(PDF) Determination of Kinetic Law for Toxic

Aug 1, 2018 - Full-Text Paper (PDF): Determination of Kinetic Law for Toxic Metals Release during Thermal ... charge via the Internet at http://pubs.a...
0 downloads 0 Views 204KB Size
Environ. Sci. Technol. 2005, 39, 9331-9336

Determination of Kinetic Law for Toxic Metals Release during Thermal Treatment of Model Waste in a Fluid-Bed Reactor J I N G L I U , * ,† S . A B A N A D E S , ‡ D. GAUTHIER,‡ G. FLAMANT,‡ CHUGUANG ZHENG,† AND JIDONG LU† State Key Laboratory of Coal Combustion, Huazhong University of Science and Technology, Wuhan, 430074, China and Processes, Materials, and Solar Energy Laboratory (CNRS-PROMES), B.P.5 Odeillo, 66125 Font-Romeu Ce´dex, France

Accumulation of toxic metals generated by thermal treatment of municipal solid waste presents a serious threat to the environment. A study was carried out to investigate the kinetic law of toxic metal release from municipal solid waste during their thermal treatment. Both direct and inverse models were developed in transient conditions. The direct mathematical model of the fluidbed reactor is based on Kunii and Levenspiel’s two-phase flow model for Geldart Group B particles. The inverse model intends to predict the metal’s rate of vaporization from its concentration in the outlet gas. The derived models were found to predict reasonably well the experimental observations. A method to derive the kinetic law of toxic metals release during fluidized bed thermal treatment of model waste from the global model and the experimental measurements is derived and illustrated. A first-order law was fitted for the mineral matrix, and a second-order law (simplified) was fitted for the realistic model waste. The kinetic law obtained in this way could be integrated in a global model of combustion of municipal solid waste in order to simulate the effects of operating parameters on the metal’s behavior.

Introduction Municipal solid waste incineration (MSWI) processes are expected to develop in the next decades, since they simultaneously permit a great reduction of waste volume (about 90%) and energy recovery by power generation. While thermal treatment is an effective method of remediating the municipal solid waste (MSW), major environmental concerns are emissions of toxic metal fumes during treatment and their possible leaching from the different residues produced by MSW incinerators. Metals will not be destroyed during high-temperature thermal treatment. They either stay in the treated waste or volatilize. The volatilized metals will later condense to form metal fumes or deposit on available surfaces. Theoretically, the process is governed by the laws of thermodynamics, reaction kinetics, and mass and heat transfer operations (1). * Corresponding author phone: (+86) 27 87545526; fax: (+86) 27 87545526; e-mail: [email protected]. † Huazhong University of Science and Technology. ‡ Processes, Materials, and Solar Energy Laboratory. 10.1021/es051042w CCC: $30.25 Published on Web 10/28/2005

 2005 American Chemical Society

The U.S. EPA has reported that toxic metals can account for almost all of the identified cancer risks from waste incineration systems (2). Compared to coal combustion, incineration of MSW yields much higher concentrations of cadmium and lead in the fly ash and far greater emissions (µg‚m-3) of arsenic, beryllium, cadmium, and lead, even though peak flame temperatures for coal combustion are likely much higher (3). It is essential to understand the release behavior of metals during the process of high-temperature thermal treatment in order to better understand their behavior and better control their emissions. Heavy metal (HM) vaporization during MSWI has been studied mainly by direct analysis of the different residues produced by combustion (4-6). The partitioning of HMs among the various residues is determined with respect to the operating conditions. Other experimental studies on the metal behavior during thermal treatment of artificial solid wastes (1, 7) showed that HMs with higher saturated vapor pressure enter the atmosphere easily after vaporization, subsequently enriching in fly ash more than in bottom ash. Mechanism studies (8, 9) showed that HM behavior in fluidized bed combustion cannot be based only on the mass balance measurements. Detailed aerosol studies indicate that HM can be released from fuel particles as bound to fine ash as well as coarse ash particles and then leaves the combustor with fly ash. The thermodynamic calculations were applied for predicting the partitioning of a metal during incineration (10). However, there is experimental evidence that equilibrium calculations overpredict the amount of metal vaporized (11). The main reason is that mass transfer limitations are not taken into account in such models and that the required assumptions (particularly the closed-system approach) are not accurate for MSWI processes. Many fluidized bed models were developed (12-15), which range from ones with pseudo-homogeneous phases to complex ones with three-phase or assemblies of bubbles. Helble (16) developed a semiempirical model of HM emission in which fundamental laboratory results as well as field emission data were included. Ho et al. (1) identified kinetic laws to describe metal behavior during fluidized bed thermal treatment of soil. This model is based on the heat and mass transfer laws and experimental results to simulate the metal vaporization process. Experiments were used to identify kinetics in the model. In summary, there are a large number of experimental data from on-site boilers and laboratory facilities. Researchers have provided qualitative mechanisms for elemental vaporization such as volatility and associations of elements. However, to our knowledge, the kinetic law on the release of metals during incineration of municipal wastes is lacking. On the basis of well-controlled experimental studies for artificial MSW, this paper identifies the kinetic law of metal vaporization from MSW during incineration. Experiments were carried out in a well-instrumented bubbling fluidized bed incinerator for measuring the metal concentration in the exhaust gas. Both direct and inverse models were developed to predict HM vaporization in transient conditions based on our previous work (17, 18). The modeling results were compared with the experimental data in order to validate the method. The main purpose of this paper is to derive kinetic laws at the particle level from the global model and the experimental data because up to now we have only compared theoretical and experimental global responses of the system (reactor scale). First, previous work related to experimental measurement and theoretical approach is VOL. 39, NO. 23, 2005 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

9331

summarized. Then the method developed to identify the kinetic law for metal release from model waste is presented. Two cases are detailed: metals in a mineral matrix and metal in an organic/mineral matrix (model waste). The results indicate that rather simple mathematical expressions of the kinetics may be derived.

Experimental Section A schematic representation of the experimental setup was given in a former paper (17). The AISI 316L S.S. reactor is a 0.105 m I.D. and 0.40 m high cylinder, topped by a 0.2 m disengaging height. The reactor was electrically heated, and K-thermocouples measured the temperature at several depths in the bed as well as at the inlet and outlet. Every measurement was recorded on a PC. Once the reactor was in steady state, a given mass of particles spiked with a metal was injected into the bed. Temperature measurements proved that because only a small quantity of such reacting particles was injected in the bed at high temperature (850 °C), the temperature does not change very much: actually, there was a decrease of about 10 °C with mineral particles but an increase of 10-20 °C with real waste due to exothermic combustion. The measurements also showed that the reactive sample penetrated deep into the bed, since the temperature change was the same whatever the height. The synthetic gas composition in the experiments (by volume) is 4.8% O2, 70.8% N2, 8.8% CO2, 400 mg‚Nm-3 SO2, 15.6% H2O, and 1000-2000 mg‚Nm-3 HCl. During experiments solid samples can be taken from the bed at given times and collected in a vessel for analysis by means of a tube plunged in the fluid bed and an aspiration pump. This method, which can be implemented only in the case of mineral matrixes, provides the temporal profile of a metal’s concentration in the mineral matrixes (q(t)). The metal concentration in exhaust gases is measured on-line by inductively coupled plasma optical emission spectrometry (ICP-OES). For that purpose, the gas outlet is connected to the nearby ICP and carried back to the ICP through a heated line (distance 6 m, maximum temperature 450 °C). The preparation of metal-spiked samples and analysis of samples were given in a former paper (17).

Modeling Methodology Direct Model. A summary of the model is given; for more details, see ref (17). The classical Kunii and Levenspiel model (19) was built up for beds of Geldart group A particles in which clouds are very thin. The modified version of Kunii and Levenspiel’s model (14) was adapted, which is especially fitted to treat beds of Geldart group B particles in which clouds are very thick, to the problem of HM vaporization (by taking into account the phenomena of HM adsorption and condensation of metal on particles). The main assumptions of the model are as follows: (1) the bed is composed of two phases, the bubbles containing a small amount of solid (subscript B in the equations) and the emulsion phase, corresponding to the rest of the bed (subscript E); (2) bubbles have a uniform diameter; (3) gas flow in the emulsion phase is at minimum fluidization, and gas is in plug flow everywhere; (4) there exists a mass transfer between both phases, and the global mass transfer coefficient is estimated from the relation (14)

Umf KBE ) 4.5 DB

9

(

-FB UB/

)

∂CBi ∂CBi ) -FBγBθrBiFp + FBKBE(CBi - CEi) + + ∂h ∂t FBγB(Fpθvspiked + Fsand(1 - θ)vsand) (2)

Emulsion phase

(

)

∂CEi ∂CEi ) -(1 - FB)(1 - mf)θrEiFp + ∂h ∂t FBKBE(CBi - CEi) + (1 - FB)(1 - mf)(Fpθvspiked + Fsand(1 - θ)vsand) (3)

-(1 - FB) Umf

where γB represents the volume fraction of solid particles dispersed in the bubbles. Its empirical value is 0.005 (14); θ is the volume fraction of reacting solid (particles spiked with heavy metal) in the bed (volume of reacting solid/total volume of solid); rEi and rBi correspond to the vaporization rates of i species in each phase (mg‚s-1‚kg-1); they are considered the same and equal to r in the following (rEi ) rBi ) r). vspiked and vsand are the adsorption rates on reacting particles and sand, respectively (mg‚s-1‚kg-1). Actually, these terms of adsorption are negligible since adsorption on sand is always very small (20, 21) and the mass of sand is much larger than the mass of reacting sample. The model parameters are provided in the Supporting Information. Inverse Model. When organic matrixes or real wastes are burning, the concentration of each metal in the outlet gas is the only experimental measurement available since the burning solids cannot be sampled. The kinetics of vaporization of a metal from the solid waste (at the particle level) can then be obtained by applying the inverse model developed below. The purpose of the inverse model is to predict the metal’s vaporization rate, knowing its concentration in the outlet gas. Thus, the vaporization rate may be determined whatever the matrix, even if it is burning. On-line analysis gives the emission intensity profile, which can be normalized (I/Imax). It is the same as that of concentration (Co* ) Co/Comax) since intensity and concentration are proportional. Thus, eqs 2 and 3 were written under dimensionless forms by dividing by Comax/FB and expressed with eqs 5 and 6. The dimensionless vaporization flux f (s-1) is related to the vaporization mass flow rate r (mg‚kg-1‚s-1)

(1)

The bed is assumed to be isothermal, so the internal phenomenon within a solid particle (heat and mass transfer, 9332

chemical reactions) would have no effect on the results. This macroscopic approach uses only the global flux of generation of species at the external surface of the particles. Therefore, it may be extended to other combustion systems in which solid-gas mass transfer occurs. Temperature measurements at different depths in the bed showed that this assumption was true for alumina (the bed’s temperature decreased about 10 °C when at 850 °C) and particles of realistic artificial waste (bed’s temperature increased 10-20 °C). The variations of the volumetric flow rate due to chemical and physical conversions were neglected. First, the metal’s vaporization flux was weak compared to the gas flow of fluidization. Second, for organic matrixes the gas flow resulting from combustion did not significantly affect the global flow rate of gas, since the mass of sample injected into the bed represented less than 1% of the total mass of the bed. Finally, the gas velocity (U) was considered constant in the bed. The model predicts the HM concentration in the bubble phase and emulsion phase as a function of bed height. For species i and an elemental step [h, h + dh], the mass balances in each phase are as follows. Bubble phase

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 39, NO. 23, 2005

f)

γBθrFp Comax/FB

The model’s equations can be written as

(4)

/ KBE ∂CBi/ 1 ∂CBi f (C /- CEi/) + ) / / / Bi ∂h ∂t U U U B

B

(5)

B

/ ∂CEi/ f 1 ∂CEi + + ) (1 - mf) ∂h Umf ∂t γBUmf FB KBE (C /- CEi/) (6) 1 - FB Umf Bi

[

Coi/) CBi/+

]

(1 - FB) ‚CEi/ FB

(7)

h)H

where the dimensionless concentrations CE* and CB* are defined as

CE/)

CE CB , CB/) Comax/FB Comax/FB

(8)

Determination of the Kinetic Law of HM Vaporization: Case of Alumina Matrix (Validation of the Method) Experimental Results. In the case of mineral matrix the metal’s concentration in the solid particle can be determined experimentally by analyzing solid samples withdrawn from the fluidized bed. The profile of experimental data of Cd concentration from intermittently sampled solid particles can be fitted as q(t) ) 439.03 + 57993.96/(1 + 100.06321(34.2179+t)), with t in minutes. The fitting results were in good agreement with the experimental data (see Figure 1.). Direct Model Simulation. The equation of Cd concentration in the solid particles (q(t)) was differentiated to get the vaporization rate (r ) dq(t)/dt). The vaporization rate was used as an inlet parameter in the direct model to calculate the Cd concentration in the outlet gas (Co), as shown in Figure 2. The values of various parameters used in the simulation are listed in Table 1. In the first few seconds the Cd concentration in the outlet gas increased sharply; when time was about 3 s Co has a maximum, and then it decreased with time. Inverse Model Simulation. Using the gaseous concentration profile at the outlet gas (Co) as an inlet parameter, the Cd vaporization flux (from porous alumina particles) from the fluidized bed was determined by applying the inverse model (see Figure 3.). The vaporization rate is relatively fast at the initial stage of the treatment but slows down and eventually levels off later. The decrease in the rate of metal vaporization is apparently caused by formation of less volatile metal compounds (CdO‚Al2O3) according to Al2O3+CdCl2+H2O f CdO‚Al2O3 + 2HCl (20, 21). From eq 4, since the parameters γB, θ, FP, Comax, and FB were known for alumina matrix, the Cd vaporization rate (r) can be calculated. Then, Cd concentration in the solid particles (qi) for each time ti can be obtained from the relation qi - qi-1 ) r‚∆t and then compared with the experimental data. Figure 4. plots the comparison between this calculated Cd concentration in the solid particles and the experimental values obtained from analysis of solid samples. The theoretical profile of Cd concentration in the solid particles is clearly consistent with the experimental results. This very good fit proves that results from on-line gas analysis can be used to estimate the rate of vaporization of metals by applying the inverse model. Kinetic Law of Cd Vaporization in Alumina. The vaporization rate (r ) dq/dt) and the profile of the metal’s concentration in solid particles (q) for each time are known. Hence, the kinetic law can be deduced from the relation between the vaporization rate (r ) dq/dt) and the profile of

FIGURE 1. Cd concentration in the solid particles (q) from alumina (q0 ) 856 ppmw; 130 g of alumina in each run; synthetic gas 850 °C; CHCl ) 1374 mg‚Nm-3).

FIGURE 2. Results of direct model. Time course of the Cd concentration in the outlet gas from alumina matrix.

TABLE 1. Values of the Model’s Parameters of the Fluidized Bed parameters (m‚s-1)

superficial gas velocity, U void fraction at minimum fluidization, Umf void fraction at normal fluidization condition, mf initial diameter of bubble, DB0 (mm) mean diameter of bubble, DB (mm) bubble’s rise velocity, UB (m‚s-1) initial height of bed, H0 (m) expanded height of bed, H (m) volume fraction of bubbles, FB global exchange coefficient, KBE (s-1)

value 0.51 0.5 0.74 8.87 27.4 0.71 0.15 0.31 0.48 27.92

the metal’s concentration in solid particles (q), as shown in Figure 5. From Figure 5 the corresponding Cd concentration decrease in the solid particles in the second part is very small (about 3 mg‚kg-1) compared to the first part (about 500 mg‚kg-1) and the time is very short (smaller than 3 s), so the second part can be neglected. The kinetic law is

dq ) 0.00241q - 0.91316 (t g 3 s), it is a first-order law dt (9) VOL. 39, NO. 23, 2005 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

9333

FIGURE 3. Results of inverse model. Time course of Cd vaporization flux (f).

FIGURE 4. Comparison between the Cd concentration in the solid particles (q) obtained from modeling and the experimental one.

FIGURE 6. On-line gaseous intensity of Cd (q0 ) 3600 ppmw), Pb (q0 ) 2980 ppmw), and Zn (q0 ) 2700 ppmw) vs duration of treatment (10 g of MSW sample in each run; synthetic gas 850 °C; CHCl ) 1700 mg‚Nm-3).

FIGURE 7. Results of inverse model. Time course of the vaporization flux of metal (f) from the model waste.

Determination of the Kinetic Law of HM Vaporization: Case of Model Waste

FIGURE 5. Cd vaporization rate (r) vs Cd concentration in the solid particles (q). or assuming

dq ) k(q - Rq0)n dt

(10)

where q0 is the initial Cd concentration in solid particles, q0 ) 856 mg‚kg-1, k ) 0.00241, R ) 0.442646, n ) 1. 9334

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 39, NO. 23, 2005

Experimental Results. The net intensity of emission (i.e., after withdrawing the background intensity), more precisely the normalized intensity (I/Imax) of spectral lines of metal, was measured versus time. Whatever the matrix, the metal’s concentration in the gas exhibited a peak almost instantaneously after the reacting sample was injected into the fluidized bed. For model waste made from real municipal solid waste, thus containing organic species, the profiles of vaporization of metals in the exhaust gas are the only data available since the burning solids cannot be sampled. Normalized profiles of the concentration of the three metals are shown in Figure 6. Inverse Model Simulation. The on-line gaseous intensity of Cd, Pb, and Zn was fitted and for each metal; the vaporization dynamics was obtained by applying the inverse model to the experimental measurements (see Figure 7). With synthetic gas containing HCl, the vaporization tendencies of Cd and Pb are similar and the release process is short. Zn vaporizes somewhat slower than Cd and Pb. This trend indicates that Zn chemically interacts with the mineral part of the sample. Kinetic Law of Cd Vaporization in Model Waste. Comparing to a mineral matrix, it is more difficult to obtain the kinetic law of HM vaporization in organic matrix (waste) because the profile of the metal’s concentration in solid

FIGURE 8. Cd concentration (q0 ) 3600 mg‚kg-1) in waste vs time.

TABLE 2. Vaporization Percentage for Different Values of a

or more accurately

2400 2600 2800 3000 3200 3400 3600 3800 4093 vaporization 58.6 64.5 68.4 73.3 78.2 83.1 88.0 92.9 100 percentage, %

particles (q) for each time cannot be measured by experiment. The HM relative concentration in outlet gas (Co) is the only measured data. Comax is not known, so the vaporization rate (r) cannot be obtained from eq 4 directly. Another method to get r has been developed. According to eq 4, f ) γBθrFp/Comax, assuming

a)

FIGURE 9. Comparison of polynomial fit of Cd vaporization rate (r) vs Cd concentration.

r ) a‚f

(11)

Comax FB‚γB‚θ‚FP

(12)

dq/dt ) 0.16926 + 0.05659q + (5.84806 × 10-6)q2 - (5.96689 × 10-9)q3.

Acknowledgments This work was supported by the National Key Basic Research and Development Program of China (No. 2002CB211602) and ADEME (French Agency for Environment and Energy Management). Portions of this work were conducted under the National Key Basic Research and Development Program of China (No. 2006CB200304) and Key Program of National Natural Science Foundation of China (No. 90410017). It was also conducted in the frame of the Sino-French Collaboratory of Chemical and Environmental Engineering (No. 2001CB711203).

Supporting Information Available γB, θ, FP, and FB are known. Therefore, for different values of a, q(t) profiles can be deduced from the relation qi - qi-1 ) r‚∆t as shown in Figure 8. Vaporization percentage is shown in Table 2 for different values of a. If the vaporization percentage is known, the value of a can be deduced, and then the kinetic law can be identified. When the realistic artificial waste is burned at 850 °C, the final Cd concentration in waste (qfinal) is 0 mg‚kg-1, so the vaporization percentage is 100%. The value of a is thus 4093. Then, from eq 4, Comax ) 645.09 mg‚Nm-3. Figure 9 shows the relation between the vaporization rate (r ) dq/dt) and the profile of the metal’s concentration in solid particles (q). Finally, the kinetic law can be obtained only from the experimental data of HM relative intensity by the inverse model as soon as q0 and qfinal in solid samples are known. This method can be applied to any matrix, whatever mineral matrix or organic matrix. From Figure 9 the kinetic law can be assessed

dq ≈ -0.59491 + 0.09155q - (2.51051 × 10-5)q2 dt or assume

dq/dt ≈ k(q - Rq0)n + β with k ) -2.51051 × 10-5, n ) 2, R ) 0.506, and β ) 82.7095

Detailed model parameters. This material is available free of charge via the Internet at http://pubs.acs.org.

Literature Cited (1) Ho, T. C.; Lee, H. T.; Shiao, C. C.; Hopper, J. R. Metal behavior during fluidized bed thermal treatment of soil. Waste Manage. 1995, 15 (5/6), 325-334. (2) Oppelt, E. T. Incineration of hazardous wastesa critical review. JAPCA 1987, 37, 558-586. (3) Greenberg, R. R.; Zoller, W. H.; Gordon, G. E. Composition and size distributions of particles released in refuse incineration. Environ. Sci. Technol. 1978, 12 (5), 566-573. (4) Linak, W. P.; Wendt, J. O. L. Toxic metal emissions from incineration: mechanisms and control. Prog. Energy Combust. Sci. 1993, 19, 145-185. (5) Wey, M. Y.; Yu, L. J.; Jou, S. I. The influence of heavy metals on the formation of organics and HCl during incinerating of PVCcontaining waste. J. Hazard. Mater. 1998, 60, 259-270. (6) Vassilev, S. V.; Braekman-Danheux, C. Characterization of refuse derived char from municipal solid waste: 2. occurrence, abundance and source of trace elements. Fuel Process. Technol. 1999, 59, 135-161. (7) Jakob, A.; Stucki, S.; Kuhn, P. Evaporation of heavy metals during the heat treatment of municipal solid waste incinerator fly ash. Environ. Sci. Technol. 1995, 29, 2429-2436. (8) Lind, T. L.; Valmari, T.; Kauppinen, E. I.; Nilsson, K.; Sfiris, G.; Kristina, N.; Maenhaut, W. Volatilization of heavy metals during circulating fluidized bed combustion of forest residue. Environ. Sci. Technol. 1999, 33, 496-502. (9) Latva-Somppi, J.; Moisio, M.; Kauppinen, E. I.; Valmari, T.; Ahonen, P.; Tapper, U.; Keskinen, J. Ash formation during VOL. 39, NO. 23, 2005 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

9335

(10)

(11) (12) (13) (14) (15) (16)

fluidized bed incineration of paper mill waste sludge. J. Aerosol Sci. 1998, 29, 461-480. Frandsen, F.; Dam-Johansen, K.; Rasmussen, P. Trace elements from combustion and gasification of coal-an equilibrium approach. Prog. Energy Combust. Sci. 1994, 20, 115138. Barton, R. G.; Clark, W. D.; Seeker, W. R. Fate of metals in waste combustion systems. Combust. Sci. Technol. 1990, 74, 327342. Davidson, J. F.; Harrison, D. Fluidised Particles; Cambridge University Press: Cambridge, 1971. Kato, K.; Wen, C. Y. Bubble assemblage model for fluidized bed catalytic reactors. Chem. Eng. Sci. 1969, 24, 1351-1369. Kunii, D.; Levenspiel, O. Fluidisation Engineering, 2nd ed.; Butterwort-Heinemann: Boston, 1991. Marias, F.; Puiggali, J. R.; Flamant, G. Modeling for simulation of fluidized-bed incineration process. AIChE J. 2001, 47 (6), 1438-1460. Helble, J. J. A Model for the air emissions of trace metallic elements from coal combustors equipped with electrostatic precipitators. Fuel Process. Technol. 2000, 63, 125-147.

9336

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 39, NO. 23, 2005

(17) Abanades, S.; Flamant, G.; Gauthier, D. The kinetics of vaporization of a heavy metal from a fluidized waste by an inverse method. Combust. Flame 2003, 134, 315-326. (18) Abanades, S.; Flamant, G.; Gauthier, D.; Tomas, S.; Huang, L. Development of an inverse method to identify the kinetics of heavy metal release during waste incineration in fluidized bed. J. Hazard. Mater. 2005, in press. (19) Kunii, D.; Levenspiel, O. Model for flow of gas through a fluidized bed. Ind. Eng. Chem. Fundam. 1968, 7 (3), 446-452. (20) Uberoi, M.; Shadman, F. High-temperature removal of cadmium compounds using sorbents. Environ. Sci. Technol. 1991, 25, 1285-1289. (21) Scotto, M. V.; Uberoi, M.; Peterson, T. W.; Shadman, F.; Wendt, J. O. L. Metal capture by sorbents in combustion processes. Fuel Process. Technol. 1994, 39, 357-372.

Received for review June 3, 2005. Revised manuscript received August 29, 2005. Accepted August 30, 2005. ES051042W