High-Temperature HCl Removal with Sorbents in a Fixed-Bed Reactor

874 Energy & Fuels 2003, 17, 874-878 High-Temperature HCl Removal with Sorbents in a Fixed-Bed Reactor Binlin Dou,*, Jinsheng Gao, Seung Wook Baek, and Xingzhong Sha Division of Aerospace Engineering, Department of Mechanical Engineering, Korea Advanced Institute of Science and Technology, 373-1 Kusong-dong Yusong-gu, Taejon 305-701, Republic of Korea, Department of Energy Chemical Engineering, East China University of Science and Technology, Meilong Road 130#, 200237 Shanghai, China Received December 17, 2001 The HCl removal with sorbents was studied in a fixed-bed reactor at 550 C. The distribution of chlorine in the fixed bed shows that the upstream end of the fixed bed was nearly saturated with the chlorine whereas the downstream end contained less than 2 wt %, and the mass-transfer zone in the bed is about 6 cm. The reaction kinetics of two sorbents with HCl indicates that there are critical times at which reaction rate controlling steps begin to change. In each of the rate-controlling steps the experimental results and model predictions are in good agreement. The results also imply that, as the reaction proceeds, pore diffusion inside the particle becomes more and more difficult, thus lowering the overall reaction rate. 1. Introduction The high-temperature molten carbonate fuel cells (MCFC) and the integrated coal gasification combined cycle (IGCC) are promising devices for future power generation plants because of their high efficiency and environmental friendship. In either case, the coalderived gas must be treated to remove impurities such as hydrogen chloride (HCl), a reactive, corrosive, and toxic gas, which is produced during gasification from chloride species in the coal. The amounts of HCl present in coal-derived gas are generally small, compared to those of sulfur compounds, nitrogen compounds, and particulate. Nevertheless, the removal of HCl vapor from the gas can be beneficial in any power plant because of the great corrosion potential of the vapor in contact with metal components. The study of the reactions between HCl and sorbents is potentially important for these technical processes. A number of processes are available for removing HCl vapor from industrial and incinerator waste gases. These processes scavenge HCl by adsorption onto activted carbon or alumina or by reaction with alkali oxides. The commercial sorbents for removal HCl are relatively expensive; meanwhile, they must operate at temperatures <400 C. Hence, inexpensive and hightemperature sorbents are need for the chlorine removal in high-temperature coal-derived gas. Despite the fact that the removal of HCl from flue gas by means of solid CaO or Ca(OH) 2 is a well-known process, 1-3 but removal * To whom correspondence should be addressed. Tel: 042-869-5754. E-mail: bldou@hotmail.com. Division of Aerospace Engineering. Department of Energy Chemical Engineering. (1) Mura, G.; Lallai, A. Chem. Eng. Sci. 1992, 47, 2407. (2) Jozewicz, W.; Chang, J. C. S.; Sedman, C. B. Environ. Progress 1990, 9, 137. (3) Weineu, C. E.; Jensen, P. I. Ind. Eng. Chem. Res. 1992, 31, 164. efficiency is not high enough, very little kinetic modeling work is available on this subject. Mura and Laliai 1 studied the kinetics of the reaction between CaO and HCl and determined that the activation energy for the chemical reaction is 45 kj mol -1 and that that for the diffusion in the solid phase is 37 kj mol -1. Weineu 3 studied hydrogen chloride reaction with lime and limestone. The kinetics of the binding reaction is governed by diffusion in the solid phase, which is proved to follow an unreacted grain-core model. But these models were proposed without regard to the conservation equation for the fixed-bed absorption. The present investigation was undertaken to develop the effective sorbents that can be used for high-temperature HCl removal. The reaction kinetics of these sorbents with HCl was also discussed in the fixed-bed reactor. 2. Experiments 2.1. Preparation of Sorbents. Sorbent ECl 1 was prepared by pelletizing the powder of alkali and alkali earth metal substance with binders and texturing agents. It contains either bentonite (5 wt %) or sodium silicate (7 wt %) as the binder. in addition, glycol (5 wt %) was added as a texturizing agent. The pellets were dried at the temperature of 100 C for 2 h, and calcined at 550 C in air for 6 h, and during calcaination, NaHCO 3, CaCO 3, Ca(OH) 2, and Mg(OH) 2 decomposed into Na 2- CO 3, CaO, and MgO, releasing CO 2 and H 2O. The evolution of these gases produced a favor sorbents, but at higher calcination temperature, the activity of sorbents decreased due to sintering of the solid. Sorbent ECl 2 was industrial catalysts for HCl removal, obtained from catalyst corporations. It was prepared with alkali or alkali earth compounds on Al 2O 3 by wet impregnation, which consists of (8 17)wt % alkali (or alkali earth) compounds on Al 2O 3. The sorbent product was crushed and sieved to 0.45-0.90 mm. The physical properties of two fresh sorbents were shown in Table 1. 10.1021/ef010294p CCC: $25.00 2003 American Chemical Society Published on Web 05/22/2003

HCl Removal with Sorbents Energy & Fuels, Vol. 17, No. 4, 2003 875 Table 1. Properties of Two Sorbents sorbent main components NaHCO 3, CaCO 3, Ca(OH) 2, Mg(OH) 2; 87% ECl 1 ECl 2 Ca(OH) 2, 11%; Al 2O 3, 89% preparation dry mixing wet impregnation pulk density (g/cm 3 ) 0.66 0.73 surface area (m 2 /g) 3.2415 127.883 pore volume (ml/g) 0.02008 0.1513 ave pore diameter (Å) 247.80 47.34 2.2. Apparatus and Procedure. The laboratory reactor system consists of a gas manifold, a fixed-bed reactor, and a chlorine analysis section. 4 In the gas manifold, the gas mixture is prepared by entraining HCl vapor with N 2. When N 2 passes through a vessel containing 20% 30% concentration HCl solution, a simulated gas with an acceptably stable HCl concentration can be obtained. The sorbent is supported by quartz wool in the center of tube in the fixed-bed reactor. The sample temperature was indirectly measured by a thermocouple. The gaseous effluent from the sorbent bed is analyzed for HCl content by dissolving the HCl vapor in a solution of NaOH. At the end of the experiment, the sorbent is cooled, removed from the reactor, and analyzed to determine its chlorine content. The HCl concentration in a solution of NaOH is analyzed chemically for Cl - by AgNO 3 titration. The chlorine content of sorbent is determined according to method of determination of chlorine in coal. 5 In a typical experiment, first the composition of simulated coal-derived gas and then the temperatures of the reactor and the HCl concentration were stabilized. During the start-up stage, the gas flow was directed away from the sample to prevent any reaction. To start the removal process, the gas flow switched to the sample section of the reactor. During HCl removal process, the HCl concentration in the effluent gas rises rapidly when the sorbent has reached certain capacity. The rapid change is termed breakthrough. The chlorine content of sorbent is based on the amount of fresh sorbent and the amount of HCl absorbed by the sorbent. A test 4 showed that the saturation chlorine contents of two sorbents are 43.02% for ECl 1 and 32.15% for ECl 2, respectively. 3. Results and Discussions 3.1. Activities of Sorbents. First, the reaction activities of two sorbents with HCl vapor were examined under a system made up of the temperature of 550 C, space velocity of 3000 h -1, and inlet HCl concentration of 1 10 3 mg/m 3. For sustained and efficient operation of IGCC or MCFC process, the feed gas must be free of contaminants such as chloride species. HCl are especially deleterious to MCFC because it can lead to sever corrosion of cathode hardware, and also react with the molten carbonate electrolyte. The allowable HCl concentration in the feed gas must be less than 1 mg/m. 3,6,7 Thus, when the dimensionless concentration value c 1 /c 0 reaches 0.001, the sorbents are regarded as being spent. The breakthrough curves of two sorbents (see Figure 1) were record until the dimensionless concentration reached the 0.05 values. Experimental errors from breakthrough curves were estimated to be around 4.8%. (4) Dou, B. L.; Gao, J. S.; Sha, X. Z. Fuel Process. Technol. 2001, 72, 23. (5) GB 3558-83. (6) Gillis, E. A. Chem. Eng. Prog. 1980, 76, 88. (7) Krinshnan, G. N.; Tong, G. T.; Wood, B. J. Report No. DOE/MC/ 30005-96/C0545; Department of Energy: Washington, DC. Figure 1. The breakthrough curves of two sorbents at 550 C. Figure 2. The chlorine content absorbed by sorbents versus time at 550 C. Figure 3. The reaction rate of sorbents as function of time at 550 C. Figure 2 shows the chlorine contents of two sorbents after exposure to reaction gas for various reaction times under the same conditions. The derivative of this plot, the rate of reaction as a function of time, is given in Figure 3. Figure 3 indicates that the reaction rate nearly keeps constant within 20 h of initial reaction period, and then the reaction rate decreases gradually. The fixed-bed experimental results indicated that ECl 1 sorbent has the longer breakthrough time. The ECl 1 sorbent was made with alkali and alkali earth metal substances having a lower porosity and a lower surface area. It displays the better adsorption capacity may be due to a combination of necessary amount of reactive component and favorable structure. Sorbent ECl 1 was made of about 87 wt % reactive component, while ECl 2 was made of 11 wt % reactive component. However, the variations in reactivity and chlorine content between two sorbents are minor. The results also show that initial chlorine content increased rapidly, but after 30 h it slows considerably. Although two sorbents are able to absorb a considerable amount of HCl, the time required for saturation is rather long. The

876 Energy & Fuels, Vol. 17, No. 4, 2003 Dou et al. Figure 4. The outlet HCl concentration vs time in the deep fixed bed. considered to be a complex process involving a large number of sub-processes. It was intended to model the performance of a fixed-bed reactor with the following assumptions: a, the fixed-bed is isothermal; b, gas velocity is constant; c, plus flow conditions for the gas phase is without axial dispersion, implying the use of a one-dimensional model along the z-axis; d, the gas phase is ideal; e, the chemical reaction rate is first order with respect to HCl. 9 The modeling of the fixed-bed reactor is based on the conservation equation derived from the mass balance on the fluid and solid phase: 10 u c z + ɛ c t +F q t ) 0 (1) When the concentration of HCl in coal-derived gas is very small, and ensuring the breakthrough pattern to be constant through the bed, a simple correlation of the gas and the solid can be obtained: 10 c c o ) q q o ) x (2) Figure 5. The distribution of chlorine content of sorbent in the deep fixed bed. effect of water vapor in the coal-derived gas on HCl removal reaction has been studied by Krinshnan, 7 which indicated that water vapor present in the coal-derived gas improves removal reaction. The effect of other contaminants such as NH 3, sulfur, and tar component in the coal-derived gas on HCl removal reaction has also studied in previous paper, 8 it is observed that these contaminants almost do not affect HCl removal. Notwithstanding, the performances of these contaminants in the HCl removal reaction are complex. 3.2. Distribution of Chlorine in the Bed. The distribution of chlorine in the bed was studied with ECl 1 sorbent at 550 C. A deep bed of sorbent pellets (1 cm diameter by 10 cm long) was exposed to a simulated coal-derived gas containing 1 10 3 mg/m 3 of HCl at a space of 1000 h -1. The HCl level in the reactor exit remained at less than 1 mg/m 3 for a period of a 14.8 h; it required about 25 h to reach the concentration of 100 mg/m 3 (see Figure 4). After reaction time of 25 h, the experiment was stopped and the spent sorbent was removed from the bed in successive layers that were analyzed for chlorine content. The results (see Figure 5) showed that the upstream end of the bed was nearly saturated with the chlorine whereas the downstream end contained less than 2 wt %. Generally, the fixedbed is divided into three zones according to chlorine content of sorbents, which are the saturated zone, masstransfer zone and void zone, respectively. There is a common phenomenon in a fixed-bed reactor: The masstransfer zone in the bed moves along the flow of reactants, leaving behind spent sorbent. The amount of active sorbent in the bed is reduced with time. In this experiment, the mass-transfer zone in the bed is about 6 cm. 3.3. Kinetics of HCl Removal Reaction. The HCl removal process in a fixed-bed reaction system is (8) Dou, B. L.; Zhang, M. C.; Gao, J. S.; Sha, X. Z. Ind. Eng. Chem. Res. 2002, 41, 4195. where c o is the inlet HCl concentration; c is the HCl concentration in the bed; q represents the chlorine content of the sorbent, which is kilograms of chlorine absorbed of per kilograms of sorbent; q o corresponds to saturation chlorine content of sorbent, which represents the saturated absorbing condition of the sorbent; and x is the conversion of sorbent. The reaction is unlikely to be controlled by the gas-film mass transfer; therefore, the rate controlling steps should be the chemical reaction or the product layer diffusion, or both. For the firstorder surface reaction, the reaction rate is defined as φ A ) 4 πr 2 k s c (3) At any given point of time the rate of diffusion of HCl per grain particle is given by φ A ) 4 πd e crr o r o - r The amount of HCl absorbed by sorbent per unit time is φ A ) 4 3 πr Fdq 3 o mdt The conversion of sorbent is also given by x ) 1 - r3 r o 3 Finally, the expression is obtained if the process is controlled by the chemical reaction at the grain surface: where τ g )Fr o /bk s c o m represents the time required for complete conversion. The influence of surface reaction may be less important than diffusion inside the particle. If the reaction (9) Fenouil, A. L.; Linn, S. Ind. Eng. Chem. Res. 1996, 35, 1024. (10) Wang, W. Y.; Ye, Z. C.; Bjerle, I. Fuel 1996, 75, 207. (4) (5) (6) τ g [1 - (1 - x) 1/3 ] ) τ g g(x) ) t (7)

HCl Removal with Sorbents Energy & Fuels, Vol. 17, No. 4, 2003 877 Table 2. Kinetic Parameters of Reaction Rate-Controlling Steps for Various Stages sorbent time (h) SSE τ g τ p δ 2 ECl 1 t < 20 0.035 76.9 20 < t < 40 0.011 118.0 190.3 0.62 t > 40 0.052 333.3 ECl 2 t < 25 0.036 142.9 25 < t <48 0.019 90.9 211.4 0.43 t > 48 0.025 166.7 is controlled by reactant diffusion through the product layer, it will follow the expression below: τ p [1-3(1 - x) 2/3 + 2(1 - x)] ) τ p p(x) ) t (8) Figure 6. g(x) versus t for two sorbents. where τ p )Fr o 2 /6bD e c o m represents the time required for complete conversion. The reaction is controlled by the combination of two mechanisms mentioned above. In this case, it is rational to derive the model following the expression below: τ g [g(x) + δ 2 p(x)] ) τ g f(x) ) t (9) δ 2 ) τ p τ g ) k s r o 6D e (10) where τ g and τ p represent the constant respectively for chemical reaction control and product layer diffusion control. τ g, τ p, and δ 2 can be estimated by minimizing the following SSE equation: Figure 7. p(x) versus t for two sorbents. N [τ g g(x i ) + τ p p(x i ) - t i ] 2 ) Q(τ g,τ p ) (11) i)1 The value of δ 2 represents the ratio of product layer diffusion resistance to chemical reaction resistance and is a form of the shrinking core reaction modulus. When δ 2, 1, the reaction can be assumed to be controlled by the rate of the intrinsic chemical reaction, when δ 2 > 10, it is reliable to assume that the reaction is under product layer diffusion control; an intermediate value of δ 2 suggests that the reaction is controlled by both chemical and product layer diffusion. 1 The sorbents studied in this work are far from being pure substances. However, experiments were conducted to elucidate the rate controlling mechanism of the sorbent and HCl reaction. The rate-controlling process is difficult to elucidate by single physical and/or chemical control. Pervious study has indicated the results of reaction kinetics of HCl with sorbents, 4 but the values of the sum of the squares of error for experiment and model in whole process were all high so that they cannot precisely illustrate the removal reaction process. The HCl removal reaction by sorbents is rather complicated. The reaction-rate-controlling steps in whole process may be divided into several stages. To establish the mechanism of HCl removal, it was tried to simulate measured value of g(x), p(x), and f(x) vs time with modesl 7, 8, and 9, respectively. The parameters calculated by this processing are shown in Table 2. which also shows those paramters used to determine the plots presented in Figures 6-8. On the basis of the mathematical model and experimental results, it is found that there are critical times at which rate-controlling steps begin to change. The Figure 8. f(x) versus t for two sorbents. linearity of g(x) as a function of time is observed at t < 20 h for sorbent ECl 1 and at t < 25 h for sorbent ECl 2 (Figure 6), which indicates in these periods the reaction rates of sorbents with HCl are mainly controlled by chemical steps. The linearity of p(x) as a function of time is observed at t > 40 h for sorbent ECl 1 and at t > 48 h for sorbent ECl 2 (Figure 7), which indicates in these periods the reaction rates of sorbents with HCl are mainly controlled byproduct layer diffusion. It is rational to obtain that the reaction rates of two sorbents with HCl are controlled by the combination of chemical and product layer diffusion in the intermediate value of these times (Figure 8). In each rate controlling steps the experimental results and model predictions are in good agreement. In fact, it is more likely from the very beginning that the chemical reaction of HCl with sorbents at the surface determines the reaction rate. The product layer diffusion is not regarded as the ratedetermining step. While the product layer proceeds, the product layer diffusion becomes increasingly rate controlling. At last, the product layer diffusion kinetically determines the reaction process. This implies that as

878 Energy & Fuels, Vol. 17, No. 4, 2003 Dou et al. the reaction proceeds, pore diffusion inside the particle becomes more and more difficult thus lowering the overall reaction rate. The diffusion inside the particle is, at any temperature, essentially dependent on porosity. This parameter can be affected by temperature and reaction degree. However, our experimental work was carried out at 550 C so that thermal sintering could not occur. Thus porosity changes are solely due to the affect of the chemical reaction. This may be the result of the great difference existing between the molar volumes of the solid reagent and the product. The decrease in the overall reaction rate is due to the decrease of particle porosity owing to the high molar volume of the solid product in comparison with that of the solid reagent. This similarity behavior has been widely studied for CaO sulfation. 11 The dependence of diffusivity on porosity was explained by referring to the random pore model. 12 4. Conclusion Two sorbents were found to be effective in removing HCl from high-temperature coal-derived gas. The distribution of chlorine in the fixed-bed shows that the upstream end of the bed was nearly saturated with the chlorine whereas the downstream end contained less than 2 wt %, and the mass-transfer zone in the bed is about 6 cm. The reaction kinetics of two sorbents with HCl indicates that there are critical times at which controlling steps of reaction rate begin to change. In each rate controlling steps the experimental results and model predictions are in good agreement. (11) Mura, G.; Lallai, A.; Olla, P. Chem. Eng. J. 1991, 46, 119. (12) Wakao, N.; Smith, J. M. Chem. Eng. Sci. 1962, 17, 834. Acknowledgment. This work is sponsored by State Key Fundamental R&D Project of China (G1999022104) and the National Natural Science Foundation of China (No. 59776017). Nomenclature b: stoichiometric ratio of sorbent to gaseous reactant c: HCl concentration in the bed (mg m -3 ) c o: HCl concentration at inlet (mg m -3 ) c 1: HCl concentration at outlet (mg m -3 ) D e: effective diffusion coefficient in a porous structure (m 2 s -1 ) f(x): conversion function under the combination of chemical control and product layer diffusion g(x): conversion function under chemical control k s: intrinsic reaction rate constant, n)1 (ms -1 ) m: molecular weight of reactant (kg kmol -1 ) p(x): conversion function under diffusion control q: chlorine content of sorbent (kg kg -1 ) q o: saturation chlorine content of sorbent (kg kg -1 ) r: radius of core (µm) r o: radius of particle (µm) SSE: sum of the squares of the errors t: time (s) T: temperature ( C) u: bed velocity (m s -1 ) x: conversion (%) z: bed depth (m) Greek Letters φ A: the rate of HCl removal in a single grain (mol/s) F: reactant density (kg m -3 ) δ 2 : ratio of diffusion to chemical reaction resistance ɛ: bed porosity τ g: characteristic time for chemical reaction control (h) τ p: characteristic time for diffusion control (h) EF010294P