November 14, White Paper. Geochemical Evolution of Fly Ash Leachate ph

Transcription

1 November 14, 2010 White Paper Geochemical Evolution of Fly Ash Leachate ph William R. Roy, Ph.D. Geochemist, Urbana, IL Introduction Solution ph is the master variable that controls the extent of leaching and solubility of many potential groundwater contaminants that can leach from coal combustion products (CCPs). Therefore, understanding the chemical reactions that control ph and their associated time-frames that occur when CCPs come in contact with water is essential. When CCPs are mixed with water, the initial ph of the mixture can range from about 4 to greater than 12. Spears and Lee (2004) generalized that about 70% of all fly ashes produce a neutral or alkaline leachate. Coal fly ash derived from U.S. bituminous coal can yield an alkaline or acidic reaction when initially mixed with water. Alkaline reactions, however, appear to be more common (cf. EPRI, 1987; Roy et al., 1984; Mattigod et al., 1990). Fly ash derived from Western U.S. lignite and subbituminous coal is typically alkaline (Groenewold et al., 1985). Talbot et al. (1978) and later Roy and Griffin (1984) generalized that alkaline leachate from fly ash was the result of the hydrolysis of metal oxides that form during the combustion of coal, viz., MO + H 2 O M OH - where M = Ca and Mg, (1) J 2 O + H 2 O 2J + + 2OH - where J = Na, K and others. (2) Acidic fly ash has received less attention. Roy and Griffin (1984) proposed that the acidity of two fly ash samples (ph 4.1) generated by the combustion of bituminous coal in Illinois was the result of sulfuric acid in the samples. Pyrite could have been the source of acid if the samples were in an oxidizing environment. Spears and Lee (2004) generalized that acidic fly ash leachate can be attributed to the sorption of SO 2 or the 1

2 condensation of sulfuric acid from the gas stream onto particle surfaces. Another possible source of acidity could stem from the hydrolysis of Al 3+ derived from aluminosilicates or aluminum sulfates. Fishman et al. (1999) proposed that leachate acidity could also be derived from the hydrolysis of Al 3+ derived from an aluminum-potassium sulfate phase. Also, acidic fly ash is more likely to occur when the ash is exposed for longer periods to exhaust gases. If calcium is used as a surrogate measurement of the amount of CaO, a strong alkaline base, and total sulfur is used as an indicator of sulfuric acid, EPRI (1987) generalized that the ratio of Ca to S can be correlated to the initial ph of water extracts of fly ash. When Ca/S ratios were less than about 2.5, the resulting extracts were acidic (Fig. 1), whereas Ca/S ratios greater than 2.5 were associated with alkaline extracts. It has been observed that the ph of ash-water systems can change with time, which can change the chemical composition of liquid phase. Talbot et al. (1978) reported that the ph values of alkaline fly ash samples derived from subbituminous coal in Montana initially ranged from ph 10.5 to 11.5 when mixed with distilled water. After about 5 days, the ph of the extracts decreased to about 8.7, and remained constant for about 6 months when the study ended. The extracts were open to the atmosphere. Talbot et al. proposed that the decrease in ph was the result of the diffusion of atmospheric CO 2, viz., - CO 2 + H 2 O = HCO 3 + H + (3) When CO 2 was excluded from a reaction vessel and replaced with nitrogen, the ph of the extract decreased from about 11 to 10 after two months of equilibration. Talbot et al. speculated that the decrease in ph in the CO 2 -free system was somehow related to the dissolution of aluminosilicates. The ph values of the extracts of two initially acidic fly ash samples of Roy and Griffin (1984) increased from 4.1 to about ph 7 after 21 to 36 days. Roy and Griffin attributed this ph evolution as the neutralization of a finite amount of acidity followed by the dissolution of matrix iron oxides and aluminosilicates. 2

3 Figure 1. Relationship between the ratio Ca/S and extract ph (EPRI, 1987). Schramke (1992) proposed that the introduction of CO 2 gas into alkaline fly ash leachate should reduce ph because when CO 2 is absorbed, calcium carbonate (calcite) can precipitate, resulting in acidity: Ca 2+ + CO 2(gas) + H 2 O CaCO 3(s) + 2H + (4) Using geochemical modeling, Schramke reported that in the absence of CO 2 gas, the equilibrium ph of distilled water in contact with portlandite (Ca(OH) 2 ) was The ph of distilled water in contact with calcite was 8.2. To test this hypothesis, Schramke (1992) mixed 50 g of four fly ash samples with 1 L of deionized water for about 300 hours in a Teflon reaction vessel at 25 C. Carbon dioxide was introduced to each reaction vessel at a gas flow rate of 200 ml/min and at a partial pressure (log fco 2 ) of As predicted, the initially high ph of each solution decreased (Fig. 2). Schramke reported that calcite was detected by X-ray diffraction in each fly ash sample after the samples were removed from the reaction vessel, which accounted for the decrease in ph. Calcite was not detected in any of the fly ash samples prior to the study. 3

4 These studies suggest that although the initial ph of fly ash leachate may be relatively acidic or strongly basic, there may be geochemical buffers that narrow this wide ph range to about ph 7 to 9. Figure 2. Change in ph of a function of time for four fly ash extracts exposed to CO 2 gas (Schramke, 1992). The objective of this study was to apply a kinetic-geochemical model to ash-water systems to gain a better understanding of geochemical buffers that influence the longterm ph of ash leachate. Methods To accomplish the goals of this study, the kinetic model REACT was used. REACT is a program that is part of the software Geochemist s Workbench (Bethke, 1996), Release The thermodynamic database thermo.com.v8.r6+ was used. For this presentation, three fly ash samples were chosen for study. The alkaline fly ash sample 159 that was investigated by Schramke (1992) was selected for this study to initially help validate the application of REACT. Then an acidic fly ash sample (I7) and an alkaline fly ash sample (W2), both described in Suloway et al. (1983), Roy and Griffin (1984), and Roy et al. (1984) were modeled. These specific ash samples were selected because of the availability of time-dependent leaching data, and detailed mineralogical and chemical characterization. In each case, 1 kg of a model rain sample (Table 1) was mixed using REACT with each model ash in proportions that matched the published studies. The model rain was designed to contain trace quantities of each solute that are required for 4

5 REACT to operate. As a first approximation, reaction rate constants were obtained from Palandri and Kharaka (2004). Surface area was estimated as 3/rρ where r is the radius of a particle, and ρ is the density of the solid phase. In many cases, however, surface areas were treated as variables, and revised to better match the experimental results. Table 1. Composition of the model rain water. Constituent Concentration ph 5.5 Eh 0.57 Volts Temperature 25º C Al mg/l Ca mg/l Cl mg/l Fe mg/l K mg/l - HCO 3 swapped with CO 2 HS mg/l Na mg/l Mg mg/l SiO mg/l 2- SO mg/l Results Ash Model 159 Based on the mineralogical data provided, a model of ash 159 was constructed (Table A1). For each model fly ash, the mineralogical composition was approximated by known reactants (those detected by x-ray diffraction), and postulated reactants which were inferred either by elemental data or by the chemical composition of the experimental leachate. In some cases, the postulated reactants were proposed to account for the chemical composition of the experimental leachate. The last category is anticipated products. The purpose of this third group is to provide some constraints to what products form, and at what rate they form. The results of REACT adding the model rain water to ash 159 indicated that the ph of the model rain increased to ph 10.5 after about 3 hours of contact with the ash (Fig. 3). Most of the lime (CaO) had reacted by this time, yielding 5

6 CaO + H 2 O Ca OH - (5) ph Time (hours) Figure 3. Change in leachate ph for ash 159. The experimental data (diamonds) are compared with the ph-time relationship predicted by REACT (solid line). REACT then predicted that the ph would decrease with time. While the model was unable to match the experimental data exactly, it did predict the overall trend. For this case, CO 2 gas was added to the leachate as a simple reactant to mimic the diffusion of - atmospheric CO 2 into solution, which generated an increase in HCO 3 (Fig. 4) and acidity, viz., - CO 2 + H 2 O HCO 3 + H + (6) REACT was also able to reproduce the concentrations of Ca in solution after the initial part of the extraction interval (Fig. 5). The model was, however, unable to reproduce the experimental data during the initial contact with water. It is possible that some portion of the Ca in solution was derived from a solid phase not included in Table A-1. The relatively slow decrease in Ca in solution resulted from the precipitation of calcite (Fig. 6) as Ca HCO 3 CaCO 3 + H + (7) 6

7 Figure 4. Concentration of bicarbonate ions in the model 159 leachate as predicted by REACT. Ca (mg/l) Time (hours) Figure 5. Concentration of Ca in ash 159 leachate as a function of time. The experimental data are compared with the results predicted by REACT. 7

8 Figure 6. Extent of calcite precipitation from ash 159 leachate as predicted by REACT. REACT was also able to match the experimental sulfate concentrations (Fig. 7). The anhydrite that was postulated to be present in the ash sample dissolved relatively quickly. Because the solubility of gypsum was not exceeded, the amount of sulfate in solution remained constant. 350 Sulfate (mg/l) Time (hours) Figure 7. Concentration of sulfate in ash 159 leachate as a function of time. The experimental data are compared with the results predicted by REACT. 8

9 Ash Model I7 As discussed in Roy et al. (1984), I7 was a fly ash sample derived from the combustion of Illinois Basin coal. A 20% slurry was periodically stirred in a 19-L Pyrex reaction vessel for 106 days. Based on the chemical and mineralogical characterization data given in Suloway et al. (1983), an ash model was constructed to mimic I7 (Table A2) to yield an acidic reaction after its initial contact with water, then a neutral reaction later in time. REACT was able to match the experimental data fairly well (Fig. 8). The initial decrease in ph was the result of adding both pyrite and sulfuric acid as postulated reactants. The decrease in ph was simulated by the presence of pyrite alone (not shown), but the inclusion of both sources of acidity yielded a closer match to the experimental data fco 2 = ph 7 fco 2 = Experimental data Time (days) Figure 8. Leachate ph as a function of time for ash I7. The experimental data are compared with the results predicted by REACT at different partial pressures of carbon dioxide. 9

10 The dissolution of lime and periclase (MgO) (Fig. 9) then increased leachate ph. The dissolution of both minerals was not complete at the end of the extraction interval, and would continue to increase ph. However, the ph of the leachate was then controlled by the formation of calcite via eq. (7). Model I7 was treated as an open system, and the experimental data best matched the results predicted by REACT when the fugacity of CO 2 was set as log fco 2 = When the fugacity of CO 2 was decreased, the ph of leachate became more basic (Fig. 8). Figure 9. Dissolution of periclase (MgO) and lime (CaO) and the precipitation of calcite as a function of time in model I7 as calculated by REACT. Because pyrite was used as a reactant in the ash model, the oxidation-reduction potential (Eh) will influence the stability of pyrite, and the subsequent production of acidity. The experimental data best matched the results predicted by REACT when Eh was set at 520 mv (Fig. 10). For example, when Eh was 850 mv, the ph of the leachate was more acidic because of the greater instability of pyrite under relatively oxidizing conditions. When Eh was 200 mv, a greater amount of pyrite was stable because of the relatively reduced state of the ash-leachate system, and therefore the initial and transient dip in ph was not well duplicated by REACT. 10

11 9 Eh = 200 mv ph mv Experimental data 5 Eh = 850 mv Time (days) Figure 10. Leachate ph as a function of time for ash I7. The experimental data are compared with the results predicted by REACT at different oxidation-reduction potentials (Eh). REACT was unable to match the experimental sulfate concentrations for ash I7 during the initial reaction, but it was able to match the long-term data fairly well (Fig. 11). During the first week of mixing in the experiment, there was more sulfate in solution than that calculated from anhydrite, sulfuric acid, and pyrite by REACT. At the end of the 106-day simulation, REACT calculated the ph of the liquid phase as 7.6, which agreed with the measured value of 7.5. At the end of the simulation, REACT indicated the solution was in equilibrium with calcite and pyrite, and near equilibrium with gypsum. 11

12 5,000 4,000 SO4 (mg/l) 3,000 2,000 1, Time (days) Figure 11. Concentration of sulfate in ash I7 leachate as a function of time. The experimental data are compared with the results predicted by REACT. It may be difficult to equilibrate CCP-leachate systems for long time periods in the laboratory for years or decades. Kinetic models, however, can be used to gain a first approximation of the long-term chemical behavior of ash-water systems such as I7. When the model I7 was run for a 100-year period, the ph of the liquid phase was calculated by REACT as 7.7 (Fig. 12). Under the conditions imposed, REACT indicated that the ph of the liquid phase would remain between ph 7 and 8. On a scale of 100 years, the initial acidity of I7 was insignificant. Figure 12. Long-term ph of the I7 extract as calculated by REACT. 12

13 Ash Model W2 As discussed in Suloway et al. (1983) and Roy et al. (1984), W2 was an alkaline fly ash sample derived from the combustion of lignite coal from North Dakota. A 20% slurry (weight:volume) was periodically stirred in a 19-L Pyrex reaction vessel for 140 days. While the reaction vessel was not stirred, it was kept closed to the atmosphere. Based on the chemical and mineralogical characterization data given in Suloway et al. (1983), an ash model was constructed to mimic W2 (Table A3) to yield a strongly alkaline reaction. It was estimated that anhydrite, lime, and periclase were the major phases known to be present in W2 that will influence the ph of the liquid phase. When REACT mixed the model rain with the model ash, it was able to reproduce the experimental ph values fairly well (Fig. 13). Unlike the first two ash samples, the ph of W2 remained constant during the 140-day extraction interval. As indicated earlier, the laboratory reaction vessel was closed other than for sampling and mixing. REACT was able to best match the experimental data by treating the model ash as a closed system: the log of the partial pressure was arbitrarily set as -13. However, if more carbon dioxide was allowed to react, the ph of the leachate was predicted to decrease to a range between ph 8 and 9 (Fig. 13). 14 ph fco 2 = -6 fco 2 = -5 fco 2 = -4 Experimental data Time (days) 150 Figure 13. Leachate ph as a function of time for ash W2. The experimental data are compared with the results predicted by REACT at different partial pressures of carbon dioxide. 13

14 At the end of the 140-day simulation, REACT calculated that the ph of the liquid phase was 12.6, which agreed well with the measured value of At the end of the simulation period, the ph of the extract was buffered by the solubility of portlandite (Ca(OH) 2 ) and it was in equilibrium with gypsum. However, this closed system is not representative of conditions that would occur in a field setting, where CO 2 is available to react with the ash. When the partial pressure of CO 2 was increased, the ph decreased and the extract was in equilibrium with calcite, gypsum, and magnesite (MgCO 3 ). When model W2 was simulated for 100 years, the ph of the extract was dependent on the partial pressure of CO 2 (Fig. 14). In the absence of CO 2, the ph of extract remained 12.6, whereas when the fugacity of CO 2 increased, ph decreased. REACT predicted that when the log of the CO 2 fugacity is in the range of -5 to -3, a ph of about 8 will develop with time. The log of atmospheric CO 2 fugacity is about -3.5 (Bethke, 1996). Figure 14. Relationship between the ph of extracts W2 and I7, and the fugacity of carbon dioxide after 100 years as calculated by REACT. 14

15 Conclusions The kinetic model REACT was able to match experimental ph data that were derived from the laboratory extractions of three fly ash samples. REACT mixed a dilute model rain water with three groups of mineral phases that were chosen to approximate the composition and reactivity of the fly ash samples. Although the initial reaction of coal ash with water may yield a relatively acidic or strongly alkaline solution, there are geochemical buffers that narrow this ph range with time. Laboratory studies and the results of the geochemical kinetic modeling presented in the current study demonstrate that carbon dioxide can buffer the ph of CCP leachate. The ph of acidic fly ash may be short-lived because the acidity is neutralized with time by the hydration of metal oxides, and the reaction path is ultimately buffered by carbon dioxide yielding a ph of 7 to 8. It appears that when strongly alkaline CCP leachate is open to atmospheric carbon dioxide, the ph of the liquid phase will decrease with time to a ph between 8 and 9. References Bethke, C. M Geochemical Reaction Modeling. Oxford University Press, Inc., NY, 397 p. EPRI Chemical characterization of fossil fuel combustion wastes. Electric Power Research Institute final report EPRI EA Fishman, N. S., C. A. Rice, G. N. Breit, R. D. Johnson Sulfur-bearing coatings on fly ash from a coal-fired power plant: composition, origin, and influence on ash alteration. Fuel, 78, Groenewold, G. H., D. J. Hassett, R. D. Koor and O. Manz Disposal of western fly ash in the northern Great Plains. Materials Research Society, Symposium Proceedings, Fall 1984, Symposium M,

16 Mattigod, S. V., D. Rai, L. E. Early, and C. C. Ainsworth Geochemical factors controlling the mobilization of inorganic constituents from fossil fuel combustion residues: I. Review of the major elements. Journal of Environmental Quality, 19, Palandri, J. L. and Y. K. Kharaka A compilation of rate parameters of watermineral interaction kinetics for application to geochemical modeling. U.S. Geological Survey Open File Report , 64 p. Roy, W. R. and R. A. Griffin Illinois Basin coal fly ashes: 2. Equilibria relationships and qualitative modeling of ash-water reactions. Environmental Science and Technology, 18, Roy, W. R., R. A. Griffin, D. R. Dickerson, and R. M. Schuller Illinois Basin coal fly ashes. 1. Chemical characterization and solubility. Environmental Science and Technology, 18, Spears, D. A. and S. Lee The geochemistry of leachates from coal ash. Geological Society, London, Special Publications, 236, Schramke, J. A Neutralization of alkaline coal fly ash leachate by CO2(g). Applied Geochemistry, 7, Suloway, J. J., W. R. Roy, T. M. Skelly, D. R. Dickerson, R. M. Schuller, and R. A. Griffin Chemical and toxicological properties of coal fly ash. Environmental Geology Notes 105, Illinois State Geological Survey, Champaign, IL 61820, 70 p. Talbot, R. W., M. A. Anderson, and A. W. Andren Qualitative model of heterogeneous equilibria in a fly ash pond. Environmental Science and Technology, 12,

17 Appendix Table A1. Ash Model 159. Known reactants Mass (g) Surface area (cm 2 /g) Rate constant (mol/cm 2 -sec) Nucleus density (cm 2 /cm 2 ) Ferrite-Mg , x ,000 Hematite , x Sillimanite (1) , x ,000 Quartz , x ,000 SiO , x 10-4 (2) 1,000 (amorphous) Postulated reactants Anhydrite 0.45 Added as a simple reactant Lime , Magnetite , x ,000 CO 2 (gas) 0.5 mmole Added as a simple reactant Anticipated products Calcite , x ,000 Gypsum x ,000 Illite , x ,000 Kaolinite , x ,000 1 Mullite is a common aluminosilicate in fly ash and was detected in this sample. However, mullite is not present in the database used by REACT. Therefore sillimanite, a high-temperature Al 2 SiO 5 phase, was used as a surrogate for mullite. 2 Coupled with an activation energy of 74.5 kjoule/mole. 17

18 Table A2. Ash Model I7. Known reactants Mass (g) Surface area (cm 2 /g) Rate constant (mol/cm 2 -sec) Nucleus density (cm 2 /cm 2 ) Anhydrite x ,000 Hematite , x Lime x Magnetite , x ,000 Quartz , x ,000 SiO , x 10-4 (1) 1,000 (amorphous) Postulated reactants Periclase x Pyrite , x H 2 SO Added as a simple reactant cutoff = 0.04 Al 2 (SO 4 ) 3 30 Added as a simple reactant cutoff = cutoff = 4.0 Thenardite Added as a simple reactant (Na 2 SO 4 ) Anticipated products Calcite , x ,000 Gypsum x ,000 Illite , x ,000 Kaolinite , x ,000 1 Coupled with an activation energy of 74.5 kjoule/mole. 18

19 Table A3. Ash Model W2. Known reactants Mass (g) Surface area (cm 2 /g) Rate constant (mol/cm 2 -sec) Nucleus density (cm 2 /cm 2 ) Periclase , x Quartz , x ,000 SiO 2 (amorphous) , x 10-4 (1) 1,000 Anhydrite 75.0 Added as a simple reactant cutoff = 13 Lime 100 Added as a simple reactant cutoff = cutoff = 1.0 Thenardite Added as a simple reactant (Na 2 SO 4 ) Anticipated products Calcite 1 x x ,000 Gypsum 1 x , x ,000 Illite , x ,000 Kaolinite , x ,000 1 Coupled with an activation energy of 74.5 kjoule/mole. 19