Ranking Low Cost Sorbents for Mercury Capture from Simulated Flue

Oct 5, 2007 - Coal fired utility boilers are the largest anthropogenic source of mercury release to the atmosphere, and mercury abatement legislation ...
0 downloads 0 Views 1MB Size
Energy & Fuels 2007, 21, 3249–3258

3249

Ranking Low Cost Sorbents for Mercury Capture from Simulated Flue Gases H. Revata Seneviratne,* Cedric Charpenteau, Anthe George, Marcos Millan, Denis R. Dugwell, and Rafael Kandiyoti Department of Chemical Engineering, Imperial College London, Prince Consort Road, London SW7 2AZ, UK ReceiVed January 19, 2007. ReVised Manuscript ReceiVed July 13, 2007

Coal fired utility boilers are the largest anthropogenic source of mercury release to the atmosphere, and mercury abatement legislation is already in place in the USA. The present study aimed to rank low cost mercury sorbents (char and activated carbon from the pyrolysis of scrap tire rubber and two coal fly ashes from UK power plants) against Norit Darco HgTM for mercury retention by using a novel bench-scale reactor. In this scheme, a fixed sorbent bed was tested for mercury capture efficiency from a simulated flue gas stream. Experiments with a gas stream of only mercury and nitrogen showed that while the coal ashes were the most effective in mercury capture, char from the pyrolysis of scrap tire rubber was as effective as the commercial sorbent Norit Darco HgTM. Tests conducted at 150 °C, with a simulated flue gas mix that included N2, NO, NO2, CO2, O2, SO2 and HCl, showed that all the sorbents captured approximately 100% of the mercury in the gas stream. The introduction of NO and NO2 was found to significantly improve the mercury capture, possibly by reactions between NOx and the mercury. Since the sorbents’ efficiency decreased with increasing test temperature, physical sorption could be the initial step in the mercury capture process. As the sorbents were only exposed to 64 ng of mercury in the gas stream, the mercury loadings on the samples were significantly less than their equilibrium capacities. The larger capacities of the activated carbons due to their more microporous structure were therefore not utilized. Although the sorbents have been characterized by BET surface area analysis and XRD analysis, further analysis is needed in order to obtain a more conclusive correlation of how the characteristics of the different sorbents correlate with the observed variations in mercury capture ability.

1. Introduction Coals contain many elements in trace amounts of parts per million (ppm). Mercury is one such trace element and is found in concentrations between 0.02 and 1.0 ppm and is readily volatilized during coal combustion.1 Since mercury is highly volatile, major portions of the mercury present in coal avoid capture by the existing particulate matter control devices.2 Although mercury has no known bio-inorganic role in life forms, it is able to enter the food chain.3 The Health and Safety Executive data sheets state that mercury is toxic to humans and acute mercury exposure could result in severe nausea, kidney damage, and death.4 It is apparent that the total Hg emissions in most developed countries have either stabilized or reduced.5 The same is true for mercury emissions from coal combustion in many of these * To whom correspondence should be addressed. E-mail: revata. [email protected]. (1) Reed, G. P.; Ergudenler, A.; Grace, J. R.; Watkinson, A. P.; Herod, A. A.; Dugwell, D. R.; Kandiyoti, R. Control of gasifer mercury emissions in a hot gas filter: the effect of temperature. Fuel 2001, 80, 623–634. (2) Ghorishi, S. B.; Sedman, C. B. Low concentration mercury sorption mechanisms and control by calcium-based sorbents: application in coalfired processes. J. Air Waste Manage. Assoc. 1998, 48, 1191–1198. (3) Richaud, R.; Lachas, H.; Collot, A.; Mannerings, A. G.; Herod, A. A.; Dugwell, D. R.; Kandiyoti, R. Trace mercury concentrations in coals and coal-derived materials determined by Atomic Absorption Spectro-photometry. Fuel 1998, 77, 359–368. (4) Health & Safety Executive. MDHS 16/2 Mercury and its inorganic divalent compounds in air. 2002, http://www.hse.gov.uk/pubns/mdhs/pdfs/ mdhs16-2.pdf. (5) Sloss, L. L. Mercury;emissions and control; Technical Report; International Energy Agency: 2002.

countries. On the other hand, the importance of mercury emissions from coal combustion relative to the total mercury emissions has increased in certain countries. This is because the more concentrated mercury emissions from other sectors (e.g., chlor-alkali plants) are more readily controlled or eliminated.5 Reducing mercury emissions by coal-fired utilities has therefore become more critical. Coal-fired utility boilers are currently the largest single source of mercury emissions in the United States, accounting for about one-third of the total anthropogenic mercury emissions.6 Data collected by the U.S. Environmental Protection Agency (EPA) indicated that during 1999 there was 75 tons of mercury input in the 900 million tons of coal used in U.S. power plants. On average, about 40% of the mercury entering the power plants was captured, while 60% was emitted. During 1999, coal-fired power plants therefore emitted ∼45 tons of mercury to the atmosphere. Recent estimates of annual global mercury emissions from all sources (both natural and anthropogenic) range from roughly 4400 to 7500 tons per year.7 Anthropogenic U.S. mercury emissions were estimated to account for roughly 3% of this global total, and U.S. coal-fired power plants are estimated to account for about 1% of the global total.7 Mercury in flue gas can be classified as particulate-bound mercury (Hgp) and gaseous mercury. The latter form may exist as either elemental or oxidized mercury.8 Particulate-bound (6) Pavlish, J. H.; Sondreal, A. E.; 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 Process. Technol. 2003, 82, 89–165. (7) U.S. Environmental Protection Agency. Clean Air Mercury Rule. 2005, http://www.epa.gov/air/mercuryrule/basic.htm.

10.1021/ef070028x CCC: $37.00  2007 American Chemical Society Published on Web 10/05/2007

3250 Energy & Fuels, Vol. 21, No. 6, 2007

mercury is effectively captured by ash collection devices, such as electrostatic precipitators (ESPs) and fabric filters (FFs). It is mainly the gaseous form of mercury that is emitted by coalfired power plants. The lifetime of elemental mercury Hg0 in the atmosphere is estimated to be up to a year; oxidized forms on the other hand have a lifetime of only a few days because of particulate settling and the solubility of oxidized mercury.6 As a result, Hg0 can be transported over transcontinental distances, while oxidized gaseous and particulate forms of mercury tend to be deposited nearer to their source. In the past, emission standards for mercury which could theoretically apply to coal-fired power plants as a group have been higher than actual emissions.5 However, on March 15, 2005, the EPA issued the first ever federal rule to permanently cap and reduce mercury emissions from coal-fired power plants.7,9 A first phase cap of 38 tons per year (tpy) will become effective in 2010, while 2018 has been set as the date for the second phase cap of 15 tpy. The EPA believes that the targets for the first phase cap could be met by taking advantage of “cobenefit” reductions. According to this scheme, the systems in place for limiting sulfur dioxide (SO2) and nitrogen oxide (NOx) emissions under the EPA’s Clean Air Interstate Rule (CAIR) would also achieve, in parallel, the required reductions in mercury emissions. Mercury emissions limited by using a capand-trade approach, where allowances could be readily transferred among all regulated facilities, were believed to be the most cost effective way of reducing these emissions from the power sector. The European Community, Australia, and Japan are also setting up programs to monitor mercury emissions from coal-fired power plants with the possibility of setting standards in the foreseeable future.5 Various control technologies to reduce mercury emission from coal-fired utilities are being developed in order to find the most appropriate method to combine with existing air pollution control devices.9 Direct injection of powdered activated carbon (PAC) upstream of a particulate control system has been intensively studied to remove vapor-phase mercury from the flue gas streams of coal-fired utilities. To date, injection of PAC has been recognized as the most promising technology as a nearterm mercury control technology. This technology, however, still needs more effort to reduce the amount of carbon injection and operating costs. The short residence time of the PAC from the injection point to the particulate control device, the existence of competing species in the flue gas that adsorb on the active sites of the carbon, and the low concentrations of mercury in the flue gases require a high ratio of carbon to mercury to be used in order for a mercury removal efficiency of 90% to be achieved.6 Commercially available activated carbon typically costs U.S. $1.00 per kg and would in practice require C/Hg injection rates in the range of 10 000:1.10 It has therefore been estimated that it would cost approximately US $30,000–50,000 (1995 prices) to remove each kilogram of mercury from flue gas using activated carbon. Consequently, for a 250 MW unit emitting 65 kg of Hg per year, it would cost between US $ 1 and 3 million to remove 50% of the mercury present.10 (8) Cao, Y.; Duan, Y.; Kellie, S.; Li, L.; Xu, W.; Riley, J. T.; Pan, W. Impact of coal chlorine on mercury speciation and emission from a 100MW utility boiler with cold-side electrostatic precipitators and low-NOx burners. Energy Fuels 2005, 19, 842–854. (9) Baek, J.-I.; Lee, S.-H. L.; Lee, J. K. Tests on Vapour-phase mercury remoVal by chemically modified heaVy-oil fly ashes in fixed bed and entrained flow reactor, Proceedings of the 22nd Annual International Pittsburgh Coal Conference, Sept 15–18, 2005. (10) Romero, C. E.; Li, Y.; Bilirgen, H.; Sarunac, N.; Levy, E. K. Modification of boiler operating conditions for mercury emissions reductions in coal-fired utility boilers. Fuel 2006, 85, 204–212.

SeneViratne et al.

In the present work, the mercury capture effectiveness of low cost sorbents derived from scrap tire rubber has been tested and ranked against commercially available Norit Darco Hg (also known as Norit Darco FGD) using a novel bench-scale fixedbed experimental reactor. This particular PAC has been extensively studied by other researchers for its mercury capture efficiency and equilibrium capacity during laboratory and largescale tests in coal-fired utility plants.6,11,12 Coal fly ash also plays a role in both the adsorption of mercury and the oxidation of elemental mercury in the flue gas.13 Therefore, this study has also investigated the possibility of reusing coal fly ashes from two UK coal-fired utility plants. 2. Low Cost Sorbents Tested Both charcoal and activated carbon prepared from scrap tire rubber were tested for their mercury capture efficiency. The char was produced by pyrolyzing tire rubber using the same scheme published by Miguel.14 A Carbolite HTR 11/150 laboratoryscale rotary furnace (diameter ∼150 mm) with an internal volume of approximately 4.5 L was therefore used in the present study. Approximately 340 g of cut scrap tire rubber pieces (less than 15 mm in size) were heated in an inert atmosphere at a rate of 10 °C/min. This inert atmosphere was maintained by having a nitrogen sweep gas flow of 500 mL/min. A holding time of half an hour at 700 °C was used for the pyrolysis runs, resulting in a char yield of 34%. In a similar manner, activated carbon was produced from scrap tire rubber by using the scheme published by Miguel et al., which results in carbons with a high surface area.15 Approximately 200 g of cut scrap tire rubber pieces (less than 15 mm in size) were heated at 5 °C/min in an inert atmosphere maintained by a nitrogen gas flow of 500 mL/min. Once the temperature reached 700 °C, this atmosphere was rapidly substituted by a flow of 500 mL/min of steam/nitrogen (80:20, v/v). This was generated by having a water flow of 0.3 mL/ min and a nitrogen gas flow of 100 mL/min. As prescribed by Miguel et al.,15 the same heating rate of 5 °C/min was maintained until an activation temperature of 925 °C was reached. Active carbon with a yield of 29% was obtained by using these conditions and a holding time of 80 min at the activation temperature. The aim of this study is to rank low cost sorbents against Norit Darco Hg; this sorbent has a mean particle size of 9–15 µm. Since other researchers have found that mercury capture may be affected by the particle size of the sorbent, the pyrolysis chars and activated carbons from scrap tire rubber were crushed and only the smallest recoverable size fraction (38–75 µm) was tested for mercury retention efficiency.16 (11) Carey, T. R.; Hargrove, O. W.; Richardson, C. F.; Chang, R.; Meserole, F. B. Factors affecting mercury control in utility flue gas using sorbent injection, Proceedings of the 90th Annual Meeting & Exhibition of the Air & Waste Management Association, 1997. (12) Bustard, J.; Durham, M.; Starns, T.; Lindsey, C.; Martin, C.; Schlager, R.; Baldrey, K. Full-scale evaluation of sorbent injection for mercury control on coal-fired power plants. Fuel Process. Technol. 2004, 85, 549–562. (13) Dunham, G. E.; DeWall, R. A.; Senior, C. L. Fixed-bed studies of the interactions between mercury and coal combustion fly ash. Fuel Process. Technol. 2003, 82, 197–213. (14) Miguel, G. S. Activated carbon produced by pyrolysis and physical activation of waste tires. Ph.D. Thesis, Imperial College, University of London, 1999. (15) Miguel, G. S.; Flower, G. D.; Sollars, C. J. Pyrolysis of tire rubber: Porosity and adsorption characteristics of pyrolytic chars. Ind. Eng. Chem. Res. 1998, 37, 2430–2435.

Low Cost Sorbents for Mercury Capture

Energy & Fuels, Vol. 21, No. 6, 2007 3251

Table 1. Particle Size Distribution of Fly Ash A and Fly Ash B particles particles particles particles particles

larger than 150 µm between 106 and 150 µm between 75 and 106 µm between 38 and 75 µm less than 38 µm

Fly ash A

Fly ash B

1% 3% 14% 21% 61%

5% 4% 5% 25% 61%

Table 2. Surface Area, Micropore Volume, and Micropore Area of the Sorbents

Norit Darco Hg tire activated carbon tire pyrolysis char Fly ash A Fly ash B

BET surface area (m2/g)

micropore volume (mm3/g)

micropore area (m2/g)

660 ( 11.48 219 ( 3.67 72 ( 0.16 4 ( 0.13 15 ( 0.17

169 60 1 2 2

367 128 5 4 5

Fly ash from two UK coal-fired power plants operated were also tested for the possibility of reusing them to capture more mercury. The fly ash A was produced from a UK coal, while fly ash B was from a Russian coal. If in the future fly ash is reinjected into the flue gas duct, it might be done without any screening to remove the larger sized ash particles. The study was therefore conducted using the as-received fly ash samples directly, i.e., without any bias toward a particular particle size. The two fly ash samples seem to have a similar particle size distribution with the majority of the particles being smaller than 38 µm (Table 1). 3. Characterization of the Sorbents The atomic radius of mercury is 1.6 Å. Therefore, these atoms would be expected to diffuse into the micropores of the sorbents tested in this study.17 Brunauer, Emmett, and Teller (BET) surface area and thickness plot (T-plot) analyses were undertaken to obtain information about the micropores in the sorbents by using a Micrometitics ASAP 2000 Surface Area Analyzer. A typical sample mass of approximately 150 mg was used to study the activated carbons. Since fly ash has a smaller surface area than activated carbon, a larger sample mass of approximately 500 mg was used to obtain accurate results for the fly ashes. Each analysis was carried out by first degassing the samples at 105 °C and then evaluating with nitrogen gas at 77 K. The results obtained are tabulated in Table 2. The percentage deviation of the BET surface area for each sample was within (5%, while the regression analysis for the T-plots for each of the samples had correlation coefficients that were approximately 0.999. When compared with the other four sorbents evaluated in this study, Norit Darco Hg had the greatest BET (660 m2/g) and micropore surface area and also the highest micropore volume. Pavlish et al.6 have also reported a similar surface area (600 m2/g) for this activated carbon from Norit America Inc. The two fly ashes had the smallest of the BET surface areas of the five sorbents tested in this study. Furthermore, the micropore areas of these ash samples were comparable to the BET surface areas. This suggested that the samples had mainly micropores with only a limited amount of mesopores and macropores. (16) Rostam-Abadi, M.; Lu, Y.; Funk, C. Properties of unburned carbons form three coal-fired power plants and their relationships to mercury capture, Proceedings of the Air Quality V;International Conference on Mercury, Trace Elements, Trace elements and Particulate matter, Sept 18– 21, 2005. (17) Dean, J. A. Lange’s Handbook of Chemistry, 13th ed.; McGrawHill Company: 1985.

Although the charcoal produced from scrap tire pyrolysis had a larger BET surface area than the two fly ash samples, the results also suggest that, for a given mass of sample, this charcoal and the fly ash samples have similar quantities of micropores. The steam activation of the scrap tire rubber char was found to significantly increase the BET surface area, micropore volume, and micropore area. Miguel14 reported similar BET surface area values for the pyrolysis charcoal (82 m2/g) and steam activated carbon (283 m2/g) produced from scrap tire rubber by using the same pyrolysis and activation conditions as the sorbents in this study. The X-ray diffraction (XRD) analysis of the sorbents was conducted with a Philips PW1700 series Automated Power Diffractometer, which used Cu KR radiation at 40 kV/40 mA with a secondary graphite crystal monochromater. Each sample was placed in a zero background substrate (single silicon crystal), and the data were obtained over a 2θ ) 5–80° range (steps of 0.04°), with a time per step of 2 s. A search match was later performed on the results by using “Philips X’Pert Graphics and Identify” software in conjunction with the International Centre for Diffraction Data (ICDD) database to identify each crystalline phase in the sorbents. The analysis of the five sorbents revealed that they contained mainly amorphous phases (Figure 1). Mullite and quartz are the two main crystalline phases in the fly ashes, although the fly ash sample also has some magnetite and lime phases present. The major crystalline phases in Norit Darco Hg were quartz and calcite, while both sorbents from scrap tire rubber had wurtzite as the primary crystalline phase. The chemical formulae for each of the compounds found in the sorbents are listed in Table 3. Iron present in coal is believed to catalyze the oxidation and therefore promote the subsequent capture of mercury.6 The magnetite in the fly ash may therefore play a role in mercury pick-up. Laboratory studies have shown that sulfur impregnation of activated carbon does improve elemental mercury adsorption capacity.18 Table 4 lists the sulfur contents for the sorbents made from scrap tire rubber. During pyrolysis of scrap tire rubber, approximately 47% (w/w basis) of the sulfur that is originally present is removed from the solid. The activation of scrap tire rubber on the other hand volatilizes 54% (w/w basis) of the initial sulfur. Most of the sulfur present in the scrap tire is organic; it is subsequently converted to highly stable metallic sulphide forms during the pyrolysis process by reaction with zinc, and other metals, that were initially present in the tire.14 XRD analysis of the sorbents derived from scrap tire rubber in the current study shows that iron sulfide and wurtzite are the main crystalline phase in the samples. Figure 1 shows that activated carbon prepared from scrap tire rubber has a more pronounced signal for wurtzite than the corresponding pyrolysis char. It could therefore be postulated that a greater percentage of the sulfur that remains on the activated carbon is consequently converted to a more crystalline form of ZnS. Thus, the pyrolyzed sorbent may have had a higher concentration of sulfur in the amorphous phase which is more capable of reacting with the mercury. (18) Hsi, H.-C.; Rood, M.; Rostam-Abadi, M.; Chen, S.; Chang, R. Effect of sulphur impregnation temperature on the properties and mercury adsorption capacities of activated carbon fibres (acfs). EnViron. Sci. Technol. 2001, 35, 2785–2791. (19) Reed, G. P. Control of trace elements in gasification. Ph.D. Thesis, Imperial College, University of London, 2000.

3252 Energy & Fuels, Vol. 21, No. 6, 2007

SeneViratne et al.

Figure 1. XRD scan results of sorbents used for mercury capture.

4. Experimental Procedure The new reactor that was used to test the sorbents is based on the experience gained from the previous hot gas cleaning project.19 The new unit is intended for operation at temperatures below 300 °C. It has been designed around three concentric borosilicate glass tubes (Figure 2). Gas tight joints are maintained by keeping all connections at the lower end of the reactor, beneath the furnace, where the temperature is moderate. The simulated flue gas stream that enters the reactor is heated by an electric furnace, as it flows upward along the annulus between the tubes of the unit. This heated gas flow is then reversed and passes downward through the sorbent bed toward the reactor outlet. A 2 mm thick Kaowool felt disc

located in the central annulus of the unit is used to support a fixed quantity of sorbent. In order to calculate the mass balance for mercury, an external gold amalgamator is connected to the reactor gas exit; this serves as a backup capture unit to pick up any residual mercury in the gas stream. At the end of each experiment, the sorbent and the amalgamator are analyzed for total mercury by using a LECO AMA 254 instrument. This analyzer quantifies the mercury present in solid samples by combustion over a catalyst in a pure oxygen flow.3 A simplified schematic diagram of the new bench-scale system that was used in this study is presented in Figure 3. As shown in this figure, mercury for the experiments is produced by using a

Low Cost Sorbents for Mercury Capture

Energy & Fuels, Vol. 21, No. 6, 2007 3253

PSA “Cavkit” 10.534 mercury generator from P S Analytical, UK. National Institute of Standards and Technology (NIST) used a similar PSA 10.534 for the generation of Hg vapor mixtures in a flowing gas stream to establish a gaseous mercury standard.20 The unit uses an oven arrangement to generate elemental mercury from a mercury impregnated inert substrate. In order to conduct a realistic investigation on the capture of mercury in power plant flue gases, mercury retention efficiencies of the sorbents have been measured in a gas stream that simulated flue gas produced from the combustion of Harworth coal. The constituents of this simulated flue gas stream are tabulated in Table 5. Flue gas from the combustion of Harworth coal contains a

mercury concentration of 29 ng/L [at normal temperature and pressure (NTP)]. The required mercury flow rate was obtained by using the Cavkit, while gas cylinders were used for the other gas constituents. The total gas flow to the reactor was kept constant at 37 mL/min (measured at NTP) which resulted in a superficial gas velocity of 0.01 m/s through the reactor section just above the sorbent bed. This velocity is similar to the air-to-cloth ratio for fabric filters installed in coal-fired power plants.21 When used in coal-fired power plants, the sorbent for mercury capture will most probably be injected into the flue gas duct after the air preheater and then be collected, along with the fly ash by the particulate control system (either an ESP or a FF). Fly ash A and B were collected from ESPs that operate at about 130 °C. In order to get a better understanding of the mercury adsorption process, test temperatures between 100 and 200 °C have been used for the experiments, including 150 °C which is probably the maximum operating temperature of cold-side ESPs. This temperature variation results in only a minor change to the superficial velocity of the gas stream and was believed to have a negligible effect on the experimental results. During a typical experimental run, approximately 100 mg of each sorbent was exposed to a simulated flue gas stream for a period of approximately 1 h. The above test conditions for the simulated flue gas flow and an experimental time of 1 h meant that approximately 64 ng of mercury flowed into the reactor during each test. The effect of transients that occurred during start-up, and at the end of each test, had a minimal effect on overall experimental error given the long length of the test. Calibration of the Cavkit from PS Analytical was verified by testing the system for an hour with the external amalgamator connected directly to the simulated flue gas inlet. At the end of each calibration test, the Leco AMA 254 instrument was used to quantify the mercury retained on the external amalgamator. The results proved that the system functioned within (10% of the required “set-point” for mercury concentration. Other researchers have found that the presence of HCl, NO, and NO2 in the gas stream, either individually or in combination, enhances the capture of mercury.6 The sorbents in the current study were thus tested under the different gaseous atmospheres listed in Table 6. The objective of this study was to test the sorbent performance in an atmosphere which simulates flue gas conditions when Harworth coal is combusted. The required conditions for the gas atmospheres Exp. 2 to Exp. 4 were therefore obtained by keeping the individual gas components (except nitrogen gas) listed in Table 5 at similar concentrations to that which was used in gas scheme Exp. 5. Additional nitrogen gas was used to make up the losses in gas flow that occurred when an individual component was excluded, and thereby ensure that the total gas flow to the reactor was maintained at the desired 37 mL/min (measured at NTP). The mercury capture efficiency is deduced from the amount of mercury captured by both the sorbent and the external amalgamator, and is defined by the following equation. Hg captured by sorbent mercury capture efficiency ) (1) total Hg to the reactor In this particular case, “Hg captured by sorbent” is the difference between the total mass of mercury on the sorbent after the experiment less the initial mass of mercury already on the sorbent before the experiment. The “total Hg to the reactor” is the sum of “Hg captured by the sorbent” and the mass of mercury retained on the external amalgamator. The external amalgamator was therefore purged before each experimental run to eliminate any residual mercury. The mass balance for each experimental run was calculated by comparing the “total Hg to the reactor” against the expected 64 ng of mercury that should have been present in the simulated flue gas stream. The mercury capture efficiencies reported in the next section are from multiple tests with the novel fixed-bed reactor. About 60%

(20) Mitchell, G. D.; Dorko, W. D. Mercury in nitrogen gas research gas mixture; Technical Report, EPA Reference DW13939860-01-0; National Institute of Standards and Technology: Gaithersburg, MD, 2004.

(21) Scheck, R. W.; Mora, R. R.; Belba, V. H.; Horney, F. A. Economics of fabric filters and electrostatic precipitators;1984; Final Report; Stearns Catalytic Corp.: Denver, CO, 1985.

Table 3. Chemical Composition of the Crystalline Phases Found in the Sorbents crystalline phase

chemical compound

calcite iron sulfide lime magnetite mullite quartz wurtzite zincite

CaCO3 FeS CaO Fe3O4 3Al2O3.2SiO2 SiO2 ZnS ZnO

Table 4. Sulfur Concentration (w/w Basis) on the Different Sorbents from Scrap Tire Rubber

scrap tire rubber tire pyrolysis char tire activated carbon

sulfur concentration

percentage of sulfur that was retaineda

1.9% 2.9% 3.0%

100% 53% 46%

a The percentage of sulfur that was retained is relative to the initial sulfur concentration on the scrap tire rubber.

Figure 2. Sectional view of the novel fixed-bed reactor for testing mercury sorbents.

3254 Energy & Fuels, Vol. 21, No. 6, 2007

SeneViratne et al.

Figure 3. Schematic diagram of the experimental setup. Table 5. Individual Gas/Vapor Concentrations and Flow Rates for the Simulated Harworth Coal Derived Flue Gas constituent

concentration

volume flow rate (mL/min at NTP)

Hg0 O2 CO2 SO2 HCl NO NO2 N2 total

3.5 ppbv 6% 12% 1800 ppmv 200 ppmv 490 ppmv 17 ppmv 82% 100%

1.3 × 10-7 2 5 7 × 10-2 7 × 10-3 2 × 10-2 6 × 10-4 30 37

Table 6. The Different Gaseous Atmospheres That Were Used in This Study gaseous atmosphere Exp. Exp. Exp. Exp. Exp.

1 2 3 4 5

constituents in the gas stream Hg0 Hg0 Hg0 Hg0 Hg0

vapor vapor vapor vapor vapor

mixed mixed mixed mixed mixed

only with with with with

with N2 N2, NO, and NO2 N2 and HCl N2, NO, NO2, O2, CO2, and SO2 N2, NO, NO2, O2, CO2, SO2, and HCl

of the results are mean values calculated from two or more experimental runs under the same test conditions. The experiments that were repeated showed that the performance of each sorbent varied within a maximum standard deviation of 10% of the calculated mean values. On this basis, it was justifiable to use the other results acquired from single tests to obtain a better understanding of each sorbent’s performance. Under the baseline test conditions with a gas flow of 37 mL/ min (measured at NTP) and the reactor heated to 150 °C, the pressure drop due to the Kaowool felt disc was approximately 30 mm of the water column (1.2 in. water column). When the sorbent bed was placed, this pressure drop (under the same test conditions) increased to approximately 65 mm of the water column (2.4 in. water column). During a full-scale sorbent injection test at a coalfired power plant in the U.S., the pressure drop across the bag house was found to vary between about 25 and 100 mm of the water

Figure 4. Mercury capture efficiencies at different test temperatures with a gas stream of only mercury vapor and nitrogen gas (Exp. 1).

column (2–4 in. water column).22 The Kaowool felt disc can therefore be considered to cause a flow resistance similar to the fabric filter bags used in coal-fired power plants.

5. Experimental Results and Discussion Effect of Test Temperature. Figures 4 and 5 present sorbent performance at different test temperatures for Exp. 1 and Exp. 2, respectively. Since only a limited quantity of fly ash B was available, it was tested only under the gas scheme Exp. 1, which studied the sorbent’s performance for retaining elemental mercury, the more difficult mercury species to capture. Under this experimental scheme, the fly ash B sample was able to capture more that 95% of the mercury in the gas stream when (22) Sjostrom, S.; Ebner, T.; Slye, R.; Chang, R.; Strohfus, M.; Pelerine, J.; Smokey, S. Full-scale eValuation of mercury control at Great RiVer Energy’s Stanton generating station using injected sorbents and a spray dryer/baghouse, Proceedings of the Air Quality III Conference, Arlington, VA, Sept 10–12, 2002.

Low Cost Sorbents for Mercury Capture

Figure 5. Mercury capture efficiencies at different test temperatures with a gas stream of mercury vapor, nitrogen oxide, nitrogen dioxide, and nitrogen gas (Exp. 2).

tested at 150 °C. The results for gas mixtures Exp. 1 and Exp. 2 show that the mercury capture efficiency decreases when the test temperature is increased. The introduction of NOx gases improved the effectiveness of the sorbents. Dunham et al.23 and Olson et al.24 have suggested that, in the presence of NO2, Hg0 may be catalytically oxidized on the surface of the sorbent. However, since the capture efficiency for Exp. 2 also reduced when the temperature was increased, this indicates that physical adsorption might be an initial step in the mercury capture when this gas scheme is used. Pavlish et al. have suggested a similar mechanism with mercury being finally bound to the sorbent by chemisorption.6 Effect of Gas Composition. Mercury capture efficiencies for four sorbents when exposed to the five gas mixtures, at a test temperature of 150 °C and a hold time of 60 min, are shown in Figure 6. When exposed to a mercury gas stream diluted in nitrogen, the efficiency of Norit Darco Hg and the sorbents prepared from scrap tire rubber were between about 40 and 50%. However, when the gas stream included NO and NO2, the performance of these sorbents improved and retained more than 95% of the mercury. Fly ash A captured practically all of the mercury during this latter gas scheme, and retained about 90% of the mercury, even under Exp. 1 conditions. As mentioned in the previous subsection, the improvement in sorbent performance when the gas scheme included NOx could be due to the mercury capture method proposed by Dunham et al.23 and Olson et al.24 They suggested that, in the presence of NO2, elemental mercury is catalytically oxidized on the surface of the sorbent. The introduction of chlorine into the gas stream improved the effectiveness of the activated carbons (Norit Darco Hg and the active carbon produced from scrap tire rubber). However, the presence of this gas had no effect on the performance of the charcoal prepared by pyrolyzing scrap tire rubber. The results therefore do suggest that Norit Darco Hg and active carbon from (23) Dunham, G. E.; Olson, E. S.; Miller, S. J. Impact of flue gas constituents on carbon sorbents, Proceedings of the Air Quality II;International Conference on Mercury, Trace Elements and Particulate Matter, Sept 19– 21, 2000. (24) Olson, E. S.; Sharma, R. K.; Miller, S. J.; Dunham, G. E. Identification of the breakthrough oxidized mercury species from sorbents in flue gas, Proceedings of the Specialty Conference on Mercury in the Environment, Sept 15–17, 1999.

Energy & Fuels, Vol. 21, No. 6, 2007 3255

Figure 6. Mercury capture efficiencies of the sorbents when tested with different gas compositions.

scrap tire rubber are more capable of retaining mercury when it is oxidized with chlorine. Miller et al. found that, during mercury break-though tests, the interaction between SO2 and NO2 impaired the capture of elemental mercury.25 SO2 may react with the catalytic sites that would have otherwise aided the retention of mercury.6 A similar outcome was therefore expected when the gas mix in Exp. 4 was used, as it contained a substantial concentration of SO2. Tests with this gas scheme showed that, compared to the results with gas scheme Exp. 2, the performance of the charcoal produced from scrap tire rubber was reduced by 15%. However, only a marginal reduction was observed for the two activated carbons. The observed high efficiency for the activated carbons could be because the experiments during the current research investigate the capture efficiency of the sorbents and not the mercury break-through capacity. Since only a limited amount of mercury entered the reactor during each test with a hold time of 1 h, there may have therefore been some sites capable of reacting with the NO2 and retaining the mercury in the gas stream. When the more complete simulated flue gas mix Exp. 5 was used, the sorbents retained practically all of the mercury that entered the reactor. Compared to the results observed for Exp. 4, the inclusion of chlorine in the gas stream may have caused the observed rise in the activated carbon’s performance for the tests with Exp. 5. The addition of HCl to the gas stream consisting of N2 and Hg0 had showed no improvement in the performance of charcoal produced from scrap tire rubber. However, this sorbent had a marked increase in its effectiveness when the gas scheme was changed from Exp. 4 to Exp. 5. The reason for this improvement is currently being investigated and will be discussed in a future publication. Experiments with the different gas schemes have shown that, when compared to the other gas constituents, inclusion of NOx into the simulated flue gas stream caused the most significant improvement in the sorbents’ performance. Varying Exposure Time. Mercury capture efficiencies of Norit Darco Hg and the two sorbents from scrap tire rubber (25) Miller, S. J.; Dunham, G. E.; Olson, E. S.; Brown, T. D. Flue gas effects on a carbon-based sorbent. Fuel Process. Technol. 2000, 65–66, 343–364 (Special Issue on Air Quality: Mercury, Trace Elements, and Particulate Matter).

3256 Energy & Fuels, Vol. 21, No. 6, 2007

Figure 7. Mercury capture efficiencies at different hold times with a gas stream of only mercury vapor and nitrogen gas. (Exp. 1) (sorbent mass was 100 mg).

examined at different hold times while exposed to the simulated flue gas at 150 °C are shown in Figure 7. When compared to the Norit Darco Hg, the charcoal from the pyrolysis of scrap tire rubber was marginally superior. All three sorbents seemed to be more effective during the initial 30–45 min of exposure to the gas stream. The mercury retention efficiency then gradually decreased and stabilized. The results do suggest that there may be a limited number of sites in the sorbents which retain mercury more effectively and these were spent when the duration of the tests were more than 45 min. However, the observation that the sorbents’ performance stabilized during the longer exposure experiments suggests the availability of sites capable of retaining mercury at a lesser efficiency. Furthermore, the maximum mercury capacity for each sorbent had not been reached, even when the sorbents were tested for 180 min. This was because the amount of mercury that entered the reactor during the experiments which ran for 180 min was only 193 ng. Consequently, at the end of these tests, mercury loading on the sorbents varied between approximately 0.7–0.4 ng/mg of sorbent, with the tire pyrolysis charcoal having the largest loading. After exposing Norit Darco Hg to the simulated flue gas for 180 min, the mercury loading was ∼0.5 ng/mg of sample; this is clearly negligible compared to the 2460–2590 ng/mg maximum mercury capacity reported by other researchers.6 The results therefore do suggest that the sorbents’ performance when examined using a gas stream of only nitrogen and mercury is not limited by their total capacity for elemental mercury. Effect of Sorbent Mass. Mercury capture efficiencies of Norit Darco Hg, fly ash A, and the two sorbents from scrap tire rubber when the mass of sample was reduced from the typical 100 mg mass to 20 mg is shown in Figure 8. The experiments were conducted at 150 °C with a gas stream of only mercury vapor and nitrogen gas and a hold time of 60 min. The ranking of the each sorbent’s effectiveness relative to each other did not change when the sample mass was reduced; therefore, fly ash A was found to be the most efficient. However, each sorbent’s performance decreased significantly when the sample mass was changed to 20 mg. Fly ash A and tire pyrolysis charcoal captured 1.9 and 2.2 times, respectively, less mercury with the smaller sample mass. The performance of Norit Darco Hg and the activated carbon produced from scrap tire reduced by approximately similar ratios of about 3.5 times. Furthermore,

SeneViratne et al.

Figure 8. Mercury capture efficiencies when sorbent masses of 20 and 100 mg were used on tests with a gas stream of only mercury vapor and nitrogen gas (Exp. 1) (hold up time 60 min).

these two activated carbons also had corresponding efficiencies for retaining mercury. The results do suggest that the processes by which elemental mercury is captured on the two activated carbons are comparable and may be different from the capture mechanism on the fly ash A and charcoal from the pyrolysis of scrap tire rubber. In each test, only about 64 ng of mercury entered the reactor; therefore, at the end of each experiment, the final mercury loadings on the 20 mg sorbent samples were considerably less than the maximum mercury capacity of Norit Darco Hg found by other researchers.6 The reduction in each sorbent’s performance when the sample mass was decreased was therefore not limited by the sorbents’ mercury capacity. On the other hand, the sorbent bed height is directly related to the mass of sorbent used in each test. Consequently, the time for which a given volume of gas was in contact with the sorbent bed would have also been reduced by 5 times when the sample mass was changed to 20 mg. The decrease in sorbent performance may therefore be related to mass transfer and kinetic limitations. However, the observation that the reduction in each sorbent’s performance was not identical suggests that other factors, which could not be ascertained by the tests conducted in this study, also seem to have a bearing on the effectiveness for retaining mercury. When comparing the different sorbents ranked in this study, the fly ash were found to be the most effective when exposed to a simulated flue gas stream of mercury vapor and nitrogen gas. However, the ash samples also had the smallest BET surface areas and micropore volumes. As noted earlier, the experimental parameters used in this study resulted in the mercury loadings being only a small fraction of the maximum mercury capacity reported for commercially available Norit Darco Hg.6 The results therefore do suggest that the limited micropore volume available in the fly ash samples was adequate for retaining most of the mercury that entered the reactor during each test. Consequently, the total micropore volume available in the highly porous Norit Darco Hg and activated carbon derived from tire rubber is barely utilized. Other researchers have found that mercury retention on fly ash increases with increasing unburnt carbon content of the ash, i.e., loss on ignition (LOI). This effect was however dependent

Low Cost Sorbents for Mercury Capture

on the type of coal being combusted.5,16,26 It has also been suggested that the oxidation of elemental mercury increases with both the LOI and the ash surface area.5 Fly ash A and fly ash B had LOI values of 8 and 16%, respectively, and this may explain why fly ash B was the more effective fly ash. Fly ashes from most coal-fired utility plants are sold as a cement replacement, and carbon contents in ash above ∼1% can adversely affect the sale value of fly ash for cement replacement in concrete, even though ASTM standards point up to 6% carbon.6 The use of ash with high LOI for retaining more mercury may therefore depend on which operation is more economically viable. The introduction of NOx into the simulated flue gas stream was found to significantly improve the sorbents’ performance. While the charcoal and activated carbon derived scrap tire rubber were not exposed to NOx gases during the production of these sorbents, the fly ash samples had already been exposed to the NOx gases in the coal-fired power plants and may have therefore been “conditioned” by these gases. This may explain why, when tested for mercury capture with gas scheme Exp. 1, the fly ash was more effective than the sorbents prepared from scrap tire rubber. When comparing the charcoal and activated carbons prepared from scrap tire rubber, the charcoal was found to be marginally more effective in retaining mercury when examined using gas scheme Exp. 1. It was postulated in the section describing the characteristics of each sorbent that, when compared to the activated carbon, the charcoal sample possibly contains a larger concentration of sulfur in the amorphous phase. This sulfur may be more capable of oxidizing the mercury present in the gas stream and therefore could be the reason why the charcoal was found to be more effective when tested with gas scheme Exp. 1. Tests with Norit Darco Hg at coal-fired utility plants have shown that it is able to capture between 60 and 90% of the mercury in the flue gas stream, depending on the type of coal and the particulate control system that is employed.27 Fabric filters in coal-fired power plants are cleaned at frequencies that are related to the pressure drop across the bag house and filterbag life. For example, the filters in the bag house at E.C. Gaston Electric Generating Plant in Alabama were cleaned at a frequency of 1.5 pulses/bag/h.28 Consequently, the filter cake of coal fly ash and sorbent particles would gradually collect on the filter over a period of 40 minutes before the bag is cleaned. The sorbents tested in at this plant would have therefore been in contact with the flue gas for a maximum time of 40 minutes. When the more complete simulated flue gas scheme Exp. 5 was used, the sorbents were found to be capable of retaining almost all of the mercury present in the inlet gas stream. There are probably several reasons why the bench-scale results differ from the tests at coal-fired utility plants. For example, on the benchscale rig, the simulated flue gas passed through a moderately deep bed of sorbent (2.4–5.6 mm), while FFs in the large-scale (26) Hassett, D. J.; Eylands, K. E. Mercury capture on coal combustion fly ash. Fuel 1999, 78, 243–248. (27) Nelson, S.; Landreth, R.; Zhou, Q.; Miller, J. Accumulated PowerPlant Mercury-RemoVal Experience with Brominated PAC Injection, Proceedings of the Combined Power Plant Air Pollutant Control Mega Symposium, Aug 30–Sept 2, 2004. (28) Bustard, C. J.; Durham, M.; Lindsey, C.; Starns, T.; Baldrey, K.; Martin, C.; Schlager, R.; Sjostrom, S.; Slye, R.; Renninger, S.; Monroe, L.; Miller, R.; Chang, R. Full-scale evaluation of mercury control with sorbent injection and COHPAC at Alabama Power E.C. Gaston. http:// epw.senate.gov/107th/Gaston_paper.pdf.

Energy & Fuels, Vol. 21, No. 6, 2007 3257

tests have a much thinner filter cake (0.8–1.6 mm).29 Sorbent injection tests at power plants with ESPs have shown that mercury sorption mainly occurs during the “in-flight” period while the flue gas flows to the ESP and in this case the flue gas does not pass through a packed bed.30 Consequently, the contact between the gas stream and the sorbent particles is much better in the laboratory test system. In addition, it is extremely difficult to replicate all of the conditions of a large coal-fired utility at bench scale. Other studies have suggested that if SO3 is present, even in ppm concentrations, it may bind to the basic sites on the sorbent and therefore restrict the adsorption of mercury.31 This gas is always present in the flue gases from coal-fired plants at low ppm levels but has not been included in the gas mix for this study to date. The next phase of the current investigation will include the addition of SO3 and moisture to the gas mix. The results obtained by these scheduled tests will be reported in a future publication. The charcoal produced from scrap tire rubber and commercially available Norit Darco Hg had similar mercury capture efficiencies to when tested using the more complete simulated flue gas scheme in Exp. 5. Furthermore, experiments with a gas stream of mercury vapor and nitrogen gas showed that the charcoal was marginally more effective than the latter activated carbon. Since the charcoal is produced from scrap tire rubber while Norit Darco Hg is made from lignite coal, the raw material cost for the first sorbent would be significantly less. Furthermore, the charcoal was carbonized at a relatively low temperature of 700 °C and did not undergo an activation process. Although no information is available on the process by which the Norit Darco Hg produced, this sorbent was most probably subjected to a steam or chemical activation process. It would therefore be cheaper to produce the charcoal from scrap tire rubber, which has a similar performance to Norit Darco Hg. 6. Summary and Conclusions The mercury capture performance of pyrolysis char and activated carbon derived from scrap tire rubber, and also coal fly ashes, has been compared against Norit Darco Hg by using a bench-scale experimental rig. Tests with a gas flow of only mercury and nitrogen revealed that chars from the pyrolysis of scrap tire rubber have a similar performance to Norit Darco Hg (a commercially available sorbent). When NO and NO2 were introduced into the simulated flue gas stream, there was a significant improvement in the mercury adsorption of all of the sorbents, possibly by the formation of a mercury nitrate. However, since the mercury capture efficiency of all of the sorbents decreases with increasing test temperature, physical sorption may be the initial mechanism by which the mercury is captured. Under the most realistic conditions tested, experiments at 150 °C with a simulated flue gas comprised of N2, NO, NO2, CO2, O2, SO2, and HCl, all four sorbents captured approximately 100% of the mercury in the gas stream. While it is recognized that the translation of laboratory-scale results to full-scale plant (29) Brown, T.; O’Dowd, W.; Reuther, R.; Smith, D. Control of mercury emissions from coal-fired power plants: A preliminary cost assessment, Proceedings of the Advanced Coal-Based Power & Environmental Systems Conference, July 21–23, 1998. (30) Bustard, C. J.; Durham, M.; Lindsey, C.; Starns, T.; Martin, C.; Schlager, R.; Sjostrom, S.; Renninger, S.; McMahon, T.; Monroe, L.; Goodman, J. M.; Miller, R. Results of actiVated carbon injection for mercury control upstream of a COHPAC fabric filter, Proceedings of the Mega Meeting: Power Plant Air Pollution Control Symposium, Washington D.C., May 19–22, 2003. (31) Holmes, M.; Dunham, G.; Edwin, O.; Mibeck, B.; Zhaung, Y.; Pavlish, J. DeVelopment of mercury control technologies; CATM Annual report 2004.

3258 Energy & Fuels, Vol. 21, No. 6, 2007

performance is not straightforward, these initial results are taken as encouraging. There was a wide variation in the BET surface area, micropore area, and micropore volume of the different sorbents. However, this study examined the capture efficiency and not the equilibrium capacity of the sorbents; thus, the modest surface area and micropore volume of the fly ash was still adequate for retaining the small quantity of mercury used in each experimental run. XRD analysis of the sorbents showed that they are mainly amorphous and have distinct differences in the crystalline phases. The main crystalline phases in the ash samples were mullite and quartz; while one sample also had some magnetite and lime phases present. The major crystalline phases in Norit Darco Hg were quartz and calcite. Wurtzite was the primary crystalline phase in char and activated carbon derived from scrap tire rubber. In the next phase of this study, SO3 and moisture will be included in the simulated flue gas stream. The inclusion of these gases may shed some light as to why laboratory results for mercury capture efficiencies are significantly different from the outcomes of tests at coal-fired power plants. More sorbent characterizations which include scanning electron microscopy with energy dispersive X-ray analysis will also be undertaken to obtain a better understanding of why some sorbents were more effective under certain simulated flue gas conditions. Fly ash from coal-fired power plants is either sold for use as a concrete admixture or disposed in landfills.32,33 Consequently, (32) Muggli, D.; Bustard, C. J.; Campbell, T.; Schlager, R.; O’Palko, A.; Chang, R.; Robers, R.; Kolbus, M.; Rees, M. TEXECON II and hightemperature reagents or sorbents for low-cost mercury remoVal, Proceedings of the Power plant air pollutant control Mega symposium, Baltimore, MD, Aug 28–31, 2006.

SeneViratne et al.

the sorbents will be exposed to rain water. The sorbents discussed in this study will therefore be tested to ascertain the leachability of the mercury that was retained, and the findings will be reported in a future publication. Since oxidized mercury is more readily retained on the sorbents, activated carbons like Sorbent Technologies’ B-PAC, B-PAC-Low Cost, and Norit Darco Hg-LH which have bromine present have exhibited significantly improved mercury removal performance when compared to untreated carbons like Darco Hg.34 Sorbents prepared from waste material and impregnated with bromine are currently being examined for their effectiveness, and the results will be reported in a future publication. Acknowledgment. Support for this study by the British Coal Utilisation Research Association (BCURA) under Contract No. B73 is gratefully acknowledged. The authors also express their gratitude to P S Analytical UK for generously providing the PSA Cavkit, without which this study would have not been possible, and to Dr. Geoffrey Fowler of the Environmental & Water Resource Engineering Section, Imperial College London, for use of the carbon activation rig. Thanks are also due to Dr. David Fitzgerald of Doosan Babcock Energy Limited, to Dr. Will Quick of E.ON UK for their helpful advice; to Dr. Graham Reed for help in designing the novel fixed-bed reactor; and to Dr. Fraser Wigley from the Royal School of Mines, Imperial College London, for advice on XRD analysis. EF070028X (33) Yudovich, Y.; Ketris, M. P. Mercury in coal: a review - Part 2. Coal use and environmental problems. Int. J. Coal Geol. 2005, 62, 135– 165. (34) Landreth, R. R.; Nelson, S., Jr.; Liu, X.; Tang, Z.; Miller, J. E.; Hoeflich, P. C.; Moore, G.; Brickett, L. A. New Full-Scale Results from B-PAC Control Trials, XPower Plant Air Pollutant Control “Mega” Symposium, Baltimore, MD, Aug 28–31, 2006.