Environ. Sci. Technol. 2003, 37, 5763-5766
Simulation and Evaluation of Elemental Mercury Concentration Increase in Flue Gas Across a Wet Scrubber JOHN C. S. CHANG* Air Pollution and Prevention Control Division, E305-03, National Risk Management Research Laboratory, Environmental Protection Agency, Research Triangle Park, North Carolina 27711 S. BEHROOZ GHORISHI ARCADIS G & M, Inc., 4915 Prospectus Drive, Durham, North Carolina 27713
Experimental data from a laboratory-scale wet scrubber simulator confirmed that oxidized mercury, Hg2+, can be reduced by aqueous S(IV) (sulfite and/or bisulfite) species and results in elemental mercury (Hg0) emissions under typical wet FGD scrubber conditions. The S(IV)-induced Hg2+ reduction and Hg0 emission mechanism can be described by a model which assumes that only a fraction of the Hg2+ can be reduced, and the rate-controlling step of the overall process is a first-order reaction involving the Hg‚ S(IV) complexes. Experimental data and model simulations predict that the Hg2+ in the flue gas can cause rapid increase of Hg0 concentration in the flue gas across a FGD scrubber. Forced oxidation can enhance Hg2+ reduction and Hg0 emission by decreasing the S(IV) concentration in the scrubbing liquor. The model predictions also indicate that flue gas Hg0 increase across a wet FGD scrubber can be reduced by decreasing the pH, increasing S(IV) concentration, and lowering the temperature.
spheric cycle of Hg (4). A potentially important step in the atmospheric cycle of Hg is the aqueous-phase (i.e., cloud and rainwater) reduction of Hg2+ by sulfite species (5, 6). The proposed mechanism involves the formation of an unstable intermediate, HgSO3, which decomposes to produce Hg+ which, in turn, is rapidly reduced to Hg0. Munthe et al. (5). proposed a first-order reaction model suggesting that the overall rate of Hg2+ reduction is proportional to the Hg(SO3)22concentration. The kinetic data show that the overall Hg2+ reduction rate is not sensitive to temperature changes from 26 to 40 °C. The reaction rate constant is inversely dependent on the S(IV) concentration (the sum of HSO31- and SO32concentrations) in the range between 0.022 and 0.46 mM. On the other hand, kinetic data generated by Loon et al. (6) indicate that the rate of Hg2+ reduction is proportional to the HgSO3 concentration and increased by more than one orderof-magnitude when the temperature was raised from 6.6 to 35 °C. However, both sets of kinetic data (5, 6) are solely based on UV spectroscopic measurements of Hg(SO3)22- or HgSO3 concentrations (5, 6), and no quantitative Hg0 emission data are reported in the literature. Furthermore, the temperature of scrubbing liquor in wet FGD scrubbers is usually between 50 and 60 °C, which is quite different from the literature conditions. The S(IV) concentrations in the scrubbing liquor can also be as high as several mM, which may quench the reduction process (5). Therefore, experimental data are needed to confirm that the Hg2+ reduction and, more importantly, the subsequent Hg0 emission can occur in wet FGD scrubbers. The objective of this research was to investigate the potential reduction of Hg2+ to Hg0 and subsequent emission of Hg0 under wet FGD scrubber operating conditions. Experiments were performed in a bench-scale, wet scrubber simulator (WSS) containing solutions as the scrubbing liquor. HgCl2-laden gases were continuously introduced into the WSS, and emissions of Hg0 from the simulator were monitored. A first-order reaction model was developed to analyze the experimental data. Another objective of the research was to evaluate the effects of operating conditions on Hg2+ reduction reactions and subsequent Hg0 emission rates and to identify conditions that can reduce Hg0 emissions from wet FGD scrubbers based on the modeling results.
Introduction Because of their high removal efficiency, reasonable cost, and wide applicability, lime/limestone wet scrubbers are widely used by U.S. coal-fired power plants for flue gas desulfurization (FGD) purposes (1). If a high removal efficiency of mercury is also achieved, wet scrubber systems can be used for cost-effective multi-pollutant control. However, field data have shown that, in general, less than 70% mercury removal efficiency is achieved by wet scrubbers. Removal efficiencies lower than 50% are reported for mercury originating from a combination of subbituminous coal and lignite (2, 3) from which elemental mercury (Hg0) is the dominant mercury species in the flue gas. Data analysis (2, 3) indicates that this is mainly due to the inability of wet scrubbers to absorb the insoluble Hg0 and the increase of Hg0 concentration in the flue gas across the scrubbers. It was speculated (2, 3) that some of the absorbed mercury (e.g., ionic form of mercury (Hg2+) such as mercuric chloride (HgCl2)) is transformed to Hg0 and desorbed in the scrubber. This speculation is partially supported by the literature data on chemical redox processes occurring in the atmo* Corresponding author phone: (919) 541-3747; fax: (919) 5412157; e-mail: Chang.John @epa.gov. 10.1021/es034352s CCC: $25.00 Published on Web 11/04/2003
2003 American Chemical Society
Experimental Methodology The schematic of the WSS system is illustrated in Figure 1. This system closely resembles the mercury adsorption and oxidation systems described in previous investigations (7, 8). Detailed descriptions of the simulated flue gas preparation and mercuric chloride (HgCl2) generation oven are included elsewhere (7, 8). In this study, the WSS replaced the fixed-bed of sorbent/oxidant. The WSS was a modified version of impingers used in EPA Method 5 (9). A schematic of the WSS is shown in Figure 2. The impinger was shortened to minimize the headspace; it had a capacity of 300 mL. A coarse frit was attached to the end of the stem to generate fine bubbles and increase mixing of gas and liquid. The WSS was located inside a water bath maintained at the desired temperatures (50.0, 55.5, and 60.0 °C). A Teflon stir-bar was placed inside the WSS and was set in motion by an air-activated stirrer located inside the water bath. All lines, connectors, and valves were made of Teflon and Pyrex and maintained at 95-100 °C (using heating tapes and temperature controllers) to prevent condensation of HgCl2 vapor. The preheater and the HgCl2 generation oven were maintained at 93 °C. Ontario Hydro analysis established the HgCl2 VOL. 37, NO. 24, 2003 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
5763
FIGURE 1. Schematic of the wet scrubber simulator (WSS) system.
FIGURE 3. Sulfite-to-bisulfite ratio effects reflected by model calculations (curves) and measured Hg0 concentrations at 1 mM S(IV) and 55.5 °C.
FIGURE 2. Schematic of the wet scrubber simulator (WSS). oven concentration at 118 ( 1 ppb; more than 99% of generated mercury was in ionic form. At the beginning of each test, 200 mL of the solution to be studied (sodium sulfite and sodium bisulfite at the desired sulfite-to-bisulfite ratio) was placed inside the WSS, which was situated in the hot water bath controlled at the desired temperature. The air-activated stirrer was used to continuously stir the content of the WSS (see Figure 1). After the WSS and its content were heated, 400 mL/min of N2 containing 118 ppb HgCl2 was introduced to the WSS to initiate the test. The WSS outlet Hg0 vapor concentration was measured continuously using a Hg0 UV analyzer (Buck 400A, detection limit of 1 ppb Hg0). Because water vapor creates interferences in the UV Hg0 analyzer, a NAFION gas sample dryer (Perma Pure, Inc.) was used to remove water vapor from the gas sample before it entered the Buck analyzer. Three sets of tests were performed. In the first set, the total S(IV) concentration (made up with sodium sulfite and sodium bisulfite) was kept at a constant value of 1 mM; the 5764
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 37, NO. 24, 2003
ratio of sulfite-to-bisulfite was varied in the range of 0 to 0.62 with temperature maintained at 55.5 °C. In the second set of tests, the sulfite-to-bisulfite ratio was kept constant at a value of 0.25, and the total S(IV) concentration was changed from 1 to 4 mM with temperature maintained at 55.5 °C. In the third set of tests, both the sulfite-to-bisulfite ratio and the total S(IV) were kept constant at 0.25 and 1 mM, respectively; the scrubber solution temperature was varied at three levels of 50.0, 55.5, and 60.0 °C, by adjusting the water bath temperature controller.
Results Hg0 Emission Data. Figures 3, 4, and 5 show the Hg0 emission curves plotted as measured Hg0 concentrations in the WSS effluent vs time for the 3 sets of experiments performed. Usually no penetration of Hg2+ through the WSS was detected by running the WSS effluent through the reduction furnace (7, 8), and the Hg0 concentration in the WSS effluent was above detection limit within a few minutes after the start of each test. The Hg0 concentration continued to increase as the experiment proceeded, but the rate of increase gradually slowed, and the Hg0 concentration seemed to approach a plateau after an hour. The experimental parameter in Figure 3 is the initial sulfite-to-bisulfite ratio. A ratio of 0 means only NaHSO3 was
Assuming the gas- and liquid-phase contents in the WSS were each well-mixed, the following material balance equation can be written to represent the mercury absorption and reduction in the WSS:
dC ) rFCin - kCV dt with initial conditions: C ) 0 when t ) 0 (2)
V
FIGURE 4. S(IV) concentration effects reflected by model calculations (curves) and measured Hg0 concentrations at sulfite-to-bisulfite ratio of 0.25 and 55.5 °C.
where V (L) is the volume of the scrubbing liquor in the WSS, C (gmol/L) is the concentration of Hg‚S(IV) complexes in the scrubbing liquor, F (gmol/min) is the nitrogen gas flow rate through the WSS, Cin (ppb) is the HgCl2 concentration in the influent nitrogen gas to the WSS, and k (min-1) is the firstorder reaction rate constant. The solution of eq 2 is:
C)
rFCin (1 - e-kt) Vk
(3)
On the basis of assumption (5) of the model mentioned above,
FCout ) kCV
(4)
where Cout (ppb) is the Hg0 concentration in the WSS effluent nitrogen gas. Substituting eq 3 into eq 4, the Hg0 concentration in WSS effluent nitrogen gas can be expressed as
Cout ) rCin(1 - e-kt) FIGURE 5. Temperature (°C) effects reflected by model calculations (curves) and measured Hg0 Concentrations at 1 mM S(IV) and sulfiteto-bisulfite ratio of 0.25. used to make up the scrubbing liquor. When a higher sulfiteto-bisulfite ratio was used, the Hg0 concentration also increased faster, which reflects the correspondingly higher Hg0 emission rate. The parameter in Figure 4 is S(IV) concentration. It is shown that the higher the S(IV) concentration, the lower the Hg0 emission rate. Figure 5 shows that when the temperature was raised, the Hg2+ reduction/ Hg0 emission rate also increased. Note that the HgCl2 concentration at the simulator inlet was maintained at a level equivalent to 118 ppb of Hg0 for all the tests. However, the Hg0 concentration in the simulator effluent never reached the 118 ppb in any of the test. Hg2+ Reduction and Hg0 Emission Model. It was found that the Hg2+ reduction and Hg0 emission in the WSS can be represented by a model as shown by eq 1 suggested by the literature, however at lower temperatures.
The model was established on the basis of the following assumptions: (1) due to its high solubility, all the HgCl2 fed into the WSS was absorbed by the scrubbing liquor; (2) in the presence of S(IV), the absorbed HgCl2 reacted with S(IV) to instantly form a number of complexes and byproducts, as evidenced by the literature data (5, 6); (3) only a fraction, r, of the HgCl2 fed into the WSS formed Hg‚S(IV) complexes, which underwent a series of chain reactions to generate Hg0; (4) the rate-controlling step of the chain reactions can be represented by a first-order reaction with respect to Hg‚S(IV) complexes and a global rate constant, k; and (5) because of its insoluble nature, all the Hg0 formed was immediately stripped off and exited the WSS with the effluent gas.
(5)
There are only two adjustable parameters, r and k, in eq 5. Using a nonlinear regression curve routine, implemented on a microcomputer, values of r and k can be obtained by fitting eq 5 to the Hg0 concentration curves as shown in Figures 3, 4, and 5. The estimated values of r and k are listed in Tables 1, 2, and 3 corresponding to Figures 3, 4, and 5, respectively. The standard deviations shown in Tables 1, 2, and 3 reflect the good correlation between model predictions and experimental data as illustrated in Figures 3, 4, and 5.
Wet Scrubber Implications The model can be used to predict Hg0 increase across wet FGD scrubbers when the Hg2+ reduction and Hg0 emission in the scrubber is dominated by the mechanism represented by eq 1. For example, by substituting the values of r and k in the fourth column of Table 1 into eq 5, the model predicts that an influent flue gas containing 1 ppb (8 µg/m3) of Hg2+ can cause rapid increases of Hg0 in the effluent flue gas when the FGD scrubber is maintain at 55.5 °C and 1 mM S(IV) with a sulfite-to-bisulfite ratio of 0.25. An increase of 0.25 ppb (2 µg/m3) Hg0 across the scrubber can occur within 21 min, and the maximum increase can be as high as 0.5 ppb (4 µg/m3) Hg0. Those predictions probably represent the worstcase scenarios because the real-world wet FGD scrubbers may deviate from the model assumptions (e.g., the Hg2+ absorption may not be 100%, the hold tank may not be wellmixed, and the absorbed mercury may be lost with the scrubber bleed or solids coprecipitation). When the scrubber influent flue gas Hg2+ concentration is relatively stable, the model parameter, r, reflects the steadystate flue gas Hg0 increase across the scrubber, and k indicates how fast the flue gas Hg0 increase reaches steady-state as shown in eq 5. The r and k values shown in Tables 1, 2, and 3 indicate that both the increase rate and the eventual steadystate flue gas Hg0 level can be reduced by decreasing the scrubbing liquor sulfite-to-bisulfite ratio, increasing the total S(IV) concentration, and lowering the temperature. Decreasing the sulfite-to-bisulfite ratio can be achieved by reducing VOL. 37, NO. 24, 2003 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
5765
TABLE 1. Estimated Valuesa of r and k of Eq 5 for Sulfite-to-Bisulfite Ratio from 0 to 0.62 and 1 mM Total S(IV) at 55.5 °C sulfite-to-bisulfite ratio
a
parameter
0
0.1
0.25
0.62
r k, min-1
0.3466 ( 0.6% 0.0223 ( 1.8%
0.3903 ( 0.5% 0.0256 ( 1.6%
0.5184 ( 1.0% 0.0333 ( 3.0%
0.7801 ( 0.9% 0.0361 ( 2.5%
Expressed as mean ( percent standard deviation.
TABLE 2. Estimated Valuesa of r and k of Eq 5 for Total S(IV) Concentration Ranging from 1 to 4 mM and 0.25 Sulfite-to-Bisulfite Ratio at 55.5 °C Total S(IV), mM
a
parameter
1
2
3
4
r k, min-1
0.5184 ( 1.0% 0.0333 ( 3.0%
0.1710 ( 0.8% 0.0234 ( 2.1%
0.0962 ( 1.5% 0.0231 ( 3.9%
0.0625 ( 2.7% 0.0116 ( 4.3%
Expressed as mean ( percent standard deviation.
TABLE 3. Estimated Valuesa of r and k of Eq 5 for Temperature Ranging from 50 to 60 °C with 1 mM S(IV) and 0.25 Sulfite-to-Bisulfite Ratio Temperature, °C parameter
50
55.5
60
r k, min-1
0.5132 ( 1.2% 0.0043 ( 1.9%
0.5184 ( 1.0% 0.0333 ( 3.0%
0.5806 ( 0.6% 0.0402 ( 2.0%
a
therefore, the Hg0 emission can theoretically be reduced by means (e.g., additives) that can inhibit the formation of Hg‚S(IV) complexes (10). Additional WSS tests are going to be conducted to confirm this assumption. Pilot- and full-scale tests are needed to validate the results from this current work which was conducted under simulated bench-scale conditions. Care should also be exercised in extrapolating or generalizing the results of this study to complicated real life applications.
Expressed as mean ( percent standard deviation.
Acknowledgments the scrubbing liquor pH. Caution should be exercised because lower pH may decrease the SO2 removal efficiency and encourage corrosion of scrubbers. Increasing the total S(IV) concentration can be implemented by reducing oxidation or adding a small amount of sodium salts (e.g., NaHSO3 or NaHCO3), which should not have any negative impact on the scrubber performance. Lowering temperature can be accomplished by installing energy recovery systems. However, Table 3 indicates that the rate constant k does not have an exponential dependence on temperature ,and the longterm, steady-state flue gas Hg0 increase across a scrubber should not be significantly reduced by lowering the temperature from 60 to 50 °C because the r value decreased by only 11.6%. The nonexponential relationship between k and temperature may be due to the interactions of complex reactions involved or the interference of mass transfer resistances. Figure 3 and Table 2 show that lowering S(IV) concentration enhances Hg2+ reduction/Hg0 emission; the model implies that the flue gas Hg0 increase would be worse in forced oxidation than in natural oxidation FGD scrubbers under similar design and operating conditions. This is due to the fact that the scrubbing liquor S(IV) concentration in forced oxidation FGD scrubbers is usually considerably lower than that in natural oxidation systems as most S(IV) in the forced oxidation scrubbing liquor is oxidized into S(VI), e.g., sulfate species. The model assumes that the Hg0 emission rate is proportional to the concentration of Hg‚S(IV) complexes;
5766
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 37, NO. 24, 2003
We acknowledge the work by Jarek Karwowski of ARCADIS G & M, Inc. for collecting the experimental data.
Literature Cited (1) Srivastava, R. K. EPA-600/R-00-93 (NTSB PB2001-101224); U.S. Environmental Protection Agency, 2000. (2) Brown, T. D.; Smith, D. N.; Hargis, R. A., Jr.; O’Dowd, W. J. J. Air Waste Manage. Assoc. 1999, 49 (June), 1-97. (3) Kilgroe, J. D.; Sedman, C. B.; Srivastava, R. K.; Ryan, J. V.; Lee, C. W.; Thorneloe, S. A. EPA-600/R-01-109 (NTSB PB2002105701);, U.S. Environmental Protection Agency, 2001. (4) Brosset, C. Water, Air, Soil Pollut. 1987, 34, 145-166. (5) Munthe, J.; Xiao, Z. F.; Lindqvist, O. Water, Air, Soil Pollu. 1991, 56, 621-630. (6) Loon, L. V.; Mader, E.; Scott, S. L. J. Phys. Chem. 2000, 104, 1621-1626. (7) Ghorishi, S. B.; Gullett, B. K. Waste Manage. Res. 1998, 16, 582593. (8) Ghorishi, S. B. EPA-600/R-98-014 (NTIS PB98-127095); U.S. Environmental Protection Agency, 1998. (9) U.S. EPA Method 5. 40 CFR Part 60, Appendix A. U.S. Environmental Protection Agency, U.S. Government Printing Office: Washington, DC, 1994. (10) Amrhein, G. T.; Kudlac, G. A.; Madden Yurchison, D. Presented at the 27th International Technical Conference on Coal Utilization & Fuel Systems, Clearwater, FL, March, 2002.
Received for review April 15, 2003. Revised manuscript received September 10, 2003. Accepted September 21, 2003. ES034352S
