irradiated with UV light at a wavelength of 254 nm to convert elemental mercury to oxidized forms (such as mercuric oxide, mercurous sulfate, and mercurous chloride). Mercury can be then collected in an existing APCD, such as a wet scrubber, ESP, and FF. Compared to conventional AC injection technology, the major advantage of PCO is that the power plant need not inject sorbents and therefore incurs no associated costs for disposal of used sorbents and possible lost resale value of the fly ash. The primary operating cost component for PCO is power consumption. It was estimated that less than 0.35% of plant power will be consumed for the 90% mercury removal with the PCO technology. The authors suggested that the process also has potential to serve as a low cost method for the removal of mercury from waste incinerator flue gases.[72] 35

52 A follow-up of the open publications relating to the applications of PCO technology finds no further progress of this technology, probably because of the operation cost, and/or feasibility in the real world situation. 2.5 Mechanism of Mercury Oxidation The mechanisms of mercury oxidation have been under investigated for many years. Mechanisms that have been proposed are usually based on some specific conditions, such as oxidation catalysts (including active sites and supports), species involved in oxidation (components in flue gas), and reaction conditions (temperature and partial pressure). In this section, a review of the mercury oxidation mechanism is presented. Homogeneous Oxidation Early studies showed that oxidized mercury species may be formed through gas-phase reactions between elemental mercury and halogens, such as Cl 2, HCl, chlorine radicals.[73-75] Sliger et al. proposed that Hg(0) oxidation occurs primarily via reaction with chlorine radicals between 400 and 700 C.[73] In this temperature range Hg+Cl reaction has a low energy barrier and occurs near the collision limit at room temperature; reaction with Cl is therefore much faster than Hg+HCl, which has a high energy barrier.[76] Lee et al. investigated the effect of chlorine species (HCl, CH 2 Cl 2 ) on the in situ Hg capture method.[74] They found that under certain conditions, the presence of chlorine enhanced the removal of Hg(0) by additional gas-phase oxidation. Although Cl 2 gas is a known strong Hg(0) oxidant (stronger than HCl gas), and forms HgCl 2 via homogeneous reactions in the gas phase, it was reported that the homogeneous Hg(0) oxidation using Cl 2 gas can be significantly degraded in the presence of SO 2.[77, 78] The modeling study by Niksa and 36

53 Fujiwara indicated that gas-phase reactions alone are not enough to account for observed extents of mercury oxidation.[58] Studies by Wang et al. revealed that the homogeneous reaction Hg + Cl 2 is too slow to generate significant Hg(0) conversion.[79] They found that the large discrepancies between published rate constants for the reaction were very likely the result of heterogeneous mercury oxidation occurring on the reactor walls. Their findings were also supported by the observation that, in addition to coal-cl, the extent of mercury oxidation is affected by loss on ignition [59] and the presence of a bag house or fabric filter.[80] A recent paper by Van Otten et al. reaffirmed that gas phase oxidation of mercury is negligible.[81] Therefore, people tend to agree that, at typical flue gas temperatures, heterogeneous oxidation of mercury dominates. Heterogeneous Oxidation Several mechanisms have been proposed for heterogeneous mercury oxidation, although none of these mechanisms is able to perfectly describe the reaction. For example, AC impregnated with halogen compounds showed both mercury adsorption and oxidation capabilities. Oslon, et al. proposed that activated carbon has two or more reaction sites to remove mercury and used the concept of zigzag carbene structures to explain the nature of the carbon site.[82] In another theoretical study, Padak et al. hypothesized that there exists Lewis base sites in the carbene structures, which might interact with Hg 2+.[83] These basic sites cannot bind Hg(0) directly but become active after reacting with HCl to form carbenium ion as shown in Figure 2.4.[84] 37

54 Figure 2.4 Mercury adsorption/oxidation mechanisms proposed by basic site model [84] Granite et al. proposed that elemental mercury oxidation could occur via a Mars-Maessen mechanism.[55] On oxide catalysts, for example, the reaction mechanism for the capture of elemental mercury in the absence of oxygen by the MnO X/TiO2 catalyst/sorbent can be written as: Hg(g) + surface Hg(ads) (2.1) Hg(ads) + MnO2 HgO(ads) + MnO (2.2) HgO(ads) + MnO2 HgMnO3 (2.3) The observation of mercury oxidation in the absence of oxygen would support this mechanism.[24] The mechanism can be described to follow the steps: (1) vapor phase elemental mercury collides with the surface of the catalyst/sorbent and is adsorbed; (2) adsorbed elemental mercury is oxidized to mercury oxide by the manganese dioxide catalyst/sorbent; (3) manganese dioxide is reduced to manganese oxide; (4) mercury oxide subsequently reacts with the catalyst/sorbent to form a binary oxide. 38

55 Schofield proposed a mechanism for the oxidation of Hg(0) to HgSO 4.[22] In a simulated flue gas containing SO 2 and Hg(0), HgSO 4 was observed to spontaneously deposit on stainless steel or platinum surfaces. In the absence of SO 2, HgO was observed to deposit. Adding HCl to the flame after deposit formation led to the removal of the deposit via reaction to HgCl 2 followed by sublimation. They therefore concluded that HgCl 2 formation in flue gases is preceded by surface reaction to form either HgO or HgSO 4, subsequently removed from the surface via reaction with HCl. Niksa et al. proposed that mercury oxidation occurs via an Eley-Rideal mechanism, where adsorbed HCl reacts with gas-phase (or weakly adsorbed) Hg(0) (reaction ).[58] Because HCl often has high gas-phase concentrations in flue gas, an Eley-Rideal reaction between adsorbed Hg(0) and gas-phase HCl is also a logical possibility, i.e. species A in Eq.(2.4) could be either Hg(0) or HCl. A(g) + surface A(ads) (2.4) A(ads) + B(g) AB(ads) (2.5) Lee, et al. reported that cupric chloride dispersed over carbon and clay substrates demonstrated excellent Hg(0) oxidation.[85-88] Carbon could effectively adsorb the resultant oxidized mercury, but a clay substrate could not adsorb it.[89] In addition, as the CuCl 2 loading on carbon increases, the adsorption capacity decreases.[87] These results indicate that Hg(0) vapor reacts with CuCl 2 and the desorbed resultant oxidized mercury is re-adsorbed onto substrate surfaces, suggesting different sites for Hg(0) oxidation and re-adsorption of the resultant oxidized mercury. Studies of cupric chloride (CuCl 2 )-impregnated materials also showed excellent Hg(0) oxidation performance in the absence of HCl throughout fixed-bed and entrained-flow tests. [85, 89] When CuCl 2 is dispersed over non-carbonaceous substrates, CuCl 2 39

56 readily oxidizes Hg(0) vapor, but its resultant oxidized mercury is not easily adsorbed onto the non-carbonaceous substrate surface. Carbon seems to be the only substrate that can effectively adsorb the oxidized mercury. However, a convincible characterization of the both fresh sorbents or catalysts by advanced technique is definitely necessary in order to understand the speciation of Hg(0). Several studies have been conducted on the speciation of critical trace elements including arsenic, chromium, and mercury adsorbed onto solid surfaces and mercury-solid surface interactions in order to investigate the speciation of individual elements.[54, 90-94] XAFS and XPS are the techniques that have been used to determine the information on the speciation and binding of mercury on the surfaces. Huggins et al. conducted an XAFS study on the sorption of Hg(0) and HgCl 2 vapors onto sulfurized, iodated, and raw activated carbons under simulated flue gas conditions.[91] The study results suggested that HgCl 2 vapor appeared to be captured by physi-sorption and Hg(0) vapor were captured by chemisorption by forming sorption complexes with the promoted chemical elements (S and I) as well as the acidic gases of HCl and SO 2 in the simulated flue gas. Additional XAFS analyses using various samples also indicated that mercury can be captured by bonding to I, Cl, S, or O anionic species on carbonaceous and noncarbonaceous surfaces.[92] An XPS study was conducted in order to determine the fate of Cl, S, and N species derived from flue gases on the surface of activated carbons.[54, 93] The study results suggested that Cl and S compete for adsorption at carbon sites with a higher adsorption preference of S primarily in the form of S(VI) in the presence of water vapor. More than one form of bound chlorine was found to be present as chloride and organic chlorine. Hutson et al. recently studied the speciation of bromine and mercury and their binding mechanisms on the carbon surface using XAFS and XPS with brominated ACs under simulated flue gas 40

57 conditions.[93] The study results indicated that bromine appears to be primarily covalently bonded to sp2 and sp3 carbon atoms with some physisorbed HBr. Hg(0) vapor then reacts with Br sites and the resultant oxidized mercury is most likely to be present as HgBr 2. To date, none of the above mechanisms has been verified as the dominant mechanism for catalytic mercury oxidation. Clearly, the fundamental aspects of heterogeneous Hg(0) oxidation need to be investigated in order to improve our understanding of the reaction and enhance our ability to predict the extent of mercury oxidation. 2.6 Kinetics Modeling of Mercury Oxidation Regardless of the dominant mechanism for mercury oxidation, it is well-known that the transformation from Hg(0) to Hg 2+ in flue gas is kinetically limited. For temperatures below ~450 C, at equilibrium, nearly all mercury should exist as Hg 2+. [95, 96] Due to the excess of chlorine-containing species (HCl and Cl 2 ), HgCl 2 is assumed to be the dominant form of Hg 2+. However, sulfate,[55] nitrate,[72] and mercuric oxide,[97] may also be formed under corresponding conditions. Obviously, the systematic knowledge of adsorption kinetics and heterogeneous mercury oxidation kinetics play a major role in the determination of the overall mercury removal, hence providing valuable information in the prediction and scaling of mercury removal processes. Kinetic studies such as the adsorption/reaction rate, adsorption/reaction constant, activation energy, etc. have been so far less conducted. With the current available data, predicting the mercury removal efficiency based on specific sorbent, or catalyst under various operating conditions is difficult. Even the data obtained before are often incomplete and sometimes contradictory. In the case of coal-fired power plants, the control of trace mercury 41

58 vapor in flue gas at a ppb level requires more fundamental understanding than the control systems of the past for pollutants at a ppm level. A more complete understanding of the reaction mechanism and kinetics will help identify candidate catalyst materials because the knowledge of the reaction kinetics can offer the predictability for the design of the catalyst beds and sorbent injection rates. Krishnan et al. studied the adsorption of elemental mercury on two kinds of thermally activated carbons and on one kind of sulfur impregnated carbon.[98] They found that the surface area has a strong effect on the adsorption capacity of all the sorbents tested. The adsorption capacity of thermal ACs decreased when temperature increased from room temperature to 140 o C. However, the sulfur impregnated carbon was temperature insensitive. Karatza et al. studied the capture of HgCl 2 by activated carbon, considered the adsorption capacity and the removal rate, and to determine reaction kinetics and adsorption kinetics/equilibrium.[99] The study was performed in a fixed-bed at laboratory scale in which simulated gas containing Hg(0) was contacted with adsorbent material. A model based on the assumption of kinetic control was used to evaluate the kinetic parameters of the process, which confirmed that activated carbon is a suitable sorbent for mercuric chloride. Ho et al. conducted an experimental and kinetic study of mercury adsorption on various activated carbons in a fixed-bed adsorber.[100] They proposed a sorption model combining a kinetic model based on the mechanisms of surface equilibrium and external mass transfer, and a material balance model based on the tank-in-series approach. Three different equilibrium expressions were used in the model, i.e., the Henry's Law, the Langmuir isotherm, and the Freundlich isotherm. In additional to kinetic simulations using the developed model, an equilibrium model was also used to simulate the thermodynamically preferred mercury species under the experimental conditions. Their experimental results indicated that the 42

59 factors affecting the adsorption efficiency include the type of activated carbon, the adsorption temperature, the inlet mercury concentration, and the gas flow rate. The developed kinetic model was found to describe well the current experimental results and those reported in the literature. For the mercury removal by catalytic oxidation, the surface-catalyzed Hg(0) oxidation kinetics has not been studied in depth. Presto, et al. reported a kinetic approach to the catalytic oxidation of mercury in flue gas.[24] They proposed a method for analyzing mercury oxidation catalyst results in a kinetic framework using the bulk reaction rate for oxidized mercury formation normalized by either the catalyst mass or surface area. Their results showed that mercury oxidation was strongly influenced by the specific experimental conditions and therefore was difficult to be used to guide the experiments under very different conditions. The difficulty in mercury oxidation kinetics investigation is not only because of lack of understanding in mechanism, but also because of the lack of accuracy in obtaining kinetic parameters due to the very low (i.e. in ppbv level) reactant concentration. It is well known that low mercury concentrations on the order of ~1 ppbv in the gas phase generates significant external mass-transfer resistance, and make mercury emissions control from coal combustion flue gases very difficult.[79] Such low concentrations of air pollutants can also cause internal mass-transfer resistances for sorbent injection particles.[58, 79, 101] It was also observed from our previous sorbent injection tests using our entrained-flow system, that both external and internal mass-transfer resistances could be significant.[86, 102] In this sense, an investigation of kinetic models considering external and internal masstransfer resistance would be helpful to create a predictive model that would allow for efficient scaling up from laboratory-scale to larger-scale studies. 43

60 2.7 Summary In summary, mercury emission control is of great importance in terms of environment protection, as well as public health. Current mercury emission control technologies cannot meet the stringent requirement as regulated by the latest EPA air pollutant control rules, mainly because of lack of fundamental understanding of adsorption and/or catalytic mechanisms and necessary kinetic modeling and reliable simulation data. The work reported in this dissertation aims to advance the efforts towards establishing the fundamental mechanistic understanding in heterogeneous catalytic oxidation reaction and adsorption by using the reaction between Hg(0) vapor and CuCl 2, and subsequent adsorption of resultant oxidized mercury (Hg 2+ ) onto sorbents or catalysts. The research work in this study will help us to fundamentally understand the speciation of mercury and copper from CuCl 2 - impregnated sorbents and catalysts to elucidate the adsorption and oxidation pathway. The outcome of this research is expected to provide systematic and scientific information on reaction and adsorption of mercury in flue gas for its control and speciation studies. 2.8 Reference 1. Bragg, L. J., Oman, J.K., Tewalt, S.J., Oman, C.J., Rega, N.H., Washington, P.M., Finkelman, R.B., Analytical data, sample locations, and descriptive information, analytical methods and sampling techniques, database perspective, and bibliographic references for selected U.S. coal samples, U.S. GEOLOGICAL SURVEY OPEN-FILE REPORT, In. 2. Toole-O'Neil, B.; Tewalt, S. J.; Finkelman, R. B.; Akers, D. J., Mercury concentration in coal unraveling the puzzle. Fuel 1999, 78, (1),

61 3. Stiegel, G. J. L., J. R.; Klett, M. G.; Maxwell, R. C.; Rutkowski, M. D., The cost of mercury removal in an IGCC plant. Final Report,. al.pdf 4. Wiedinmyer, C.; Friedli, H., Mercury Emission Estimates from Fires: An Initial Inventory for the United States. Environmental Science & Technology 2007, 41, (23), "National Emission Standards for Hazardous Air Pollutants From Coal- and Oil-Fired Electric Utility Steam Generating Units and Standards of Performance for Fossil-Fuel-Fired Electric Utility, Industrial-Commercial-Institutional, and Small Industrial-Commercial- Institutional Steam Generating Units; Proposed Rule." 40 CFR Parts 60 and 63. Federal Register 76:85 (May 1, 2011) p EPA, Integrated Planning Model (IPM) v.4.10 Model Runs Galbreath, K. C.; Zygarlicke, C. J., Mercury Speciation in Coal Combustion and Gasification Flue Gases. Environmental Science & Technology 1996, 30, (8), Senior, C. L.; Sarofim, A. F.; Zeng, T. F.; Helble, J. J.; Mamani-Paco, R., Gas-phase transformations of mercury in coal-fired power plants. Fuel Processing Technology 2000, 63, (2-3), Yudovich, Y. E.; Ketris, M. P., Mercury in coal: a review Part 2. Coal use and environmental problems. International Journal of Coal Geology 2005, 62, (3), Meij, R.; Vredenbregt, L. H. J.; Winkel, H. t., The Fate and Behavior of Mercury in Coal- Fired Power Plants. Journal of the Air & Waste Management Association 2002, 52, (8), Park, K. S.; Seo, Y. C.; Lee, S. J.; Lee, J. H., Emission and speciation of mercury from various combustion sources. Powder Technology 2008, 180, (1 2), Zhuang, Y.; Zygarlicke, C. J.; Galbreath, K. C.; Thompson, J. S.; Holmes, M. J.; Pavlish, J. H., Kinetic transformation of mercury in coal combustion flue gas in a bench-scale entrainedflow reactor. Fuel Processing Technology 2004, 85, (6 7), Kilgroe, J. D. S., C. B.; Srivastava, R. K.; Ryan, J. V.; Lee, C. W.; Thorneloe, S. A.,, Control of mercury emissions from coal-fired electric utility boilers: interim report. In EPA- 45

62 600/R-01/109; U.S. Environmental Protection Agency, Ed. National Risk Management Research Laboratory: Research Triangle Park, NC, 2001, Cao, Y.; Duan, Y.; Kellie, S.; Li, L.; Xu, W.; Riley, J. T.; Pan, W.-P.; Chu, P.; Mehta, A. K.; Carty, R., Impact of Coal Chlorine on Mercury Speciation and Emission from a 100-MW Utility Boiler with Cold-Side Electrostatic Precipitators and Low-NOx Burners. Energy & Fuels 2005, 19, (3), Bock, J., Hocquel, M.J.T., Unterberger, S., Hein, K.R.G., In Mercury Oxidation Across SCR Catalysts Of Flue Gas With Varying HCl Concentration, Mega Symposium and Air & Waste Management Association, Specialty Conference, Washington, DC, May 19-20, 2003, 2003; Washington, DC, May 19-20, 2003, Lee, C. W., Srivastava, R. K., Ghorishi, S. B., Hastings, T. W., Stevens, F. M. In Study of Speciation of Mercury under Simulated SCR NOx Emission Control Conditions,Specialty Conference,, Mega Symposium and Air & Waste Management Association,, Washington, DC, May 19-20, 2003, 2003; Washington, DC, May 19-20, 2003, Pavlish, J. H.; Sondreal, E. A.; Mann, M. D.; Olson, E. S.; Galbreath, K. C.; Laudal, D. L.; Benson, S. A., Status review of mercury control options for coal-fired power plants. Fuel Processing Technology 2003, 82, (2 3), EPA/600/R-10/006, Control of Mercury Emissions from Coal Fired Electric Utility Boilers: An Update, Feb, 2010, In. 19. Pavlish, J. H.; Holmes, M. J.; Benson, S. A.; Crocker, C. R.; Galbreath, K. C., Application of sorbents for mercury control for utilities burning lignite coal. Fuel Processing Technology 2004, 85, (6 7), DOE, U. S. Annual Energy Outlook 2007 with Projections to 2030; Report No. DOE/EIA-0383(2007); Energy Information Administration.: In. 21. National Emission Standards for Hazardous Air Pollutants From Coal- and Oil-Fired Electric Utility Steam Generating Units and Standards of Performance for Fossil-Fuel-Fired Electric Utility, Industrial-Commercial-Institutional, and Small Industrial-Commercial- Institutional Steam Generating Units. 40 CFR Parts 60 and 63. Federal Register 77:32 (Feb 16, 2012) p

63 22. Schofield, K., Fuel-Mercury Combustion Emissions: An Important Heterogeneous Mechanism and an Overall Review of its Implications. Environmental Science & Technology 2008, 42, (24), Lide, D. R., CRC Handbook Chemistry and Physics, 85th Edition CRC Press: Boca Raton, FL: Presto, A. A.; Granite, E. J., Survey of Catalysts for Oxidation of Mercury in Flue Gas. Environmental Science & Technology 2006, 40, (18), Kamata, H.; Ueno, S.-i.; Naito, T.; Yukimura, A., Mercury Oxidation over the V2O5(WO3)/TiO2 Commercial SCR Catalyst. Industrial & Engineering Chemistry Research 2008, 47, (21), He, S.; Zhou, J.; Zhu, Y.; Luo, Z.; Ni, M.; Cen, K., Mercury Oxidation over a Vanadiabased Selective Catalytic Reduction Catalyst. Energy & Fuels 2008, 23, (1), Chu, P., Laudal, D., Brickett, L., Lee, C.W., Power Plant Evaluation of the Effect of SCR Technology on Mercury,. In Joint EPRI DOE EPA Combined Utility Air Pollution Control Symposium, The Mega Symposium,, Washington, D.C., May 19-22, Winberg, S., Winthum, J., Tseng, S., Locke, J., In Evaluation of Mercury Emissions from Coal-Fired Facilities with SCR-FGD Systems,, DOE/NETL Mercury Control Technology R&D Program Review,, Pittsburgh, PA, July 14-15, 2004, Pittsburgh, PA, July 14-15, Renninger, S., Farthing, G., Ghorishi, S.B., Teets, C., Neureuter, J., In Effects of SCR Catalyst, Ammonia Injection and Sodium Hydrosulfide on the Speciation and Removal of Mercury within a Forced-Oxidized Limestone Scrubber,, Joint EPRI DOE EPA Combined Utility Air Pollution Control Symposium, The Mega Symposium,, Washington, D.C., August 30-September 2, 2004, Washington, D.C., August 30-September 2, Senior, C. L., and Linjewile, T., In Oxidation of Mercury Across SCR Catalysts in Coal- Fired Power Plants,, DOE/NETL Mercury Control Technology R&D Program Review,, Pittsburgh, PA, July 14-15, 2004, Pittsburgh, PA, July 14-15, Sakulpitakphon, T.; Hower, J. C.; Trimble, A. S.; Schram, W. H.; Thomas, G. A., Mercury Capture by Fly Ash: Study of the Combustion of a High-Mercury Coal at a Utility Boiler. Energy & Fuels 2000, 14, (3), Senior, C. L.; Johnson, S. A., Impact of Carbon-in-Ash on Mercury Removal across Particulate Control Devices in Coal-Fired Power Plants. Energy & Fuels 2005, 19, (3),

64 33. Niksa, S., Fujiwara, N., In The Impact of Wet FGD Scrubbing On Hg Emissions From Coal-Fired Power Stations,, Joint EPRI DOE EPA Combined Utility Air Pollution Control Symposium, The Mega Symposium,, Washington, D.C., August 30-September 2, 2004, Washington, D.C., August 30-September 2, EPA-600/R , Control of Mercury Emissions from Coal-Fired Electric Utility Boilers: Interim Report,. In Air Pollution Prevention and Control Division, N. R. M. R. L., Office of Research and Development,, Ed. Research Triangle Park, NC, April U.S. EPA, Control of Mercury Emissions from Coal-Fired Electric Utility Boilers; EPA Air Pollution Prevention and Control Division,. In National Risk Management Research Laboratory, Ed. ORD: Research Triangle Park, NC, Norit Darco-FGD Specifications. (March 12, 2012), 37. Activated Lignite HOK Specifications. (March 12, 2012), 38. JV Task 90 - Activated Carbon Production from North Dakota Lignite, Final Report for DOE/NETL, EERC Report No EERC-03-08, Norit Darco-Hg Specifications. (March 12, 2012), 40. Nelson, S., Landreth, R., Zhou, Q., Miller, J. In Accumulated Power-Plant Mercury- Removal Experience with Brominated PAC Injection,, Joint EPRI DOE EPA Combined Utility Air Pollution Control Symposium, The Mega Symposium,, Washington, D.C., August 30- September 2, 2004, Washington, D.C., August 30-September 2, Freeman, M., Pennline, H., Granite, E., Hargris, R., O Dowd, W.,, A Technique to Control Mercury From Flue Gas: The Thief Process,. DOE/NETL Mercury Control Technology R&D Program Review,. 42. Ley, T., Ebner, T., Fisher, K., Slye, R., Patton, R., Chang, R., In Assessment of Low-Cost Novel Sorbents for Coal-Fired Power Plant Mercury Control,, Joint EPRI DOE EPA Combined Utility Air Pollution Control Symposium, The Mega Symposium,, Washington, D.C., August 30-September 2, 2004, Washington, D.C., August 30-September 2,

65 43. Abu-Daabes, M. A.; Pinto, N. G., Synthesis and characterization of a nano-structured sorbent for the direct removal of mercury vapor from flue gases by chelation. Chemical Engineering Science 2005, 60, (7), Ji, L. Novel Nano-Structured Sorbents for Elemental and Oxidized Mercury Removal from Flue Gas,. PhD Dissertation,, University of Cincinnati, Lee, Y., Park, J., Kim, J., Min, Y., Jurng, J., Kim, J., Lee, T.,, Comparison of mercury removal efficiency from a simulated exhaust gas by several types of TiO2 under various light sources. Chem. Lett. 2004, 33, (1), Eswaran, S.; Stenger, H. G., Understanding Mercury Conversion in Selective Catalytic Reduction (SCR) Catalysts. Energy & Fuels 2005, 19, (6), Benson, S. A.; Laumb, J. D.; Crocker, C. R.; Pavlish, J. H., SCR catalyst performance in flue gases derived from subbituminous and lignite coals. Fuel Processing Technology 2005, 86, (5), Richardson, C.; Machalek, T.; Miller, S.; Dene, C.; Chang, R., Effect of NOx Control Processes on Mercury Speciation in Utility Flue Gas. Journal of the Air & Waste Management Association 2002, 52, (8), Tang, Q., Zhang, Z., Zhu, W., Cao, Z.,, SO2 and NO selective adsorption properties of coal-based activated carbons. Fuel 2005, 84, (4), Rubel, A. M.; Stencel, J. M., The effect of low-concentration SO2 on the adsorption of NO from gas over activated carbon. Fuel 1997, 76, (6), Davini, P., SO2 adsorption by activated carbons with various burnoffs obtained from a bituminous coal. Carbon 2001, 39, (9), Lizzio, A. A.; DeBarr, J. A., Mechanism of SO2 Removal by Carbon. Energy & Fuels 1997, 11, (2), Mangun, C. L.; Benak, K. R.; Economy, J.; Foster, K. L., Surface chemistry, pore sizes and adsorption properties of activated carbon fibers and precursors treated with ammonia. Carbon 2001, 39, (12), Laumb, J. D.; Benson, S. A.; Olson, E. A., X-ray photoelectron spectroscopy analysis of mercury sorbent surface chemistry. Fuel Processing Technology 2004, 85, (6-7), Granite, E. J.; Pennline, H. W.; Hargis, R. A., Novel Sorbents for Mercury Removal from Flue Gas. Industrial & Engineering Chemistry Research 2000, 39, (4),

66 56. Carey, T. R.; Jr, O. W. H.; Richardson, C. F.; Chang, R.; Meserole, F. B., Factors Affecting Mercury Control in Utility Flue Gas Using Activated Carbon. Journal of the Air & Waste Management Association 1998, 48, (12), Serre, S. D.; Silcox, G. D., Adsorption of Elemental Mercury on the Residual Carbon in Coal Fly Ash. Industrial & Engineering Chemistry Research 2000, 39, (6), Niksa, S.; Fujiwara, N., Predicting extents of mercury oxidation in coal-derived flue gases. Journal of the Air & Waste Management Association 2005, 55, (7), Gibb, W. H.; Clarke, F.; Mehta, A. K., The fate of coal mercury during combustion. Fuel Processing Technology 2000, 65 66, (0), Fujiwara, N.; Fujita, Y.; Tomura, K.; Moritomi, H.; Tuji, T.; Takasu, S.; Niksa, S., Mercury transformations in the exhausts from lab-scale coal flames. Fuel 2002, 81, (16), Olson, E. S.; Miller, S. J.; Sharma, R. K.; Dunham, G. E.; Benson, S. A., Catalytic effects of carbon sorbents for mercury capture. Journal of Hazardous Materials 2000, 74, (1 2), Kellie, S.; Cao, Y.; Duan, Y.; Li, L.; Chu, P.; Mehta, A.; Carty, R.; Riley, J. T.; Pan, W.- P., Factors Affecting Mercury Speciation in a 100-MW Coal-Fired Boiler with Low-NOx Burners. Energy & Fuels 2005, 19, (3), Galbreath, K. C.; Zygarlicke, C. J., Mercury transformations in coal combustion flue gas. Fuel Processing Technology 2000, 65, Ghorishi, S. L., C.; Jozewicz, W.; Kilgroe, J., Effects of fly ash transition metal content and flue gas HCl/SO2 ratio on mercury speciation in waste combustion.. Environ. Eng. Sci. 2005, 22, (2), Zhao, Y.; Mann, M. D.; Pavlish, J. H.; Mibeck, B. A. F.; Dunham, G. E.; Olson, E. S., Application of Gold Catalyst for Mercury Oxidation by Chlorine. Environmental Science & Technology 2006, 40, (5), Hargrove, O. C., T.; Richardson, C.; Skarupa, R.; Meserole, F.; Rhudy, R.; Brown, T. Enhanced Control of Mercury and Other HAPs by Innovative Modifications to Wet FGD Processes, Report to DOE/FETC; U. S. Department of Energy Agreement No. DEAC22-95PC95260; Radian International: 1997;

67 67. Enhanced Control of Mercury by Wet Flue Gas Desulfurization Systems Site 1 Results, Report to U. S. DOE/NETL; U. S. Department of Energy Agreement No. DE-AC22-95PC95260; Radian International: Blythe, G. M., S.; Richardson, C.; Searcy, K. Enhanced Control of Mercury by Wet Flue Gas Desulfurization SystemssSite 2 Results, Report to U. S. DOE/NETL; U. S. Department of Energy Agreement No. DE-AC22-95PC95260; Radian International: 2000; Blythe, G. M., B.; Miller, S.; Richardson, C.; Richardson, M. Enhanced Control of Mercury by Wet Flue Gas Desulfurizations Site 3 Topical Report, Report to U. S.DOE/NETL;U. S. Department of Energy Agreement No. DE-AC22-95PC95260; URS Corporation: 2001; Dunham, G. E.; DeWall, R. A.; Senior, C. L., Fixed-bed studies of the interactions between mercury and coal combustion fly ash. Fuel Processing Technology 2003, 82, (2 3), Powerspan Corp, Multi-Pollutant Control Technologyfor Coal-Fired Power Plants McLarnon, C. R.; Granite, E. J.; Pennline, H. W., The PCO process for photochemical removal of mercury from flue gas. Fuel Processing Technology 2005, 87, (1), Sliger, R. N.; Kramlich, J. C.; Marinov, N. M., Towards the development of a chemical kinetic model for the homogeneous oxidation of mercury by chlorine species. Fuel Processing Technology 2000, 65, Lee, T. G.; Hedrick, E.; Biswas, P., Hg reactions in the presence of chlorine species: Homogeneous gas phase and heterogeneous gas-solid phase. Journal of the Air & Waste Management Association 2002, 52, (11), Hall, B.; Schager, P.; Lindqvist, O., Chemical Reactions of Mercury in Combustion Flue Gases. Water Air and Soil Pollution 1991, 56, Hranisavljevic, J.; Fontijn, A., Kinetics of Ground-State Cd Reactions with Cl2, O2, and HCl over Wide Temperature Ranges. The Journal of Physical Chemistry A 1997, 101, (12), Niksa, S.; Helble, J. J.; Fujiwara, N., Kinetic Modeling of Homogeneous Mercury Oxidation: The Importance of NO and H2O in Predicting Oxidation in Coal-Derived Systems. Environmental Science & Technology 2001, 35, (18),

68 78. Agarwal, H.; Stenger, H. G.; Wu, S.; Fan, Z., Effects of H2O, SO2, and NO on Homogeneous Hg Oxidation by Cl2. Energy & Fuels 2006, 20, (3), Wang, J. S.; Anthony, E. J., An analysis of the reaction rate for mercury vapor and chlorine. Chemical Engineering & Technology 2005, 28, (5), Amrhein, G. B., R.; Downs, W.; Holmes, M.; Kudlac, G.; Madden, D. Advanced Emissions Control Development Program, Phase III Final Report to U. S. DOE/FETC and OCDO; U. S. Department of Energy Agreement No. DE-FC22-94PC ; McDermott Technology:; Van Otten, B.; Buitrago, P. A.; Senior, C. L.; Silcox, G. D., Gas-Phase Oxidation of Mercury by Bromine and Chlorine in Flue Gas. Energy & Fuels 2011, 25, (8), Olson, E. S. L., J. D.; Benson, S. A.; Dunham, G. E.; Sharma, R. K.; Mibeck, B. A.; Miller, S. J.; Holmes, M. J.; Pavlish, J. H., Chemical mechanisms in mercury emission control technologies.. J. Phys. IV France 2003, 107, Padak, B.; Brunetti, M.; Lewis, A.; Wilcox, J., Mercury binding on activated carbon. Environmental Progress 2006, 25, (4), Olson, E. S., Mibeck, B. A. In Oxidation kinetics of mercury in flue gas, Prepr. Pap.-Am. Chem. Soc., Div. Fuel Chem. 50, 68-69, 2005; Lee, S. S.; Lee, J. Y.; Keener, T. C., Bench-Scale Studies of In-Duct Mercury Capture Using Cupric Chloride-impregnated Carbons. Environmental Science & Technology 2009, 43, (8), Lee, S.-S.; Lee, J.-Y.; Keener, T. C., The effect of methods of preparation on the performance of cupric chloride-impregnated sorbents for the removal of mercury from flue gases. Fuel 2009, 88, (10), Lee, S.-S.; Lee, J.-Y.; Khang, S.-J.; Keener, T. C., Modeling of Mercury Oxidation and Adsorption by Cupric Chloride-Impregnated Carbon Sorbents. Industrial & Engineering Chemistry Research 2009, 48, (19), Lee, S.-S.; Lee, J.-Y.; Keener, T. C., Novel sorbents for mercury emissions control from coal-fired power plants. Journal of the Chinese Institute of Chemical Engineers 2008, 39, (2),

69 89. Lee, J. Y.; Ju, Y. H.; Keener, T. C.; Varma, R. S., Development of cost-effective noncarbon sorbents for Hg-0 removal from coal-fired power plants. Environmental Science & Technology 2006, 40, (8), Huffman, G. P.; Huggins, F. E.; Shah, N.; Zhao, J.; Lu, F., Investigation of environmentally critical elements in coal and ash by XAFS spectroscopy. Fuel 1993, 72, (5), Huggins, F. E.; Huffman, G. P.; Dunham, G. E.; Senior, C. L., XAFS Examination of Mercury Sorption on Three Activated Carbons. Energy & Fuels 1998, 13, (1), Huggins, F. E.; Yap, N.; Huffman, G. P.; Senior, C. L., XAFS characterization of mercury captured from combustion gases on sorbents at low temperatures. Fuel Processing Technology 2003, 82, (2-3), Hutson, N. D.; Attwood, B. C.; Scheckel, K. G., XAS and XPS characterization of mercury binding on brominated activated carbon. Environmental Science & Technology 2007, 41, (5), Shah, P.; Strezov, V.; Prince, K.; Nelson, P. F., Speciation of As, Cr, Se and Hg under coal fired power station conditions. Fuel 2008, 87, (10-11), Wang, J.; Clements, B.; Zanganeh, K., An interpretation of flue-gas mercury speciation data from a kinetic point of view. Fuel 2003, 82, (8), Gerasimov, G., Investigation of the Behavior of Mercury Compounds in Coal Combustion Products. Journal of Engineering Physics and Thermophysics 2005, 78, (4), Granite, E. J.; Pennline, H. W., Photochemical Removal of Mercury from Flue Gas. Industrial & Engineering Chemistry Research 2002, 41, (22), Krishnan, S. V.; Gullett, B. K.; Jozewicz, W., Sorption of Elemental Mercury by Activated Carbons. Environmental Science & Technology 1994, 28, (8), Karatza, D., Kinetics of adsorption of mercuric chloride vapors on sulfur impregnated activated carbon. Combustion Science and Technology 1996, 112, Ho, T. C.; Kobayashi, N.; Lee, Y.; Lin, J.; Hopper, J. R., Experimental and Kinetic Study of Mercury Adsorption on Various Activated Carbons in a Fixed-Bed Adsorber. Environmental Engineering Science 2004, 21, (1),

70 101. Li, J.; Yan, N.; Qu, Z.; Qiao, S.; Yang, S.; Guo, Y.; Liu, P.; Jia, J., Catalytic Oxidation of Elemental Mercury over the Modified Catalyst Mn/α-Al2O3 at Lower Temperatures. Environmental Science & Technology 2009, 44, (1), Lee, S. S.; Lee, J. Y.; Keener, T. C., Performance of Copper Chloride-Impregnated Sorbents on Mercury Vapor Control in an Entrained-Flow Reactor System. Journal of the Air & Waste Management Association 2008, 58, (11),

71 Chapter 3 Preparation and Characterization of CuCl 2 -Impregnated Activated Carbon for Mercury Emissions Control * 3.1 Introduction In March 2011, the U.S. EPA issued a new proposed rule, the Mercury and Air Toxics Standards (MATS), which regulates mercury, other toxic heavy metals, and acid gases from coal- and oil-fired utility, industrial, commercial, and institutional power plants.[1] This new rule, which has been finalized on December 21, 2011, proposes to reduce mercury emissions by >90% starting from 2016 by adopting a Maximum Achievable Control Technology (MACT) approach, instead of a Cap & Trade approach adopted by the rescinded CAMR. The implementation of this new rule will make it necessary to more strictly control elemental mercury emissions. In July 2010 prior to the MATS, the U.S. EPA also issued a new proposed rule, the Transport Rule, which replaces the 2005 Clean Air Interstate Rule (CAIR) and will start to more strictly regulate sulfur dioxide (SO 2 ) and nitrogen oxide (NO x ) emissions from power plants in 28 states from 2012.[2] As a result, additional installations of flue gas desulfurization (FGD) and NO x control systems are highly anticipated. From a practical standpoint, sorbent injection is currently the most mature technology applicable to most coal-fired power plants.[3-7] It has been shown that raw activated carbon (AC) does not directly adsorb Hg(0) vapor. Raw AC rather adsorbs HCl gas from flue gas and heterogeneously oxidizes Hg(0) vapor to form oxidized mercury bound to the surface.[3-7] Among chemically-promoted AC sorbents, halogenated ACs have been reported to significantly enhance Hg(0) adsorption capability of AC.[3-5, 7, 8] These chemically-treated AC sorbents have demonstrated good performance from * Part of the content in this chapter has been published in Fuel, 2012, 93,

72 the coal combustion flue gases of subbituminous and lignite coals with relatively low HCl concentrations (e.g., < ~5 ppmv) in the flue gas stream. Several studies[9-14] have been conducted using the X-ray Absorption Fine Structure (XAFS) and X-ray Photoelectron Spectroscopy (XPS) techniques in order to investigate the speciation and binding of mercury on the surfaces. An XAFS study on the sorption of Hg(0) and HgCl 2 vapors onto sulfurized, iodated, and raw activated carbons under simulated flue gas conditions suggested that HgCl 2 vapor appeared to be captured by physi-sorption and that Hg(0) vapor was captured by chemisorption by forming sorption complexes with the promoted chemical elements (S and I) as well as the acidic gases of HCl and SO 2 in simulated flue gases.[10] Additional XAFS analyses also indicated that mercury can be captured by binding to I, Cl, S, or O anionic species on carbonaceous and non-carbonaceous surfaces. Another XPS study result suggested that Cl and S compete for adsorption at carbon sites with a higher adsorption preference of S primarily in the form of S(VI) in the presence of water vapor.[11] More than one form of bound chlorine was found to be present as chloride and organic chlorine. Another XAFS and XPS study on brominated ACs indicated that bromine appears to be primarily covalently bonded to sp2 and sp3 carbon atoms with some physisorbed HBr.[12] Hg(0) vapor then reacts with Br sites and the resultant oxidized mercury is most likely to be present as HgBr 2. Our previous studies have demonstrated that cupric chloride dispersed over carbon and clay substrates possesses excellent Hg(0) oxidation performance.[15-18] Carbon could effectively adsorb the resultant oxidized mercury, but a clay substrate could not adsorb it.[19, 20] In addition, as the CuCl 2 loading on carbon increases, the adsorption capacity decreases.[19] These results indicate that Hg(0) vapor reacts with CuCl 2 and the desorbed resultant oxidized 56

73 mercury is re-adsorbed onto substrate surfaces, suggesting different sites for Hg(0) oxidation and re-adsorption of the resultant oxidized mercury. The objective of this study is to identify the resultant oxidized mercury species formed from the reaction between Hg(0) vapor and CuCl 2 - impregnated activated carbon (CuCl 2 -AC) by means of a set of advanced characterization techniques, including XAFS (XANES, EXAFS), XPS, and XRD. 3.2 Experimental Section Sorbent Preparation, Characterization, and Mercury Loading onto Sorbents Three sorbents were purchased or prepared in order to investigate the mercury speciation: (1) DARCO-HG raw AC; (2) CuCl 2 -impregnated DARCO-HG; and (3) HCl-impregnated DARCO-HG. DARCO HG is a lignite coal-based AC specifically for the removal of mercury in coal fired utility flue gas emission streams.[21] It has an open pore structure and fine particle size which permits rapid adsorption in a short contact time. According to the provider (Norit Americas), DARCO HG has a mean particle size of 23μm, a iodine number of ~570 mg/g, total sulfur ~1.2 %(wt), bulk density ~0.51 g/ml, and nominal surface area 600 m 2 /g. Our characterization showed that as-purchased DARCO HG sample has BET surface area of 472 m 2 /g, which is very close to the result (480 m 2 /g) measured by other researchers.[22] This sample contains about 1.1% (wt) sulfur and undetectable chloride, which is similar to the DARCO FGD sample as reported in the literature [11] and is the same as the number shown in the datasheet provided by Norit company. The element analysis results for DARCO HG (both from literature and this work) are shown in Table 3.1. Neither SEM/EDX nor XPS can detect chlorine in the samples. Our results are in a good agreement with the data reported in the 57

74 literature. An SEM image of the DARCO-HG sample shows irregular shape and size of the particles, and the amorphous geometry of the surface (Figure 3.1). Table 3.1. Element analysis for DARCO HG activated carbon using SEM/EDX (this work), and XPS (Reference [11]) Element C O Mg Al Si S Cl Ca Total SEM/EDX (mol, %) 86.8± ± ± ± ± ±0.1 ND 1.7± XPS (mol, %) ND a SEM/EDX (wt, %) 78.8± ± ± ± ± ±0.3 ND 5.1± Norit Datasheet (wt,%) 1.1 a: the other 0.8% of elements are N and Fe. Figure 3.1 SEM image of DARCO HG activated carbon 58

75 The detailed synthesis techniques used for the second sorbent are described in our previous studies.[16, 18] The HCl-impregnated solution was prepared in a 0.3 N HCl solution using procedures similar to those used for the CuCl 2 -impregnated sorbents. The prepared sorbents were characterized for surface area, pore volume, and pore size distribution with a surface area and porosimetry system (ASAP 2020, Micromeritics), and the characterization data are summarized in Table 3.2. The concentrations of CuCl 2 impregnated onto sorbents were determined from the difference before and after the impregnation process. The concentrations of copper and chloride in solutions were determined using an atomic absorption spectrophotometer (AAnalyst 300, Perkin-Elmer) and ion chromatography (DX-600, Dionex), respectively. SEM- EDAX (XL30 ESEM, FEI) was also used to investigate the morphology of the surface and the concentration of Cl residing on HCl-impregnated AC and the sulfur concentration on DARCO- HG. The concentration of Cl impregnated onto raw DARCO-HG was found to be approximately 1.0% (wt). Table 3.2. Particle size, surface area, and pore volume of sorbents used in this study Mean Particle BET Surface area Pore volume Diameter (μm) (m 2 /g) (ml/g) Raw DARCO-HG %(wt) CuCl 2 -AC %(wt) CuCl 2 -AC %(wt) HCl-AC* * Increase in surface area, pore volume after acid-wash is primarily due to the removal of ash. After all the sorbents were prepared, they were loaded with Hg(0) vapor using a fixedbed reactor system. A high inlet Hg(0) concentration of ~3 µg/l was introduced to the system in 59

76 order to saturate the sorbent in a short period of time. The total gas flow rate was 500 ml/min. Approximately 100 mg of the sorbent was exposed to the Hg(0)-laden gas stream until a sorbent reached an equilibrium capacity. The fixed-bed reactor was placed in an oven maintained at 140 C, which is a typical sorbent injection temperature between an air preheater and a particulate control device in coal-fired power plants. All sorbents were saturated with Hg(0) vapor at 140 C except for one raw AC sample (namely, Raw AC-N 2 ). The raw AC sorbent was exposed to Hg(0) vapor in nitrogen flow at room temperature since no mercury adsorption was observed at 140 C. An online Hg analyzer (UV-1201S with mercury analysis kit, Shimadzu Corp., Columbia, MD) was used to observe breakthrough curves and determine an equilibrium adsorption point. Four different carrier gases were used in this study: (1) N 2 (Nitrogen % UHP, Wright Brothers, Inc., OH); (2) O 2 (Oxygen 99.99%); (3) 10 ppmv HCl in N 2 ; and (4) 10 ppmv HCl in O 2. The last two carrier gases were selected in order to investigate the capability of HCl and O 2 in Hg(0) oxidation TGA/MS Characterization Thermal Gravimetric Analysis-Mass Spectrometry (TGA, TA Instruments Q5000IR, and MS, Pfeiffer-Vacuum Thermostar) was used to evaluate the thermal stability of pure CuCl 2, raw AC, and fresh 10% CuCl 2 /AC. Approximately 10 mg of a sample was used to ramp from room temperature to 800 C with a heating rate of 5 C/min under nitrogen flow (99.999% UHP, Wright Brothers, Inc. OH) at a rate of 100 ml/min. 60

77 3.2.3 XAFS Analysis Hg L III (12,284 ev) edge X-ray absorption experiments were performed using the 20- BM-B Beamline at Sector 20 of the Advanced Photon Source (APS) at Argonne National Laboratory (ANL, Argonne, IL). A Si (111) monochromator was used, and energy calibration was performed using Au metal foil (Au L III edge is 11,919 ev). An array of selected reference mercury compounds including HgO, HgSO 4, HgCl 2, Hg 2 Cl 2 and HgS (all were purchased from Sigma-Aldrich, except for Hg 2 Cl 2, which was purchased from Fisher Scientific), as well as mercury-containing spent sorbent samples were examined. The reference mercury compounds were dispersed on Kapton (polyimide) tape and the tape was folded. The spent sorbent samples were loaded onto thin Teflon sample holders between layers of Kapton tape. XAFS data were collected in both transmission and fluorescence modes, and the fluorescent data were obtained using a 13-element Ge detector. For reference mercury compounds, the transmission data were applied to data processing. For spent sorbent samples, the fluorescence data had better signal-tonoise ratio, and was thus applied. The vertical slit width was 0.5 mm at 30 m, and the energy resolution (ΔE/E) was The step size in the near edge region was 0.2 ev. The XAFS spectroscopy used in this study is known to have a detection limit of ~30 ppmw, and is expected to detect mercury captured in most of the sorbents used in this study. At least five scans were performed in order to improve the signal-to-noise ratio, and neither X-ray induced redox reaction nor sample damage was observed XANES Data Analysis Mercury XAFS data were processed using Athena and Artemis interfaces to the IFFEFIT program.[23] In many cases, it is difficult to distinguish between oscillatory signals and noises 61

78 from a single scan. Therefore, multiple scans were used to improve the signal-to-noises ratio and obtain the average absorption coefficient, μ(e), dependent on absorption energy.[24] Mercury XANES data were analyzed by taking the first derivative, dμ/de, from which inflection point peaks were obtained. When the data analysis of the XANES region shows two inflection points, the difference between the two inflection points is defined as inflection point difference (IPD). In addition, the relative intensity of each inflection point peak is defined as peak height ratio (PHR) by taking the ratio of the height between a low energy peak and a high energy peak. The definitions and analysis of these two parameters are well described elsewhere, and they vary with the oxidation state of an absorbing atom and its neighbor atoms, which can be used to identify mercury species.[10] EXAFS Data Analysis The oscillations above the absorption edge belong to the EXAFS region, and are analyzed by converting the measured absorption coefficient, μ(e), into the EXAFS fine-structure function or the EXAFS, χ(k), which is the oscillations as a function of k, the wave number of photoelectron. XANES is often used as a simple fingerprint to identify chemical species. Although quantitative XANES analysis are still rare, EXAFS data can be modeled with theoretical phase and amplitude functions in order to estimate the bonding distance and coordination number.[25] The single scattering version of the EXAFS equation is, k Shells NS k 0e e kr 2R / 2 k f k sin 2kR k (3.1) where N is the coordination number, R is the distance between the adsorbing atom and its neighbor atom, σ 2 is the mean-square disorder in R (Debye-Waller factor), and ΔE 0 is the energy 62

79 shift. These structural information (N, R, σ 2, and ΔE 0 ) can be determined by refining them to make the model EXAFS equation match with experimental data. The f(k), (k), and λ(k) in the equation are scattering properties of the neighboring atom, and are generated by FEFF.[23] They can be used to distinguish O and S neighbor atoms, but do not have sufficient sensitivity to distinguish between S and Cl neighbor atoms. The amplitude reduction factor, S 2 0, is usually 2 regarded as a constant parameter (typically in the range of 0.7 < S 0 < 1.1). It can be estimated from analysis of reference mercury compounds with their known coordination numbers. The χ(k) is often multiplied by a power of k in order to emphasize the high k oscillations. Hence, the Fourier-transformed EXAFS, χ(r), is obtained by applying the χ(k) so as to distinguish the frequency components of a signal and to obtain the radial structure function (RSF). The RSF provides the structural information on the bonding distance between the absorbing (i.e. mercury) atom and its neighbor atoms. It can be used to isolate and identify different coordination shells around the absorbing atom. In normal conditions, the real bond distance between an absorption atom and a coordination shell is larger than the RSF peak position due to the scattering phase shift (i.e. the term (k) in the EXAFS equation).[26] XPS and XRD Characterization XPS was performed on a Kratos Axis Ultra, with a monochromated Al Kα source ( ev), using a concentric hemispherical analyzer. The applied power to the X-ray gun was 130 W (I=10 ma, V=13 kv), the base pressure was Torr, and the analyzer pass energy was 20 ev. The resolution of the spectra was ~0.2 ev and the step size was 0.1 ev. The carbon 1s peak at ev was used for referencing the collected spectra. 63

80 XRD patterns were recorded on a X Pert Pro MPD X-ray diffractometer using Cu Kα radiation (wavelength Å). An aluminum holder was used to support the catalyst samples. The scanning range was from 10 to 60 (2θ) with a step size of 0.02 and a step time of 0.5 second. 3.3 Results and Discussion TGA-MS Characterization The thermal stability of pure CuCl 2, raw AC, and fresh 10% CuCl 2 /AC samples were examined by TGA-MS, and the results are shown in Figure 3.2. The weight loss of 21% before 100 C from pure CuCl 2 is derived from the vaporization of dihydrate. The sample exhibited weight losses within the temperature range of C. It is noted that there are two stages within the temperature range of C: the first stage of C and the second stage of C. There was more chlorine detected by MS at the first stage than the second stage. The result suggests that CuCl 2 may release chlorine gas in the first stage and leave CuCl as a solid, and then the remaining CuCl starts to melt (melting point: 422 C) and evaporate in the second stage. The TGA profile of raw AC shows a slow trend of weight loss, mainly coming from the carboxyl group on the AC surface. Those carboxyl groups were released from the AC surface as the form of and CO 2, as detected by on-line mass spectroscopy. The 10% (wt) CuCl 2 /AC sample shows TGA profile more similar to the raw AC. Due to the highly dispersed CuCl 2 on AC, the CuCl 2 on AC surface does not show the same behavior as of pure CuCl 2. Additionally, the weight loss difference between raw AC and 10% (wt) CuCl 2 /AC sample does not match the total 64

81 TGA Weight (%) MS Ion intensity (na) weight of CuCl 2 on AC, indicating that there could be strong interaction of CuCl 2 between the functional groups on raw AC surface TGA Raw AC TGA Fresh 10% CuCl2-AC TGA CuCl MS Raw AC m/e=44 (CO2) MS Fresh 10% CuCl2-AC m/e=44 (CO2) MS CuCl2 m/e=35 (Cl) Temperature ( C) Figure 3.2 TGA results of CuCl 2 and 10%CuCl 2 impregnated sorbents XANES Analysis Results The reference mercury compounds and spent sorbent samples used in this study are listed in Table 3.3. The Hg L III -edge XANES spectra are shown in Figure

82 Table 3.3 XANES features for reference mercury compounds and spent samples Reference mercury compounds Spent sorbent samples c Inflection Point Peak Height Difference (IPD) Ratio (PHR) (ev) Hg(0) (liquid) 0 b b HgO (red) HgSO HgNO a Hg 2 Cl HgCl HgS (red) HgBr a HgI b 0.90 b Raw AC-N Raw AC-O Raw AC-N 2 with 10 ppmv HCl Raw AC-O 2 with 10 ppmv HCl a: these values are from reference 12; b: these values are from reference 10; and %(wt) CuCl 2 -AC-N %(wt) CuCl 2 -AC-O %(wt) CuCl 2 -AC-N %(wt) HCl-AC-N c: spent sorbent is labeled as sorbent-carrier gas. For instance, Raw AC-O 2 with 10ppmv HCl is a raw activated carbon sorbent that has been tested under O 2 with 10 ppmv HCl gas. All samples were prepared at 140 C except for Raw AC-N 2 that was prepared at room temperature. 66

83 Normalized (E) HgO HgSO 4 HgCl 2 Hg 2 Cl 2 HgO HgSO 4 HgS HgCl 2 Raw AC- N 2 Hg 2 Cl 2 HgS Raw AC- O 2 Raw AC- N 2, HCl d /de Raw AC- N 2 Raw AC- O 2, HCl 1%CuCl 2 -AC - N 2 1%CuCl 2 -AC - O 2 Raw AC- O 2 Raw AC- N 2, HCl Raw AC- O 2, HCl 10%CuCl 2 -AC - N 2 1%CuCl 2 -AC - N 2 1%HCl-AC - N 2 1%CuCl 2 -AC - O 2 10%CuCl 2 -AC - N 2 1%HCl-AC - N E(eV) E(eV) Figure 3.3 The XANES portion of Hg L III -edge XAFS spectra (left) and the first derivative of the spectra (right). The most notable feature of oxidized mercury on the Hg L III -edge XANES spectra is the existence of two inflection points, which appear with two distinct peaks in the first derivative spectra, dμ/de. In contrast, elemental mercury exhibits a single peak, which appears at the lower energy peak of oxidized mercury.[10] In general, when a mercury atom is surrounded by relatively a large number of small anions such as mercury oxide and mercury acetate, IPD values 67

84 become large. On the other hand, when mercury atom is surrounded by relatively a small number of large anions such as mercury halides and sulfides, IPD values become small. As summarized in Table 3.3, the oxygen-containing reference mercury compounds (i.e., HgO, HgNO 3, HgSO 4 ) showed large IPD values, while reference mercury halides (i.e. Hg 2 Cl 2, HgCl 2, HgBr 2, HgI 2 ) and HgS showed small IPD values. This observation is well agreeable with the result reported in a previous study.[10] The tested spent sorbent samples include raw AC, CuCl 2 -AC, and HCl-AC in a N 2 or O 2 carrier gas in combination with HCl gas. All the spent sorbent samples (prepared with different substrates and tested under different carrier gas conditions) showed two inflection points. This is an indication that oxidized mercury compounds appeared with IPD values between 7.5 and 8.2 ev. When these IPD values are compared with those of the reference mercury compounds, it turns out that the potential mercury compounds formed onto the samples could be HgS, HgCl 2, and/or Hg 2 Cl 2, not oxygen-containing mercury compounds such as HgO or HgSO 4. This result indicates that when Hg(0) vapor is adsorbed onto the surface, mercury is unlikely to be associated with oxygen, regardless of substrate, presence of dopant, and carrier gas conditions. This result is also in a good agreement with previous findings.[9, 10, 12] It is interesting to note that the formation of oxygen-containing mercury compounds (i.e. HgO or HgSO 4 ) is not evidenced by the IPD values even for the raw AC (DARCO-HG) sample prepared under O 2 carrier gas in the absence of HCl gas (i.e. Raw AC-O 2 ). This observation suggests that Hg(0) vapor does not react with O 2 to produce HgO or HgSO 4 on the carbon surface at 140 C. The elemental analysis data using SEM-EDAX on DARCO-HG sample obtained from three different spots showed that the DARCO-HG sample contains ~ %(wt) sulfur. No chlorine was detected from any of the three spots. The technical datasheet provided 68

85 by Norit Americas also shows that DARCO-HG contains the total sulfur of up to 1.1%(wt). Among the reference mercury compounds listed in Table 3.3, the IPD and PHR values obtained for the Raw-AC-N 2 and Raw-AC-O 2 samples are close to those of HgS (red). These data suggest that if oxidized mercury is formed onto DARCO-HG, mercury is more likely to be bound to sulfur species. The peak height ratio in conjunction with IPD also provides the information regarding the identification of mercury species.[10] When both elemental and oxidized mercury are present, the two-peak spectrum would be more asymmetric. The asymmetric peak height ratio would indicate the presence of elemental mercury in the sample.[10] All the spent sorbent samples doped with HCl and CuCl 2 listed in Table 3.2 had the peak height ratio values in the range of The two DARCO-HG samples showed the peak height ratio values of 1.22 and These results suggest that the physi-sorption of Hg(0) onto raw activated carbon is not likely to be the dominant adsorption mechanism even in N 2 carrier gas at room temperature. A comparison of the IPD and PHR values of the reference mercury compounds with those of HCl- and CuCl 2 -doped sorbent samples suggests that the dominant resultant mercury species could be HgCl 2 and/or Hg 2 Cl EXAFS Analysis and Modeling Results The EXAFS, χ(k), and its Fourier transformed spectra, χ(r), of all samples are shown in Figure 3.4 Among the reference mercury compounds, HgO, HgCl 2 and HgS have a single distinctive peak corresponding to their first shell around the center mercury atom in the χ(r) profiles, whereas HgSO 4 and Hg 2 Cl 2 show multiple distinctive peaks. Compared to the reference 69

86 mercury compounds, all the spent sorbent samples show only one strong distinctive peak in their χ(r) profiles. The first shell RSF peak position values obtained from the spectra are summarized in Tables 3.4. It can be seen that mercury halides and mercury sulfide have larger RSF values than oxygen-containing mercury compounds. In addition, the RSF values increase with an increase of the size of halides, in the order of HgI 2 > HgBr 2 > HgCl 2. These data are consistent with those reported in previous studies.[10, 12] The RSF peak position values of all the spent sorbents ranged from 1.84 to These values ruled out the possibility of the formation of oxygencontaining mercury compounds (i.e. HgO and HgSO 4 ) and narrowed down a potential candidate pool to HgCl 2, Hg 2 Cl 2, and HgS. In the EXAFS data modeling, the scattering property functions of f(k), (k), and λ(k) in the EXAFS equation were calculated from Cl, S, or O atom based on an assumption that one of the above atoms bonds with the mercury atom. O atom was re-considered in the EXAFS first shell modeling. The structural parameters of N, R, σ 2, and ΔE 0 were determined such that the best fit of χ(k) matches with the EXAFS data. The crystalline structure data including the coordination number (N) and the bonding distance (R) in each coordination shell of some reference mercury compounds were calculated and summarized in Table

87 Hg O Hg S HgSO 4 HgO HgCl 2 Hg 2 Cl 2 HgS Hg O Hg Cl Hg Hg Hg Cl Hg Cl Hg S HgSO 4 HgO HgCl 2 Hg 2 Cl 2 k 2 X(k)(Å -2 ) Raw AC- N 2 Raw AC- O 2 X(R)(Å -2 ) HgS Raw AC- N 2, HCl Raw AC- N 2 Raw AC- O 2, HCl Raw AC- O 2 Raw AC- N 2, HCl 1%CuCl 2 -AC - N 2 Raw AC- O 2, HCl 1%CuCl 2 -AC - O 2 1%CuCl 2 -AC - N 2 10%CuCl 2 -AC - N 2 1%CuCl 2 -AC - O 2 1%HCl-AC - N 2 10%CuCl 2 -AC - N 2 1%HCl-AC - N k(å -1 ) Figure 3.4 k-weighted EXAFS spectra, k 2 χ(k), and the Fourier transformed EXAFS spectra χ(r) (shown in amplitude χ(r) ). R(Å) The XAFS-derived bonding distances (R) in terms of coordination shell and number are in accordance with the reported results.[27] The coordination numbers and bonding distances for all the sorbent samples are also summarized in Table 3.4. Based on the above RSF peak position results, HgO and HgSO 4 are not likely to be formed on all sorbent samples. Therefore when 71

88 coordination number was calculated for all spent sorbent samples, a value 0.9 was selected as an amplitude reduction factor (S 2 0 ) from the values of HgCl 2, Hg 2 Cl 2, and HgS. It was found that the values of the bonding distance (R) were roughly 0.5 Å larger than the RSF peak position values. Table 3.4. EXAFS results for reference mercury compounds RSF Peak Coordinated atoms Position and numbers (Å) HgO 1.66 First shell 2 O atoms XAFS-derived R(Å) 2.05 Reference[27] R(Å) 2.05 S Second shell 4 O atoms HgSO First shell 8 O atom Second shell 6 S atoms 3.35 HgCl First shell 2 Cl atoms Second shell 4 Cl atom Hg 2 Cl First shell 1 Cl atom Second shell 1 Hg atom Third shell 4 Cl atoms HgS 1.91 First shell 2 S atoms Second shell 2 S atoms The mercury bonding in the molecules of HgCl 2, Hg 2 Cl 2, and HgS is covalent bonding. When the samples listed in Table 3.5 are assumed to be associated with S and Cl atoms, the first shell bonding distances were measured to be Å for HgS and Å for HgCl 2 compared to the reported values of 2.37 Å for HgS and 2.29 Å for HgCl 2 shown in Table

89 Therefore, the XAFS-derived bonding distances from the first shell fitting are in a good agreement with the reference data. The crystalline structures of the reference mercury compounds have multiple shells around Hg atom due to a regular arrangement of its neighbor atoms. As a result, multiple peaks appear in the χ(r) profiles of the reference mercury compounds. In contrast, only one strong peak is present in the χ(r) profiles of all the spent sorbent samples, indicating that any longer bonding may be highly disordered and/or the Hg containing particles are very small. Table 3.5. EXAFS results for sorbent samples Sample RSF Peak First shell Position (Å) Coordinated atom N R (Å) σ 2 ΔE 0 Raw AC-N S Raw AC-O S Raw AC-N 2, 10 ppmv HCl 1.84 Cl Raw AC-O 2, 10 ppmv HCl 1.87 Cl %(wt) CuCl 2 -AC-N Cl %(wt) CuCl 2 -AC-O Cl %(wt) CuCl 2 -AC-N Cl %(wt) Cl-AC-N Cl Notes: S 0 2 = 0.9; N: coordination number ± 25%; R: refined atomic distance ± 0.02 Å; σ 2 : Debye-Waller factor (a measure of the amount of disorder) ±25%; and ΔE 0 : energy shift. Figure 3.5 shows the experimental Hg EXAFS spectra and their theoretical fittings in χ(r) and χ(k) for Raw-AC-O 2 sample and 10% CuCl 2 -AC-N 2, respectively (other samples show 73

90 similar spectra). The fitting procedures were performed over the range of 1~3 Å for R and 2~12 Å -1 for k. The EXAFS first shell modeling result for raw AC suggests that mercury is more likely to be associated with S atom, but less likely to be with O atom. The modeling did not include potential Cl association since Cl was not found from the previous SEM-EDAX result. Based on the above analysis, HgS is very likely to be formed on raw AC samples in the absence of HCl gas. The modeling result for 10% CuCl 2 -AC-N 2 with Cl and S atoms also gives a good fitting result and suggests no noticeable difference in association with Cl and S atoms, whereas the curve fitting with O atom does not match well with the experimental profile. When the sorbent samples are assumed to be associated with S and Cl atoms, the first shell bonding distances were calculated to be Å for HgS and Å for HgCl 2 compared to the reported values of 2.37 Å for HgS and 2.29 Å for HgCl 2. Judging from the EXAFS result, one cannot determine whether mercury is bound to S or Cl on CuCl 2 -AC only. However, the peak height ratio result obtained from the XANES analysis on HgS suggests that the oxidized mercury found from CuCl 2 -AC be unlikely to be associated with S atom. 74

91 RawAC-O 2 First shell fit with S First shell fit with O X(R) (Å -2 ) k 2 X(R) (Å -2 ) k(å -1 ) 0 2 R(Å) 4 6 (a) 10% CuCl 2 -AC-N 2 First shell fit with Cl First shell fit with S First shell fit with O k 2 X(k) (Å -2 ) X(R) (Å -2 ) k(å -1 ) R(Å) (b) Figure 3.5 Experimental Hg L III -edge EXAFS spectra and their fitting results for (a) Raw AC-O 2 and (b) 10% CuCl 2 -AC-N 2. 75

92 For both the Raw AC-N 2 and Raw AC-O 2 samples, the coordination number was 1.8 and the bonding distance was 2.32Å. The EXAFS modeling results suggest that Hg is coordinated with two S atoms with a bonding distance of 2.32Å, which is in agreement with the previous results.[28] For other sorbents, the EXAFS fitting results show that Cl is very likely to be bound to Hg with a coordination number between 2.0 and 2.6 in the first shell. It also suggests that the oxidized mercury formed on the AC surfaces might exist in the form of HgCl 2 with a coordination number of 2 rather than Hg 2 Cl 2 with a coordination number of 1. Thermodynamic calculations also indicate that HgCl 2 is a thermodynamically favorable form of mercury species compared to Hg 2 Cl 2. Table 3.6 summarizes the thermodynamic values for possible reactions that may occur between elemental mercury vapor and CuCl 2 with or without oxygen as a reactant. The enthalpy and Gibbs free energy formation values for each compound were obtained from the NIST-JANAF thermochemical tables.[29] Table 3.6. Thermodynamic values for possible mercury oxidation reactions on sorbents used in this study Temperature ΔH 0 ΔG 0 Equilibrium Equilibrium Reaction ( C) (kj) (kj) Constant,K Conversion, % Hg(0)(g) + 2CuCl 2 (s) = HgCl 2 (g) CuCl(s) Hg(0)(g) + CuCl 2 (s) = 0.5Hg 2 Cl 2 (g) CuCl(s) Hg(0)(g) + CuCl 2 (s) + 0.5O 2 (g) = HgCl 2 (g) + CuO(s) Hg(0)(g) + 0.5CuCl 2 (s) O 2 (g) = Hg 2 Cl 2 (g) + 0.5CuO(s) Note: The enthalpy and Gibbs free energy formation values for each compound were obtained from the NIST-JANAF thermochemical tables.[29] 76

93 3.3.4 XPS and XRD Results Figure 3.6 is Cu 2p XPS profiles of fresh 10%CuCl 2 -AC sample and spent sample. Both the fresh and spent samples show the coexistence of Cu 2+ and Cu + species. However, the XPS profiles do not show significant difference between the fresh and spent samples. It seems that XPS cannot provide insight information of the probably because of the highly dispersed Cu species on AC surface. Binding energy = 934.5eV Cu 2+ Binding energy = 931.5eV Cu + CuCl2 CuCl Fresh 10%CuCl2-AC Spent 10%CuCl2-AC Binding energy Figure 3.6 Cu 2p XPS of CuCl 2, CuCl, fresh and spent 10%CuCl2-AC samples We also obtained the XRD patterns for pure CuCl 2, CuCl, raw AC, fresh and spent sorbents, as shown in Figure 3.7 Both patterns of fresh and spent CuCl 2 -AC sorbents do not show distinguishable peaks, which can be assigned to copper species, except for the peaks from the raw AC. The XRD results once again suggest that CuCl 2 might be loaded on activated 77

94 Intensity [a.u.] carbon surface with well-dispersed monolayer or sub-monolayer forms and was not readily detectable with XRD. CuCl 2 CuCl Fresh 10%CuCl 2 -AC Spent 10%CuCl 2 -AC Raw AC Figure 3.7 X-ray diffraction patterns of CuCl 2 -impregnated sorbents 3.4 Conclusions Mercury XAFS analysis on spent sorbents provides valuable information on the identification of potential mercury compounds formed onto the sorbent surface. XANES and EXAFS have been used to determine mercury compounds formed on AC sorbents. Our XANES study on raw and CuCl 2 -impregnated AC sorbents suggests that little or no elemental mercury is formed onto any spent sorbents and the chemisorption of elemental mercury vapor is very likely to be the dominant mechanism consistent with the previous study results obtained from raw and other chemically promoted AC sorbents. HgS is most likely to be formed when Hg(0) vapor is adsorbed onto DARCO-HG raw carbonic which contains sulfur under both nitrogen and oxygen 78

95 carrier gases. In addition, HgCl 2 is very likely to be a major oxidation reaction product when CuCl 2 and HCl were impregnated onto raw AC regardless of a carrier gas (i.e. N 2 or O 2 ). Additional EXAFS experimental results and their fitting calculations support that the resultant oxidized mercury could be either HgS for raw AC or HgCl 2 as a result of the reaction with CuCl 2 for CuCl 2 -AC. 3.5 Acknowledgements The XAFS experiments were performed using the 20-BM Beamline at the Sector 20 of Advanced Photon Source (APS) at Argonne National Laboratory (ANL, Argonne, IL). Use of the Advanced Photon Source is also supported by the U.S. Department of Energy, Office of Science, Office of Basic Energy Sciences, under Contract DE-AC02-06CH We appreciate Ms. Bala Lingaraju for her assistance in sample preparation and data acquisition for XAFS measurements. We also greatly appreciate Dr. Dale L. Brewe of Argonne National Laboratory for his assistance in the measurements, data processing, interpretation, and helpful comments. This study was funded by the College of Engineering and Applied Science at the University of Cincinnati through faculty start-up funds. 3.6 Reference 1. "Federal Implementation Plans To Reduce Interstate Transport of Fine Particulate Matter and Ozone." 40 CFR Parts 51, 52, 72, 78, and 97. Federal Register 75:147 (August 2, 2010) p "National Emission Standards for Hazardous Air Pollutants From Coal- and Oil-Fired Electric Utility Steam Generating Units and Standards of Performance for Fossil-Fuel-Fired Electric Utility, Industrial-Commercial-Institutional, and Small Industrial-Commercial- 79

96 Institutional Steam Generating Units; Proposed Rule." 40 CFR Parts 60 and 63. Federal Register 76:85 (May 1, 2011) p Pavlish, J. H.; Sondreal, E. A.; Mann, M. D.; Olson, E. S.; Galbreath, K. C.; Laudal, D. L.; Benson, S. A., State review of mercury control options for coal-fired power plants. Fuel Processing Technology 2003, 82, (2-3), Srivastava, R. K.; Hutson, N.; Martin, B.; Princiotta, F.; Staudt, J., Control of mercury emissions from coal-fired in electric utility boilers. Environmental Science & Technology 2006, 40, (5), Jones, A. P.; Hoffmann, J. W.; Smith, D. N.; Feeley, T. J.; Murphy, J. T., DOE/NETL's Phase II Mercury Control Technology Field Testing Program: Preliminary Economic Analysis of Activated Carbon Injection. Environmental Science & Technology 2007, 41, (4), Sjostrom, S.; Durham, M.; Bustard, C. J.; Martin, C., Activated carbon injection for mercury control: Overview. Fuel 2010, 89, (6), Granite, E. J. Halogenated Sorbents for Mercury Control. Presented at the DOE/NETL Mercury Control Technology R&D Program Review, Pittsburgh, PA. July 12-14, pm.pdf. 8. Maroto-Valer, M. M.; Zhang, Y.; Granite, E. J.; Tang, Z.; Pennline, H. W., Effect of porous structure and surface functionality on the mercury capacity of a fly ash carbon and its activated sample. Fuel 2005, 84, (1), Huggins, F. E.; Huffman, G. P.; Dunham, G. E.; Senior, C. L., XAFS Examination of Mercury Sorption on Three Activated Carbons. Energy & Fuels 1998, 13, (1), Huggins, F. E.; Yap, N.; Huffman, G. P.; Senior, C. L., XAFS characterization of mercury captured from combustion gases on sorbents at low temperatures. Fuel Processing Technology 2003, 82, (2-3), Laumb, J. D.; Benson, S. A.; Olson, E. A., X-ray photoelectron spectroscopy analysis of mercury sorbent surface chemistry. Fuel Processing Technology 2004, 85, (6-7), Hutson, N. D.; Attwood, B. C.; Scheckel, K. G., XAS and XPS characterization of mercury binding on brominated activated carbon. Environmental Science & Technology 2007, 41, (5),

97 13. Huffman, G. P.; Huggins, F. E.; Shah, N.; Zhao, J.; Lu, F., Investigation of environmentally critical elements in coal and ash by XAFS spectroscopy. Fuel 1993, 72, (5), Shah, P.; Strezov, V.; Prince, K.; Nelson, P. F., Speciation of As, Cr, Se and Hg under coal fired power station conditions. Fuel 2008, 87, (10-11), Lee, J. Y.; Ju, Y. H.; Keener, T. C.; Varma, R. S., Development of cost-effective noncarbon sorbents for Hg-0 removal from coal-fired power plants. Environmental Science & Technology 2006, 40, (8), Lee, S. S.; Lee, J. Y.; Keener, T. C., Performance of Copper Chloride-Impregnated Sorbents on Mercury Vapor Control in an Entrained-Flow Reactor System. Journal of the Air & Waste Management Association 2008, 58, (11), Lee, S.-S.; Lee, J.-Y.; Keener, T. C., The effect of methods of preparation on the performance of cupric chloride-impregnated sorbents for the removal of mercury from flue gases. Fuel 2009, 88, (10), Lee, W. J.; Bae, G. N., Removal of Elemental Mercury (Hg(O)) by Nanosized V2O5/TiO2 Catalysts. Environmental Science & Technology 2009, 43, (5), Lee, S.-S.; Lee, J.-Y.; Khang, S.-J.; Keener, T. C., Modeling of Mercury Oxidation and Adsorption by Cupric Chloride-Impregnated Carbon Sorbents. Industrial & Engineering Chemistry Research 2009, 48, (19), Lee, S.-S.; Lee, J.-Y.; Keener, T. C., Novel sorbents for mercury emissions control from coal-fired power plants. Journal of the Chinese Institute of Chemical Engineers 2008, 39, (2), Norit Darco-Hg Specifications. (March 12, 2012), 22. Klasson, K. T.; Lima, I. M.; Boihem Jr., L. L.; Wartelle, L. H., Feasibility of mercury removal from simulated flue gas by activated chars made from poultry manures,. J. Environ. Manag. 2010, 91, Ravel, B.; Newville, M., ATHENA, ARTEMIS, HEPHAESTUS: data analysis for X-ray absorption spectroscopy using IFEFFIT. Journal of Synchrotron Radiation 2005, 12, (4),

98 24. Newwille, M. Fundamentals of XAFS, Revision 1.7 July 23, Bunker, B., Introduction to XAFS: a practical guide to X-ray absorption fine structure spectroscopy. Cambridge University Press: Teo, B. K., EXAFS: Basic Principles and Data Analysis. Springer-Verlag: Wells, A. F., Structural Inorganic Chemistry. Clarendon Press: Oxford, Lennie, A. R.; Charnock, J. M.; Pattrick, R. A. D., Structure of mercury(ii) sulfur complexes by EXAFS spectroscopic measurements. Chemical Geology 2003, 199, (3-4), NIST-JANAF Thermochemical Tables. American Chemcial Society: Woodbury, NY,

99 Chapter 4 Modeling and Simulation of Mercuric Chloride Adsorption onto CuCl 2 -Impregnated Activated Carbon Sorbents 4.1 Introduction Many previous studies have shown that raw AC does not directly adsorb Hg(0) vapor. Rather, raw AC adsorbs HCl gas from flue gas and heterogeneously oxidizes Hg(0) vapor to form oxidized mercury bound to the surface.[1, 2] Among chemically-promoted activated carbon sorbents, halogenated activated carbons have been reported to significantly enhance Hg(0) vapor adsorption.[1-5] These chemically-treated activated carbon sorbents have demonstrated excellent performance in the adsorption of both elemental and oxidized forms of mercury vapor from the coal combustion flue gases of subbituminous and lignite coals with relatively low HCl concentrations (e.g., < ~5 ppmv). [6] The cupric chloride (CuCl 2 )-impregnated activated carbon (CuCl 2 -AC) demonstrated excellent Hg(0) oxidation performance in our previous work. [7-9] It was also found that as the CuCl 2 loading on carbon increases, the adsorption capacity decreases. [10] After extensive characterization work, we confirm that, on CuCl 2 -AC sorbents, Hg(0) vapor reacts with CuCl 2 impregnated on carbon surface first and the resultant oxidized mercury is subsequently readsorbed onto active carbon (i.e. CuCl 2 -free) surface.[9] It is known that mercury conversions in flue gas are kinetically, and not thermodynamically controlled.[11] A fundamental understanding of adsorption mechanisms plays a critical role ultimately in the prediction of mercury removal for sorbent injection. It was found that HgCl 2 vapor appeared to be captured by physi-sorption and that Hg(0) vapor was captured by chemisorption through forming sorption complexes with the promoted chemical elements as well as the acidic gases of HCl and SO 2 in 83

100 simulated flue gases.[9, 12] However, kinetic studies for the adsorption/reaction rate, adsorption/reaction activation energy, etc. have been so far less conducted. Due to the lack of kinetic data, it is difficult to predict a mercury removal efficiency for a physical or chemical sorbent under various different operating conditions. Karatza et al. discussed the influence of sulfur compounds on the capture of HgCl 2 by activated carbon, considering both the adsorption capacity and adsorption rate, to determine adsorption kinetics/equilibrium.[13] The study was performed in a lab-scale fixed bed in which a simulated flue gas containing HgCl 2 was introduced to a sulfur-impregnated activated carbon adsorbent. A model based on the assumption of adsorption kinetic control was used to evaluate the adsorption and desorption kinetic parameters. Scala applied the data from Karatza et al. s studies to conduct simulations of HgCl 2 capture by activated carbon for sorbent injection process in an in-flight duct, and a fabric filter, respectively.[14, 15] Scala et al. also reported a modeling study on Hg(0) adsorption on AC both in fabric filter and eletrostatic precipitator.[16] They suggested that the process of Hg(0) adsorption onto AC is schematized as two steps, namely (1) mass transfer from the bulk gas to the external surface of the AC particle through the gas boundary layer; and (2) surface adsorption within the particle. The first step can be treated by means of an external mass transfer coefficient determined by the particle Sherwood number. However, for the second step, there exists considerable uncertainty on the mechanism of elemental mercury adsorption on RAW or impregnated AC particles.[16] The paper applied Langmuir theory due to its simple expression. Other researchers also studied Hg(0) adsorption on various ACs.[13, 17, 18] For example, Ho et al. conducted an experimental and kinetic study of elemental mercury adsorption on various activated carbons in a fixed-bed adsorber.[18] They proposed an adsorption model comprising 84

101 adsorption kinetics, surface equilibrium, external mass transfer, and a material balance following a tank-in-series approach. They compared three equilibrium adsorption isotherms of the Henry's Law, the Langmuir isotherm, and the Freundlich isotherm, and found that all three surface equilibrium isotherms appear to describe the adsorption profiles equally well due to a very low mercury concentration level in the flue gas. Skodras et al. reported their work of elementary mercury adsorption on activated carbon.[19] They evaluated several models taking into account intra-particle diffusion, comprised of pseudo-first-order, pseudo-second-order, and the Elovich adsorption kinetic equations. They found that the most accurate prediction of experimental data was obtained by using a second-order adsorption kinetic equation, and chemisorption seems to be a rate-controlling step in adsorption. Generally, mercuric chloride vapor adsorption on AC is thought to be physical adsorption.[13, 20] Several adsorption isotherms have been used as summarized in Table 4.1.[10, 21-23] Table 4.1. Summary of adsorption isotherms Isotherm type Isotherm equation Adsorption type Langmuir isotherm Brunauer Emmett Teller (BET) q * q max K L C * 1 K C * C Bq0 * CS q * * C C (1 )(1 ( B 1) ) C C Henry s law * * Freundlich isotherm q K S H C q K C * n * F L * S Monolayer adsorption Multilayer adsorption Adsorption at low coverage Empirical isotherm Rectangular isotherm q*=constant Irreversible adsorption q: HgCl 2 loading on sorbent (g HgCl 2 /g sorbent); C: HgCl 2 concentration in gas phase ( g HgCl 2 /m 3 flue gas); *: Equilibrium 85

102 The local adsorption kinetics for HgCl 2 vapor adsorbed onto carbon sorbent was investigated using different models as summarized below. Equilibrium theory [10] Equilibrium isotherm is expressed in a general form by: * q f ( C) (4.1) Assuming that local adsorption equilibrium can be instantaneously attained within the pores between the gas and solid phases of HgCl 2, we have a local adsorption rate, r A. r A * dq q C p p (4.2) dt C t Linear adsorption model It is convenient to assume that the adsorption rate is proportional to the concentration difference as a driving force, as Lin et al. proposed.[24] r A dq * * p pk( q q) pkkl( C C ) (4.3) dt k: Adsorption rate constant K L : Adsorption equilibrium constant The solution to the differential equations for this case can be found in references.[10, 25] Langmuir adsorption model Several studies assumed that the HgCl 2 adsorption onto AC sorbent follows Langmuir adsorption due to its simple expression and success in correlating experimental adsorption data.[13-15] The adsorption rate is given by 86

103 r A dq p p ( k1( qmax q) C k2q) (4.4) dt The Langmuir theory will be employed in this study. Additionally, mercury concentrations in various flue gases are very low (usually on the order of ~1 ppbv), which can generate significant external and internal mass-transfer resistances for sorbent injection.[26, 27] Therefore, both external and internal (particularly for big sorbent particles) mass-transfer resistances should be considered when determining adsorption kinetic parameters over CuCl 2 -AC in a fixed-bed reactor. Axial diffusion along the reactor length has also been taken into account in mercuric chloride adsorption on CuCl 2 -AC. Simulation results are compared with the experimental data obtained from fixed-bed tests under different initial HgCl 2 concentrations in the range of 5 to 20 ppbv on three AC sorbents. 4.2 Experimental Section A lab-scale fixed-bed system was used for the mercury adsorption study. The fixed-bed used for the experiments of HgCl 2 adsorption is shown in Figure 4.1, and the reactor parameters and operating conditions are summarized in Table 4.2. Three different sorbents (i.e. DARCO- HG activated carbon, and two CuCl 2 -AC: 4%(wt)and 10%(wt)) were tested under different inlet HgCl 2 concentrations (i.e. 5, 10, 15, and 20 ppbv). A 20-mg sorbent mixed with 4 g of inert quartz sand (with diameter ~300 µm) was loaded in the fixed-bed reactor with an inner diameter of 12 mm. HgCl 2 vapor in pure N 2 flow was brought into the fixed-bed reactor. From each experimental run, breakthrough curves were obtained by measuring HgCl 2 concentrations in the outlet stream from the fixed-bed reactor. It was assumed that thermodynamic equilibrium was reached when the outlet HgCl 2 concentration was equal to its inlet concentration. The length of reaction zone was 20 mm. Before loading a sorbent, a blank test was performed and showed 87

104 negligible HgCl 2 adsorption on the internal wall of either borosilicate reactor or Teflon tubing. In order to reduce the effect of pore diffusion resistance, all carbonic sorbents were sieved with 635 mesh (i.e. 20 μm). An average diameter of the tested activated carbon was 15 μm. HgCl2 permeation tube 1 L/min TC TC (Control Temperature at 140 degree C ) Fixed-bed reactor in an oven ID = 12mm Sorbent N2 Teflon o-ring Bypass 1 m glass fiber filter sitting on a fritted quartz disc 0.7 m Glass fiber filter Hood Bubble meter 1 M (w/v) KCl solution Figure 4.1 Schematic of HgCl 2 adsorption fixed-bed reactor system The outlet mercury speciation was measured by the Ontario Hydro method.[28] A 1M KCl impinger solution was used to capture oxidized mercury, and 4%(w/v) KMnO 4 /10%(v/v) H 2 SO 4 impinger solution was used to capture Hg(0). Oxidized mercury and Hg(0) concentrations in the effluent gas stream were determined by analyzing those solutions using a cold vapor atomic absorption spectrophotometer (CVAAS, Model 400A, Buck Scientific Inc., East Norwalk, CT). More detailed information on the system and experiments are described in our previous study.[7] The quantity of HgCl 2 adsorbed on a sorbent was measured by digesting after each run with aqua regia and then analyzing the solution by CVAA as described in the Ontario Hydro method.[28] 88

105 Table 4.2. Summary of fixed-bed testing conditions for HgCl 2 adsorption Item Reactor diameter Flow rate Flow mode Test Conditions 12 mm 1 L/min at 20 C and 1atm Down-flow Operating temperature 140 C Gas velocity in an empty reactor 0.18 m/sec at 140 C Gas Inlet Hg(0) concentration Sorbent amount Average sorbent particle size Sorbent packed height Nitrogen, 5 ppbv~20 ppbv 20 mg 15 μm 20 mm Silica bed porosity Kinetic Model of HgCl 2 Vapor Adsorption on Activated Carbon Model Assumptions In this work, the kinetic model of HgCl 2 vapor adsorption on AC in a fixed-bed system is developed. The adsorption process is supposed to be a series of three steps: 1) external mass transfer from the bulk gas-phase to the external surface of a sorbent particle; 2) internal mass transfer from the external carbon surface to the interior of the particle through particle pores by 89

106 means of Knudsen diffusion; and 3) surface adsorption on the internal surface area of a sorbent. The model is developed based on the following assumptions: 1). The gas travels in plug flow in the fixed-bed system; 2). Activated carbon particles are spherical, and all the particles have the same size and are uniformly dispersed in the fixed bed. 3). The heat effect associated with HgCl 2 adsorption is neglected (i.e. isothermal) due to the trace level concentrations. Therefore, the temperature is regarded to be constant and uniform through the system. 4). The pressure losses are negligible, so the gas velocity is constant along the duct. The pressure drop is estimated to be ~1 kpa according to Darcy equation, so the assumption is reasonable. 5). Mercury adsorption on the duct wall is negligible in steady-state operation (i.e., equilibrium conditions are reached between the gas phase and duct walls so that no net exchange of mercury is present. This was also experimentally confirmed.) Model Equations Intrinsic adsorption rate expression The adsorption rate is supposed to follow the Langmuir theory, given by, r A dq p p( k1( qmax q) C k2q) (4.5) dt ra: HgCl 2 adsorption rate, g HgCl 2 adsorbed/(m 3 sorbent particle s) q: HgCl 2 adsorbed onto sorbent, g HgCl 2 /g sorbent C: HgCl 2 concentration in the gas phase of internal pore, g HgCl 2 /m 3 ρ p : Sorbent particle density, g sorbent/m 3 90

107 -1 k 1 : adsorption rate constant, m 3 g -1 s k 2 : desorption rate constant, s -1 At equilibrium (r A =0), the rate equation leads to Langmuir isotherm. q KC * q (4.6) 1 KC * * max K: Equilibrium constant, K=k 1 /k 2 *: Equilibrium state q max : Maximum adsorption capacity Mass balance in a single spherical sorbent [29] Figure 4.2 is a schematic of a single spherical model. The mass transfer within the porous particle is well described in equation (4.7). r r+dr R p Figure 4.2 Mass transfer over a single spherical model C 1 C t r r r ε p : Particle porosity. 2 p ( r D ) 2 e ra (4.7) D e : Effective pore diffusion coefficient, m 2 /s. r A : HgCl 2 adsorption rate in equation (4.5). 91

108 Mass Balance in a Fixed-bed Adsorber A shell mass balance approach was used to describe the mass transfer of HgCl 2 inside a fixed-bed system as shown in Figure 4.3. Figure 4.3 Setting up the mass balance in a fixed-bed system. By taking a thin slice of fixed-bed reactor, mass balance for HgCl 2 gives 2 CB CB CB b u Dm r 2 A, obs (1 b) t z z (4.8) C B : HgCl 2 concentration in the bulk gas phase, g HgCl 2 /m 3 ε b : Bed porosity u: Superficial velocity, m/s D m : Molecular diffusivity, m 2 /s r A,obs : observed adsorption rate given by Rp 2 ra 4 r dr dc 3 dc A, obs 4 p e r R p e r Rp R dr Rp dr p Rp r R D D (4.9) 3 3 Initial conditions and boundary conditions I.C. C(z, r, 0)=0, C B (z, 0)=0, q (z, r, 0)=0 B.C. C B (0,t)=C 0, C r 0; q r r 0 r

109 C KG ( C B( z, t ) C ( z, R p, t )) r D r Rp e K G: Mass-transfer coefficient, m/s By solving the partial differential equations of Eqs. 4.5, 4.7, and 4.8, the concentration profiles C(z, r, t), C B (z, t), q(z, r, t) can be obtained. C B (t, z=l) at the length of the fixed-bed is a breakthrough curve Parameters in PDEs The model has been applied to several activated carbons (including DARCO-HG AC, and two CuCl 2 -AC sorbents (4% and 10% by wt)). Adsorptive and physical properties of the selected sorbents are listed in Table 4.3. Table 4.3. Properties of the selected sorbents sorbent ρ p (kg/m 3 ) ε p d pore (nm) DARCO-HG %(wt) CuCl 2 -DARCO-HG %(wt) CuCl 2 -DARCO-HG 1, d pore : average pore size diameter of a sorbent particle. The effective pore diffusion coefficient (D e ) is given by a combination of molecular and Knudsen diffusivities, Eq. (4.10): 1 D e p 1 ( D p m 1 D Kn ) (4.10) 93

110 The particle pore tortuosity (τ p ) is estimated to be 1/ε p.[14] The molecular diffusivity of mercury species in gas environment (e.g. air, flue gas, or simulated flue gas) was reported to be between 0.172~ m 2 /s in the literature at ~140 o C (Table 4.4).[30, 31] In this study, a value of m 2 /s was used based on the Chapman-Enskog theory shown in Equation (4.11).[32] D m T (4.11) M HgCl M 2 N2 P HgCl2,N 2 Hg, N2 where T is the gas phase temperature in K; M HgCl2 and M N2 are the molecule weights in g/gmol; P is the atmospheric pressure in atm; σ HgCl2,N2 is the average collision diameter between mercury and nitrogen molecules in angstrom, which is the average kinetic diameter of HgCl 2 and N 2 molecules; Ω HgCl2,N2 is the dimensionless collision integral (the value is temperaturedependent, but usually of order 1). Table 4.4 Molecular diffusivity of mercury species in the gas phase Mercury Molecular diffusivity Gas Temperature, Reference species (10-4 m 2 /s) C Hg(0) air 140 [30] Hg(0) flue gas 140 [31] HgCl N This work 94

111 Knudsen diffusivity was estimated to be m 2 /s using [33] D Kn d pore 8RT (4.12) M 3 HgCl 2 where T is the gas phase temperature in K and M HgCl2 is the molecule weight of HgCl 2 in kg/gmol. Mass transfer coefficient,[34] K g D Sh 2R m (4.13) p where Sh is Sherwood number, given by Sh 1/ 2 1/ Re Sc 4.4 Results and Discussion Breakthrough Curves and Langmuir Isotherms Breakthrough curves were obtained for three different sorbents under four different inlet HgCl 2 concentrations (5, 10, 15, and 20 ppbv) as shown in Figure 4.4. All the curves are expressed as the outlet HgCl 2 concentration with respect to time (hr). Note that the solid curves are only used for visual guide. For a given inlet HgCl 2 concentration, raw activated carbon takes the longest time until saturation, followed by 4%(wt) and 10%(wt) CuCl 2 -AC sorbents. For a specific sorbent, the higher the inlet HgCl 2 concentration, the shorter the saturation time. The model, including the partial differential equations of Eqs. 4.5, 4.7, and 4.8, has been developed to describe the breakthrough processes. 95

112 (a) (b) 96

113 (c) Figure 4.4 Breakthrough curves of sorbents with different inlet mercury concentrations, (a) Raw-AC, (b) 4%(wt) CuCl 2 -AC, and (c) 10%(wt) CuCl 2 -AC The saturation of HgCl 2 adsorption onto the sorbent is considered to reach a thermodynamic equilibrium condition. The amount of HgCl 2 adsorbed onto a sorbent at a saturation point (q*) was estimated from a breakthrough curve, and was also measured using a digestion method, the difference of which was less than 15%. The values of q* obtained from different inlet HgCl 2 concentrations give the adsorption isotherms depicted in Figure 4.5, which have a characteristic Langmuir shape. The Langmuir parameters of the maximum adsorption, q* max, and Langmuir adsorption constant, K, were determined by taking a reciprocal of Eq. (4.6) and listed in Table 4.5. The Langmuir isotherms clearly show that the higher the CuCl 2 loading on activated carbon, the lower the capacity of HgCl 2 adsorption. This indicates that only the carbon site (i.e. CuCl 2 free site) is responsible for HgCl 2 adsorption. Langmuir adsorption 97

114 q*, g HgCl 2 /g sorbent constant (K) increases with an increase in CuCl 2 loading. Usually, high K corresponds to high binding energy of adsorption.[29] C*, mg/m x x x Raw AC 4% CuCl 2 -AC 10% CuCl 2 -AC Simulated Languir isotherm C*, ppbv Figure 4.5 Langmuir adsorption isotherms of HgCl 2 adsorption on AC Table 4.5 Langmuir parameters determined from three sorbents q* max (g/g) (m 3 /g) Raw AC %(wt) CuCl 2 -AC %(wt) CuCl 2 -AC K Simulation results A set of the coupled partial differential equations (PDEs) (Eq. 4.5, 4.7, 4.8) were solved using the finite element method under the platform of COMSOL Multiphysics (Version 4.2a). COMSOL Multiphysics is a commercial finite element solver for various physics and 98

115 engineering applications.[35] COMSOL Multiphysics allows for modeling and simulation for coupled PDE systems. The PDEs can be entered directly or using the weak formulation. The simulation in this work is mainly conducted with Mathematics PDE Module. Parameter Estimation of Adsorption Kinetic Constant The values of K and q max were determined by the Langmuir isotherm (in Table 4.5). The only unknown parameter in the PDE equations is k 1 (adsorption kinetic constant), which can be estimated by fitting the model calculation to the experimental breakthrough curves. Then the desorption kinetic constant k 2 was determined using k 2 = k 1 /K. Figure 4.6 shows the fitting process by taking the raw AC at HgCl 2 inlet concentration C B_in =10 ppbv as an example. The kinetic constants determined for adsorption k 1 and desorption k 2 that give the best fit are listed in Table 4.6. It is observed that the kinetic constants vary with sorbents, but are independent of inlet HgCl 2 concentrations. It is interesting to note that the adsorption kinetic constant k 1 vary insignificantly in terms of different sorbents. The desorption kinetic constant k 2, however, decreases with an increase in CuCl 2 loading, due to an increase in K indicating an increase in adsorption binding energy. Table 4.6 Kinetic parameters for the adsorbing materials k 1 k 2 (m 3 g -1 s -1 ) (s -1 ) Raw AC %CuCl 2 -AC %CuCl 2 -AC

116 Figure 4.6 Adsorption constant (k 1 ) estimation by fitting simulation results with experimental breakthrough data (Raw AC, C B _ in =10 ppbv, d p =15 μm) The breakthrough curves calculated from Eqs. (4.7) and (4.8) are plotted in Figure 4.7(ac). It can be seen from the Figure that the simulation results are in good agreement with the experimental data. The deviation between simulations and experimental data are probably caused by the realities that the sorbent particle is neither spherical nor uniform in size, and the uncertainty of measurement. 100

117 (a) (b) 101

118 (c) Figure 4.7 Simulation results for (a) Raw-AC, (b) 4%CuCl 2 -AC, and (c) 10%CuCl 2 -AC (d p =15 μm) HgCl 2 Concentration Profiles The solutions to the model allow us to extensively examine the HgCl 2 concentration profiles both inside the sorbent particle and in the bulk gas phase along the fixed-bed length. Here we only report the simulation results by using Raw AC as a demonstration, with the inlet HgCl 2 concentration of 5 ppbv and a particle diameter of 15 µm. Figure 4.8 (a, b) shows the results of HgCl 2 concentrations inside the raw AC particle (C)after 50 hours of the adsorption process in the entire fixed bed, and HgCl 2 concentration profiles inside the raw AC particle positioned at z/l=0.2 of the fixed-bed length at different adsorption times. Figure 4.9 (a,b) reports the HgCl 2 uptake (q) with respect to radius of the sorbent after 50 hours of the adsorption process, and HgCl 2 uptake (radius profiles) inside the raw AC particle positioned at z/l=0.2 of 102

119 the fixed-bed length. The clearly show that the internal HgCl 2 concentration gradient is not significant because the particle size used in this example is not big (i.e. 15 µm) and thus the pore diffusion effect is not significant, which will be discussed in the following section. (a) (b) Figure 4.8 HgCl 2 concentration (C) inside raw AC particle along the particle radius, (a) at the time of 50 hr and (b) at the position of z/l=0.2 (Raw AC, C B _ in =5 ppbv, d p =15 μm). (a) (b) Figure 4.9 HgCl 2 uptake (q) on raw AC along the particle radius, (a) at the time of 50 hrs and (b) at the position of z/l=0.2 (Raw AC, C B _ in =5 ppbv, d p =15 μm). 103

120 (a) (b) Figure 4.10 (a) HgCl 2 concentration in the bulk gas phase and (b) HgCl 2 uptake on the sorbents along fixed-bed length (Raw AC, C B _ in =5 ppbv, d p =15 μm). Figure 4.10 (a, b) shows the HgCl 2 concentration (C B ) profiles in the bulk gas phase and the HgCl 2 uptake (cumulative q) profiles on the sorbent along the fixed-bed length. The outlet 104

121 concentration results from the simulation are in agreement with the experimental breakthrough data Effects of Pore Diffusion Particle size is an important factor that can affect the rate of HgCl 2 adsorption on sorbents and thus ultimately the rate of sorbent injection. By selecting different particle sizes, we can study the adsorption behavior (e.g. breakthrough curve, HgCl 2 concentration radius profiles inside sorbent particles) of HgCl 2 on sorbents. Figure 4.11 reports the breakthrough curves from the Raw AC sorbents with different particle size (an inlet HgCl 2 concentration of 10 ppbv was used in the computation). It can be seen from the Figure that the computed breakthrough curves obtained from sorbents with particle sizes between µm are the best fits when compared with the experimental data, which were obtained from a sorbent with an average particle size of 15 µm. Larger particle sizes make the sorbents take longer time to reach saturation and the breakthrough curves begin early, but increase slowly. Figure 4.11 Breakthrough curves at different particle sizes (Raw AC, C B _ in =10 ppbv) 105

122 Figure 4.12 also show the effects of different particle sizes on HgCl 2 adsorption at the position of z/l=0.2 after 50 hours. The HgCl 2 concentration gradient increases greatly with an increase in particle size (from the flat curve of 10 µm particle colored in dark blue to the sloping curve of 40 µm particle in light blue). The result evidently shows that for sorbents with large particle sizes, pore diffusion is a rate-controlling step that slows downs the adsorption rate. A set of complete HgCl 2 concentration profiles inside the sorbent particle (Raw AC) with different sizes (10, 20, 30, and 40 µm) along the whole fixed-bed length are depicted in Figure Figure 4.12 HgCl 2 concentration profile inside particle with different particle sizes after 50 hrs (Raw AC, C B _ in =10 ppbv) 106

123 (a) (b) (c) (d) Figure 4.13 HgCl 2 concentration profiles inside sorbent particles with different sizes along the fixed-bed length after 50 hrs, (a) 10 µm, (b) 20 µm, (c) 30 µm, and (d) 40 µm Effects of Axial Diffusion and Film Resistance In order to investigate the effects of mass-transfer resistance on the mercury adsorption of activated carbons, we also perform the calculation by reasonably simplifying the diffusion pathways of the HgCl 2 molecules adsorption on activated carbons. That is, on one hand, when 107

124 the superficial velocity is large enough, the effect of external resistance can be ruled out; on the other hand, when the particle size is small enough, the internal resistance is negligible, thus, C bulk C C (4.14) surface pore C is uniform in the pores of sorbent, and then it is not the function of r. Therefore, Eqs. (4.5) and (4.7) are not required. Plug dq ra obs ra in equation (4.8), p dt C b t C u Z dq p dt ( b 1 ) (4.15) in which dq k1 ( qmax q) C k2q (4.16) dt I.C. C 0, q 0 at t=0, B. C. C C B _ in at z=0 This approach allows us to investigate the effects of axial diffusion and film resistance on HgCl 2 adsorption on several activated carbon sorbents by comparing the simulation results calculated from complete and simplified models. We hereby define Simplified Model, as schematically shown in Figure The term of a/v is spherical area in unit of m 2 /m 3 particle. Model (i) ignoring both external diffusion and axial diffusion; Model (ii) only considering axial diffusion; Model (iii) only considering external film diffusion; and Model (iv) full set of PDEs considering both external diffusion and internal diffusion. 108

125 Figure 4.14 Four models describing adsorption kinetics with different assumptions For a demonstration purpose, HgCl 2 adsorption on raw AC was selected with 10 ppbv inlet HgCl 2 concentration and an operation temperature of 140 C. Figure 4.15 shows typical breakthrough curves calculated from four different models. The breakthrough curves from the four models are very close to one another, indicating that axial diffusion and film resistance do not significantly affect HgCl 2 adsorption conducted under the experimental conditions. 109

126 Figure 4.15 A comparison of simulation results obtained from four models taking into account the effects of axial diffusion and film resistance Comparison of pore diffusion and surface adsorption The simulation also allows us to determine the controlling step of HgCl 2 adsorption by comparing the concentration driving forces of pore diffusion and surface adsorption. As discussed in the last Section, the external diffusion resistance is negligible, and the bulk HgCl 2 concentration is the same as surface concentration of the sorbent particle (e.g. C B = C at r = R p ). Figures 4.16 (a) and (b) show the HgCl 2 concentration profiles inside the sorbent particle with the particle diameter of 15 µm and 40 µm, respectively. The difference between surface concentration and interior concentration (C B C) is a driving force for pore diffusion (the distance between the dashed line and solid line in Figure 4.16). C* is defined as the HgCl 2 110

127 concentration in equilibrium with solid HgCl 2 concentration on sorbent (q), given by the Langmuir equation: C* q K( q max q) Therefore, (C C*) represents the driving force for surface adsorption, shown as the distance between the dotted line and solid line in Figure By comparing (C B C) and (C C*), we can determine the contributions of pore diffusion resistance and surface adsorption resistance. For small particle (i.e. d p =10 μm), (C B C) is less than (C C*), which means the pore diffusion resistance is less significant than the surface adsorption resistance; hence the surface adsorption resistance is a major rate-limiting step for adsorption. For large particles (i.e. d p =30 μm), (C B C) is much greater than (C C*), which means that pore diffusion is a controlling step in the entire adsorption process. (a) (b) Figure 4.16 HgCl 2 concentration profiles inside particle with particle sizes of (a) d p =10 μm and (b) d p =30 μm (Raw AC, C B _ in =5 ppbv, z/l=0.2, Time=40 hr) 111

128 4.4.6 Model and Simulation for Entrained-flow System Similar approaches can be used to simulate the adsorption behavior of HgCl 2 on sorbents in an entrained-flow system. An entrained-flow reactor is a system in which the sorbents (e.g. by point ejection) flow together with flue gas. In-flight state of the sorbent is schematically shown in Figure Figure 4.17 Schematic of HgCl 2 adsorption in an entrained-flow system The equations for a local adsorption rate (Eq. 4.5) and a mass balance taken for a single sorbent particle (Eq. 4.7) in a fixed-bed system are also valid for an entrain-flow system. However, the mercury mass balance for the moving sorbent and gas phase in ductwork changes as shown below. This equation is written by adopting the Eulerian approach. dcb 3mAC C De dt R r p p ( r Rp, t) (4.17) m AC : AC loading per unit volume of bulk gas phase, g AC/m 3. Initial conditions: C(r, t=0)=0, C B, t=0)=0, q (r, t=0)=0 Boundary conditions: C B (0,t)=C B_in, C r q 0; 0 r ( r 0, t) ( r 0, t) 112

129 C KG ( C B( t ) C ( R p, t )) r D ( r Rp, t) e By solving the coupled PDEs (i.e. Eqs. 4.5, 4.7 & 4.17), the concentration profiles C(r, t), C B ( t), q(r, t) can be obtained. Figure 4.18 shows that the bulk HgCl 2 concentration decreases with respect to sorbent residence time in the ductwork. The figure also reports that the smaller the particle size, the higher the HgCl 2 removal. When the particle size is smaller than 10 µm, further reducing the particle size results in no significant changes in HgCl 2 removal (see blue and green curves in Figure 4.18). Figure 4.19 gives the HgCl 2 concentration profiles inside the particle for different particle sizes, which shows a similar trend to the results obtained from the fixed-bed system (Figure 4.10(a) and Figure 4.12). Figure 4.18 Bulk-phase HgCl 2 concentration (C B ) as a function of sorbent residence time in the ductwork for different particle sizes, (Raw AC, C B _ in =10 ppbv, m AC =10 g/m 3 ) 113

130 Figure 4.19 HgCl 2 concentration profile inside particle for different particle sizes after 1 s (Raw AC, C B _ in =10 ppbv, m AC =10 g/m 3 ) Figure 4.20 reports HgCl 2 removal with respect to sorbent loading in an entrained-flow system with a sorbent residence time of 10 s with different particle sizes. Fine particles possess high HgCl 2 removal efficiency at low sorbent loadings due to the small pore diffusion resistance. For example, in order to achieve 90% removal of HgCl 2 with a sorbent residence time of 10 s, 10 g raw AC per cubic meter flue gas need to be injected into the duct for 20 µm particles. The value increases to 30 g per cubic meter for 80 µm particles. Figure 4.21 reports the HgCl 2 removal efficiency of different sorbents in an entrain-flow system. CuCl 2 -AC sorbents, which are capable of converting Hg(0) into Hg 2+, show lower removal efficiencies due to less active carbon sites for HgCl 2 adsorption. This indicates that the injection loading of CuCl 2 -AC sorbents need to be controlled depending on mercury speciation in flue gas. 114

131 Figure 4.20 HgCl 2 removal with respect to sorbent loading in entrained flow system for different particle sizes (Raw AC, C B _ in =10 ppbv, time=10 s) Figure 4.21 HgCl 2 removal with respect to sorbent loading in entrained flow system for different sorbents (C B _ in =10 ppbv, dp =20 µm, time=10 s) 115

132 4.5 Conclusions A detailed kinetic model for HgCl 2 adsorption on CuCl 2 -impregnated AC sorbents in a fixed-bed reactor system is presented in this chapter. In the model, mercury mass balances were taken in both gaseous and the adsorbed phases along the reactor length and inside the sorbent particle, by taking into account external and internal mass-transfer resistances and adsorption kinetics. A formulated set of coupled partial differential equations (PDEs) was solved by finite element method using COMSOL Multiphysics. The model has been applied to a commercial AC sorbent (DARCO HG), and two CuCl 2 -impregnated AC sorbents. HgCl 2 adsorption experiments were carried out by varying HgCl 2 concentrations in the gas phase. The range of the operating parameters in a fixed-bed reactor (based on the lab-scale experiments) has been chosen in order to simulate the adsorption behavior in the typical flue gas conditions. The Langmuir adsorption equilibrium constants of three sorbents were calculated to be m 3 /g for raw AC, m 3 /g for 4%(wt) CuCl 2 -AC, m 3 /g for 10%(wt) CuCl 2 -AC, respectively. The kinetic adsorption constants were estimated by fitting the model simulation results with experimental data. The breakthrough data from experiments were in good agreement with the simulation results from the modified kinetic model. Pore diffusion resistance was predicted to significantly increase with an increase in sorbent particle size. For large particles (> 30 µm), pore diffusion controls the total adsorption rate. For the sorbents used in this work (d p =15 µm), both pore diffusion and adsorption step was found to be the ratelimiting step. External mass transfer resistance turns out to be negligible under the experimental conditions. 116

133 The performance of HgCl 2 adsorption removal was also predicted in an entrained-flow system using a modified model. Smaller particles exhibit higher HgCl 2 removal efficiency when the sorbent loading and residence time are the same. However, simulation results show that sorbents with particle size less than 10 µm predicted a similar HgCl 2 removal efficiency at the same sorbent loading and residence time. A HgCl 2 removal efficiency of >90% can be achieved by using 10 g (raw AC)/m 3 (flue gas) loading of sorbent (particle size 20 µm) with a residence time of 10 s. 4.6 Reference 1. Sjostrom, S.; Durham, M.; Bustard, C. J.; Martin, C., Activated carbon injection for mercury control: Overview. Fuel 2010, 89, (6), Presto, A. A.; Granite, E. J., Survey of Catalysts for Oxidation of Mercury in Flue Gas. Environmental Science & Technology 2006, 40, (18), Jones, A. P.; Hoffmann, J. W.; Smith, D. N.; Feeley, T. J.; Murphy, J. T., DOE/NETL's Phase II Mercury Control Technology Field Testing Program: Preliminary Economic Analysis of Activated Carbon Injection. Environmental Science & Technology 2007, 41, (4), Pavlish, J. H.; Sondreal, E. A.; Mann, M. D.; Olson, E. S.; Galbreath, K. C.; Laudal, D. L.; Benson, S. A., Status review of mercury control options for coal-fired power plants. Fuel Processing Technology 2003, 82, (2 3), Srivastava, R. K.; Hutson, N.; Martin, B.; Princiotta, F.; Staudt, J., Control of mercury emissions from coal-fired in electric utility boilers. Environmental Science & Technology 2006, 40, (5), Yang, H.; Xu, Z.; Fan, M.; Bland, A. E.; Judkins, R. R., Adsorbents for capturing mercury in coal-fired boiler flue gas. Journal of Hazardous Materials 2007, 146, (1 2), Lee, J. Y.; Ju, Y. H.; Keener, T. C.; Varma, R. S., Development of cost-effective noncarbon sorbents for Hg-0 removal from coal-fired power plants. Environmental Science & Technology 2006, 40, (8),

134 8. Lee, S. S.; Lee, J. Y.; Keener, T. C., Bench-Scale Studies of In-Duct Mercury Capture Using Cupric Chloride-impregnated Carbons. Environmental Science & Technology 2009, 43, (8), Li, X.; Lee, J.-Y.; Heald, S., XAFS characterization of mercury captured on cupric chloride-impregnated sorbents. Fuel 2012, 93, Lee, S.-S.; Lee, J.-Y.; Khang, S.-J.; Keener, T. C., Modeling of Mercury Oxidation and Adsorption by Cupric Chloride-Impregnated Carbon Sorbents. Industrial & Engineering Chemistry Research 2009, 48, (19), Wang, J.; Clements, B.; Zanganeh, K., An interpretation of flue-gas mercury speciation data from a kinetic point of view. Fuel 2003, 82, (8), Huggins, F. E.; Yap, N.; Huffman, G. P.; Senior, C. L., XAFS characterization of mercury captured from combustion gases on sorbents at low temperatures. Fuel Processing Technology 2003, 82, (2-3), Karatza, D., Kinetics of adsorption of mercuric chloride vapors on sulfur impregnated activated carbon. Combustion Science and Technology 1996, 112, Scala, F., Simulation of mercury capture by activated carbon injection in incinerator flue gas. 1. In-duct removal. Environmental Science & Technology 2001, 35, (21), Scala, F., Simulation of mercury capture by activated carbon injection in incinerator flue gas. 2. fabric filter removal. Environmental Science & Technology 2001, 35, (21), Scala, F.; Clack, H. L., Mercury emissions from coal combustion: modeling and comparison of Hg capture in a fabric filter versus an electrostatic precipitator. J Hazard Mater 2008, 152, Krishnan, S. V.; Gullett, B. K.; Jozewicz, W., Sorption of Elemental Mercury by Activated Carbons. Environmental Science & Technology 1994, 28, (8), Ho, T. C.; Kobayashi, N.; Lee, Y.; Lin, J.; Hopper, J. R., Experimental and Kinetic Study of Mercury Adsorption on Various Activated Carbons in a Fixed-Bed Adsorber. Environmental Engineering Science 2004, 21, (1), Skodras, G.; Diamantopoulou, I.; Pantoleontos, G.; Sakellaropoulos, G. P., Kinetic studies of elemental mercury adsorption in activated carbon fixed bed reactor. Journal of Hazardous Materials 2008, 158, (1),

135 20. Lee, S.-S.; Lee, J.-Y.; Keener, T. C., The effect of methods of preparation on the performance of cupric chloride-impregnated sorbents for the removal of mercury from flue gases. Fuel 2009, 88, (10), Li, Y. H.; Lee, C. W.; Gullett, B. K., Importance of activated carbon's oxygen surface functional groups on elemental mercury adsorption. Fuel 2003, 82, (4), Li, J.; Yan, N.; Qu, Z.; Qiao, S.; Yang, S.; Guo, Y.; Liu, P.; Jia, J., Catalytic Oxidation of Elemental Mercury over the Modified Catalyst Mn/α-Al2O3 at Lower Temperatures. Environmental Science & Technology 2010, 44, (1), Diamantopoulou, I.; Skodras, G.; Sakellaropoulos, G. P., Sorption of Mercury by Activated Carbon in the Presence of Flue Gas Components.. Fuel Processing Technology 2010, 91, Lin, H. Y.; Yuan, C. Y.; Chen, W. C.; Hung, C. H., Determination of the adsorptive capacity and adsorption isotherm of vapor-phase mercury chloride on powdered AC using thermogravimetric analysis. J. Air & Waste Management Association 2006, 56, Chen, W. C.; Lin, H. Y.; Yuan, C. Y.; Hung, C. H., Kinetic modeling on the adsorption of vapor-phase mercury chloride on activated carbon by thermogravimetric analysis.. J. Air & Waste Management Association 2009, 59, Niksa, S.; Fujiwara, N., Predicting extents of mercury oxidation in coal-derived flue gases. Journal of the Air & Waste Management Association 2005, 55, (7), Wang, J. S.; Anthony, E. J., An analysis of the reaction rate for mercury vapor and chlorine. Chemical Engineering & Technology 2005, 28, (5), ASTM, Standard Test Method for Elemental, Oxidized, Particle-Bound and Total Mercury in Flue Gas Generated from Coal-Fired Stationary Sources (Ontario Hydro Method). In 2008; Vol. D Ruthven, M. D., Principles of adsorption and adsorption processes. Wiley: New York, 1984, pp Chen, S.; Rostam-Abadi, M.; Chang, R. In Mercury removal from combustion flue gas by activated carbon injection: mass transfer effects,, Prepr. Pap.- Am. Chem. Soc., Div. Fuel Chem., 1996; 1996; pp Cussler, E. L., Diffusion: Mass Transfer in Fluid System Cambridge University Press: New York,

136 32. Bird, R. B.; Stewart, W. E.; Lightfoot, E. N., Transport Phenomena, 2nd Edition. Wiley: 2001, pp Welty, J. R.; Wicks, C. E.; Wilson, R. E.; Rorrer, G. L., Fundamentals of Momentum, Heat and Mass Transfer (5th ed.). Wiley and Sons: Levenspiel, O., Chemical Reaction Engineering. 3rd ed.; John Wiley & Sons: Hoboken, NJ, 1999, pp COMSOL, 120

137 Chapter 5 Oxidation of Elemental Mercury Vapor over CuCl 2 /α-al 2 O 3 Catalysts for Mercury Emissions Control * 5.1 Introduction As has been pointed out, the U.S. EPA s new rule proposes to reduce mercury emissions by >90% starting from 2016 by adopting a Maximum Achievable Control Technology (MACT) approach, instead of a Cap & Trade approach adopted by the rescinded Clean Air Mercury Rule (CAMR).[1] Apparently, more strictly control of elemental mercury (Hg(0)) emissions will have to be implemented in most existing coal-fired power plants. The U.S. EPA estimates that the installation of activated carbon injection systems will increase from current 48 gigawatts (GW) to 99 GW by 2020 (according to the ToxR policy case).[2] In July 2010 prior to the MATS, the U.S. EPA also issued a new proposed rule, the Transport Rule, which replaces the 2005 Clean Air Interstate Rule (CAIR) and will start to regulate sulfur dioxide (SO 2 ) and nitrogen oxide (NO x ) emissions from power plants in 28 states from 2012.[3, 4] The U.S. EPA also estimates that a total GW of Flue Gas Desulfurization (FGD) and GW of Selective Catalytic Reduction (SCR) units of a total 373 GW to be generated from coal combustion would be operative by 2020 in order to meet the Transport Rule requirements (according to the TR SB Limited Trading model).[2] Since the use of Powder River Basin (PRB) subbituminous coal, which generates higher percentages of elementary mercury vapor, is increasing,[5] and the proposed Transport Rule is very likely to increase the installation of wet FGD systems (>95% for SO 2 control on a basis of * Part of the content in this chapter has been submitted to the Environmental Science & Technology for publication. 121

138 total electricity generation) and SCR units for large coal-fired power plants. In this context, heterogeneous Hg(0) oxidation using catalysts or oxidants is highly expected to play a critical role in future mercury emissions control in the U.S.[6-8] Most mercury compounds that can be formed in typical coal combustion flue gas are known to be weakly bonded, and only HgCl 2 and HgBr 2 are considered to have significant bond strengths.[9] Among these oxidized mercury species, HgCl 2 has high solubility in water (i.e g/100 g water at 25 C) and other oxidized forms have very low solubility.[10] Therefore, HgCl 2 is the most desirable oxidized form for capture in wet FGD systems. To date, noble metals and metal oxides have been primarily studied for heterogeneous catalytic Hg(0) oxidation. Noble metal-based catalysts including gold (Au), platinum (Pt), and palladium (Pd) were evaluated as Hg(0)-specific catalysts, and Pd has been reported to display promising results.[7, 11-13] These have shown limited success in the absence of very low concentrations of HCl or Cl 2 gas. Recently, various metal oxide-based Hg(0) catalysts including V 2 O 5, MoO 3, Cr 2 O 3, MnO x, CeO 2, Co 3 O 4 and RuO 2 have been studied for the development of a Hg(0)-specific or modified SCR catalyst.[14-18] However, even in the presence of HCl gas, many of these metal oxide catalysts exhibited limited Hg(0) oxidation at a low HCl level typically found in subbituminous or lignite coal combustion flue gas (e.g. 10 ppmv) due to competitive adsorption of SO 2, NH 3, HCl, and Hg(0) gases onto active metal oxide sites. A RuO 2 -modified SCR catalyst has been recently reported to demonstrate excellent Hg(0) vapor oxidation performance with good tolerance to SO 2 and NH 3.[18] All of these previous studies indicate that noble metals and some of the metal oxides can oxidize Hg(0) via the Deacon reaction followed by successive Hg(0) oxidation with atomic chlorine (Cl) and/or molecular chlorine (Cl 2 ) over the metal surfaces. However, the adsorption of HCl gas molecules onto metal 122

139 oxide surfaces is very competitive, and is most likely a rate-limiting step for Hg(0) oxidation via the Deacon reaction. 0 2HCl 1/ 2O Cl H O Deacon reaction, H 28.4 kj/mol (5.1) Cl Hg(0) Cl HgCl HgCl 2 or HgCl Cl2 HgCl 2 Cl (5.2) Our previous studies on cupric chloride (CuCl 2 )-impregnated materials showed excellent Hg(0) oxidation performance in the absence of HCl gas throughout fixed-bed and entrained-flow tests.[19, 20] When CuCl 2 is dispersed over non-carbonaceous substrates, CuCl 2 readily oxidizes Hg(0) vapor, but its resultant oxidized mercury is not easily adsorbed onto the noncarbonaceous substrate surface. Carbon seems to be the only substrate that can effectively adsorb the oxidized mercury. The resultant oxidized mercury species have recently been found to be primarily mercuric chloride (HgCl 2 ) using X-ray Absorption Fine Structure (XAFS) spectroscopy.[21] Based on these findings, the objective of this study is to investigate the performance of CuCl 2 /α-al 2 O 3 catalyst on Hg(0) oxidation under different gases at a typical flue gas temperature after the air preheater, 140 C, and the oxidation mechanism of CuCl 2 on Hg(0) vapor. 5.2 Experimental Section Catalyst Preparation There are many factors that can influence the performance of CuCl 2 /α-al 2 O 3 catalyst on Hg(0) oxidation such as CuCl 2 loading, dispersion, copper speciation, surface area, and pore volume. In this study, nonporous α-al 2 O 3 was selected as a support in order to probe the surface reaction taking place between Hg(0) vapor and CuCl 2, and 10%(wt) CuCl 2 was used throughout the study. A CuCl 2 /α-al 2 O 3 catalyst was prepared by impregnating CuCl 2 2H 2 O (Aldrich, 123

140 >99.99%) onto α-al 2 O 3 (Alfa Aesar aluminum oxide-43862, ϕ1/8 pellets) in the aqueous phase following the incipient wetness method. After impregnation, the sample was dried at 100 C for 8 hrs. The amount of CuCl 2 loading used for the catalyst in this study was 10%(wt). The synthesized catalyst was ground and sieved to ~40μm particles for the characterization and performance evaluation of Hg(0) oxidation Evaluation of CuCl 2 /α-al 2 O 3 Catalyst A lab-scale fixed-bed system was used for the performance evaluation of Hg(0) oxidation and preparation of a spent catalyst for X-ray Photoelectron Spectroscopy (XPS) and X-ray Diffraction (XRD) analysis. 25 mg of the catalyst was mixed with 4 g quartz sand and then was loaded into the fixed-bed reactor with an inner diameter of 12 mm. The length of reaction zone was ~20 mm. Before loading a catalyst, a blank test has been performed and showed negligible Hg(0) adsorption on the internal wall of either borosilicate reactor or Teflon tubing. The outlet mercury speciation was measured by the Ontario Hydro method. A 1 M KCl impinger solution was used to capture oxidized mercury. A 4%(w/v) KMnO 4 /10%(v/v) H 2 SO 4 impinger solution was used to capture Hg(0). Oxidized mercury and Hg(0) concentrations in the effluent gas stream were determined by analyzing those solutions using a cold vapor atomic absorption spectrophotometer (CVAAS, Model 400A, Buck Scientific Inc., East Norwalk, CT). More detailed information on the system and experiments was described in our previous study.[22] The inlet Hg(0) concentration was 0.25 mg/nm 3 (= 30 ppbv) in 1 L/min of carrier gas flow, and the reactor was placed in an oven maintained at 140 C, targeting an installation of the Hg(0) catalytic bed after the air preheater. Four different gas conditions were used in this study: (1) N 2 (99.999% UHP, Wright Brothers, Inc.); (2) 6% O 2 (balanced with N 2 ); (3) 10 ppmv HCl in O 2 (6% 124

141 O 2 balanced with N 2, Wright Brothers, Inc.); and (4) 2,000 ppmv SO 2 and 10 ppmv HCl in O 2 (6% O 2 balanced with N 2, Wright Brothers, Inc.). When spent catalysts were prepared for characterization, the catalysts were not mixed with quartz sand and an inlet Hg(0) vapor concentration was raised to ~2 mg/nm 3 (= 240 ppbv) in order to saturate the catalyst in a short period of time Characterization Thermal Gravimetric Analysis-Mass Spectrometry (TGA, TA Instruments Q5000IR, and MS, Pfeiffer-Vacuum Thermostar) was used to evaluate the thermal stability of pure CuCl 2 and CuCl 2 /α-al 2 O 3. Approximately 10 mg of a sample was used to ramp from room temperature to 800 C with a heating rate of 5 C/min under nitrogen flow (99.999% UHP) at a rate of 100 ml/min. The H 2 -Temperature Programmed Reduction (TPR) experiments were also performed using the Micromeritics Autochem 2910 automated catalyst characterization system with ~50 mg of samples. The samples were pretreated at 300 C for 1 h in ultra-pure argon gas at a flow rate of 30 ml/min. After the furnace temperature decreased to ~50 C, the feed gas (10%(v) H 2 balanced with AR) was fed into the system at a flow rate of 30 ml/min. H 2 -TPR runs were performed by heating the sample up to 600 C at a linear heating rate of 6.5 C/min. The amount of hydrogen gas consumed during the TPR experiment was measured by a built-in thermal conductive detector (TCD). XPS (Kratos Axis Ultra, with a mono-chromatid Al Kα source ( ev)) measurements were used to examine copper speciation with a concentric hemispherical analyzer. The power applied to the x-ray gun was 130 W (I=10 ma, V=13 kv), and the base vacuum pressure was Torr, and the pass energy of the analyzer was set to 20 ev. The resolution of the spectra was ~0.2 ev with the step size of 0.1 ev. The carbon 1s peak 125

142 at ev was used to charge reference all binding energies. XRD data were obtained using Cu Kα radiation with a wavelength of Å (X Pert Pro MPD X-ray diffractometer). An aluminum holder was used to support the catalyst samples. The scanning range was from (2θ) with a step size of 0.02 and a step time of 0.5 s. 5.3 Results and Discussion Hg(0) Oxidation over CuCl 2 /α-al 2 O 3 Catalyst The Hg(0) oxidation over 10%(wt) CuCl 2 /α-al 2 O 3 catalyst evaluated under different gas conditions are shown in Figure 5.1, and their mercury mass balances are summarized in Table 5.1. Before the performance evaluation, it was confirmed that unsupported alumina did not adsorb Hg(0) vapor at 140 C. The mercury mass balance closures obtained from the impinger analysis are found to be within ~±3.1% of the inlet Hg(0) concentration. The amount of mercury determined from the digestion of spent catalysts was found to be less than 0.2% of the total amount of inlet Hg(0) added over the evaluation period, all of which indicate that the adsorption of both Hg(0) and oxidized mercury (Hg 2+ ) onto the internal surfaces of the reactor system is negligible and almost all mercury left the reactor system. The results clearly show that the catalyst can oxidize almost all Hg(0) vapor during the first 24 hrs under any carrier gas conditions. The 10%(wt) CuCl 2 / -Al 2 O 3 catalyst lost its oxidation capability after ~120 hrs under a nitrogen carrier gas. Based on the molecular weight of CuCl 2 (MW=134.5) and 30 ppbv Hg(0) vapor in 1 L/min of a total flow rate, the Hg(0) oxidation reaction is estimated to complete with CuCl 2 in ~120 hrs when the reaction takes place at a stoichiometric ratio of 2 between CuCl 2 and Hg(0) (i.e. 2CuCl 2 : Hg(0)). This mercury speciation result also corresponds to our previous result obtained from CuCl 2 doped onto 126

143 Hg(0),out Hg(0),in activated carbon, and corroborates our proposed reaction scheme reported in our previous study.[19, 23] When Hg(0) vapor was introduced in N 2 and O 2 gases, O 2 gas was found to have the capability to promote Hg(0) oxidation compared to N 2 gas. However, the Hg(0) oxidation started to decrease and eventually stopped under both carrier gases. Our previous study also showed that CO 2, H 2 O, and NO gases did not significantly affect the Hg(0) oxidation performance. On the other hand, it is very interesting to note that Hg(0) vapor was almost completely oxidized in the presence of 10 ppmv HCl and 6%(v) O 2 gas balanced with N 2 gas, even in the presence of 2,000 ppmv SO 2 gas over 140 hrs of the performance evaluations. SO 2 gas has been reported to adversely affect Hg(0) oxidation over many metal oxide-based catalysts.[15, 16, 24] These results indicate that the CuCl 2 /α-al 2 O 3 catalyst has excellent resistance to SO 2 and can be used to oxidize Hg(0) even at high SO 2 concentrations. 100% Pure N2 6% (v) O2 balanced with N2 10 ppmv HCl + 6% (v) O2 in N2 10 ppmv HCl + 2,000 ppmv SO2 + 6% (v) O2 in N Time (hr) Figure 5.1 Breakthrough curves of Hg(0) by 10%(wt) CuCl 2 /α-al 2 O 3 catalyst under different gas conditions at 140 C. 127

144 Table 5.1. Mercury speciation results from 10%(wt) CuCl 2 / -Al 2 O 3 fixed-bed tests Time (hr) Hg species (%) Inlet Hg Under N 2 Outlet Hg ±1 16±2 62±3 93±3 97±2 Outlet Hg 2+ 92±2 96±2 93±2 79±2 36±2 8±2 2±2 Total outlet Hg 92±2 96±2 96±3 95±4 98±5 101±5 99±4 Time (hr) Hg species Under 6%(v) (%) O 2 Under 6%(v) O ppmv HCl in N 2 Under O 2, 10 ppmv HCl + 2,000 ppmv SO 2 in N 2 Inlet Hg Outlet Hg 0 0 1±1 2±1 6±3 19±2 35±4 54±4 Outlet Hg 2+ 94±2 97±3 94±3 91±2 79±3 67±2 45±2 Total outlet Hg 94±2 98±4 96±4 97±5 98±5 102±6 99±6 Time (hr) Hg species (%) Inlet Hg Outlet Hg 0 0 1±1 2±1 1±1 2±1 1±1 3±1 Outlet Hg 2+ 94±3 97±2 94±1 96±2 96±3 100±2 96±3 Total outlet Hg 94±3 98±3 96±2 97±3 98±4 101±3 99±4 Time (hr) Hg species (%) Inlet Hg Outlet Hg 0 1±1 2±1 0 2±1 2±1 3±2 2±1 Outlet Hg 2+ 95±4 95±2 96±1 95±2 99±2 97±2 97±3 Total outlet Hg 96±5 97±3 96±1 97±3 101±3 100±4 99±4 Note: Inlet Hg 0 concentration = 30±2 ppbv 128

145 These notable performance differences in Hg(0) oxidation over the CuCl 2 /α-al 2 O 3 catalyst under different gaseous conditions prompted us to conduct the following catalyst characterization study on the spent catalyst in order to understand the Hg(0) oxidation mechanisms Characterization Results of CuCl 2 /α-al 2 O 3 Catalyst BET Surface Area and Pore Volume Analysis The α-al 2 O 3 substrate pellet has a macrospore structure with very small surface area (0.25 m 2 /g) and pore volume (0.26 ml/g). The SEM images in Figure 5.2 show that the α- alumina powder is a non-porous material primarily with external surface areas, and CuCl 2 forms crystallite with a size less than 1 μm between the alumina pellets. 129

146 Figure 5.2. SEM images of α-al 2 O 3 pellet before and after CuCl 2 impregnation. TGA-MS Analysis The thermal stability of pure CuCl, CuCl 2 2H 2 O and CuCl 2 /α-al 2 O 3 catalyst was examined by TGA-MS, and the results are shown in Figure 5.3. The weight loss of 21% before 100 C from both pure CuCl 2 2H 2 O and CuCl 2 /α-al 2 O 3 catalyst is derived from the vaporization of dihydrate. Both samples exhibited weight losses within the temperature range of C. It is noted that there are two stages within the temperature range of C: the first stage of C and the second stage of C. More chlorine was detected by MS during the first stage than the second stage for the CuCl 2 2H 2 O sample, whereas the MS ion peak for the CuCl sample shows no chlorine signal. Between 500 and 550 C, the weight loss profile for CuCl 2 2H 2 O follows that for CuCl. 130

147 TGA Weight (%) MS Ion intensity (na) TGA TGA 80 TGA % CuCl2/Al2O m/e=18(h2o) m/e=35(cl) CuCl2 2H2O CuCl Temperature ( C) Figure 5.3. TGA-MS results for CuCl 2 2H 2 O and 10%(wt) CuCl 2 /α-al 2 O 3 catalyst. These result suggest that CuCl 2 turn into CuCl at 500 C after releasing chlorine between 350 and 500 C and, and then the remaining CuCl starts to evaporate without any chlorine release. Since alumina is thermally very stable up to 800 C, the 10%(wt) CuCl 2 / -Al 2 O 3 sample lost all CuCl 2 (i.e. 10% weight) during the ramping process. TPR Analysis The H 2 -TPR profiles (Figure 5.4) for CuCl 2 /α-al 2 O 3, CuCl 2 2H 2 O, and CuCl were obtained in order to evaluate the redox performance of the CuCl 2 /α-al 2 O 3 catalyst in terms of temperature. The TPR profiles for unsupported CuCl 2 2H 2 O and CuCl (Aldrich, anhydrous, 97%) show two peaks: one major peak at ~430 C and one minor peak at ~550 C for CuCl 2 ; and 131

148 one minor peak at ~380 C and one major peak at ~550 C for CuCl, respectively. The first lowtemperature major peak for CuCl 2 2H 2 O results from the reduction of Cu(II) to Cu(I), and the second minor peak is derived from the reduction of Cu(I) to Cu(0). The CuCl TPR profile shows a low-temperature minor peak and a high-temperature major peak. The low-temperature minor peak indicates the reduction of Cu(II) to Cu(I), and seems to come from a Cu(II) species included in the unsupported CuCl sample with 97% purity. 10% CuCl2/ -Al2O3 CuCl2 CuCl H2 consumption( a. u.) Temperature O C Figure 5.4. H 2 -TPR profiles for 10%(wt) CuCl 2 /α-al 2 O 3, CuCl 2 2H 2 O, and CuCl On the other hand, the 10%(wt) CuCl 2 /α-al 2 O 3 catalyst showed two distinct peaks at the temperatures lower than unsupported CuCl 2 2H 2 O and CuCl, which is in agreement with a previous study reported by Rocco, et al.[25] The lower copper reduction temperature for the supported catalyst may be derived from the increased surface/volume ratio compared to unsupported CuCl 2. The low- and high-temperature peaks indicate the two-step reduction of Cu(II) to Cu(I) and Cu(I) to Cu(0), respectively. The CuCl 2 supported on alumina distinctively showed a two-step reduction process at temperatures lower than unsupported CuCl 2 and CuCl. 132

149 XRD Characterization As presented in the performance evaluation section, the CuCl 2 /α-al 2 O 3 sample can oxidize Hg(0) vapor regardless of the presence of O 2 and HCl gases. In order to investigate the Hg(0) oxidation mechanism, XRD and XPS techniques were employed to identify the copper compounds in the crystalline phase and to examine the oxidation state of copper during the Hg(0) oxidation process. First, the XRD patterns of several samples including unsupported CuCl 2, CuCl, α-al 2 O 3 substrate, fresh CuCl 2 /α-al 2 O 3, and spent CuCl 2 /α-al 2 O 3 prepared in different Hg(0)-laden nitrogen gases were examined as shown in Figure 5.5. The XRD pattern of the fresh 10%(wt) CuCl 2 /α-al 2 O 3 sample was superimposed on those of CuCl 2 2H 2 O and α-al 2 O 3, indicating that the impregnated CuCl 2 2H 2 O predominantly has its original crystalline CuCl 2 structure. On the other hand, the spent CuCl 2 / -Al 2 O 3 showed a distinguishable difference in the XRD pattern. In N 2 flow (Figure 4(b)), the peaks derived from CuCl 2 2H 2 O became attenuated, and the CuCl peaks started to appear, indicating that the CuCl 2 2H 2 O on alumina was converted into CuCl as the Hg(0) vapor reacted with CuCl 2. On the other hand, when 10 ppmv HCl and 6%(v) O 2 were used in N 2 flow (Figure 5.5), CuCl peaks became weak at 2θ = 28.9 and disappeared at 47.2 and 56.1 while CuCl 2 2H 2 O peaks were as noticeable as the fresh catalyst shown in Figure 5.5. The presence of 2000 ppmv SO 2 did not give any noticeable difference in XRD patterns, substantiating that SO 2 does not give an impact on Hg(0) oxidation over CuCl 2 / - Al 2 O

150 Intensity [a.u.] CuCl2 2H2O CuCl CuCl2 2H2O CuCl Cu(OH)Cl - Al2O3 - Al2O3 (a) (b) CuCl2 2H2O CuCl - Al2O3 CuCl, CuCl2 2H2O - Al2O3 (c) CuCl2 2H2O CuCl - Al2O3 (d) CuCl2 2H2O CuCl - Al2O3 (e) # # Cu2OCl2 CuCl (f) CuCl2 CuCl Figure 5.5. X-ray diffraction patterns of CuCl 2 /α-al 2 O 3 obtained under different gases. (a) Fresh 10% CuCl 2 / -Al 2 O 3 ; (b) Spent 10% CuCl 2 / -Al 2 O 3 (carrier gas: N 2 ); (c) Spent 10% CuCl 2 / - Al 2 O 3 (carrier gas: 6% O ppmv HCl in N 2 ); (d) Spent 10% CuCl 2 / -Al 2 O 3 (carrier gas: 6%(v) O ppmv HCl+2,000 ppmv SO 2 in N 2 ); (e) Spent CuCl (treated in O 2 ); (f) Spent CuCl (treated in 10 ppmv HCl+O 2 ). CuCl 2 has been studied as a candidate catalyst for Cl 2 production via HCl oxidation in the presence of O 2 and HCl called the Deacon reaction.[26] In order to further investigate a 134

151 regeneration possibility of CuCl, fresh CuCl exposed to O 2 and/or HCl gas(is) was also examined with XRD. Upon the exposure of fresh CuCl to only O 2 gas, two new peaks appeared at 2θ = 17.6 and 35.6, and were attributed to the copper ox chloride species, Cu 2 OCl 2 (PDF- ICDD ). When fresh CuCl was treated under 10 ppmv HCl in O 2 ) gas, an anhydrated CuCl 2 phase (PDF-ICDD ) was identified from the spent CuCl sample. XPS Characterization XPS characterization was also used to examine the copper speciation over the CuCl 2 /α- Al 2 O 3 catalyst. Figure 5.6 shows the high resolution Cu 2p 3/2 XPS characterization result for unsupported CuCl 2, CuCl, and 10% CuCl 2 /α-al 2 O 3 catalyst before and after Hg(0) oxidation. The standard CuCl 2 and CuCl XPS profiles show that the Cu 2p 3/2 characteristic binding energy is ev for Cu(II) in CuCl 2, and ev for Cu(I) in CuCl, respectively. When the XPS result for a fresh CuCl 2 /α-al 2 O 3 sample is compared with fresh CuCl 2 and CuCl, the oxidation state of copper in the fresh CuCl 2 /α-al 2 O 3 sample is found to be predominantly Cu(II) with little Cu(I). However, after Hg(0) oxidation in N 2 gas, the major copper oxidation state of the spent CuCl 2 /α-al 2 O 3 sample turned out to be Cu(I), and the Cu(II) oxidation peak intensity started to be attenuated as the Hg(0) reaction time increased. It is interesting to note that the XPS result obtained from the spent CuCl 2 /α-al 2 O 3 sample under 10 ppmv HCl in O 2 gas showed primarily the Cu(II) oxidation state of CuCl 2. These XPS results also suggest that CuCl 2 is converted into CuCl during Hg(0) oxidation in N 2 flow, but CuCl 2 can continue to be regenerated for Hg(0) oxidation in 10 ppmv HCl in O 2 gas. In addition, the presence of 500 ppmv SO 2 did not make any significant difference as shown in Figure

152 Binding energy = 934.5eV Cu 2+ Binding energy = 931.5eV CuCl 2 CuCl Cu + (a) (b_1) (b_2) (c) (d) (e) (f) Binding Energy Figure 5.6. Cu 2p 3/2 high resolution XPS results for 10%(wt) CuCl 2 /α-al 2 O 3 obtained before and after Hg(0) oxidation under different gases. (a) fresh 10% CuCl 2 / -Al 2 O 3 ; (b_1) spent 10% CuCl 2 / -Al 2 O 3 (carrier gas: N 2, after 30 hrs); (b_2) spent 10% CuCl 2 / -Al 2 O 3 (carrier gas: N 2, after 60 hrs); (c) spent 10% CuCl 2 / -Al 2 O 3 (carrier gas: 6%(v) O ppmv HCl, after 30 hrs); (d) spent 10% CuCl 2 / -Al 2 O 3 (carrier gas: 6%(v) O ppmv HCl+2,000 ppmv SO 2, after 30 hrs); (e) spent CuCl (treated in 6%(v) O 2, after 30 hrs); (f) spent CuCl (treated in 10 ppmv HCl in 6%(v) O 2, after 30 hrs). 136

153 C1s Cl2p Al2p Fresh 10%CuCl2-alumina O1s Cu2p C1s Cl2p Al2p Spent 10%CuCl2-alumina O1s Cu2p Binding energy (ev) Figure 5.7. Survey XPS of 10%CuCl 2 -alumina before and after mercury oxidation It was also observed from survey XPS (Figure 5.7) that the peak intensities for Cu, O, and Al did not change from the CuCl 2 /α-al 2 O 3 samples before and after Hg(0) oxidation with the exception of Cl. The Cl content on the spent sample decreased because Cl was consumed to oxidize Hg(0) vapor and the resultant HgCl 2 did not adsorb and leave the sample. Table 5.2 shows the change in atomic ratios during the Hg(0) reaction over CuCl 2 /α- Al 2 O 3 determined from an XPS curve fitting method based on all Cu 2p peaks. After 60 hrs of the Hg(0) reaction, an atomic ratio of Cu to Cl decreased from 1:2 to 1:1.1, while almost all Cu(II) was converted into Cu(I). This result also supports the conversion of CuCl 2 into CuCl during the reaction. All of these XPS results are in good agreement with the XRD results. 137

154 Table 5.2 Results of quantitative XPS analysis for 10%(wt) CuCl 2 /α-al 2 O 3 Atomic ratio of Cu 2+ to Cu 1+ Atomic ratio of Cu to Cl Reference CuCl 1:1 Atomic ratio of O to Al Reference CuCl 2 1:2 Fresh 10% CuCl 2 /α-al 2 O 3 4:1 1:2 3:2 Spent 10% CuCl 2 /α-al 2 O 3 0.5:1 1:1.5 3:2 (reaction time 30 hr) Spent 10% CuCl 2 /α-al 2 O 3 0.2:1 1:1.1 3:2 (reaction time 60 hr) A fresh CuCl sample was also examined for XPS after treated with O 2 and 10 ppmv HCl in O 2 flow. Upon the exposure of CuCl to O 2 gas, most Cu(I) was turned into Cu(II). However, the binding energy peak for Cu(II) appears at ev, and is slightly different from the binding energy peak of ev for CuCl 2. The Cu(II) peak may be derived from a copper species associated with an oxygen atom, such as Cu 2 OCl 2 found from the above XRD examination. On the other hand, a CuCl sample exposed to 10 ppmv HCl in O 2 gas for 30 hrs shows the Cu(II) peak matching with that of CuCl Catalytic Oxidation of Hg(0) over CuCl 2 Using the Deacon Reaction The XRD and XPS results for the CuCl 2 /α-al 2 O 3 sample presented above clearly show that the impregnated CuCl 2 exists in a crystalline structure, and has the capability to oxidize Hg(0) vapor regardless of the presence of HCl gas. In the absence of HCl and O 2 gases, it acts as a semi-catalyst that consumes lattice chlorine of CuCl 2 for Hg(0) oxidation, and CuCl 2 is reduced 138

155 to CuCl via Reaction (5.3). This suggests that Hg(0) oxidation over CuCl 2 should take place via Mars-Maessen mechanism by which adsorbed Hg(0) would react with lattice Cl in CuCl 2.[7] When fresh CuCl was exposed to O 2 gas, Cu 2 OCl 2 was detected by XRD. From XPS, it was noted that the Cu(II) peak position of CuCl exposed to O 2 is slightly different from that of CuCl 2, but the XPS peak position of Cu 2 OCl 2 could not be found from the NIST XPS database. There is a general agreement that CuCl forms Cu 2 OCl 2 by oxidation with O 2 gas and is subsequently converted into CuCl 2 with HCl gas.[26, 27] It is not clear at this moment how Hg(0) oxidation is promoted over CuCl 2 by O 2 gas as shown in Figure 5.1. However, it was reported that the oxygen adsorption could lower the most energy-demanding atomic chlorine recombination step to liberate Cl 2 gas, and it may promote Hg(0) oxidation.[26] In this case, CuCl 2 works as a redox catalyst by oxidizing Hg(0) vapor while reduced CuCl is re-oxidized to CuCl 2 via the overall Reaction (5.6). The entire schematic of the Hg(0) reaction over the CuCl 2 /α-al 2 O 3 catalyst using the Deacon reaction is also summarized in Figure 5.8. Our recent study using X-ray Absorption Fine Structure (XAFS) spectroscopy found that the resultant oxidized mercury species is primarily mercuric chloride (HgCl 2 ) regardless of the presence of HCl and O 2 gases.[21] Hg(0) 2CuCl 2 HgCl 2 2CuCl (5.3) 2CuCl 1 2O (5.4) 2 Cu 2OCl2 Cu 2OCl2 2HCl 2CuCl 2 H2O (5.5) CuCl Overall : Hg(0) 2HCl 1 2O2 2 HgCl 2 H2O (5.6) 139

156 Hg(0) 2CuCl 2 H 2 O HgCl 2 2CuCl 2HCl + 1/2O 2 Hg(0) + 2HCl +1/2O 2 HgCl 2 +H 2 O Figure 5.8. Schematic of Hg(0) oxidation reaction over CuCl 2 /α-al 2 O 3. As shown in Reaction (5.1), the overall Deacon reaction is an exothermic reaction. The HCl uptake reaction is exothermic and thermodynamically favorable at low temperatures between 100 and 250 C. However, the Cl 2 release step is endothermic and favors a hightemperature window between 300 and 360 C. Thus in a two-stage process, a CuCl 2 -based catalyst is typically operated at ~350 C for Cl 2 production with unfavorable equilibrium limits for its faster kinetics.[27] The CuCl 2 -based catalyst has also showed a propensity to adsorb the produced Cl 2 gas compared with the RuO 2 -based catalyst.[26] Our XPS and XRD results indicate that Hg(0) vapor can continue to be oxidized over a CuCl 2 -based catalyst in the presence of HCl and O 2 gases by following the first step of HCl conversion into Cl of the Deacon reaction. This step will replenish the empty Cl lattices in CuCl 2 while Cl in occupied lattices is consumed for the conversion of Hg(0) vapor into HgCl 2 and Cl 2 gas production is suppressed at such a low temperature range. From this perspective, pre-chlorinated CuCl 2 has potential as an Hg(0) oxidation catalyst with good resistance to SO 2 in a thermodynamically favorable lowtemperature window after the air preheater. The acidic sites available in CuCl 2 are thought to 140

157 make the CuCl 2 catalyst more resistant to SO 2 than other metal oxide-based catalysts, but this needs to be further investigated. The actual space velocity inside the fixed-bed reactor used in this study was ~40,000 hr -1 at 140 C and was high enough to be realized in a honeycomb or plate catalyst bed. The ratelimiting step on the CuCl 2 -based catalyst in the Deacon process appears to be the re-chlorination step. The replenishment of the lattice Cl on the CuCl 2 surface under various flue gas components seems to be the key to the success of the CuCl 2 -based catalyst. When CuCl 2 was impregnated onto activated carbon, it also showed an excellent sorbent capability, and thus is expected to be able to be used as a catalyst and a sorbent with non-carbonaceous and carbonaceous substrates. Additional performance evaluations and surface analyses under various flue gas conditions and with different supports need to be further studied (Chapter 6) in order to obtain the characteristics of Hg(0) oxidation and reaction kinetic expressions for the prediction of Hg(0) oxidation and the design of a catalyst bed under various conditions. 5.4 Conclusion The CuCl 2 /α-al 2 O 3 catalyst possesses high activity for the oxidation of Hg(0) to Hg 2+, with an excellent stability under the environment similar to the flue gas from coal-fired power plant. The CuCl 2 /α-al 2 O 3 catalysts (both fresh and spent) were characterized by SEM-EDX, TGA-MS, TPR, XRD, and XPS. The CuCl 2 crystallites formed onto α-al 2 O 3 was found to be very stable up to 300 C, and undergoes the thermal reduction process from Cu(II) to Cu(0) via Cu(I). In the absence of HCl and O 2 gases, CuCl 2 was found to follow a Mars-Maessen mechanism by consuming lattice chlorine of CuCl 2 for Hg(0) oxidation and to be reduced to 141

158 CuCl. In the presence of 10 ppmv HCl, 2,000 ppmv SO 2, and 6% O 2 gases, the CuCl 2 /α-al 2 O 3 sample works as an Hg(0) oxidation catalyst exhibiting >90% conversion with good resistance to SO 2 at 140 C. The reduced CuCl was found to be re-chlorinated to CuCl 2 under HCl and O 2 gases by following the Deacon reaction. CuCl 2 is expected to be able to be used as a catalyst and a sorbent by impregnating onto non-carbonaceous and carbonaceous substrates in a temperature window after the air preheater. 5.5 Reference 1. Yang, R. T., Kikkinides, E.S.,, New sorbents for olefin/paraffin separations by adsorption via pi -complexation. AIChE Journal 1995, 41, (3), EPA, Integrated Planning Model (IPM) v.4.10 Model Runs Skodras, G.; Diamantopoulou, I.; Pantoleontos, G.; Sakellaropoulos, G. P., Kinetic studies of elemental mercury adsorption in activated carbon fixed bed reactor. Journal of Hazardous Materials 2008, 158, (1), Sable, S. P.; de Jong, W.; Spliethoff, H., Combined Homo- and Heterogeneous Model for Mercury Speciation in Pulverized Fuel Combustion Flue Gases. Energy & Fuels 2007, 22, (1), DOE, U. S. Annual Energy Outlook 2007 with Projections to 2030; Report No. DOE/EIA-0383(2007); Energy Information Administration.: Jones, A. P.; Hoffmann, J. W.; Smith, D. N.; Feeley, T. J.; Murphy, J. T., DOE/NETL's Phase II Mercury Control Technology Field Testing Program: Preliminary Economic Analysis of Activated Carbon Injection. Environmental Science & Technology 2007, 41, (4), Presto, A. A.; Granite, E. J., Survey of Catalysts for Oxidation of Mercury in Flue Gas. Environmental Science & Technology 2006, 40, (18),

159 8. Srivastava, R. K.; Hutson, N.; Martin, B.; Princiotta, F.; Staudt, J., Control of mercury emissions from coal-fired in electric utility boilers. Environmental Science & Technology 2006, 40, (5), Schofield, K., Fuel-Mercury Combustion Emissions: An Important Heterogeneous Mechanism and an Overall Review of its Implications. Environmental Science & Technology 2008, 42, (24), Lide, D. R., CRC Handbook Chemistry and Physics, 85th Edition CRC Press: Boca Raton, FL: Zhao, Y.; Mann, M. D.; Pavlish, J. H.; Mibeck, B. A. F.; Dunham, G. E.; Olson, E. S., Application of Gold Catalyst for Mercury Oxidation by Chlorine. Environmental Science & Technology 2006, 40, (5), Hrdlicka, J. A.; Seames, W. S.; Mann, M. D.; Muggli, D. S.; Horabik, C. A., Mercury Oxidation in Flue Gas Using Gold and Palladium Catalysts on Fabric Filters. Environmental Science & Technology 2008, 42, (17), Presto, A. A.; Granite, E. J., Noble Metal Catalysts for Mercury Oxidation in Utility Flue Gas. Platinum Metals Review 2008, 52, Kamata, H.; Ueno, S.-i.; Sato, N.; Naito, T., Mercury oxidation by hydrochloric acid over TiO2 supported metal oxide catalysts in coal combustion flue gas. Fuel Processing Technology 2009, 90, (7-8), Qiao, S.; Chen, J.; Li, J.; Qu, Z.; Liu, P.; Yan, N.; Jia, J., Adsorption and Catalytic Oxidation of Gaseous Elemental Mercury in Flue Gas over MnOx/Alumina. Industrial & Engineering Chemistry Research 2009, 48, (7), Li, J.; Yan, N.; Qu, Z.; Qiao, S.; Yang, S.; Guo, Y.; Liu, P.; Jia, J., Catalytic Oxidation of Elemental Mercury over the Modified Catalyst Mn/α-Al2O3 at Lower Temperatures. Environmental Science & Technology 2010, 44, (1), Liu, Y.; Wang, Y.; Wang, H.; Wu, Z., Catalytic oxidation of gas-phase mercury over Co/TiO2 catalysts prepared by sol-gel method. Catalysis Communications 2011, 12, (14), Yan, N.; Chen, W.; Chen, J.; Qu, Z.; Guo, Y.; Yang, S.; Jia, J., Significance of RuO2 Modified SCR Catalyst for Elemental Mercury Oxidation in Coal-fired Flue Gas. Environmental Science & Technology 2011, 45, (13),

160 19. Lee, S.-S.; Lee, J.-Y.; Keener, T. C., Mercury oxidation and adsorption characteristics of chemically promoted activated carbon sorbents. Fuel Processing Technology 2009, 90, (10), Lee, S. S.; Lee, J. Y.; Keener, T. C., Bench-Scale Studies of In-Duct Mercury Capture Using Cupric Chloride-impregnated Carbons. Environmental Science & Technology 2009, 43, (8), Li, X.; Lee, J.-Y.; Heald, S., XAFS characterization of mercury captured on cupric chloride-impregnated sorbents. Fuel 2012, 93, Lee, J. Y.; Ju, Y. H.; Keener, T. C.; Varma, R. S., Development of cost-effective noncarbon sorbents for Hg-0 removal from coal-fired power plants. Environmental Science & Technology 2006, 40, (8), Lee, S.-S.; Lee, J.-Y.; Khang, S.-J.; Keener, T. C., Modeling of Mercury Oxidation and Adsorption by Cupric Chloride-Impregnated Carbon Sorbents. Industrial & Engineering Chemistry Research 2009, 48, (19), Cao, Y.; Gao, Z.; Zhu, J.; Wang, Q.; Huang, Y.; Chiu, C.; Parker, B.; Chu, P.; Pan, W.-p., Impacts of Halogen Additions on Mercury Oxidation, in A Slipstream Selective Catalyst Reduction (SCR), Reactor When Burning Sub-Bituminous Coal. Environmental Science & Technology 2007, 42, (1), Rouco, A. J., TPR study of Al2O3- and SiO2-supported CuCl2 catalysts. Applied Catalysis A: General 1994, 117, (2), Amrute, A. P.; Mondelli, C.; Hevia, M. A. G.; P rez-ram rez, J., Temporal Analysis of Products Study of HCl Oxidation on Copper- and Ruthenium-Based Catalysts. The Journal of Physical Chemistry C 2011, 115, (4), Pan, H. Y.; Minet, R. G.; Benson, S. W.; Tsotsis, T. T., Process for Converting Hydrogen Chloride to Chlorine. Industrial & Engineering Chemistry Research 2002, 33, (12),

161 Chapter 6 Oxidation of Elemental Mercury Vapor over γ-al 2 O 3 Supported CuCl 2 Catalysts for Mercury Emission Control 6.1 Introduction Our previous studies of cupric chloride (CuCl 2 )-impregnated materials showed excellent Hg(0) oxidation performance both in the presence and absence of HCl gas throughout fixed-bed and entrained-flow tests.[1-3] When CuCl 2 is dispersed over non-carbonaceous substrates, CuCl 2 readily oxidizes Hg(0) vapor, but its resultant oxidized mercury is not easily adsorbed onto the non-carbonaceous substrate surface. The resultant oxidized mercury species has recently been found to be primarily mercuric chloride (HgCl 2 ) using X-ray Absorption Fine Structure (XAFS) spectroscopy.[4] In our recent reaction mechanistic study, non-porous α- Al 2 O 3 was used as a substrate to focus on the reactions taking place between Hg(0) vapor and CuCl 2 surface. The CuCl 2 /α-al 2 O 3 catalyst was found to follow a Mars-Maessen mechanism in the absence of HCl and O 2 gases, by consuming the lattice chlorine of CuCl 2 for Hg(0) oxidation and to be reduced to CuCl. In the presence of 10 ppmv HCl and 6% O 2 gases, the CuCl 2 /α-al 2 O 3 sample was found to work as an Hg(0) oxidation catalyst exhibiting >90% conversion with good resistance to as high as 2,000 ppmv SO 2 at 140 C. The reduced CuCl was found to be rechlorinated to CuCl 2 under HCl and O 2 gases by following the Deacon reaction.[3] Obviously, substrates supporting CuCl 2 play a critical role in Hg(0) oxidation and/or adsorption. Since the α-al 2 O 3 -supported CuCl 2 catalyst showed good Hg(0) oxidation performance, it was hypothesized that dispersing CuCl 2 on non-carbonaceous substrates with high surface areas would significantly enhance the Hg(0) oxidation performance. In this work, we chose a porous 145

162 γ-al 2 O 3 as a substrate to prepare a series of CuCl 2 /γ-al 2 O 3 catalysts with different CuCl 2 loadings ( %(wt)) for Hg(0) oxidation. CuCl 2 /γ-al 2 O 3 catalyst has been studied for the oxy-chlorination reaction of hydrocarbons.[5-15] The previous studies of the CuCl 2 /γ-al 2 O 3 catalyst show that there are two different families of Cu species formed on the catalyst surface. For low CuCl 2 loading (i.e. 3.5 %(wt)) catalysts, several hypotheses about copper speciation have been proposed before, including (1) a mono-dispersed positive copper oxide;[11] (2) a monolayer of interacting species;[13] (3) atomically dispersed species interacting via oxygen bonds with the γ-al 2 O 3 support;[9] (4) an attached bi-dimensional phase strongly interacting with the support;[14] (5) surface copper aluminates through isolated copper ions occupying octahedral vacancies of the alumina surface.[12, 15] For high CuCl 2 loading(i.e. >3.5 %(wt)) catalysts, two main Cu compounds have been identified, i.e. CuCl 2, and Cu 2 (OH) 3 Cl.[8,12] However, the nature of the copper species formed on the CuCl 2 /γ-al 2 O 3 catalysts has not been satisfactorily understood, and none of the above hypotheses has been convincingly accepted so far probably due to the complexity of the copper species and the limitations of the characterization techniques. As we investigate a potential application of CuCl 2 /γ-al 2 O 3 for elemental mercury oxidation, it is hardly found a detailed understanding of the copper speciation on both low- and high-loading CuCl 2 /γ-al 2 O 3 catalysts in the open literatures. Therefore, the objective of this study is to investigate the copper speciation and performance evaluations of CuCl 2 /γ-al 2 O 3 catalysts for Hg(0) oxidation under different gases at 140 C, a typical flue gas temperature after the air preheater. Several characterization techniques, including BET, SEM-EDAX, TGA-MS, TPR, XRD, and XPS, have been employed to study copper speciation from fresh and spent 146

163 CuCl 2 /γ-al 2 O 3 catalysts. An understanding of the copper speciation on these catalysts are essential in uncovering the Hg(0) oxidation mechanism over CuCl 2 /γ-al 2 O 3 catalysts. 6.2 Experimental Section Catalyst Preparation A γ-al 2 O 3 substrate was supplied by Alfa Aesar (aluminum oxide-43855, ϕ1/8 pellets). The pellets were first ground and sieved to ~40 μm particles. These particles were then dried at 300 C for 2 hrs and cooled down for CuCl 2 loading. Five CuCl 2 /γ-al 2 O 3 samples (with CuCl 2 loadings of 1.0, 3.5, 5.0, 10, and 15 %(wt), respectively) were prepared by impregnating CuCl 2 2H 2 O (Sigma, >98% purity) onto γ-al 2 O 3 in the aqueous phase following the incipient wetness method. Please note that these loadings were prepared based on the amount of CuCl 2 excluding the amount of dihydrate (2H 2 O). The content of CuCl 2 in the catalyst was varied by controlling the concentration of CuCl 2 dissolved in the aqueous solution. After impregnation, the samples were dried at room temperature under a dry air flow for 24 hrs, and further dried at 100 o C for 2 hrs. Samples with 1%(wt) and 10%(wt) CuCl 2 loadings onto α-al 2 O 3 substrate (Alfa Aesar aluminum oxide-43862, ϕ1/8 pellets) were also prepared for a comparison between different substrates Performance Evaluation of CuCl 2 /γ-al 2 O 3 Catalyst A lab-scale fixed-bed system was used for the performance evaluation of Hg(0) oxidation and preparation of a spent catalyst for X-ray Photoelectron Spectroscopy (XPS) and X-ray Diffraction (XRD) analysis. 25 mg of the catalyst was mixed with 4 g of quartz sand and then the mixture was loaded into a fixed-bed reactor with an inner diameter of 12 mm. The length of 147

164 reaction zone was ~20 mm. The detailed information on the system and experiments were described in our previous study.[1, 3] The inlet Hg(0) concentration was 250 g/nm 3 (= 30 ppbv) in 1 L/min of carrier gas flow, and the reactor was placed in an oven maintained at 140 C, targeting an installation of the Hg(0) catalytic bed after the air preheater. Four different gas conditions were used in this study: (1) N 2 (99.999% UHP, Wright Brothers, Inc.); (2) 6% (v) O 2 balanced with N 2 (Wright Brothers, Inc.); (3) 10 ppmv HCl in 6%(v) O 2 balanced with N 2 (Wright Brothers, Inc.); and (4) 2,000 ppmv SO 2 and 10 ppmv HCl in 6%(v) O 2 balanced with N 2. When spent catalysts were prepared for characterization, the catalysts were not mixed with quartz sand and the inlet Hg(0) vapor concentration was raised to ~2 mg/nm 3 (= 240 ppbv) in order to saturate the catalyst in a short period of time Characterization XRD data were obtained using Cu Kα radiation with a wavelength of Å (X Pert Pro MPD X-ray diffractometer). An aluminum holder was used to support the catalyst samples. The scanning range was from (2θ) with a step size of 0.02 and a step time of 0.5 s. XPS (Kratos Axis Ultra, with a monochromated Al Kα source ( ev)) measurements were used to examine copper speciation with a concentric hemispherical analyzer. The power applied to the X-ray gun was 130 W (I=10 ma, V=13 kv), and the base vacuum pressure was Torr, and the pass energy of the analyzer was set to 20 ev. The resolution of the spectra was ~0.2 ev with the step size of 0.1 ev. The carbon 1s peak at ev was used to charge reference all binding energies. Thermal Gravimetric Analysis (TGA, TA Instruments Q5000IR) was used to evaluate the thermal stability of pure CuCl 2 and CuCl 2 /γ-al 2 O 3. Approximately 10 mg of a sample was used to ramp from room temperature to 800 C with a heating rate of 5 148

165 C/min under nitrogen flow (99.999% UHP) at a rate of 100 ml/min. The H 2 -Temperature Programmed Reduction (TPR) experiments were also performed using the Micromeritics Autochem 2910 automated catalyst characterization system with ~50 mg of samples. The samples were pretreated at 300 C for 1 hr in ultra-pure argon gas at a flow rate of 30 ml/min. After the furnace temperature decreased to ~50 C, the feed gas (10%(v) H 2 balanced with AR) was fed into the system at a flow rate of 30 ml/min. H 2 -TPR runs were performed by heating the sample up to 600 C at a linear heating rate of 6.5 C/min. The amount of hydrogen gas consumed during the TPR experiment was measured by a built-in thermal conductive detector (TCD). 6.3 Results and Discussion Performance Evaluation of CuCl 2 /γ-al 2 O 3 Catalyst on Hg(0) Oxidation The CuCl 2 /γ-al 2 O 3 catalysts with different CuCl 2 loadings were evaluated in the presence of N 2 gas. The catalyst performances for Hg(0) oxidation are shown in Figure 6.1. The results clearly show that the catalysts with CuCl 2 loading less than 3.5%(wt) exhibit little catalytic activity in Hg(0) oxidation. For example, the breakthrough of Hg(0) oxidation over the 1.0%(wt) CuCl 2 /γ-al 2 O 3 catalyst occurred in less than 4 hrs (with little oxidation activity). The 3.5%(wt) CuCl 2 /γ-al 2 O 3 catalyst has a breakthrough curve similar to 1.0%(wt) CuCl 2 /α-al 2 O 3 (in ~30 hrs). On the other hand, 10.0% (wt) CuCl 2 /γ-al 2 O 3 catalyst showed Hg(0) conversion of >90% within the first 30 hrs even under N 2 flow. It was saturated after ~130 hrs similar to 10%(wt) CuCl 2 /α- Al 2 O 3.[3] Apparently, as CuCl 2 loading increases, the inert species (i.e. copper aluminates, as reported in the literature [12]) derived from the initial contact of CuCl 2 and γ-al 2 O 3 surface was 149

166 Hg(0)out Hg(0) in overlapped by extra CuCl 2. The presence of multiple phases of copper species on high loading CuCl 2 /γ-al 2 O 3 catalyst will be characterized and discussed in the following sections % Inlet Hg(0) Conc.: 30ppbv Carrier gas: Nitrogen 1% CuCl2/ -Al2O3 3.5% CuCl2/ -Al2O3 10% CuCl2/ -Al2O3 1% CuCl2/ -Al2O3 10% CuCl2/ -Al2O Time (hr) Figure 6.1 Catalytic performances of CuCl 2 /γ-al 2 O 3 catalysts with different CuCl 2 loadings under N 2 gas at 140 C. The performances of the 10% (wt) CuCl 2 /γ-al 2 O 3 catalyst evaluated under different gas conditions are shown in Figure 6.2. It can be seen from Figure 6.2 that the catalyst can oxidize almost all Hg(0) vapor during the first 24 hrs under any carrier gas conditions. When Hg(0) vapor was introduced in N 2 and O 2 gases, O 2 gas was found to have the capability to promote Hg(0) oxidation compared to N 2 gas. However, Hg(0) oxidation started to decrease and eventually stopped under both carrier gases. On the other hand, it is very interesting to note that Hg(0) vapor was almost completely oxidized in the presence of 10 ppmv HCl and 6%(v) O 2 gas balanced with N 2 gas, regardless of the presence of 2,000 ppmv SO 2 gas over 140 hrs of the performance evaluations. These results indicate that the CuCl 2 /γ-al 2 O 3 catalyst can be used to 150

167 oxidize Hg(0) vapor even under high SO 2 concentrations. The notable differences of the Hg(0) oxidation performance over the CuCl 2 /γ-al 2 O 3 catalysts with different CuCl 2 loadings and gas conditions prompted us to conduct the following characterization study of the fresh and spent samples in order to understand the Hg(0) oxidation mechanisms. CHg(0)out / CHg(0) in (%) N2 6% (v) O2 in N2 10 ppmv HCl + 6% (v) O2 in N2 10 ppmv HCl + 2,000 ppmv SO2 + 6% (v) O2 in N Time (hr) Figure 6.2 Catalytic performance of 10%(wt) CuCl 2 /γ-al 2 O 3 catalyst under different gas conditions at 140 C (30 ppbv inlet Hg(0) concentration and 25 mg catalyst) Characterization of CuCl 2 /γ-al 2 O 3 Catalyst BET Analysis and SEM Images The γ-al 2 O 3 substrate powder used in this study has a mesoporous (medium pore diameter ~7 nm) structure with a large surface area (220 m 2 /g) and pore volume (0.62 cm 3 /g). The α-al 2 O 3 substrate powder has a macropore structure with very small surface area (0.25 m 2 /g) and pore volume (0.26 cm 3 /g). The SEM image in Figure 6.3a shows that the γ-al 2 O 3 powder is 151

168 in an amorphous state. No visible CuCl 2 particles can be observed from the surface, indicating a highly dispersed CuCl 2 phase, While CuCl 2 /α-al 2 O 3 sample (Figure 6.3b) shows visible CuCl 2 crystallites deposited between α-al 2 O 3 particles. (a) (b) Figure 6.3 SEM images of CuCl 2 /γ-al 2 O 3 and CuCl 2 /α-al 2 O 3 with the same 10%(wt) loading. 152