Removal of Hydrogen Sulfide by Iron-Rich Soil: Application of the Deactivation Kinetic Model for Fitting Breakthrough Curve

Tzu-Hsing Ko1*, Hsin-Ta Hsueh2

1 Department of Hospitality and Tourism Management, Kao Fong College of Digital Contents, 38, Chang-Ji Township, Pingtung County 908, Taiwan 2 Sustainable Environment Research Center and University Center for Bioscience and Biotechnology, National Cheng Kung University, Tainan 701, Taiwan

ABSTRACT

In this study a deactivation kinetic model was used to predict the breakthrough curve in a noncatalytic gas-solid reaction. The iron-rich soil was tested to react with H2S under a reducing atmosphere at high temperature. The results indicated that the deactivation kinetic model can be well fitted to the breakthrough curve in the experimental range. The breakthrough curves were accurately predicted by the model, and provide useful information for the time to reload the solid materials in the reaction. The activation energy of the reaction of iron-rich soil and H2S was experimentally calculated to be about 34 kJ/mol and 131 kJ/mol, respectively for the deactivation kinetic model I (m = 0, n = 1) and model II (m = 1, n = 1). Both of the deactivation kinetic models can fit the experimental results. The order of H2S in the deactivation model probably ranged from zero to one.

Keywords: Deactivation kinetic model; Iron-rich soil; H2S; Breakthrough curve; Activation energy.

INTRODUCTION During the reaction of H2S and metal oxides, a dense sulfide layer is expected to be formed on the reactive oxide. The gas-solid noncatalytic reaction is a very important Diffusion resistance through this layer causes a significant process in chemical industrials, such as the reaction of decrease in the reactivity of the metal oxide (Kyotani et CaO and SO2, and the sulfurization reaction between metal al., 1989; Kyotani et al., 1989). In addition, significant oxides and H2S, that have been widely applied in the coal changes in pore structure, active surface area, and active gasification cleanup process. Many metal oxides, including site distribution during the sulfurization of the metal oxide ZnO, CuO, Fe2O3 and MnOx have been widely used will cause a significant deactivation of the solid. Deactivation because of their thermodynamic superiority with H2S models proposed in the literature for gas-solid reactions under high temperature (Slimane and Abbasian, 2000; with significant changes of activity of the solid due to Alonso and Palacios, 2004; Kim and Park, 2010; Cheah et textural changes, as well as product layer diffusion resistance al., 2011). The most attention was mainly aimed at the during reaction, were reported to be quite successful in development of effective metal oxides for H2S removal. predicting conversion time data (Dogu, 1981; Balci et al., Discussion of kinetic analysis for gas-solid noncatalytic 1993; Yasyerli et al., 1996; Bandyopadhyay et al., 1999; reaction is relatively lack. Most of kinetic analysis is built Yasyerli et al., 2003). on the basis of the shrinking core model (Homma et al., Our previous studies shown that the natural soils and 2005; Kar and Evans, 2008). Prior to fitting the shrinking contaminated soils have a highly reaction ability with H2S core model, the experimental work has to be operated with a and thus be used for many times after regeneration process thermogravimetric analysis (TGA). Although the shrinking (Ko, 2008, 2011). The iron species is the major active core model is appropriate to fit the kinetic results, a corrosion metal to react with H2S. Unfortunately, the information of effect should be considered when the TGA is used to carry kinetic performance is lack due to instrument limitation. out. Therefore, the main objective was to evaluate the feasibility of the deactivation kinetic model and obtained a series of kinetic parameters to predict the breakthrough time.

* Corresponding author. Tel.: +886-8-7626365 ext. 6220; MATERIALS AND METHODS Fax: +886-8-7629907 E-mail address: [email protected] The tested soil was collected from the field of the

1356 Ko and Hsueh, Aerosol and Air Quality Research, 12: 1355–1361, 2012 campus of the National Pingtung University of Science expected to cause a drop in the reaction rate. One would and Technology. Soils were sampled at a depth of 0–15 cm also expect it to cause significant changes in the pore from a site. After sampling, unwanted materials, such as structure, active surface areas and activity per unit area of leaves, tree root and blinding were removed from soil solid reactant with reaction extent. All of these changes sample and then dried at room temperature for a week. The cause a decrease in the activity of the solid reactant with collected soils were ground with an agate mortar and sieved time. As reported in the previous literatures, the deactivation to pass through a 2-mm sieve. The physical and chemical model works well for gas-solid reactions (Suyadal et al., property was tabulated in Table 1. 2000; Yasyerli et al., 2002; Kopac and Kocabas, 2004). In The experiment of this study was carried out using a fixed- this model, the effects of the factors on the diminishing bed reactor near atmospheric pressure. The experimental rate of sulfur fixation were combined in a deactivation rate system consisted of three parts: (i) a coal gasified gas term. To simulate the removal of H2S by sorbents, the simulation system; (ii) a desulfurization reactor system; and following assumptions were made: (iii) an exiting gas analyzing system. The composition of (i) The sulfidation is operated under isothermal conditions. the simulation coal gas involved was 1 vol % H2S, 15 vol % (ii) The external mass-transfer limitations are neglected. H2, and balanced N2. To avoid the formation of by-products (iii) The pseudo-steady state is assumed. during the kinetic experiments, only H2 and H2S were (iv) The deactivation of the sorbent is first-order with respect considered in this study. Gases were supplied from gas to the solid active sites, while zero-order for the cylinders and flow rates were monitored through mass flow concentration of H2S and can be described as follows: controllers. All mass flow controllers were monitored accurately by an IR soap bubble meter and the concentration da kCmn a  a aexp  kt (m = 0, n = 1) (1) of all species calculated at the condition of STP. Prior to dt d0d entering the reactor the gases were conducted in a mixing pipe to confirm that the mixture gas was turbulent flow. With the pseudo-steady state assumption, the isothermal The reactor consisted of a quartz tube, 1.6 cm i.d., 2.0 cm species conservation equation for the reactant gas H S is o.d., and 150 cm long, located inside an electric furnace. 2 expressed as follows: Quartz fibers were set in the reactor in order to support the soil samples. Weight hourly space velocity (WHSV) was dC controlled at 4000 mL/hr/g to avoid severe pressure drops QkCa000 (2) and channeling flow effect, and provided enough retention dW time. Two K-type thermocouples were inserted exactly into the reactor near the positions on the top and bottom of the Integrating Eq. (2), the following equation can be obtained sorbent packing to measure and control the inlet and outlet temperatures. Before sorption proceeding, a pure nitrogen CWdCCka  ka gas (purity 99.99%) was fed into the reactor for 30 minutes 00dlnWW    (3) CQ000 C Q at 773 K in order to remove adsorbed water and impure C0 0  materials, which coated on the surface of the sorbent. In addition, blank breakthrough experiments were also Combining with Eq. (1) and Eq. (3), Eq. (4) can be executed under the same conditions and verified that no obtained reaction was taking place anywhere between H2S and the lines/reactor. The inlet and outlet concentration of H S was 2  kW  analyzed by an on-line gas chromatograph (Shimadzu, GC- 0 CC0dexp exp  kt (4) 14B) equipped with a flame photometry detector (FPD) and  Q0  fitted with a GS-Q capillary column. A six-port sampling with 0.5 mL sampling loop was used to sample the inlet Arranging Eq (4), the following can be obtained: and outlet concentration of H2S. The removal experiment was terminated when the outlet H2S concentration from the CkW00 reactor approached the inlet concentration of H2S. In this ln ln( ) ln ktd (5) study, the breakthrough time was defined as the time from CQ0 the beginning of the sorption to the point outlet H2S concentration reached 100 ppm. Thus, if ln[lnCo/C] is plotted versus time, a straight line should be obtained with a slope equal to –kd and intercept Deactivation Model for Kinetic Study giving ln[koW/Qo], from which ko can be obtained. The formation of a dense product layer over the solid To obtain analytical solutions of Eq. (1) and Eq. (2) by reactant creates an additional diffusion resistance and is taking n = m = 1, an iterative procedure was applied. In

Table 1. Some chemical and physical properties of iron-rich soil Depth (cm) pH Clay Texture CEC (cmol/kg) OM Total iron oxides (g/kg) Iron-rich soil 0–15 4.6 25% Sandy clay loam 24.4 2.0% 163.63

Ko and Hsueh, Aerosol and Air Quality Research, 12: 1355–1361, 2012 1357 this procedure, the Eq. (4) is substituted into Eq. (1), and the To obtain an authentic experimental data, particle size first correction for the activity is obtained by the integration samples ranged between 90–110 mesh were collected to of this equation. The following approximate expression was carry out the deactivation kinetic model. For the external mass then derived (deactivation model type II, n = m = 1). transfer experiments, as shown in Fig. 1(b), a remarkable decline was found when the WHSV was controlled at 15,000 mL/hr/g. The external mass transfer resistance heavily C 1 expkW0d / Q (1 exp( kt )) exp exp(ktd ) affected the overall reaction, thus the optimal WHSV should Ckt(1 exp( )) 0dbe controlled between the ranges of 1,000–10,000 mL/hr/g/. (6) To minimize the external mass transfer resistance the kinetic study was conducted at a WHSV of 4,000 mL/hr/g. Nomenclature a: Activity of the solid reactant Deactivation Kinetic Type I Model (m = 0, n = 1) 3 Qo: Gas flow rate (m /min) The temperature dependence of the H2S breakthrough C: Outlet concentration of H2S (kmol/m) curves for the iron-rich soil is presented in Fig. 2. Co: Inlet concentration of H2S (kmol/m) According to the breakthrough curves, the relationship of kd: Deactivation rate constant (L/min) time and ln(lnCo/C) could be easily obtained at various 3 2 ko: Initial sorption rate constant (m /kg/min) temperatures, as shown in Fig. 3. The R values for all cases t: Time (min) were better than 0.98. The initial reaction rate constants, W: Active species mass (kg) ko, were calculated from intercept in Fig. 3. Meanwhile, a −1 straight line was attained by plotting ln ko versus T . Via its RESULTS AND DISCUSSION intercept and slope, the values of frequency factor and the apparent activation energy were calculated from Arrhenius Prior to investigating the feasibility of the deactivation relationship. The frequency factor and activation energy 11 kinetic model for the reaction of iron-rich soil and H2S, the were 2.31 × 10 and 131.51 kJ/mol, respectively. effects of external and internal mass transfer resistances have to be considered in order to understand their influence. Deactivation kinetic type II model (m = 1, n = 1) The performance of particle size and WHSV are the major Unlike type I model, the relationship between ko and Co indicators for determination of internal and external mass is a complex and nonlinear equation. To obtain parameters transfer in kinetic study. As shown in Fig. 1(a), the sulfur in Eq. (6), the regression fitting was performed using Eq. sorption capacity decreased with increasing particle size. (6) in the Sigma Plot software. Fig. 4 shows the regression Small particle size enhanced the sulfur sorption capacity results for the experimental data by the deactivation type II compared to larger ones, implying that the internal mass model. The R2 values for all cases were better than 0.99. transfer resistance should be considered if the particle size Likewise, the activation energy could be obtained from ranges from 10–60 meshes. On the other hand, the sulfur Arrhenius relationship. The frequency factor and activation sorption capacity maintained constantly while the particle energy were 2.49 × 107 and, 34.02 kJ/mol, respectively. size ranged from 80–120 meshes, indicating that the internal To further establish the fitness of the deactivation type I mass transfer resistance could be ignored within this range. and II models, three sets of reaction were performed at 598,

2.0 1.9 (a) (b) 1.9

1.8 1.8

1.7

1.6 1.7

1.5

1.4 1.6 Sulfur Capacity (g-sulfur/100g-soil) Capacity Sulfur 1.3 Sulfur sorption (g-S/100g-soil) sorption capacity Sulfur

1.2 1.5 10-30 40-60 80-100 100-120 0 2000 4000 6000 8000 10000 12000 14000 16000 Particle size (mesh) WHSV (m hr-1g-1)

Fig. 1. Effect of the mass transfer resistance experiments (a) particle size (b) WHSV. Inlet H2S: 1%, H2: 15% and balanced N2.

1358 Ko and Hsueh, Aerosol and Air Quality Research, 12: 1355–1361, 2012

12x103

10x103

S (ppm) 8x10 2

6x103

4x103

573K Outlet concentration of H 623K 2x103 673K 723K 773K

0 0 5 10 15 20 25 30 35 40 45 Duration time (minutes)

Fig. 2. Breakthrough curves of the reaction between iron-rich soil and H2S. Inlet H2S: 1%, H2: 15% and balanced N2, WHSV: 4,000 mL/hr/g with a flow of 100 mL/min.

3 22 2 573K R =0.997 lnKo=39.981-15821/T 623K R2=0.991 R 2=0.9102 2 2 673K R =0.988 Ea=131.51 kJ/mole 20 A=2.31*1017 723K R2=0.997 2 1 773K R =0.982

18 0 /C)) o o

-1 k 16 ln ln(ln(C -2 14 -3

12 -4

-5 10 2 4 6 8 10 12 14 16 18 20 22 1.2 1.3 1.4 1.5 1.6 1.7 1.8 t (minutes) T -1*103 (K-1)

Fig. 3. The relationship of the time and ln(lnCo/C) at various temperatures by a deactivation model (m = 0, n = 1 ) and Arrhenius equation fitting result.

653, and 723 K as well as simulated the model to predict around 30–100 kJ/mol (Ranade and Harrison, 1981; their breakthrough behaviors in Fig. 5. As can be seen, the Tamhankar et al., 1981; Sa et al., 1989; Pineda et al., R2 values were higher than 0.99, indicating the deactivation 1995). The difference between the type I and type II was kinetic type I and type II models could accurately predict the order of outlet concentration of H2S (represent by C in the breakthrough behaviors for the reaction of iron-rich the deactivation kinetic model, kmol/m3). Although there soil and H2S. In particular, the breakthrough times were was a significant change in activation energy for both two accurately predicted, which provided useful information models, their activation energies appeared to be accepted for the time to change loading materials in actual operating because they ranged among the previous reported studies. In condition. addition, the fitting results also presented well predictions The reasonable values for the typical noncatalytic gas- for both two models. It was speculated that the actual order solid reaction of iron oxides and H2S were reported to be of C may be ranged between zero to one, which resulted in

Ko and Hsueh, Aerosol and Air Quality Research, 12: 1355–1361, 2012 1359

12x103

10x103

8x103 S (ppm) 2

6x103

4x103

2 Outlet concentration of H of concentration Outlet 573K R =0.998 2 2x103 623K R =0.999 673K R2=0.999 723K R2=0.997 773K R2=0.999 0 0 5 10 15 20 25 30 35 40 45 Duration time (minutes) 12.0

lnko=17.032-4.0918/T R2=0.991 Ea=34.02kJ/mole A=2.49*107 11.5

11.0 o k ln

10.5

10.0

9.5 1.2 1.3 1.4 1.5 1.6 1.7 1.8 T-1*103 (K-1) Fig. 4. Regression fittings of the experimental data by a deactivation model (m = n = 1) under various temperatures and Arrhenius equation fitting result. Inlet H2S: 1%, H2: 15% and balanced N2, WHSV: 4,000 mL/hr/g with a flow of 100 mL/min. the nearly perfective fitting results for both models. for deactivation kinetic model I (m = 0, n = 1) and model II (m = 1, n = 1). The breakthrough curves were accurately CONCLUSIONS predicted which provided useful information for the time to reload the solid materials in the reaction. The deactivation kinetic model was employed to fit the breakthrough curve for the reaction of H2S and iron-rich ACKNOWLEDGMENTS soil. Two types of deactivation kinetic model for describing the order of reactant and H2S concentration was used to The authors gratefully acknowledge the National Science predicted the breakthrough curve at various reaction Council, Republic of China, for the financial support under temperatures. Results showed that the activation energy of grant NSC-97-2221-E-041-011. We are grateful to Prof. Hsin the reaction of iron-rich soil and H2S was experimentally Chu, the Department of Environmental Engineering, National calculated about 34 kJ/mol and 131 kJ/mol, respectively Chen Kung University, for providing experimental analyses.

1360 Ko and Hsueh, Aerosol and Air Quality Research, 12: 1355–1361, 2012

723K 653K 598K type I model 723K R2=0.996 type I model 653K R2=0.999 type I model 598K R2=0.997 12x103

10x103

8x103

6x103

4x103

2x103 S (ppm)

2 0

type II model 723K R2=0.996 type II model 653K R2=0.998 type II model 598K R2=0.997 12x103

Outletconcentration of H 10x103

8x103

6x103

4x103

2x103

0 0 5 10 15 20 25 30 35 40 45 50 55 Duration time (minutes) Fig. 5. Regression fittings of the experimental with results predicted by deactivation type I and type II model under various temperatures. Inlet H2S: 1%, H2: 15% and balanced N2, WHSV: 4,000 mL/hr/g with a flow of 100 mL/min.

REFERENCES Cheah, S.F., Olstad, J.L., Jablonski, W.S. and Magrini-Bair, K.A. (2011). Regenerable Manganese-Based Sorbent for Alonso, F. and Palacios, J.M. (2004). Synthesis and Surface Cleanup of Simulated Biomass-Derived Syngas. Energy Properties of Zinc Ferrite Species in Supported Sorbents Fuels 25: 379–387. for Coal Gas Desulphurization. Fuel Process. Technol. Dogu, T. (1981). The Importance of Pore Structure and 86: 191–203. Diffusion in the Kinetics of Gas-Solid Non-Catalytic Balci, S., Dogu, T. and Yuücel, H. (1993). Pyrolysis Kinetics Reactions. Chem. Eng. J. 21: 213–222. of Lignocellulosic Materials. Ind. Eng. Chem. Res. 32: Homma, S., Ogata, S., Koga, J. and Matsumoto, S. (2005). 2573–2579. Gas-Solid Reaction Model for a Shrinking Spherical Bandyopadhyay, S., Chowdhury, R. and Biswas, G.K. (1999). Particle with Unreacted Shrinking Core. Chem. Eng. Sci. Thermal Deactivation Studies of Coconut Shell Pyrolysis. 60: 4971–4980. Can. J. Chem. Eng. 77: 1028–1036. Kar, P. and Evans, J.W. (2008). A Shrinking Core Model

Ko and Hsueh, Aerosol and Air Quality Research, 12: 1355–1361, 2012 1361

for the Electro-Deoxidation of Metal Oxides in Molten (1989). High-Temperature Desulfurization Using Zinc Halide Salts. Electrochim. Acta 53: 5260–5265. Ferrite: Solid Structural Property Changes. Chem. Eng. Kim, K. and Park, N. (2010). Removal of Hydrogen Sulfide Sci. 2: 215–224. from a Steam-Hydrogasifier Product Gas by Zinc Oxide Slimane, R.B. and Abbasian, J. (2000). Regenerable Mixed Sorbent: Effect of Non-Steam Gas Components. J. Ind. Metal Oxide Sorbents for Coal Gas Desulfurization at Eng. Chem. 16: 967–972. Moderate Temperature. Adv. Environ. Res. 4: 147–162. Ko, T.H. (2008). Removal of Hydrogen Sulfur from Coal Suyadal, Y., Erol, M. and Oguz, H. (2000). Deactivation Derived Gas by Iron Oxides in Various Oxisols. Environ. Model for the Adsorption of Trichloroethylene Vapor on Eng. Sci. 25: 969–973. an Activated Carbon Bed. Ind. Eng. Chem. Res. 39: 724– Ko, T.H. (2011). Application of Zn-contaminated Soil: 730. Feasibility Study on the Removal of H2S from Hot Coal Tamhankar, S.S., Hasatani, M. and Wen, C.Y. (1981). Derived Gas. Environ. Chem. Lett. 9: 77–82. Kinetic Studies on the Reactions Involved in the Hot Gas Kopac, T. and Kocabas, S. (2004). Deactivation Models Desulfurization. Chem. Eng. Sci. 36: 1181–1191. for Sulfur Dioxide Adsorption on Silica Gel. Adv. Yasyerli, N., Dogu, T., Dogu, G. and Ar, I. (1996). Environ. Res. 8: 417–424. Deactivation Model for Textural Effects on Kinetics of Kyotani, T., Kawashima, H. and Tomita, A. (1989). High- Gas-Solid Noncatalytic Reactions “Char Gasification with Temperature Desulfurization Reaction with Cu-Containing CO2”. Chem. Eng. Sci. 51: 2523–2528. Sorbents. Environ. Sci. Technol. 23: 218–223. Yasyerli, N., Dogu, G. and Ar, I. (2001). Activities of Kyotani, T., Kawashima, H., Tomita, A., Palmer, A. and Copper Oxide and Cu-V and Cu-Mo Mixed Oxides for Furimsky, E. (1989). Removal of H2S from Hot Gas in H2S Removal in the Presence and Absence of Hydrogen the Presence of Cu-Containing Sorbents. Fuel 68: 74–80. and Predictions of a Deactivation Model. Ind. Eng. Chem. Pineda, M., Fierro, J.L., Palacios, J.M., Cilleruelo, C. and Res. 40: 5206–5214. Ibarra, J.V. (1995). Kinetic Behavior and Reactivity of Yasyerli, S., Ar, I., Dogu, G. and Dogu, T. (2002). Removal Zinc Ferrites for Hot Coal Gas Desulfurization. Ind. Eng. of Hydrogen Sulfide by Clinoptilolite in a Fixed Bed Chem. Res. 30: 6171–6178. Adsorber. Chem. Eng. Process. 41: 785–792. Ranade, P.V. and Harrison, D.P. (1981). The Variable Property Grain Model Applied to the ZnO-H2S Reaction. Chem. Eng. Sci. 36: 1079–1090. Received for review, March 12, 2012 Sa, L.N., Focht, G.D., Ranade, P.V. and Harrison, D.P. Accepted, July 1, 2012