Environ. Sci. Technol. 2002, 36, 3879-3884
Kinetics of Heavy Metal Vaporization from Model Wastes in a Fluidized Bed STE ´ PHANE ABANADES,* GILLES FLAMANT, AND DANIEL GAUTHIER Institut de Science et de Ge´nie des Mate´riaux et Proce´de´s (CNRS-IMP), B.P. 5 Odeillo, 66125 Font-Romeu Ce´dex, France
Metal vaporization experiments were carried out in an atmospheric fluidized bed to study the influence of operating conditions on the extent of heavy metal (HM) release in fumes from municipal solid waste incinerators. Model wastes spiked with compounds of Pb, Cd, and Zn were used. The parameters studied were temperature, treatment duration, matrix of the model waste (mineral and organic), HM initial speciation, and gas composition (N2, air, air + HCl, gas mixture simulating the incinerators). The extent of vaporization was measured by solid sample analysis and online analysis of the gaseous effluent, after customization of the ICP technique for gas analysis. The results indicate the metal vaporization rate is very high initially and then slows. The results with mineral matrices give the decreasing order of volatility Cd > Pb > Zn, but in industrial incinerators Zn volatilizes slightly more than Pb. Temperature (especially for porous alumina) and mineral matrix have a strong influence on the HM vaporization, but HCl concentration and HM initial speciation do not. The gas composition and the initial metal concentration are significant parameters. The matrix influence clearly denoted the mass transfer limitations in the vaporization process from mineral matrix.
1. Introduction Municipal solid waste incineration (MSWI) simultaneously reduces the waste volume (∼90%) and generates power. Yet one of the major environmental concerns is the emission of toxic heavy metals (HM) during this thermal treatment and their possible leaching from the different residues produced by municipal solid waste (MSW) incinerators: bottom ash, boiler ash, filter ash, and dusts collected by the flue gas cleaning processes. The recycling or the disposal of the ultimate residues, in which HM are concentrated, is problematic. To improve the ashes and make their reuse possible, the physicochemical processes occurring in an incinerator and involving HM must be understood (1). A fraction of the toxic metal compounds vaporizes during the heat treatment, and then it condenses and forms particulate during the flue gas cooling or deposits on available surfaces (2). The classical air pollution control devices (APCD) may not always be sufficiently efficient for collecting the sub-micrometer metal particles and gaseous metals. Thus, the sub-micrometer particles and vaporized metals, which are extremely hazardous, may be emitted into the environment. * Corresponding author telephone: (+33) (0)4-68-30-77-00; fax: (+33) (0)4-68-30-29-40; e-mail:
[email protected]. 10.1021/es020037e CCC: $22.00 Published on Web 08/02/2002
2002 American Chemical Society
The HM involved in the volatilization process are metals with a low boiling point, such as Hg, but also semivolatile ones, which can be significantly enriched in fly ash. Cd and Hg are readily volatilized during the waste incineration and differ in their subsequent behavior: most Cd condenses on fly ash particulate (70% of the total input is found in the APCD dust), whereas 90% of Hg remains gaseous throughout the entire process, thus requiring specific abatement technologies. Pb and Zn remain principally in the bottom ash (60 and 50%, respectively), but a significant fraction is vaporized and condensed like Cd (respectively, 35 and 45%), whereas most of the Cu, Cr, Co, and Ni stay in the bottom ash (>90%) (3). The HM partitioning among the various residues depends on the MSW composition, on the physicochemical properties of the metals and/or their compounds, and on the incinerator operating conditions (4). The waste composition, which varies with the location and the period, and the HM properties, which determine their mobility, cannot be modified to prevent their release. To master the metal emission, it is thus essential to determine the relationship between the HM vaporization and the operating parameters of the process, such as temperature, gas composition, and residence time. Previous studies indicate that the degree of metal volatilization, which clearly depends on the metal, is a complex function of many factors: the initial metallic speciation and concentration, the nature of the matrix, the treatment temperature and duration, the gas composition and air flow rate, and the presence of other species such as chlorine, sulfur, or aluminosilicate compounds (5-10). Most studies intend to reduce or suppress HM volatilization during combustion, whereas others use sorbents to capture them through various scavenging mechanisms (11-15). So far, the fate of HM was studied mainly through thermodynamic equilibrium calculations (16-19); the main parameters influencing the final HM partitioning were identified; among them, the oxygen and the chlorine contents are the more significant. In a burning load of waste, some regions may be locally oxygen-depleted (reducing conditions), which affects the HM speciation and its release in the gas phase. Chlorine, due to plastic and cooking salt in MSW, enhances HM volatilization because of the formation of volatile metallic chlorides. The objective of this study, developed in a fluid bed for better control of the temperature and mass transfer into gas, is to evaluate the effects of operating conditions on metal vaporization kinetics and to understand the physical and chemical phenomena controlling HM release in the combustion zone. The behavior of three metals of most concern (Cd, Pb, and Zn) is studied. The parameters considered are treatment duration (solid residence time), temperature, matrix of model waste (mineral or organic), gas composition (air, air + HCl, N2 + HCl, a synthetic gas measured in industrial incinerators), initial HM concentration, and speciation. This study is also interesting for a process under development (20). In this process, in which HM vaporize nearly completely from the ash particles, the ash collected on-line is treated in a fluidized bed (at 900 °C with 10% HCl).
2. Experimental Section 2.1. Experimental Setup. The experimental setup is shown in Figure 1; it was described in detail by Abanades (21). The fluidized bed reactor is made of AISI 316L stainless steel, 4.5 × 10-3 m thick. It is a 0.105 m i.d. and 0.4 m high cylinder, topped by a 0.2 m disengaging height. The bed is composed VOL. 36, NO. 17, 2002 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
3879
FIGURE 1. Experimental setup. of 1.6 kg of sand of mean diameter 0.7 × 10-3 m, in which a given mass of reactive HM-spiked particles is injected at thermal steady state. The fluidizing gas, which can be air or a mixture of air, nitrogen, and carbon dioxide, is preheated through a series of two electrical resistances (Osram Sylvania in-line heater). The preheater exit temperature is set at 850 °C, and it is controlled by an exposed junction K-thermocouple linked to a phase angle fired controller (Chauvin Arnoux Statop 4897). For obvious resistance protection reasons, corrosive agents are added afterward. They can be SO2 and diluted hydrochloric acid, which is injected through a water-cooled three-wall probe. HCl is titrated in the feed solution to obtain the desired HCl/H2O mass ratio in the fluidizing gas. The reactor is electrically heated by two half-cylinder radiative shells (Kanthal Superthal SHC) 0.2 m high, fed by a low induction current transformer (JLL) and piloted by a power switch (Eurotherm TC 3001). The bed temperature is controlled by a K-thermocouple welded on the wall and linked to a regulator/programmer (Eurotherm 818). Alumina-silica blocks (Kerlane 50 Prismo RX1), stapled onto an aluminum shell, insulate the whole facility, and the insulation thickness ranges between 0.30 and 0.40 m. Gas flow rates are measured by mass flowmeters (Brooks Instruments 5800). All data (temperatures and flow rates) are recorded and stored on a PC using ADAM acquisition modules and GENIE software (Advantec). By means of a device based on compressed air and tubing plunged into the bed, reactive particles can be injected when the bed is at steady state, and solid samples can be aspirated at given times. Then, the metal concentration in samples is measured by inductively coupled plasma optical emission spectroscopy (ICP-OES) after manual sorting and microwaveassisted (CEM Mars 5) acid digestion. This leads to the concentration versus time profile (vaporization kinetics) for given conditions. In addition, HM concentration in exhaust gases is measured on-line by ICP-OES. To do so, the gas outlet is connected to the nearby ICP (Jobin Yvon JY 38S) by heated tubing (T ) 400 °C). The gas to analyze is aspirated through a set including a primary sampler (membrane pump) and a secondary sampler (peristaltic pump); the ICP was 3880
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 36, NO. 17, 2002
TABLE 1. Properties of Mineral Matrices particle diameter (mm) density (kg‚m-3) specific surface area (m2‚g-1) porous volume (cm3‚g-1)
alumina
sepiolite
1.6-2 1348 104 0.45
1-1.6 2880 60 0.37
previously adapted for analyzing heavy metals in gas according to Trassy and Diemiaszonek’s technique (22). 2.2. Metal-Spiked Samples. Three metals, representing metal families that behave differently during incineration, were considered both one by one and simultaneously: Cd, Pb, and Zn. Because MSW samples do not exist in a convenient form for fluidizing them, the metals were spiked into both mineral and organic matrices by impregnating the latter in aqueous solutions (during 5 h) and drying them at 80 °C (during at least 24 h). Before impregnation, mineral matrices were calcined for 4 h at 850 °C to eliminate volatile substances (moisture) and to stabilize their properties (density and porosity), which are listed in Table 1. They were porous alumina (diameter ) 1.6-2 mm), one of the main components of ash, and sepiolite (diameter ) 1-1.6 mm), more precisely a mixture of sepiolite [Mg4Si6O15(OH)2‚6H2O (85%)] and dolomite [CaMg(CO3)2 (15%)]. The initial metal concentration q0 was measured by ICP-OES after microwaveassisted acid digestion of matrix samples (a mixture of 9 mL of 85% H3PO4 and 2.5 mL of H2O and a mixture of 8 mL of 40% HF and 4 mL of 69% HNO3 were, respectively, used to dissolve 0.5 g samples of alumina and sepiolite). The initial metal concentration obtained this way in the model waste was reproducible and was in the concentration range corresponding to that measured in MSW (cf. Figures 2-5). The metal distribution in the porous mineral matrices was determined by X-ray microanalysis: the concentration is homogeneous in alumina particles, whereas it decreases from the periphery to the core of sepiolite particles. Two organic matrices were also studied: activated coal and polyacrylamide (diameter ) 0.6-1.6 mm). In the latter, classically used for holding back water in plant pots, the desired metal concentration could be adjusted very easily.
FIGURE 2. Vaporization percentage of Cd and Pb from alumina versus time at 850 °C.
FIGURE 3. Vaporization percentage of Cd, Pb, and Zn from sepiolite versus time at 850 °C.
FIGURE 4. Influence of oxygen content on the vaporization kinetics of Cd from alumina at 850 °C.
times for chemical analysis to measure the metallic concentration versus time profile q(t) (vaporization kinetics). Analysis was done by ICP-OES after manual sorting of reactive particles in the aspirated sample and their microwaveassisted acid digestion. These analyses gave the extent of metal vaporization, that is, the vaporization percentage of HM [defined as 1 - q(t)/q0]. Then, the vaporization rate profile (dq(t)/dt) was deduced. In several experiments, on-line analysis measured the time course of the HM concentration in exhaust gases. Trassy and Diemiaszonek (22) developed this method based on the coupling of a sampling line with an ICP-OES analyzer. The method is detailed by Abanades (21).
3. Results 3.1. Vaporization Kinetics of Cd, Pb, and Zn. The vaporization kinetics from solid could be studied quantitatively only in the case of mineral matrices because the samples are not destroyed during the heat treatment. With porous alumina, the following decreasing order of volatility is observed: Cd > Pb . Zn. The maximum percentages of vaporization obtained are 55% for Cd, 22% for Pb, and Zn, and the percentage of vaporization depends on the chlorine content in the gas (Figure 3). Therefore, the matrix clearly has a strong influence on the HM vaporization extent: the release is much larger from sepiolite than from alumina, especially in the presence of chlorine. 3.2. Influence of the Fluidizing Gas Composition. The influence of the gas composition was studied by considering three cases: fluidization with air, fluidization with air or synthetic gas containing HCl and water, and fluidization with nitrogen containing chlorine and water. 3.2.1. Experiments with Air. Experiments using air as fluidizing gas lasted between 90 and 120 min. HM vaporization was not significant at either 600 or 850 °C (maximum process temperature). Thermogravimetry confirmed that Cd and Pb begin vaporizing from alumina at temperatures higher than the operating temperature in the fluidized bed (respectively, 1165 and 1220 °C), whereas Zn is not vaporized at even 1500 °C. This is due to oxidation reactions occurring at high temperature. Actually, in an oxidizing atmosphere, the initial metallic species, trapped in the material as chloride, reacts either with oxygen or with water adsorbed on the matrix to form the corresponding oxide. Cd can also react with alumina (13, 14, 24, 25) according to
Al2O3 + CdCl2 + H2O f CdO‚Al2O3 + 2HCl FIGURE 5. Influence of temperature on the vaporization kinetics of Cd from alumina. 2.3. Experimental Procedure. Batch experiments were carried out. For each run, the fluid bed was first loaded with 1.6 kg of silica sand (particle diameter ) 600-800 µm). When at steady state (i.e., desired temperature reached and steady operating conditions), a given amount of model waste (130 g) was injected into the bed. Experiments were performed with particles spiked with metallic compounds, and several matrices were considered: alumina, sepiolite, activated coal, and polyacrylamide. After the injection of the mineral particles, solid samples were aspirated from the bed at given
The same reaction prevails with Zn, which explains its high stability in alumina. 3.2.2. Experiments with Air or Synthetic Gas Containing HCl and Water. Experiments were carried out by adding HCl and water in the fluidizing gas that may be air or a synthetic gas mixture (O2, 4.8%; N2, 70.8%; CO2, 8.8%; H2O, 15.6%; SO2, 400 mg/Rm3), simulating incineration gaseous conditions. The main tendencies observed for CdCl2 release from alumina are illustrated in Figure 2 for 90-min experiments. The vaporization percentage can reach 56% and levels off with time. HCl addition is required to volatilize the metal, but its concentration does not seem to significantly affect the final vaporization percentage (at least in the range studied). VOL. 36, NO. 17, 2002 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
3881
Moreover, the extent of vaporization seems to be less when the initial HM concentration is higher. When the synthetic gas is used as fluidizing gas instead of air, the initial vaporization rate is higher (Figure 4). This means the vaporization kinetics slow when the oxygen content in the gas increases. 3.2.3. Experiments with N2 Containing HCl and Water. When the fluidizing gas contains no oxygen (pyrolysis conditions), both the vaporization percentage and the vaporization rate increase at the beginning of the treatment. Nevertheless, the final vaporization percentage remains unchanged as illustrated in Figure 4. The oxygen content has a strong influence on the initial HM vaporization rate. When the oxygen amount is high (in the case of air), the chlorinating reaction is kinetically hindered, which delays the HM release. 3.3. Influence of Temperature. The influence of temperature depends on the mineral matrix. With alumina, the kinetics of HM vaporization increases with temperature as shown in Figure 5. The higher the temperature, the higher the rate of metal vaporization at the initial stage of the treatment and the higher the final vaporization percentage. Inversely, temperature has no significant influence on the metal release from sepiolite. 3.4. Influence of HM Initial Speciation. Experiments were performed with alumina matrix spiked with cadmium chloride and sulfate and with sepiolite spiked with lead chloride, nitrate, and acetate. The HM initial speciation has no apparent effect on the vaporization kinetics. 3.5. Simultaneous Presence of the Three Metals: Cd, Pb, and Zn. Cd, Pb, and Zn were introduced simultaneously as their chlorides in the mineral matrices, and the results vary greatly depending on the matrix. In the case of sepiolite matrix, all three metals are released simultaneously, and each metal behaves as if it were the only metal in the matrix. Therefore, the simultaneous presence of several metals has no influence on the metal release. On the contrary, no metal vaporizes from alumina. This comes from the formation of solid ZnAl2O4, which fills the pores, thereby inhibiting the vaporization of any other metal. This was confirmed by experiments on alumina samples spiked with Cd and Pb only: the vaporization of both metals is similar to that when they are alone in the matrix. 3.6. Results of On-line Analysis. The on-line analysis of gases was implemented when both mineral and organic matrices, spiked with respectively, Cd, Pb, and Zn chlorides, were used. In each case, after the reactive, metal-spiked particles were injected, the evolution of the net emission intensity Inet (background intensity deduced; units of counts per second) of metallic spectral lines was measured continuously by the customized ICP-OES system. The analytical wavelengths used for the detection of Cd, Pb, and Zn were 226.50, 220.35, and 213.85 nm, respectively. The emission intensity is proportional to the gaseous metal concentration because both the gas composition and the sampling flow are constant with time. Therefore, this method gives a good qualitative overview of the transient metal concentration in the outlet gas. A calibration method is required to obtain absolute metal concentrations, and it will be implemented in the next step of the study. The gaseous metal concentration exhibits a peak almost instantaneously after the reactive sample is injected into the fluid bed, whatever the matrix. We compared the data resulting from the solid and the emitted gas analysis for the sepiolite-spiked sample. We checked that vaporization rate (from the solid analysis) and gaseous metal concentration follow the same tendency (a decrease with time). However, the quantity of metal released in the gas is less than the quantity of metal vaporized from the particles. This discrepancy points out the possibility of interaction between the sand bed and heavy metals. For organic matrices (see 3882
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 36, NO. 17, 2002
FIGURE 6. On-line gaseous emission intensity of Cd (q0 ) 1500 ppmw), Pb (q0 ) 3000 ppmw), and Zn (q0 ) 5000 ppmw) versus time (20 g of activated coal in each run; synthetic gas, 850 °C; CHCl ) 1425 mg/Nm3). Figure 6 in the case of activated coal), the metal release period is short and the vaporization process occurs during the organic matter combustion. However, the plots show that Zn release is slower than that of other metals. This behavior may come from reducing conditions within burning particles, and actually, according to thermodynamics (17, 21), the volatilized species would be Zn(g) instead of ZnCl2(g). 3.7. Characterization of Mineral Matrices with X-ray Microanalysis (XRM). The objective of this analysis is to determine the initial HM distribution in the mineral matrices in order to explain the variations of HM behavior with the matrix. Sections of spiked-mineral particles were examined with an electron microprobe CAMECA Microbeam, operating in wavelength-dispersive scanning (WDS) mode. The particle morphology was observed with the scanning electron microscope (SEM). X-ray spectra were acquired to study the HM location and distribution in the particle (mapping of HM distribution). In the case of alumina, the metallic species is uniformly distributed inside the particle after the sample preparation by impregnation. This distribution happens because the material porosity is high enough to allow metal adsorption in the whole particle. For sepiolite, the HM concentration decreases from the surface to the particle center, which means that the diffusion of the HM solution inside the matrix is much slower.
4. Interpretation and Discussion of Experimental Results The experimental study shows that the HM behavior depends strongly on the matrix in which they are trapped. The aim of this section is to explain the discrepancies of vaporization trends according to the matrix. We report the major possible phenomena during the HM release that control the vaporization process. First, HM vaporization is highly time dependent. The rate of vaporization is maximum at the initial stage of the treatment, and then it slows and levels off with time. With mineral matrices, the vaporization extent of HM is limited whatever the operating conditions, which means that the chemical environment influences the HM speciation. Indeed, oxidation reactions or interactions with the mineral matter result in the formation of stable metallic species. HCl must be present in the system to form chloride species and then to obtain metal vaporization. Metallic chloride is the only species vaporized because the initial speciation has no effect on the vaporization kinetics. For mineral matrices, the experimental trends found for HM vaporization are the following: (i) Cd vaporization is significant, even when the metal is included in a mineral matrix that can interact to form stable compounds (alumina case).
(ii) Pb release is limited when vaporizing from alumina even with chlorine excess, which is a tendency observed during MSWI. According to thermodynamics, Pb is expected to vaporize as its chloride like Cd does (26). Both transfer phenomena and the chloride formation kinetics may control its release. (iii) Zn vaporization from the considered mineral matrices is very low, whereas Zn vaporizes slightly more than Pb during MSWI. Actually, Zn is particularly stabilized in mineral matrices. It must be found as aluminate, because the volatility of Zn from alumina is negligible, but also as silicate, because Zn release from sepiolite is weak. Thermodynamic equilibrium calculations predict this behavior for Zn, and they show the existence of the aluminate as major and that of the silicate as minor at incineration temperatures. Zn vaporization depends on the reducing conditions and on the chlorine content in the system (23). The experimental results point out the great influence of the matrix on the vaporization kinetics of HM. The process of sorption (either physical or chemical) within the matrix has important effects on the HM behavior, because it determines the binding force between the substrate and the metallic species. Therefore, the surface-contaminant interactions (adsorption effects) must be investigated closely to understand the role of the substrate in the vaporization kinetics of HM. In particular, the difference between multiple layer and monolayer coverage must be considered to explain the discrepancies between the vaporization kinetics from alumina and from sepiolite. For alumina, the metal is homogeneously distributed in the particle as shown by XRM, and temperature affects the vaporization kinetics, whereas HCl content does not. The chemical sorption inside the pores, coupled with the internal diffusion of gaseous metal species, controls the vaporization process (19). The HM vaporization slows with time because of the high-temperature chemical reactions, the existence of which was shown by Masseron et al. (24) (formation of the binary oxides CdO‚Al2O3 and ZnO‚Al2O3). For sepiolite, a gradient of the metal concentration was observed with XRM. Therefore, to obtain the same total amount of HM as with alumina, the relatively low surface area of sepiolite requires a multilayer coating of the metallic species. Because the metal is more concentrated near the particle external surface, the multilayer coating may induce a physical adsorption (reversible phenomenon) in this zone. Then, when the temperature increases, the metal chloride desorbs and is released in the gas phase. Therefore, two consecutive dominant steps can describe the process controlling the HM release from sepiolite, as follows: (i) Surface phenomena control the vaporization kinetics during the first part of the treatment ( ∼10 min); the metal release comes from the chloride desorption at high temperature. The reaction of chloride formation chemically controls this step, which particularly explains the influence of temperature and HCl content in the gas. (ii) Then, when residence time increases, the internal diffusion becomes dominant, because most peripheral metal has been released. The metal can be partly stabilized within the matrix (possible formation of silicate compounds), which explains its limited release and the vaporization rate decrease. In the case of organic matrices, the vaporization kinetics is high as soon as the sample has been injected. When the matrix burns, the particles shrink rapidly, and the metallic compound is released in the gas stream without any limitation because internal transfers obviously do not control the process. Nevertheless, the chlorine content in the gas significantly affects the release duration and the proportion of metal vaporized. The proportion is much lower when the gas contains no HCl because elemental gaseous metals are released mainly during the organic material burning, because
of reducing conditions. Once in the gaseous phase, most HM is oxidized in an atmosphere without HCl (due to the presence of oxygen in the gas). Then, HM condenses on fly ash particles (which are trapped in the cyclone) or it deposits on the reactor walls. On the contrary, when HCl is present, metallic chlorides are rapidly formed and the gas easily carries them away. This study showed that the types of matrix and processes involved are of great importance. The nature of metalsubstrate interactions (physical or chemical) and the initial HM distribution vary with the matrix, which determines the HM behavior and the vaporization extent. On the one hand, mass transfer limitations control the HM vaporization when introduced in a porous mineral material, which favors metallic reactions in solid phase and limits the vaporization extent. On the other hand, HM are quickly released when contained in an organic, thus burning, matrix. If another matrix is considered, knowing its properties and composition (which determine possible interactions with metals), it is possible to explain the HM behavior because the main limiting phenomena can be pointed out. The results may be quantitatively different with another kind of matrix, but the vaporization tendency and its sensitivity to operating conditions will be similar to those obtained with either matrix involved in this study. The next step of the study will involve a more complex matrix issued from real municipal solid wastes, including both mineral and organic phases.
Acknowledgments This study was supported by ADEME (Agence de l’Environnement et de la Maıˆtrise de l’Energie) and the CNRS-ECODEV program. We acknowledge R. Flamand for XRM analysis and C. Trassy for fruitful advice about the on-line analysis method.
Literature Cited (1) Chandler, A. J.; Eighmy, T. T.; Hartle´n, J.; Hjelmar, O.; Kosson, D. S.; Sawell, S. E.; van der Sloot, H. A.; Vehlow, J. Municipal Solid Waste Incinerator Residues; The International Ash Working Group (IAWG); Elsevier Science: Amsterdam, The Netherlands, 1997. (2) Barton, R. G.; Clark, W. D.; Seeker, W. R. Combust. Sci. Technol. 1990, 74, 327-342. (3) Le Goux, J. Y.; Le Douce, C. L’incine´ration des De´chets Me´nagers, (Household Waste Incineration); Economica: Paris, France, 1995. (4) Brunner, P. H.; Monch, H. Waste Manage. Res. 1986, 4, 105-119. (5) Linak, W. P.; Wendt, J. O. L. Prog. Energy Combust. Sci. 1993, 19, 145-185. (6) Morf, L. S.; Brunner, P. H.; Spaun, S. Waste Manage. Res. 2000, 18, 4-15. (7) Eddings, E. G.; Lighty, J. S.; Kozinski, J. A. Environ. Sci. Technol. 1994, 28, 1791-1800. (8) Ho, T. C.; Lee, H. T.; Shiao, C. C.; Hopper, J. R.; Bostick, W. D. Waste Manage. 1995, 15 (5/6), 325-334. (9) Jakob, A.; Stucki, S.; Kuhn, P. Environ. Sci. Technol. 1995, 29, 2429-2436. (10) Belevi, H.; Langmeier, M. Environ. Sci. Technol. 2000, 34, 25072512. (11) Punjak, W. A.; Uberoi, M.; Shadman, F. AIChE J. 1989, 35, 11861194. (12) Uberoi, M.; Shadman, F. AIChE J. 1990, 36, 307-309. (13) Uberoi, M.; Shadman, F. Environ. Sci. Technol. 1991, 25, 12851289. (14) Scotto, M. V.; Uberoi, M.; Peterson, T. W.; Shadman, F.; Wendt, J. O. L. Fuel Process. Technol. 1994, 39, 357-372. (15) Linak, W. P.; Srivastava, R. K.; Wendt, J. O. L. Combust. Flame 1995, 100, 241-250. (16) Wu, C. Y.; Biswas, P. Combust. Flame 1993, 93, 31-41. (17) Frandsen, F.; Dam-Johansen, K.; Rasmussen, P. Prog. Energy Combust. Sci. 1994, 20, 115-138. (18) Durlak, S. K.; Biswas, P.; Shi, J. J. Hazard. Mater. 1997, 56, 1-20. (19) Abanades, S.; Flamant, G.; Gauthier, D. J. Hazard. Mater. 2001, 88 (1), 75-94. (20) Mantelet, J. Technol. Int. 1999, 56, 37-38. (21) Abanades, S. Ph.D. Dissertation, University of Perpignan, 2001. VOL. 36, NO. 17, 2002 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
3883
(22) Trassy, C.; Diemiaszonek, R. High Temp. Chem. Processes 1994, 3, 449-458. (23) Verhulst, D.; Buekens, A.; Spencer, P.; Eriksson, G. Environ. Sci. Technol. 1996, 30, 50-56. (24) Masseron, R.; Gadiou, R.; Delfosse, L. Environ. Sci. Technol. 1999, 33, 3634-3640. (25) Ho, T. C.; Chuang, T. C.; Chelluri, S.; Lee, Y.; Hopper, J. R. Waste Manage. 2001, 21, 435-441.
3884
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 36, NO. 17, 2002
(26) Owens, T. M.; Wu, C. Y.; Biswas, P. Chem. Eng. Commun. 1995, 133, 31-52.
Received for review February 18, 2002. Revised manuscript received June 20, 2002. Accepted July 2, 2002. ES020037E
