Tetra Sulfide for Mercury Vapor Capture - American Chemical Society

2290

Energy & Fuels 2008, 22, 2290–2298

Copper-Doped Silica Materials Silanized With Bis-(Triethoxy Silyl Propyl)-Tetra Sulfide for Mercury Vapor Capture D. E. Meyer,† N. Meeks,† S. Sikdar,‡ N. D. Hutson,‡ D. Hua,§ and D. Bhattacharyya*,† Department of Chemical and Materials Engineering, UniVersity of Kentucky, Lexington, Kentucky 40506, U.S. EnVironmental Protection Agency, and J. M. Huber Corporation ReceiVed March 17, 2008. ReVised Manuscript ReceiVed May 9, 2008

The use of Cu-S sites for Hg capture from the gas phase has been successfully applied to a silica-based platform using an S4 organic polysulfane and copper sulfate. The maximum fixed-bed equilibrium capacity achieved using these materials was 19 789 µg Hg · g-1 sorbent for a material with 2.5 wt % Cu and 6 wt % S. An optimal S level was determined to be around 3 wt % because enhancement of capacity was only 18% when increasing from this 3 to 6 wt %. The rate of adsorption in pure beds ranged from 0.6 to 1.6 µg Hg · min-1 depending on the inlet concentration. Differences in breakthrough times suggest that material deposition is not uniform. When compared to two other platforms, commercially available Darco HG-LH and previously tested Fe-Cu-S4 nanoaggregates, the Si-1 material performed the best in fixed-bed testing. During entrained-flow testing, a steady-state Hg removal of 82% was achieved using Si-1 at injection rates of both 6 × 10-5 and 1.2 × 10-4 g · L-1 · h-1. The lack of increase in Hg removal when the injection rate is doubled suggests that pore accessibility is the rate-controlling step during dynamic Hg capture. A calculation of the approximate pore usage based on injection testing helped confirm this observation. During injection testing, the performance of Si-1 was only diminished 10% when exposed to 20 ppm SO3. This is an encouraging result for flue-gas applications where SO3 levels range from 1 to 40 ppm. Testing demonstrated that Si-1 is stable when exposed to leaching conditions after concrete blending and cement impregnation. This is an important aspect to consider for injection because the sale of fly ash for concrete is a key cost-recovery tool for power plants.

Introduction With the rising concerns regarding Hg accumulation within the environment, a great deal of research has been focused on its safe removal from key sources such as the flue-gas of coalfired power-plants.1 Such a task may require mercury-specific control technology for compliance.2 Studies have shown that the Hg in the flue gas occurs as either elemental mercury (Hg0), oxidized mercury (Hg2+, usually as HgCl2), or particulate-bound mercury (Hgp).3–5 Of the three forms, the Hg0 vapor poses the most challenge for removal because it is relatively insoluble in water and is not readily captured simultaneously in other air pollution control equipment.3–5 As a result, there has been considerable work in the development and testing of innovative and cost-effective control technologies for the elemental form.2 The two main approaches that have been studied are either the direct capture of Hg0 by injection of sorbents or the use of oxidation catalysts to transform Hg0 to Hg2+ for removal during * Corresponding author. Phone: 859-257-2794. Fax: 859-323-1929. E-mail: [email protected]. † University of Kentucky. ‡ U.S. Environmental Protection Agency. § J. M. Huber Corporation. (1) EPA: Mercury - Controlling Power Plant Emissions. http://www.epa.gov/mercury/control_emissions/index.htm (accessed January 2008). (2) Srivastava, R. K.; Hutson, N. D.; Martin, G. B.; Princiotta, F.; Staudt, J. EnViron. Sci. Technol. 2005, 41, 1385–1393. (3) Lu, D. Y.; Granatstein, D. L.; Rose, D. J. Ind. Eng. Chem. Res. 2004, 43, 5400–5404. (4) Pavageau, M.-P.; Pecheyran, C.; Krupp, E. M.; Morin, A.; Donard, O. F. X. EnViron. Sci. Technol. 2002, 36, 1561–1573. (5) Presto, A. A.; Granite, E. J. EnViron. Sci. Technol. 2006, 40, 5601– 5609.

SO2 scrubbing processes.5–12 Of the two, removal by sorbent injection has received the most attention. The most widely studied material for injection is powdered activated carbon (PAC).8–11 Hsi et al. examined sulfurimpregnated activated carbon fibers to determine the effects of sulfur speciation and impregnation temperature on Hg capture.10 For an inlet Hg concentration of 50 ppmv, they found that a capacity of 11.3 mg Hg · g-1 could be achieved at 135 °C when sulfur impregnation occurred at 400 °C. In addition, it was determined that the sulfur deposited as elemental sulfur was most likely responsible for approximately 70% of the Hg sorption, although organic sulfur was not specifically tested. While it does possess the ability to capture mercury, traditional activated carbon has been found to possess disadvantages, such as a high cost of material and loss of Hg capacity at elevated temperatures.6 In addition, the performance of traditional activated carbon is known to be significantly impacted by the presence of sulfur trioxide.13 To improve these performance deficiencies, a halogenated site, often bromine, has been (6) Lee, J.-Y.; Ju, Y.; Keener, T. C.; Varma, R. S. EnViron. Sci. Technol. 2006, 40, 2714–2720. (7) Pitoniak, E.; Wu, C.-Y.; Mazyck, D. W.; Powers, K. W.; Sigmund, W. EnViron. Sci. Technol. 2005, 39, 1269–1274. (8) Huggins, F. E.; Huffman, G. P.; Dunham, G. E.; Senior, C. L. Energy Fuels 1999, 13, 114–121. (9) Liu, W.; Vidic, R. D.; Brown, T. D. EnViron. Sci. Technol. 2000, 34, 154–159. (10) Hsi, H.-C.; Rood, M. J.; Rostam-Abadi, M.; Chen, S.; Chang, R. EnViron. Sci. Technol. 2001, 35, 2785–2791. (11) Feng, W.; Borguet, E.; Vidic, R. D. Carbon 2006, 44, 2998–3004. (12) Zhao, Y.; Mann, M. D.; Pavlish, J. H.; Mibeck, B. A. F.; Dunham, G. E.; Olson, E. S. EnViron. Sci. Technol. 2006, 40, 1603–1608. (13) Presto, A. A.; Granite, E. J. EnViron. Sci. Technol. 2007, 41, 6579– 6584.

10.1021/ef8001873 CCC: $40.75  2008 American Chemical Society Published on Web 06/26/2008

Copper-Doped Silica for Mercury Vapor Capture

deposited on traditional PACs. However, the removal of the PAC occurs in the ESP, contaminating the fly ash that is normally sold for concrete production to help recover plant costs. Therefore, the development of suitable noncarbon sorbents for Hg capture is of great benefit. The ability of iron (Fe) nanoaggregates modified with organic polysulfanes (R-SN-R) and trace quantities of copper (Cu) to remove Hg0 has been demonstrated. Organic polysulfanes are a good source for deposition of S because they can be formed in situ if desired through a variety of reaction sequences.14 The amount of S can be tailored to the desired application because “N” can theoretically have any value greater than 2. Makkuni and co-workers reported a capacity of 5.3 mg Hg · g-1 sorbent (2.6 × 10-2 mmol Hg · g-1 sorbent) for Fe nanoaggregates containing 2.0 wt % Cu and 6 wt % organic S (as bis-(triethoxy silyl propyl)-tetra sulfide, N ) 4) at 140 °C.15 Complete removal of Hg0 at inlet concentration of 367 ppb was sustained for >100 min at a rate of 6.2 µg Hg0 · g-1 sorbent · min-1. Our previous work examined the influence of composition on Fe nanoaggregates doped with copper and functionalized with organic sulfur (S4) to probe the mechanism of capture.16 For a sulfur content of 4 wt %, the total capacity of the Fe-based sorbents increased with increasing copper from 170 µg Hg · g-1 sorbent, with no copper doping, to 2.7 mg Hg · g-1 sorbent at 1.2 wt % Cu. However, the capacity of nonsulfur doped aggregates was only determined to be 180 µg Hg · g-1 sorbent. This led to the conclusion that a combination of both elements is necessary for the development of high-capacity sorbents. For this type of sorbent, the Fe nanoparticles serve only as a support platform for copper deposition and silanization. Therefore, it is possible to extend the use of the Cu/S site to other suitable noncarbon materials. The most obvious choice based on the use of silane chemistry is a silica material because of the large quantity of hydroxyl groups (3-5 silanol groups per nm2) on the surface that are available for functionalization. The use of functionalized silica materials for mercury capture has been investigated thoroughly for liquid phase mercury capture applications, yielding key insights regarding the factors impacting the adsorption process.17–19 This typically involves the deposition of a thiol-containing silane such as 3-mercaptopropyltrimethoxysilane (MPTMS), which leads to capacities of 200-400 mg Hg · g-1 sorbent (1-2 mmol Hg · g-1 sorbent) based on the strong affinity of Hg2+ for thiol. Walcarius and co-workers studied the diffusion and capture of Hg2+ in various MPTMS-functionalized commercial silica gels with average particles sizes from 60-150 µm and average pore diameters between 4 and 15 nm.17 They determined that initial average diffusion coefficients for Hg within the particles were 3.5-10 × 10-10 cm2 · s-1, 4 orders of magnitude less than in the liquid phase. The rate of diffusion was influenced mainly by pore size and functional group density. At greater than 50% loading, the diffusion process was reported to slow down as a result of the Hg accumulation within the pores narrowing the pore diameter. Bibby and Mercier used a one-step process during which fresh mesostructured silicas were prepared and functionalized with MPTMS simultaneously.18 The resulting materials were spheri(14) Steudel, R. Chem. ReV. 2002, 102, 3905–3946. (15) Makkuni, A.; Varma, R. S.; Sikdar, S. K.; Bhattacharyya, D. Ind. Eng. Chem. Res. 2007, 46, 1305–1315. (16) Meyer, D. E.; Sikdar, S. K.; Hutson, N. D.; Bhattacharyya, D. Energy Fuels 2007, 21, 2688–2697. (17) Walcarius, A.; Etienne, M.; Lebeau, B. Chem. Mater. 2003, 15, 2161–2173. (18) Bibby, A.; Mercier, L. Chem. Mater. 2002, 14, 1591–1597. (19) Olkhovyk, O.; Jaroniec, M. Ind. Eng. Chem. Res. 2007, 46, 1745– 1751.

Energy & Fuels, Vol. 22, No. 4, 2008 2291

cal with a uniform mean particle size on the range of 10-12 µm and an average pore size of 2-3 nm. The Hg diffusion coefficients reported for these materials were on the range of 10-10-10-11 cm2 · s-1. Interestingly, the rate of diffusion was found to increase with both thiol loading and Hg uptake. This suggests that the initial penetration of Hg2+ into a small pore size can be a rate limiting step during the adsorption process. This problem is then overcome as Hg accumulates within the pores through a mechanism of ion permeation and displacement (site-jumping). Olkhovyk and Jaroneic have created periodic mesoporous mesosilicates (PMOs) using S4 functionalization.19 The high surface area materials (678-705 m2 · g-1) had moderate Hg2+ capacities of 183-275 mg Hg · g-1 sorbent using 3 mol % of the S4 site. The capture of Hg0 from the gas-phase is more challenging to understand because the mechanism of capture is more complex than the simple Hg2+-thiol interaction encountered in the liquid-phase. For example, Hutson and co-workers have determined that Hg0 capture by both brominated and conventional PACs results in the formation of oxidized Hg species at the particle surface.20 This means an additional reaction step must be included in the process of adsorption from the bulk phase. Eswaran and co-workers have studied the gas-phase adsorption of Hg using activated carbon, char, and mordenite.21 They found that the rate of adsorption increased with temperature, inlet Hg concentration, and acid gas concentration. The presence of acid gases (SO2 and NO) is believed to increase the surface reactivity of the activated carbon and char. Without these gases present, mordenite was found to promote Hg oxidation. Inlet Hg concentration had the greatest impact on the mordenite, a zeolite, which suggests that the mass transfer limitations are dominant with this material. It is logical to assume that the pore size and tortuosity will still ultimately control the rate of adsorption in mesoporous materials. Therefore, understanding these mass transfer phenomena in the gasphase is important because the short contact times encountered in gas-phase Hg applications will require materials that possess rapid rates of adsorption, or high dynamic capacities. The current work will examine the gas-phase adsorption of Hg0 by functionalized mesoporous silicas to determine how material properties (pore structure/size, site density) impact the adsorption process. The materials will utilize the previously reported Cu-S site with a slight variation occurring through the use of copper sulfate as a copper source. As with the Fe nanoaggregates, the sulfur is tethered to the silica surface as organic sulfur via traditional dry silanization chemistry using a silane material containing a tetrasulfide (S4) center. The objectives of this work are: (1) examine the impact of both S and Cu on Hg0 removal during fixed-bed contact at moderate temperature (140 °C) to better understand the mechanism of Hg capture; (2) perform vapor-phase Hg capture during entrained flow at 140 °C in a simulated flue gas stream; (3) compare results of fixed-bed and entrained-flow sorbent testing to determine what factors influence dynamic Hg capture; and (4) assess the potential environmental impact of sorbent-bound Hg using leaching studies. For comparison, results for a (20) Hutson, N. D.; Attwood, B. C.; Scheckel, K. G. EnViron. Sci. Technol. 2007, 41, 1747–1752. (21) Eswaran, S.; Stenger, H. G.; Fan, Z. Energy Fuels 2007, 21, 852– 857.

2292 Energy & Fuels, Vol. 22, No. 4, 2008

Figure 1. Experimental setup for fixed-bed testing of Hg sorbents.

commercial halogenated PAC, Norit Darco HG-LH, and Fe nanoaggregates functionalized with Cu/S are also included. Experimental Methods Silica samples were provided by the J. M. Huber Corporation. For most materials, mesoporous silica gel particles were doped with copper (as copper sulfate, CuSO4) and silanized with bis-(triethoxy silyl propyl)-tetra sulfide (S4) using dry silanization techniques. The materials were conditioned prior to Hg exposure using ultrahigh purity nitrogen (N2) at 140 °C for 30-60 min. A variety of silicabased samples were provided to examine the impact of composition on Hg capture. Surface Area Measurements by N2 Sorption. A 100-mg sample was prepared by outgassing for 6 h at 140 °C under N2 refluxing. The N2 adsorption isotherm at 77 K was then measured using a Micromeritics Tristar 3000 pore volume analyzer. The total surface area was then calculated using the Brunauer-Emmett-Teller (BET) method. Scanning Electron Microscopy (SEM) with Energy Dispersive X-Ray (EDS) Analysis. Scanning electron microscope images of select sample were acquired on a Hitachi Model S-3200-N scope using a working distance of 15 mm and source voltage of 20 kV. A sample of Si-1 after mercury saturation was pressed onto carbon tape affixed to an Al stage (15 mm diameter). Elemental probing was performed using EDS. Fixed-Bed Mercury Vapor Adsorption. The total adsorption capacity of all functionalized silicas was determined using a fixedbed system reported previously by our group.16 The experimental setup is shown in Figure 1. The packed-bed consisted of a 500-mg sorbent bed suspended in one arm of a 0.9 cm I.D. glass U-tube using approximately 0.15 g of glass wool to prevent particle loss under gas flow. The U-tube was maintained at 140 °C using a silicon oil bath and temperature controller. Mercury vapor was generated using a mercury permeation tube (VICI Metronics) that is capable of producing 1596 ng · min-1 of Hg0 at 100 °C. The permeation tube was supported on glass beads packed in a glass U-tube and maintained at 100 °C using a silicon oil bath. Ultrahigh purity nitrogen, N2 (99.999%, Scott Gross), was used as the carrier gas and was fed into the system at 60 mL · min-1 using a Brooks Model 5850E mass flow control valve and Model 0152 controller. At these conditions, the concentration of Hg0 in the inlet to the packed bed should be 26,600 µg · m-3 (2.97 ppm). The empty-bed residence time based on a bed height of approximately 1 cm was 0.65 s. The concentration of Hg used was three orders-of-magnitude larger than the concentration of Hg in most flue-gases (2 ppb). This provided rapid attainment of equilibrium capacities using moderate time periods. It is reasonable to expect that the Hg inlet concentration used will suppress mass transfer resistance during fixed-bed testing because of the large concentration gradient. However, the fixed-

Meyer et al. bed values obtained are used in conjunction with entrained-flow test results to assess sorbent performance. Therefore, exposure to typical Hg levels observed in power-plant flue gases is taken into account. The system was equipped with four 3-way valves that allow for bypass of the Hg source, the packed bed, or both. Qualitative realtime monitoring of Hg was achieved using a Buck Scientific Cold Vapor Mercury Analyzer 400. The detector supplied an output signal in terms of the millivolt response to the change in percent transmittance at a 253.6-nm wavelength to a coupled computer with a sampling interval of 0.5 s. Typical tests began by first bypassing the Hg source and feeding nitrogen through the packed bed to establish a zero baseline for the detector. The nitrogen flow was then routed to the Hg permeation tube and fed to the packed bed until a minimum of 90% exhaustion was achieved. A 20-mg sample of the contents of the bed was digested overnight using 40 mL of an aqua regia medium (4 M HCl and 1 M HNO3). The amount of total Hg was then quantified using inductively coupled plasma (ICP) elemental analysis. Mercury Analysis Using Inductively-Coupled Plasma (ICP) Analysis. Digested samples were analyzed for Hg using a Varian Vista-Pro CCD Simultaneous ICP-OES (optical emission spectrometer) under the supervision of the University of Kentucky’s Environmental Research and Training Laboratory. Analysis of Hg was performed at wavelengths of 194.164 and 253.652 nm. These wavelengths were selected because of their lack of susceptibility to Cu interference. The instrument was programmed to obtain readings in triplicate for each sample. An acidic rinse cycle was used between samples to avoid carry-over. The detector was calibrated to a lower limit of 0.5 ppm Hg, with a linear trend existing over the entire calibration range. The lower detectable limit was 0.2 ppm Hg. The upper calibration limit was selected based on the anticipated amount of Hg captured for a given sorbent and ranged from 6-10 ppm Hg. Standards were prepared by digesting fresh sorbent (no Hg exposure) in the same aqua regia medium and spiking with appropriate quantities of a known Hg standard (Fisher) to produce the desired concentrations. Prior to analysis of digested samples, known samples were first prepared and analyzed to verify the accuracy of the machine. Both a 1.5 and 0.5 ppm known sample routinely gave errors of less than 1%. Digested samples of exhausted beds were split in half, with a 0.5 ppm Hg spike added to one sample to verify the analysis by method of additions. This method was checked using a 0.5 ppm known sample, which yielded a recovery error of 9%. Leaching Studies. The loss of Cu and Hg from a selected sorbent was examined both before and after impregnation in concrete. Concrete samples were prepared using a Quikrete Fast-Setting cement mix that had been degraveled. A typical concrete blend consisted of 65 wt % cement, 34.5 wt % fly ash obtained from a Reliant Energy coal-fired power plant, and 0.5 wt % of Hg-saturated sorbent. This composition was based on typical standards for fly ash in concrete.22 The dry materials were thoroughly mixed and then prepared for casting through the addition of deionized, ultrafiltered water to obtain a suitable working viscosity. After pouring, the concrete slug was allowed to dry and set overnight. Leaching of sorbents after Hg exposure were performed by placing 50 mg of sorbent in 20 mL of deionized ultrafiltered water with an adjusted pH of 4.0 through the addition of nitric acid. The leaching was carried out under mixing for 12 h. Concrete leaching studies were conducted using EPA Method 1312 Synthetic Precipitation Leaching Procedure.23 Concrete slugs were placed in deionized ultrafiltered water with an adjusted pH of 4.2 through the addition of a solution of sulfuric and nitric acid (60 vol % sulfuric and 40 vol % nitric). The liquid-to-solid ratio was 20:1. Leaching was allowed to occur under mixing for 24 h. In both cases, samples were prepared for Hg and Cu analysis by ICP. The analysis (22) Golightly, D. W.; Sun, P.; Cheng, C.-M.; Taerakul, P.; Walker, H. W.; Weavers, L. K.; Golden, D. M. EnViron. Sci. Technol. 2005, 39, 5689–5693. (23) Method 1312: Synthetic Precipitation Leaching Procedure. http:// www.epa.gov/SW-846/pdfs/1312.pdf (accessed January 2008).



Figure 2. Diagram of the entrained flow reactor (EFR) used for movingbed (injection) testing. Table 1. Experimental Conditions for Entrained-Flow Mercury Capture Experiments target

unit

HCl SO2 NO CO2 CO O2 H2 O N2 Hg0 (g) total flow temperature

ppm ppm ppm vol % ppm vol % vol % vol % µg/nm3 L/min °C

wet basis

dry basis

10 500 200 3.5 ∼5 6.8 6.8 balance ∼21.5 14 140

10.8 538 215 3.8 ∼5 7.3 NA balance ∼20 13 140

remark from gas cylinders from gas cylinders from gas cylinders from methane burner from methane burner from methane burner from methane burner from cylinders/burner Hg0 permeation tube std cond, 1 atm, 20 °C 139-144 °C

of Cu by ICP followed the same procedures used for Hg analysis given above with the exception of the selected wavelength for analysis (213.598, 219.227, 224.700). Moving-Bed (Injected) Vapor-Phase Hg Capture in Simulated Flue Gas. In-flight mercury capture experiments were conducted in the U.S. EPA’s entrained flow reactor (EFR, Research Triangle Park, NC). A detailed description of the EFR setup has been provided previously.16 A schematic of the basic system is shown in Figure 2. Briefly, the system consisted of a 332 cm by 4 cm ID Pyrex contactor that was heated by three electric tube furnaces (Lindberg, USA) to maintain a constant controlled temperature. A water-cooled methane gas burner provided combustion flue gases (CO, CO2, H2O, and O2) while other flue gas components (N2, SO2, NO, HCl) were introduced into the reactor at constant concentrations using compressed gases and mass flow controllers. Elemental mercury vapor (Hg0) was generated using a permeation device (Dynacalibrator, VICI Metronics) with a N2 carrier and subsequently mixed with the simulated flue gas before entering the EFR. The sorbent was entrained into the reactor using a nitrogen purge stream. The sorbent feeding assembly consisted of a gas supply manifold, a feed tube, and a syringe pump mounted to a vibrating plate. For all tests, the sorbent was diluted with an inert material (diatomaceous earth (DE) flour) at varying mixtures depending upon the desired sorbent injection rate. This dilution allowed the syringe pump to perform in the middle of its operating range and helped to maintain a steady sorbent addition rate. Although the EFR is capable of simulating a variety of conditions, the gases were mixed to simulate the flue gas that would be expected from the combustion of a low-sulfur Western sub-bituminous coal (e.g., such as that from Wyoming’s Powder River Basin). The experimental conditions are given in Table 1. Experiments were initiated by heating the reactor to the desired temperature and performing the necessary calibration checks for continuous emissions monitoring. An advanced Hg continuous emission monitor (CEM) Nippon DM-6B (NIC, Japan) provided the concentration and speciation (elemental vs oxidized) of Hg in the simulated coal combustion flue gas using a 10 s sampling interval. Baseline measurements were collected before beginning injection of the sorbent. The sorbent-DE mixture was added

Figure 3. Nitrogen adsorption isotherms (77 K) for the bare silica precursor (Si), copper-impregnated silica (Si-Cu), and fully functionalized Si-1 (Si-Cu-S4) showing a type 4, multilayer adsorption.

continuously for at least 20 min before stopping the injection and diverting the gas stream to bypass the online Hg analysis. The system was cooled and the reactor walls and tubing were thoroughly cleaned between experiments to prevent “memory effects” due to any accumulated sorbent on the reactor or tubing walls.

Results and Discussion Material Characterization. Nitrogen adsorption at 77 K was performed on the blank silica precursor, the Cu-impregnated silica, and S-functionalized Cu-impregnated silica (Si-1) to examine the impact of functionalization on pore accessibility. The results for nitrogen adsorption as a function of relative pressure for all three materials are shown in Figure 3. All three isotherms exhibit type IV behavior with condensation occurring in the mesopores at elevated pressures. The blank precursor, Si, and functionalized material, Si-Cu-S4, have the same shape. As expected, the functionalized material has a decreased volume in comparison. Interestingly, the copper-impregnated material (Si-Cu) exhibited a large increase in the volume of gas adsorbed. This is the result of enhanced surface roughness after impregnation with CuSO4. The Barrett-Joyner-Halenda (BJH) average pore size after functionalization with both Cu and S4 is 2.9 nm. Fixed-Bed Hg0 Capture. A summary of results for fixedbed testing using pure sorbent beds is shown in Table 2. The total capacity of materials ranges from 9738 to 19 789 µg Hg · g-1 sorbent. For even the low end of the range, this is an 84% increase in capacity for Cu/S materials when compared to the capacity of 5.3 mg · g-1 sorbent reported by Makkuni and co-workers using Fe nanoaggregates as a support with similar levels of Cu and S. The highest capacity was achieved using 2.5 wt % Cu and 6 wt % S. A better understanding of the role of Cu and S can be gained through comparison of the various materials. The immediate impact of S is observed when comparing Si-2 and Si-3. For a constant level of Cu, the capacity of materials increases 72% as the S level is increased from 1 to 3 wt %. Similarly, when comparing Si-1 and Si-4, the capacity increases 18% as the S level is increased from 3 to 6 wt %. These observations suggest that an optimal S level exists around 3 wt % because enhancement of capacity is much less pronounced beyond these levels. Variation of Cu at the levels tested showed no real impact. This can be seen when comparing Si-2 and Si-4 where the capacity is essentially unchanged while the Cu level is increased from 3 to 5 wt %. On the basis of


Meyer et al.

Table 2. Summary of Results for Fixed-Bed Testing material

wt % Cu

wt>% S

Si-1 Si-2 Si-3 Si-4 Si-1a Fe-Cu-S4a Darco HG-LHa

2.5 5 5 3 2.5 1.2

6 3 1 3 6 3.5

a

inlet Hg Conc (ppm)

linear rate of adsorption (µg Hg · min-1)

equilibrium capacity (µg · g-1 sorbent)

breakthrough onset (min)

tfinal (min)

3.0 2.2 2.0 1.2 4.3 4.5 3.1

1.6 1.2 1.1 0.6 2.4 2.4 1.7

19789 16777 9738 16800 19789 2730 2942

30 4440 3510 6978 20 7 10.5

12577 10091 6749 17443 910 30 40

Blended-bed exposure involving 10-50 mg sorbent in 1 g of inert sand.

previous work with Fe nanoaggregates, the minimal required level of Cu was around 1 wt %. Therefore, the optimal Cu level for the silica platform is approximately 3 wt %. The existence of optimal levels is sensible considering that elevated levels of functionalization will result in pore blocking and increased resistance to mass transfer. The increased capacity of Cu/S materials using a silica support as compared to Fe nanoaggregates can be attributed to the increased number of surface groups for silanization and a much larger surface area for capture. However, some questions remain regarding the effect of Cu species on Hg capture. The silicabased materials were prepared using CuSO4 as the source of copper while Fe nanoaggregates had Cu0 deposited on the surface. As part of this project, materials were also prepared using alternative counter-ions, including hydroxide and nitrate, as well as noncopper doped samples to determine the importance of Cu for Hg separation. The alternative copper materials were found to have significantly decreased capacities for Hg of less than 1,000 µg · g-1 sorbent using similar Cu/S compositions. The noncopper materials could only achieve a maximum capacity of approximately 400 µg Hg · g-1 sorbent using the same S loading. Therefore, the presence of copper as CuSO4 appears to have a beneficial impact on Hg capture. The reason for the enhanced capacity when using sulfate might be based on a thermal transition that occurs for this material at approximately 140 °C after 45 min of heating. The transition is accompanied by a change in color from blue to dark brown. Interestingly, the materials had low Hg sorption capacities unless heated to this transition point. If the Cu and S are forming a copper sulfide complex, the material will have an increased affinity for oxidation of Hg. The role of silanized S4 during this transition is not well understood. The dimensionless breakthrough curves for the materials are shown in Figure 4 where the concentration (C) at a given time (t) has been normalized using the inlet concentration (C0) and the time has been normalized using the equilibrium time (tfinal). The most interesting material shown is Si-1 because it maintained near complete capture of 3 ppm Hg0 for only 10% of its run time and then underwent a steady linear breakthrough. This is not what was expected given that this material exhibited the highest capacity. The other materials were able to maintain complete capture for approximately 50% of the run time, but had sharper breakthroughs resulting in lower capacities. It should be noted that these materials were running at slightly lower inlet concentrations (Figure 4) which could have allowed for a longer period of maximum capture. The variations in inlet concentration (1-3 ppm Hg) were the result of slight pressure build-ups in the system when the packed-bed was brought online, causing suppression of Hg vapor generation in the emitter tube. Of the three, Si-4 had the least severe breakthrough, followed by Si2, and finally Si-3. The difference in breakthrough shape for Si-2 and Si-3 is the result of the increased level of S on Si-2. The fact that the materials with approximately 3 wt % Cu underwent more gradual breakthroughs than those with 5 wt %

Figure 4. Hg vapor breakthrough during fixed-bed contact at 140 °C with various sulfur-functionalized, Cu-impregnated Si samples at 140 °C (Si-1 ) 2.5 wt % Cu, 6 wt % S; Si-2 ) 5 wt % Cu, 3 wt % S; Si-3 ) 5 wt % Cu, 1 wt % S; Si-4 ) 3 wt % Cu, 3 wt % S). The inlet concentration, C0, is defined for each run because it was known to vary over the range 1-3 ppm (10.5-22.6 mg Hg · m-3) for the various runs as a result of Hg vapor suppression in the emitter tube caused by pressure build-up in the presence of the fixed bed.

Figure 5. Examination of the dimensionless adsorption profiles at 140 °C for various functionalized silicas.

Cu is a reflection of differences in available surface area resulting from pore blockage at elevated Cu loading. When developing materials for entrained-flow Hg capture, it is important to consider the dynamic capture, or rate of adsorption for the materials because of the short contact times involved. This is best visualized by examining the plot of capacity (q) as a function of time. The dimensionless plot of this form is shown in Figure 5 where the capacity has been normalized using the equilibrium capacity (q∞) for each material. The values of q/q∞ were obtained by integrating the trace curve data in Figure 4. All four materials exhibit similar favorable adsorption behavior and maintain a linear rate of adsorption for a significant portion of their respective run times. To better compare the materials, the initial linear rate has been calculated for each material based on the actual inlet concentration as


Figure 6. SEM image of Si-1 showing a nonuniform distribution of elements based on the EDX analysis for: (A) 3 wt % Si, 4 wt % Cu, 32 wt % S, 53 wt % Hg; (B) 41 wt % Si, 1.5 wt % Cu, 20 wt % S, 3 wt % Hg; and (C) 54 wt % Si, 1.6 wt % Cu, 16 wt % S, 4 wt % Hg.

determined by the ICP digestion results. These values represent the maximum possible rate of adsorption and are also compiled in Table 2 along with the time during which these rates were sustained. For comparison of platforms, values have also been included for blended-bed experiments using Si-1, Fe/Cu/S nanoaggregates (1.2 wt % Cu and 4 wt % S), and Darco HGLH PAC. For the Si-based materials, the rate of adsorption in pure beds ranges from 0.6 to 1.6 µg Hg · min-1 based on variations in the inlet concentration from 1.2 to 3.0 ppm. The Si-1 sample had the shortest time for maximum capture with the onset of breakthrough occurring after 30 min. The other materials were able to maintain the maximum rate for much longer times ranging from 3510 min for Si-3 to 6978 min for Si-4. At first glance, it is possible that one might conclude that these materials were able to maintain maximum capture longer because of the reduced inlet concentration. However, the variations in breakthrough time are not consistent with the changes in inlet concentration. Instead, it is more likely that differences in material deposition (Cu and S) are the reason. Analysis of Si-1 after Hg capture (Figure 6) using SEM/EDX confirms this explanation. The smallest particles (point “A” in Figure 6) have key-component compositions of 3 wt % Si, 4 wt % Cu, 32 wt % S, and 53 wt % Hg. The 10-µm particle (point “B”) had a much different key-component composition of 41 wt % Si, 1.5 wt % Cu; 20 wt % S, and only 3 wt % Hg. The largest particle (point “C”) is also primarily functionalized Si with a composition of 54 wt % Si, 1.6 wt % Cu, 16 wt % S, and 4 wt % Hg. The EDX results from point A suggest that a portion of the Hg removed accumulates as some form of a mercury-sulfur precipitate because 85% of the particle mass is comprised of these elements. When comparing the blendedbed results of the three different platforms, the Si-based material yields the best results by maintaining a maximum rate of adsorption of 2.4 µg Hg · min-1 for 20 min. The Fe-Cu-S nanoaggregates maintained the same rate for only 7 min. This is understandable when considering the much smaller surface area available for capture when compared to the mesoporous


silica. Interestingly, the Darco HG-LH, which has a similar surface area to the silica, maintained a smaller maximum rate of 1.7 µg Hg · min-1 for only 10.5 min. Leaching Studies. The use of sorbents for Hg capture by injection must be evaluated with regard to their potential environmental impact because they will be removed with the fly ash from coal-fired power plants. This is most important when considering the use of fly ash in concrete as a means of cost recovery for plant operations. If the use of a particular Hg sorbent would prevent the sale of fly ash for cost recovery, the actual cost of Hg removal using this material would be greatly increased.24,25 A benefit of silica-based platforms when compared to PACs is that silica is already used in concrete blends to both reduce cost and improve mechanical properties.26 In order to continue the use of fly ash in concrete, a sorbent must demonstrate the ability to maintain chemical stability when exposed to a leaching environment and not degrade the quality of concrete when used.27 The results for Hg leaching from Si-1 after Hg exposure were at the detectable limits of the machine, only 0.2 ppm Hg. This corresponds to a 0.4% loss of the total Hg captured and indicates a possible strong affinity of Hg for the Cu/S site. The loss of Cu was much greater at 15.2% of the maximum possible, with 8.2 ppm leached for the given conditions. The amount of Cu leached merits some concern because it is approximately six times greater than the action level (AL) of 1.3 mg · L-1 established for Cu by the U.S. EPA. The actual compositions of concrete slugs tested are given in Table 3. For sample 3, the saturated sorbent was blended with fresh sorbent to better approximate power-plant applications where only a fraction of the total capacity is utilized during injection. The results for concrete leaching of both Cu and Hg are also given in Table 3 along with the maximum possible quantity of each. The quantities of both the Cu and Hg were below the detectable limit. For Hg, these results are consistent with leaching of the sorbent after Hg exposure and support the use of these materials in concrete processing. The improved Cu results in the presence of the concrete matrix suggest that concrete impregnation can be beneficial with regard to material disposal because it reduces the amount of leached Cu to well below the AL. However, further testing of cement/sorbent blends needs to be performed to determine if the Cu and Hg leaching is affected by the presence of the air entraining admixture chemicals added to concrete for mechanical stability. Comparison of Hg0 Capture During Fixed-Bed and Entrained-Flow Contact. The results for entrained-flow Hg capture using Si-1 are shown in Figure 7. Results for Darco HG-LH and Fe-Cu-S nanoaggregates are shown for comparison. A steady-state condition is assumed to exist at the endpoint of injection testing. At a modest injection rate of 6 × 10-5 g · L-1 · h-1, the steady-state removal of Hg for Si-1 was 82% (C0 ) 19.6 µg · m-3). For the same injection rate, the removal of Hg for Darco HG-LH and Fe-Cu-S nanoaggregates were 94% (C0 ) 17.0 µg · m-3) and 36% (C0 ) 17.8 µg · m-3), respectively. The larger removal by Si-1 when compared to the Fe-Cu-S nanoaggregates should be anticipated based on the fixed-bed capacities of the two platforms. On the basis of the (24) Bustard, J.; Durham, M.; Starns, T.; Lindsey, C.; Martin, C.; Schlager, R.; Baldrey, K. Fuel Process. Technol. 2004, 85, 549–562. (25) Jones, A. P.; Hoffmann, J. W.; Smith, D. N.; Feeley, T. J. III; Murphy, J. T. EnViron. Sci. Technol. 2007, 41, 1365–1371. (26) Developed by Committee E-701, Materials for Concrete Construction, Charles K. Nmai (Chairman) ACI Education Bulletin E3-01: Cementitious Materials For Concrete; American Concrete Institute: Farmington Hills, MI, 2001. (27) Senior, C.; Bustard, C. J.; Durham, M.; Baldrey, K.; Michaud, D. Fuel Process. Technol. 2004, 85, 601–612.


Meyer et al.

Table 3. Copper and Mercury Loss from Concrete/Fly Ash/Sorbent Mix for Leaching at pH ) 4.2 exhausted Si-1 sample 1 sample 2 sample 3 a

concrete mass (mg)

cement (wt %)

fly ash (wt %)

sorbent (wt %)

max Cu (mg)

11675 9577 9581

0 71.7 65 65

0 28.3 34.5 34.5

100 0 0.5 0.5b

1.1 0 1.1 1.1

leached Cu (mg) 0.2 0 BDLa BDL

max Hg 0.99 0 1.0 0.1

leached Hg (mg) 0.004 0 BDL BDL

BDL ) below detectable limit. b The 50-mg sorbent sample consisted of 5 mg of saturated sorbent blended with 45 mg of fresh sorbent.

Figure 7. Comparison of results for Hg removal from simulated flue gas at 140 °C during entrained-flow using various platforms (Si-1 (∆), C0 ) 2.0 ppb; Fe-Cu-S (0), C0 ) 1.8 ppb; Darco HG-LH (O), C0 ) 1.7 ppb) at an injection rate of 6 × 10-5 g · L-1 · h-1 (0.05 g · h-1).

steady-state Hg removal, Darco HG-LH would appear to have a slightly better performance than Si-1. However, the two materials actually have the same capacity of 268.8 µg Hg · g-1 sorbent, with variations in removal occurring because of differences in the inlet Hg concentration. One should expect Si-1 to actually remove more Hg than Darco HG-LH if the 9-fold increase in fixed-bed capacity when comparing Si-1 to Darco HG-LH is considered. However, injection involves a gasto-particle concentration gradient that is three orders-ofmagnitude less than what was employed during fixed-bed testing. Therefore, it is likely that mass transfer resistance with regard to pore diffusion is the controlling aspect of Hg capture during injection using mesoporous materials. Proof of this is based on the identical performances of the two materials, which possess similar surface area and average pore size. Additional tests were performed using Si-1 to determine the impact of both the mass injection rate and the presence of SO3 on Hg capture. The results are shown in Figure 8. Two trials were made at a mass injection rate of 1.2 × 10-4 g · L-1 · h-1, double the original test rate. The resulting steady-state removal of Hg ranged from 82-100%. The lower end of the range is the same as the results for the original mass injection rate, which is further support that the adsorption process is mass transfer controlled. The diatomaceous earth/sorbent blend used for trial 2 was subsequently used for Hg capture in the presence of 20 ppm SO3 at a mass injection rate of 1.2 × 10-4 g · L-1 · h-1. The SO3 was generated by injecting a dilute sulfuric acid solution into a tube furnace. This level is representative of the typical 1-40 ppm of SO3 that is found in most flue gases.13 It is believed that competitive sorption is the reason that PACs suffer a significant decrease in Hg capture when SO3 is present because the concentration of SO3 is much larger than Hg. The steady-state removal of Hg for Si-1 at this concentration of SO3 was marginally impacted, with a 10% decrease from 100 to 90%. This implies that the Cu/S site is more selective for Hg than SO3, which is beneficial when considering power-plant applications.

Figure 8. Results for Hg removal from simulated flue gas during entrained-flow using Si-1 at various injection rates (6 × 10-5 (0.05) and 1.2 × 10-4 g · L-1 · h-1 (0.1 g · h-1)) and with 20 ppm SO3.

As a means of comparison, the results of fixed-bed and entrained-flow testing are summarized on both a mass-normalized and surface-area-normalized basis in Table 4. The use of surface-area-normalization provides understanding regarding the impact of particle structure on Hg capture. On a massnormalized basis, Si-1 has the largest capacity in both modes of contact. The Fe-Cu-S nanoaggregates, which possess much less surface area per unit mass, had the smallest capacities of the three platforms. Darco HG-LH actually maintained a significant capacity (268.8 µg Hg · g-1) during injection despite having a modest fixed-bed capacity (2942 µg Hg · g-1). When the capacities are reported using surface-area-normalization, the Fe-Cu-S nanoaggregates demonstrate a better utilization of available surface area, especially during entrained-flow contact. While the nanoaggregates have a capacity of 4.4 µg Hg · m-2, the mesoporous Si-1 and Darco HG-LH have much smaller capacities of only 0.7 and 0.9 µg Hg · m-2, respectively. This means that a substantial portion of the internal surface area of the mesoporous materials is not accessible during injection involving short residence times. Of the two mesoporous materials, Darco HG-LH was able to use 9.1% of its total capacity as opposed to only 3.9% for Si-1, indicating a more accessible pore structure for the activated carbon. The lack of rapid pore accessibility when using Si-1 is most likely the result of the presence of a dense S4 network on the particle surface. Further characterization of this material is necessary to verify this. Our previous work involving the study of Fe-Cu-S nanoaggregates led to a conclusion that mass-based comparison might not be a suitable criterion when discussing different platforms.16 This was based on the fact that variations in density will lead to different quantities of particles being injected for the same mass rate. To further analyze this claim, a series of case studies for Hg capture during entrained-flow contact involving the three platforms was performed. The results are summarized in Table 5. The necessary physical data for each material is listed first. A 20-µm particle diameter (dp) was assumed for each platform because this value is near the midpoint of the particle size ranges



Table 4. Summary of Sorbent Performance Data for Fixed and Entrained-Flow Contact fixed-bed capacity

steady state (SS) injection capacity

surface area (m2 · g-1)

µg Hg · g-1

µg Hg · m-2

µg Hg · g-1

µg Hg · m-2

capacity used during injection (%)

391.3 29.6 305.9

19789 2726 2942

50.6 92.1 9.6

268.8 107.5 268.8

0.7 4.4 0.9

1.4% 3.9% 9.1%

Si-1 Fe-Cu-S Darco HG-LH

Table 5. Study of Site Utilization Based on Particle Injection Si-1 dp, µm Fp, g · cm-3 ABET, m2 · g-1 q∞ mg Hg · g-1 sorbent (experimental value)

20 2 391.3 19.8

Darco HG-LH

Fe-Cu-S4

20 0.6 305.9 2.9

20 7.96 29.6 2.7

qEX )

case 1: single particle NP, particles 1 1 θ, sites · m-2 1.46E+17 2.89E+16 qEX, mol Hg · particle-1 3.04E-16 6.03E-17 SS removal, % 7.31E-05 1.45E-05 (dq/dt)AVG, µg Hg · min-1 2.05E-07 4.06E-08

1 2.77E+17 5.78E-16 1.39E-04 3.89E-07

case 2: total external capture 1.37E+06 6.90E+06 NP, particles MINJ, g · L-1 · h-1 2.75E-03 4.16E-03 θ, sites · m-2 1.46E+17 2.89E+16 qEX, µg Hg · g-1 7.28 4.81 SS removal, % 100 100 -1 (dq/dt)AVG, µg Hg · min 2.80E-01 2.80E-01

7.20E+05 5.76E-03 2.77E+17 3.48 100 2.80E-01

case 3: experimental results 2.96E+04 9.88E+04 NP, particles MINJ, g · L-1 · h-1 6.0E-05 6.0E-05 θ, sites · m-2 1.46E+17 2.89E+16 qEX, µg Hg · g-1 7.28 4.81 SS removal, % 82 94 (dq/dt)AVG, µg Hg · min-1 6.07E-03 4.01E-03 261.52 263.94 qPORE, µg · g-1 sorbent utilized APORE, m2 · g-1 5.39 27.44

7.45E+03 6.0E-05 2.77E+17 3.48 72 2.90E-03 104.18 1.13

for all three materials. The particle density (FP g · cm-3), BET surface area (ABET m2 · g-1), and equilibrium capacity (q∞ mg Hg · g-1) are the experimental values discussed previously. The system is the same as what has been given in Figure 2 with a residence time (τ) of 17.9 s and inlet concentration (C0) of 20 µg Hg · m-3. For simplification, it has been assumed that gasphase mass transfer is negligible such that the bulk phase concentration of Hg exists at the particle surface. If modeling the actual transport process, this assumption would not be valid and a mass transfer coefficient would need to be calculated using a suitable correlation for the Sherwood number. However, the following calculations are not intended to describe the actual mass transfer process. Instead, these are used as a means by which to assess the impact of material density on mass-based sorbent injection while providing additional insight regarding pore utilization during Hg adsorption. Case 1 involves the calculation of Hg removal based on the use of only the external surface area of a single particle (Np ) 1). The site density, θ (sites · m-2), was calculated as follows: θ)

NAq∞ 1000mHgABET

× 1017, 2.89 × 1016, and 2.77 × 1017 sites · m-2, respectively. Using this parameter, the external Hg capacity of the particle (qEX (mol Hg · particle-1)) was calculated as

(1)

where NA is Avogadro’s number and mHg is the atomic mass of Hg. The factor of 1000 is needed for unit consistency. Two key assumptions made with this calculation are: (1) the sites for Hg sorption are homogeneously distributed throughout the material and (2) Hg sorption occurs as a monolayer on the surface. This implies that Hg can fully access the pore network of the mesoporous materials during fixed-bed contact. The values of θ for Si-1, Darco Hg-LH, and Fe-Cu-S4 are 1.46

AEXθ NA

(2)

where the external surface area (AEX (m2 · particle-1)) was approximated based on the geometric surface are (πdP2) of a smooth sphere with the same particle diameter. The values of qEX for Si-1, Darco Hg-LH, and Fe-Cu-S4 are 3.04 × 10-16, 6.03 × 10-17, and 5.78 × 10-16 mol Hg · particle-1, respectively. The steady-state removal (SS removal (%)) for each material was then calculated using qEX and the inlet concentration of Hg to the system. The steady state removals for Si-1, Darco Hg-LH, and Fe-Cu-S4 are 7.31 × 10-5, 1.45 × 10-5, and 1.39 × 10-4%, respectively. Finally, the average dynamic capture (µg Hg · min-1) was calculated using the following: 1 × 106qEXNPmHg dq ) dt AVG τ 60

(3)

where the τ is the residence time in seconds, 1 × 106 is a unit conversion for grams to micrograms, and 60 is a unit conversion for seconds to minutes. These values for Si-1, Darco Hg-LH, and Fe-Cu-S4 are 2.05 × 10-7, 4.06 × 10-8, and 3.89 × 10-7 µg Hg · min-1, respectively. The significance of these calculations is that they allow for a direct comparison of Hg removal on a particle basis in the absence of the issues associated with mass-normalization. The primary difference for the materials is the site density, θ. Si-1 and the Fe nanoaggregates have approximately the same value of θ and, therefore, exhibit similar performances. This is most evident when comparing the average dynamic capture rate because the rate for Darco HG-LH is an order-of-magnitude less than the rates for Si-1 and Fe-Cu-S4. For case 2, the results for case 1 were used to solve for the necessary value of Np that would provide 100% removal of Hg. The resulting values for Si-1, Darco Hg-LH, and Fe-Cu-S4 are 1.37 × 106, 6.90 × 106, and 7.20 × 105 particles, respectively. As one should expect based on the performance of a single particle, the largest number of particles is required for the Darco HG-LH while the Fe nanoaggregates needed the minimum number. On a mass-injection basis, this trend does not hold, with Fe nanoaggregates requiring the most mass of 5.76 × 10-3 g · L-1 · h-1 during injection. Darco HG-LH is next with a mass injection of 4.16 × 10-3 g · L-1 · h-1. Si-1 actually required the least mass, 2.75 × 10-3 g · L-1 · h-1, because of its much higher equilibrium capacity. Thus, the original question of mass-based criterion for comparing sorbents has some validity when comparing materials of similar capacity. For all cases, the average dynamic capture rate is 2.8 × 10-1 µg Hg · g-1 sorbent, much less than the maximum rate obtained during fixedbed testing. For case 3, the actual experimental data for a mass injection rate of 6.0 × 10-5 g · L-1 · h-1 were applied to determine the approximate extent of pore usage. For the given experimental parameters, the rate of adsorption for Si-1 and Darco HG-LH


is 2.24 × 10-1 µg Hg · min-1 while the rate for Fe-Cu-S4 is 60% less at 8.97 × 10-2 µg Hg · min-1. However, less than 2% of the total capture occurred on external surface area. On the basis of the experimental Hg removal for Si-1, Darco HG-LH, and Fe-Cu-S4, an additional quantity of Hg (qPORE (µg Hg · g-1 sorbent)) must be captured on the internal surface area (APORE (m2 · g-1)) of the particles. The value of qPORE for Si-1, Darco HG-LH, and Fe-Cu-S4 is 261.52, 263.94, and 104.18 µg Hg · g-1 sorbent, respectively. Equation 2 can be resolved for area as a function of capacity. If the value of qPORE is used, the corresponding area after appropriate unit conversions is APORE. The value of APORE for Si-1, Darco HG-LH, and Fe-Cu-S4 is 5.4, 27.4, and 1.1 m2 · g-1, respectively. The Fe nanoaggregates have the least amount of usage of internal surface area, which should be expected because the nanoaggregates are not a mesoporous material. In fact, previous results found that the capture during injection for nanoaggregates was strongly dependent on the quantity of material injected, unlike Si-1.16 Interestingly, Si-1 utilizes a much smaller portion of pore area when compared to Darco HG-LH based on the total quantity of surface area available determined using BET analysis. However, this result is consistent with the much larger value of θ for Si-1. If the internal area is partially blocked by the S network, optimization of functionalization could improve the performance of Si-1 during injection by increasing the quantity of readily accessible pore area.

Meyer et al.

bed testing. During entrained-flow testing, the steady-state removal of Hg for Si-1 was 82% for injection rates of 6 × 10-5 and 1.2 × 10-4 g · L-1 · h-1. The lack of increase in Hg removal when the injection rate is doubled suggests that pore accessibility is the rate-controlling step during dynamic Hg capture. A calculation of the approximate pore usage based on injection testing helped confirm this observation. During injection testing, the performance of Si-1 was only diminished 10% when exposed to 20 ppm SO3. This is an encouraging result for flue-gas applications where SO3 levels range from 1-40 ppm. On the other hand, PACs are reported to have low Hg0 sorption capacities in the presence of SO3. An additional potential benefit of using Si-based platforms for power-plant injections is the stable nature of the material when exposed to leaching conditions after concrete impregnation. This is an important aspect to consider for injection because the sale of fly ash for concrete blending is a key cost-recovery tool for power-plants. Acknowledgment. This work was made possible with funding provided by the J. M. Huber Corporation. Mercury analysis by ICP for fixed-bed testing was carried out at the University of Kentucky’s Environmental Research and Training Laboratory. Dr. Yongxin Zhao of Arcadis, an on-site contractor to the National Risk Management Research Laboratory at Research Triangle Park, performed the tests in the EPA’s entrained flow reactor.

Nomenclature Conclusions The use of Cu-S sites for Hg capture at 140 °C from the gas phase has been successfully applied to a Silica-based platform using an S4 organic polysulfane and CuSO4. The maximum fixed-bed equilibrium capacity achieved using these materials was 19,789 µg Hg · g-1 sorbent for a material with 2.5 wt % Cu and 6 wt % S. An optimal S level was determined to exist around 3 wt % because enhancement of capacity was only 18% when increasing from this 3 to 6 wt %. This is believed to be the result of pore blocking and increased resistance to mass transfer at elevated levels of functionalization. For the Si-based materials, the rate of adsorption in pure beds ranges from 0.6 to 1.6 µg Hg · min-1 based on variations in the inlet concentration from 1.2 to 3.0 ppm. Differences in breakthrough times suggest that material deposition is not uniform. When compared to two other platforms, commercially available Darco HG-LH and previously tested Fe-Cu-S4 nanoaggregates, the Si-1 material performed the best in fixed-

ABET ) sorbent surface area determined by BET method, m2 · g-1 AEX ) geometric external surface area of a spherical particle, m2 · g-1 APORE ) pore area utilized during Hg capture, m2 · g-1 C ) outlet Hg concentration, µg Hg · m-3 C0 ) inlet Hg concentration, µg Hg · m-3 mHg ) atomic mass of Hg, 200.45 g · mol-1 NA ) avogadro’s Number, 6.02 × 1023 atoms · mol-1 NP ) number of particles, particles q ) time-specific sorbent capacity, µg Hg · g-1 sorbent qEX ) Hg capacity of the external surface of a sorbent, µg Hg · g-1 sorbent qPORE ) utilized Hg capacity of the pores of a sorbent, µg Hg · g-1 sorbent q∞ ) fixed-bed equilibrium Hg capacity of a sorbent, µg Hg · g-1 sorbent t ) exposure time, min tfinal ) equilibrium exposure time, min θ ) Hg site density, sites · m-2 τ ) contactor residence time, s EF8001873

Tetra Sulfide for Mercury Vapor Capture - American Chemical Society

Recommend Documents