Environ. Sci. Technol. 2001, 35, 1406-1410
Fate of Selenium in Coal Combustion: Volatilization and Speciation in the Flue Gas RONG YAN, DANIEL GAUTHIER,* GILLES FLAMANT, AND GILLES PERAUDEAU Institut de Science et de Ge´nie des Mate´riaux et Proce´de´s, CNRS-IMP, B.P.5 Odeillo, 66125 Font-Romeu, France JIDONG LU AND CHUGUANG ZHENG National Key Laboratory on Coal Combustion, Huazhong University of Science & Technology, Wuhan 430074, China
In light of Title I of the Clean Air Act Amendments of 1990, selenium will most probably be considered for regulation in the electric power industry. This has generated interest for removing this element from fossil-fired flue gas. This study deals with coal combustion: selenium volatilization and its speciation in the cooled flue gas were investigated to better understand its chemical behavior to validate the thermodynamic approach to such complex systems and to begin developing emission control strategies. Se volatility is influenced by several factors such as temperature, residence time, fuel type, particle size, and Se speciation of the fuels, as well as the forms of the Se in the spiked coal/coke. Spiked coke and coal samples were burned in a thermobalance, and atomic Se and its dioxide were identified in the cooled combustion flue gas by X-ray photoelectron spectroscopy (XPS). A thermodynamic calculation was applied to a complex system including 54 elements and 3200 species that describes the coal combustion. Several theoretical predictions concerning Se behavior, such as its speciation in flue gas, agreed well with experiments, which supports using thermodynamics for predicting trace element chemistry in combustion systems.
1. Introduction Selenium (Se) is one of the most volatile trace elements in coal, and it is largely released in vapor phase to the atmosphere. Its partition between the vapor and the condensed forms depends on such factors as initial concentration and combination in coal, design and operating conditions of the combustor, air pollution control facility, etc. Up to 59% of the in-stack Se was found as vapor Se and SeO2 (1, 2), and its concentration in the vapor phase was about 25 µg/m3. Andren et al. (3), using a very efficient electrostatic precipitator, reported that nearly 93% of the in-stack Se is emitted in vapor phase. There is much concern about Se because of its toxicological impacts, although it is an essential trace element for humans and other animals (4). Coal-fired power plants have been identified as major sources of emissions of trace elements including Se. In light of Title I of the Clean Air * Corresponding author phone: (33) 468307700; fax: (33) 468302940; e-mail:
[email protected]. 1406
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 35, NO. 7, 2001
Amendments of 1990, the environmental regulations will most probably become more drastic for these emissions soon. This probability has generated interest in developing technologies for removing Se from fossil-fired flue gas. Current metal control technologies use conventional air pollution control devices (APCDs). However, they do not always control all elements effectively and consistently, especially volatile elements such as selenium, which is partially released as vapors. It is necessary now to develop alternative technologies to improve emission control of harmful elements. A first way to better control trace element emissions is to reduce or even suppress their volatilization during combustion. In some cases, this can be achieved by modifying the combustion conditions and/or by pretreating the coal. The other way to control harmful element emissions is to use sorbents to scavenge them; in this case, the element chemical speciation must be known for selecting the best sorbent. Clearly, both ways require a better knowledge of the Se volatilization process during coal combustion. Se speciation in the flue gas has been a subject of intense controversy and speculation. Se is similar to sulfur in most of its chemistry; its possible oxidation states are -2, 0, +2, +4, and +6. Andren et al. (3) used a series of chemical tests to conclude that nearly all the Se in the cooled flue gas (about 150 °C) is elemental Se. Oppositely, Davison et al. (5) concluded from thermodynamics that it is SeO2 and that as much as 80 µg/m3 of Se can exist at 25 °C as SeO2. Oehm et al. (6) reported that the dominant form of Se in atmospheric aerosols is selenious acid, and they suggested that this acid results from the reaction of SeO2 with moisture. Dismukes (7) indicated that Se exists as SeO2 whatever the combustion environment and that it may be partially reduced to Se by SO2 in the stack (lower temperature). Our previous thermodynamic study (8) established the major Se species, as well as Hg and As species, in the coal-fired flue gas from 400 to 1800 K. Gaseous Se, SeO, and SeO2 were found to be dominant at high temperature, and H2Se(g) and SeCl2(g) were the major species at low temperature. The objective of this study is to experimentally validate thermodynamic predictions of Se behavior in coal combustion. The following method was used: tests were performed with Se-spiked coal and coke samples, and the Se speciation in the flue gas was determined. Then these results were compared with the predictions of thermodynamics applied to the model (Se-spiked) sample. A favorable comparison would thus justify the thermodynamic approach for predicting the behavior of real combustion systems, since experiments dealing with trace elements in real combustion systems are unfeasible. The study focused on the following: (i) The dynamics of Se volatilization during coal and coke combustion in a thermobalance. (ii) The influence of the Se speciation in the original Sespiked coke and coal samples (identified by X-ray diffraction, XRD). (iii) The speciation of the Se captured in the flue gas by membrane filters (identified by X-ray photoelectron spectroscopy, XPS). (iv) The thermodynamic prediction of the major Se speciation in the cooled flue gas accounting for the composition of the spiked sample and the comparison with experimental results. (v) The thermodynamics of real combustion systems containing much less Se (1000-10 000 times less). 10.1021/es0001005 CCC: $20.00
2001 American Chemical Society Published on Web 03/02/2001
2. Experimental Procedures First, Se-loaded coke/coal samples were prepared by impregnating coke and coal in a highly concentrated Se solution. Then, combustion experiments were conducted in a thermobalance (SETARAM B70), and the Se volatilization was evaluated. The speciations of the Se-added compounds were identified by XRD (X’Pert, PHILIPS), used to interpret the Se volatilization process. Finally, the vaporized Se and other compounds in the cooled flue gas were trapped on an alumina membrane filter (Anodisc 25, WHATMAN), and their chemical speciations were identified by XPS (CAMECA-RIBER SIA250). The results were compared with our thermodynamic predictions. Se contents were analyzed by ICP-AES (JOBIN YVON JY38S) after microwave-accelerated digestion of the samples (MARS 5, CEM Co.). The related experimental operations are described hereafter one by one. 2.1 Sample Preparation. Four samples were considered: one coal sample (Kromdraai coal) and one coke (Francoke) in three size fractions: 50-80 µm, 80-120 µm, and 100-125 µm. To highlight the main results of this study and to simplify the related discussions, the complete results are presented in most cases for one sample only (coke 50-80 µm); the influence of the fuel composition is discussed separately. Information data concerning the chosen coke and coal are given in Tables 1 and 2, Supporting Information. Their Se original contents ranged from 0.5 to 1.2 µg/g. 2.2 Addition of Se Species into Samples. The four solid samples were impregnated in highly concentrated Se solutions containing Na2SeO3 or Se2Cl2/SeCl4. These solutions were prepared from the reaction of solid SeO2 either with H2O and NaOH to form Na2SeO3 or with concentrated HCl to form Se2Cl2/SeCl4 (see method 1, Supporting Information). After 24-h impregnation, the samples were filtered and dried at low temperature (80%) at temperature lower than 420 °C (test duration 30 min) and that it was found in the filter as Se and SeO2. Surprisingly, tests performed with SeO2spiked sample did not give any deposit on the filter containing Se or Cl. This may be due to vapor condensation before the filter, but further experiments are required to propose more explanations. These experiments show that XPS is a good way to identify the trace vapor Se compounds in the flue gas after their capture. This technique is more adapted for determining vapor metal speciation than wet-chemistry tests, which involve a series of chemical treatments (absorption, scrubbing, reaction, concentrating, and transferring, etc.) that may change the original metal speciation in the flue gas. 3.4 Thermodynamic Prediction of Se Species in the CoalFired Flue Gas. A complex chemical system containing 54 elements and more than 3200 species was studied before (11, 15). Significant interactions concerning Se were found. The objective of our analysis is to compare the Se speciation of spiked samples obtained experimentally with that calculated by thermodynamics. On the basis of this comparison, the prediction of Se speciation in real combustion systems by thermodynamic approach should be supported. We simulated the chemical composition of Se-spiked (in Na2SeO3 form) coke. The thermodynamic calculation was VOL. 35, NO. 7, 2001 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
1409
the probable interaction of Se with the mineral matter of coal, which reduces the volatilization. Under reducing conditions, H2Se(g) and SeO(g) are the dominant species at 1200 K, but numerous interactions between Se and other trace elements may occur at lower temperatures to form PbSe, GeSe, Tl2Se, and AgSe in variable concentration. The complex chemical reactions occurring during combustion obviously determine the trace element partitioning. With no doubt, thermodynamic calculations are very helpful for a better understanding of the speciation of Se and probably of other trace elements. Regarding Se behavior during coal combustion, several experimental results and thermodynamic predictions agree well: (i) The Se speciation in the flue gas was identified by XPS and predicted by thermodynamics; both give Se and SeO2. (ii) Chlorine increases the Se volatilization; it was predicted by thermodynamics and confirmed experimentally. (iii) Interaction of Se with the mineral matter of the coal matrix is probable: it was predicted by thermodynamics, and it is a coherent interpretation of the experimentally found difference between spiked coke and coal behaviors.
Acknowledgments This work was supported by ECSC (European Community of Steel and Coal) through a subcontract between CERCHAR and CNRS. D. Pe´rarnau, R. Berjoan, F. Sibieude, and E. Beˆche are thanked for their help in XPS and XRD analyses.
Supporting Information Available In this section, the basic properties of the coke and coal are presented, and the methods of impregnation of samples and of Se analysis are described. Two interesting results from the thermogravimetric study and from the thermodynamic study are plotted. This material is available free of charge via the Internet at http://pubs.acs.org.
Literature Cited FIGURE 4. Thermodynamics of Se equilibrium distribution in the flue gas from Se-added coke combustion (a) under oxidizing conditions, (b) under reducing conditions. (54 elements and 3200 species in the system, Se: 28.4 mol and Na: 2.925 mol). carried out by increasing the molar numbers of Se and Na to 28.4 and 2.925 mol, respectively (with respect to standard coke), corresponding to their contents in the spiked coke (10 000 ppm Se and 0.45 wt % Na2O, respectively). Other computation parameters were the same as before (15). The Se dominant species are plotted versus temperature in Figure 4. Under oxidizing conditions (Figure 4a), the dominant gaseous species is SeO2, which is present at temperatures as low as about 500 K (227 °C), whereas solid SeO2 is dominant when T < 500 K. Under reducing conditions (Figure 4b), gaseous H2Se is dominant beyond 500 K. Below this temperature, atomic Se, i.e., Se (cr,l), is nearly the only existing species. These thermodynamic predictions are in good agreement on one hand with our experimental results since we detected both atomic Se and SeO2 in the cooled flue gas from the combustion of Se-spiked coke, and on the other hand, with measurements in combustors (1, 2). These findings show that thermodynamic prediction of Se behavior in combustion and incineration systems is valid. Such calculation is illustrated by Figure 2, Supporting Information, for real coal composition. Under oxidizing conditions, SeO(g) and SeO2(g) are the most probable species, and Se interacts with the mineral matrix, forming CoSeO3 and NiSeO3. This behavior supports our observation about
1410
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 35, NO. 7, 2001
(1) Germani, M. S.; Zoller, W. H. Environ. Sci. Technol. 1988, 22, 1079. (2) Bolten, J. G. EPRI Report EA-5358, 1987. (3) Andren, A. W.; Klein, D. H.; Talmi, Y. Environ. Sci., Technol. 1975, 9, 9. (4) Goyer, R. A. Casarett and Doull’s Toxicology the Basic Science of Poisons, 4th ed., Amdur, M. O., Doul, J., Klaassen, C. D., Eds., Pergamon Press: New York, 1991. (5) Davison, R. L.; Natusch, D. F.s.; Wallace, J. R.; Evans, C. A. Environ. Sci. Technol. 1974, 8, 1107. (6) Oehm, G. J.; Crisp, P. T.; Ellis, J. J. Air Waste Manage. Assoc. 1991, 41, 190. (7) Dismukes, E. B. Fuel Process. Technol. 1994, 39, 403. (8) Yan, R.; Gauthier, D.; Flamant, G. Combust. Flame 2000, 120, 49. (9) Querol, X.; Fernandez, T. J. L.; Lopez, S. A. Fuel 1995, 74, 331. (10) Raask, E. Fuel 1985, 11, 97. (11) Yan, R., Ph.D. Dissertation, University of Perpignan, France, 1999. (12) Querol, X.; Ferna`ndez-Turiel, J. L.; Lo`pez-Soler, A. Mineral. Mag. 1994, 58, 119. (13) Hanawalt, J. D. Powder Diffraction File- Inorganic MaterialsSearch Manual (Hanawalt), SMH29 (Alphabetical Index), SMA29, JCPDS, International Center for Diffraction Data, 1979. (14) Feng, B.; Yan, R.; Zheng, C. G. Dev. Chem. Eng. Mineral Proc. 1999, 7 (3/4), 387. (15) Yan, R.; Gauthier, D.; Flamant, G. Combust. Flame 2000, to be published.
Received for review May 15, 2000. Revised manuscript received January 2, 2001. Accepted January 2, 2001. ES0001005
