Kinetic Study on High-Temperature H2S Removal over Mn-Based Regenerable Sorbent Using Deactivation Model.

The kinetics of high-temperature H2S removal over Mn/Al sorbents prepared by co-precipitation method was investigated in a fixed-bed reactor using a deactivation model. The initial sorption rate constant (k 0), deactivation rate constant (k d), apparent activation energy (E a), and deactivation energy (E d) were obtained. The k 0 and k d values of Mn/Al sorbents are much higher than those of pure Mn2O3. This indicates that Mn/Al sorbents have higher reactivity on the removal of H2S and less diffusion resistance caused by the formation of the sulfided product. The E a and E d values for the sorbent with the Mn content (wt %) of 35.4% are 38.18 and 31.05 kJ/mol, respectively. The deactivation model gives excellent predictions for the H2S breakthrough curves in the sulfidation-regeneration process.

Introduction

The removal of H2S from syngas produced in the gasification of coal, biomass, municipal solid waste, and so forth is an essential step in the processes using syngas as feedstocks or fuels, in which H2S may cause severe corrosion of downstream equipment as well as sulfur oxide emission.[1−3] The currently commercial desulfurization process uses an amine solution to absorb sulfur hydride for the syngas at near-ambient temperature.[4−7] This process leads to substantial thermal efficiency loss on account of the cooling-down and heating-up of hot syngas.

Hot gas desulfurization with metal oxide sorbents has been widely investigated in the past few decades. The sorbents containing zinc,[8,9] ferric,[10,11] or copper[12−14] have been extensively reported for hot gas desulfurization. However, these metal oxides could only work effectively at temperatures of <600 °C because they are prone to be reduced into a metallic state or form metal carbides at higher temperatures.

Mn-based sorbents have been developed, and they showed high sulfur capacity, high mechanical stability, high thermal stability, and fast initial reaction rate for H2S removal at 850 °C which is close to the gasifier temperature.[15−20] To attain a high thermal efficiency, sulfur removal at a high temperature is the most preferable choice. On the other hand, high-temperature desulfurization might lead to additional large savings because the heat exchange equipment is omitted.[21] In our earlier work,[22] Mn/Al sorbents were prepared by a co-precipitation method for 850 °C H2S removal. We found that the used sorbents could be easily regenerated by diluted air or steam. The performance of these sorbents appeared to be stable over multiple cycles, which may meet the requirements of high-temperature desulfurization.

In order to scale up and commercialize hot coal gas desulfurization, the kinetic analysis of H2S removal should be investigated. The removal of H2S with metal oxide sorbents is a typical noncatalytic gas–solid reaction. Many kinetic models have been proposed to describe the kinetics, such as the unreacted shrinking core model (SCM),[23] deactivation model (DM),[24] deactivation kinetic model (DKM),[25,26] and so forth. The SCM assumes that the reaction occurs at a sharp interface between the reacted outer surface and the unreacted interior core. It is suitable for solid sorbents with low porosity. The DM is reported to be successful in predicting the conversion–time data for gas–solid reactions. To modify the DM, Yasyerli et al.[27−29] introduced the concentration dependence of the deactivation term and applied it to describe the removal of H2S over a variety of sorbents, which has excellent predictions for the H2S breakthrough curves. Hong et al.[25,26] hold the view that the DM model is not suitable for all complicated desulfurization reactions because the reaction order of H2S and the sorbent is assumed to be 1. They established the DKM based on the elementary stoichiometric equation of the desulfurization reaction. In earlier works,[23,30] it was found that the reaction order of H2S and Mn-based sorbents is 1, which is in accordance with the assumption of Yasyerli.[27−29]

In this work, we focus on the kinetic behaviors of the Mn/Al sorbents prepared by the co-precipitation method. The activity and regenerability of the Mn/Al sorbents were tested in a fixed-bed reactor, and the sorption rate parameters were evaluated by the analysis of the H2S breakthrough curves using the DM modified by Yasyerli. This work is essential for the scale-up and commercialization of high-temperature desulfurization using a Mn-based regenerable sorbent.

Deactivation Model

In the removal of H2S, significant changes in the pore structure, active surface area, and activity per unit area of the sorbent have been caused by the formation of a dense product layer with the reaction extent. DM has not considered the detailed characteristic parameters of the solid sorbent in such a microscopic way as SCM but in a macroscopic way. In the DM, the effects of the textural variation (pore structure, active surface area, and activity per unit area) of the sorbent and an additional diffusion resistance caused by the formation of the dense product layer were combined in an activity term. The change in the rate of the activity of the sorbent was written aswhere kd is the deactivation rate constant. With the pseudo-steady-state assumption, the species conservation equation for the reactant gas H2S in a fixed-bed reactor was expressed aswhere k0 is the initial reaction rate constant. The following equation for the H2S breakthrough curves was then derived by an iterative procedure.

The rate constants kd and k0 can be evaluated by the regression analysis of the H2S breakthrough curve.

Results and Disscussion

H2S Sorption Results with Mn/Al Sorbents with Different Mn Contents

The H2S breakthrough data over the sorbents with different Mn contents at 850 °C and the breakthrough curves predicted by DM are presented in Figure . In the initial period, the outlet H2S concentration is lower than the detection limits of GC, and the removal ratios of H2S of all sorbents are essentially 100%. Mn oxides could reduce H2S from 1% to <5 ppm at 850 °C. As the Mn content increased, the breakthrough curves shift to longer times, indicating a higher sorption capacity. As shown in Table , the sulfur capacity of the sorbent increases with an increase in the Mn content. The S/Mn molar ratios of all the saturated sorbents are between 0.88 and 0.96, indicating the approximate utilization ratio of the active component.

Figure 1

Experimental data and calculated H2S breakthrough curves for samples with different Mn contents; T = 850 °C; GHSV = 11 942 h–1.

Table 1

Rate Parameters Evaluated from the Breakthrough Data for Samples with Different Mn Contents

samples	1	2	3	4	5
kd (min–1)	0.5579	0.4404	0.4350	0.3962	0.0554
k0 (m3·kg–1·min–1)	2.7363	4.3460	5.3797	6.7697	1.2907
R2	0.9997	0.9998	0.9999	1.0000	0.9881
sulfur capacity (g S/100 g sorbent)	7	15	18	25	36
S/Mn molar ratio	0.88	0.96	0.92	0.92	0.89

For samples 1–4, regression analyses of the kinetic model are of good agreement with the experimental data, and the breakthrough curves are all very sharp. The obtained correlative coefficients (R2) listed in Table are close to 1. For sample 5, pure Mn2O3, the curve has a significant tail in the breakthrough period, and the completion of the breakthrough curve takes quite some time.

The rate parameters evaluated from the kinetic model are listed in Table . For samples 1–4, the deactivation rate constants kd range from 0.5579 to 0.3962 min–1, which are of the same order of magnitude. The initial reaction rate constants k0 increase from 2.7363 to 6.7697 m3·kg–1·min–1 when the Mn content increases from 13.7 to 46.5%. The k0 values obtained with sample 3 and sample 4 are higher than the corresponding values reported with the Zn–Mn (4.36 m3·kg–1·min–1), V–Mn (4.04 m3·kg–1·min–1), and Fe–Mn (3.31 m3·kg–1·min–1) sorbents.[27] The initial reaction rate constant k0 is closely correlated with the diffusion resistance of H2S molecules, and the active sites reacted with H2S. For the supported sorbents prepared by the sol–gel or impregnation method, the rate constants often decline with the increase of active species because the specific surface or active sites on the surface decreased with the incremental loading amount of metal oxide.[25] However, for sorbents prepared by the co-precipitation method in this work, a good dispersion of Mn–Al in bulk phase is achieved. With the increase of Mn content, the diffusion resistance of H2S molecules in Mn–Al sorbents declines and the number of active sites that reacted with H2S rises.

For sample 5, pure Mn2O3, kd is 1 order of magnitude lower than those of samples 1–4, and k0 is the lowest, which indicates the significant increase of diffusion resistance caused by the formation of the sulfided product.

Effect of GHSV on Desulfurization Reaction over Mn/Al Sorbent

A set of sulfidation tests was conducted over sample 3 with different flow rates ranging from 50 to 125 mL/min. The corresponding GHSVs range from 11 942 to 29 855 h–1. Figure shows the experimental data obtained with different GHSVs and the breakthrough curves predicted by DM. It could be seen that the breakthrough curves calculated using the kinetic model are almost identical with the experimental data.

Figure 2

Experimental data and the calculated H2S breakthrough curves for sample 3 at different GHSVs (h–1); T = 850 °C.

The sulfur capacities and the results of regression analysis are shown in Table . The sulfur capacity almost keeps stable at different GHSVs, indicating that external mass-transfer resistances can be neglected within the ranges of 11 942–29 855 h–1 at 850 °C. Both the deactivation rate constant kd and the initial reaction rate constant k0 do not depend on GHSV. The R2 values are all higher than the critical values, which means that all regressions are significant.[31]

Table 2

Sulfur Capacities and Rate Parameters Evaluated from the Breakthrough Data for Sample 3 at Different GHSVs (h–1)

Q × 106 (m3·min–1)	50	75	100	125
GHSV (h–1)	11942	17913	23884	29855
kd (min–1)	0.4350	0.4775	0.5484	0.4668
k0 (m3·kg–1·min–1)	5.3797	5.8013	6.7450	6.1366
R2	0.9999	1.0000	0.9999	1.0000
sulfur capacity (g S/100 g sorbent)	18	18	18	19

As expected, the breakthrough time becomes shorter with increased GHSV, whereas the outlet H2S concentration before breakthrough is lower than the detection limits of GC at all GHSVs, and the sulfur capacity almost maintains constant. From the viewpoint of industrialization, it is favorable for the extensive output to increase GHSV.

Effects of Particle Size on Desulfurization Reaction over Sample 3

To investigate the effects of internal diffusion on H2S removal,[32] another set of experiments was carried out over sample 3 with different particle sizes in the range of 80–200 mesh. Figure shows the data obtained in the desulfurization reaction and the breakthrough curves predicted by DM. The rate parameters evaluated from the kinetic model are listed in Table . In the particle size of 80–200 mesh (75–180 μm), the breakthrough curves coincide well. Further, the rate constants kd and k0 do not show significant changes. The internal diffusion is considered to have little effects in this range of particle size.

Figure 3

Experimental data and the calculated H2S breakthrough curves for sample 3 with different particle sizes; T = 850 °C; GHSV = 11 942 h–1.

Table 3

Rate Parameters Evaluated from the Breakthrough Data for Sample 3 with Different Particle Sizes

mesh number	80–200	80–100	100–120	120–170	170–200
particle sizes (μm)	75–180	150–180	120–150	90–120	75–90
kd (min–1)	0.4350	0.4832	0.4268	0.4245	0.4862
k0 (m3·kg–1·min–1)	5.3797	5.9198	5.2399	5.2602	5.9727
R2	0.9999	0.9999	1.0000	0.9999	0.9999

Effects of Temperature on Desulfurization Reaction and Estimation of Activation Energies

Sample 3 was sulfided at different temperatures ranging from 450–950 °C. Figure shows the experimental data and the calculated H2S breakthrough curves related to the desulfurization reaction temperature. The evaluated rate parameters and sulfur capacities at different temperatures are listed in Table .

Figure 4

Experimental data and calculated H2S breakthrough curves for sample 3 at different temperatures; GHSV = 11 942 h–1.

Table 4

Rate Parameters and Sulfur Capacities Evaluated from the Breakthrough Data for Sample 3 at Different Temperatures

temperature (°C)	450	650	850	950
kd (min–1)	0.0745	0.2676	0.4350	0.6646
k0 (m3·kg–1·min–1)	0.5876	2.8551	5.3797	8.4316
R2	0.9958	0.9985	0.9999	1.0000
sulfur capacity (g S/100 g sorbent)	13	16	18	18

As shown in Figure , the breakthrough time at 450 °C is the shortest, and the sulfur capacity decreases to 13 g S/100 g sorbent. Furthermore, the relative concentrations of H2S before breakthrough at 450 °C are between 0.009 and 0.019 (85–150 ppm), as seen in the magnified patterns of Figure . It is commonly believed that Mn oxides can reduce the amount of H2S to a level below 50 ppm at 400–1000 °C.[21] However, the Mn/Al sorbents prepared by the co-precipitation method have less surface active sites than those prepared by impregnation, which have a negative effect on the desulfurization performance at mid-temperature. With the increase of temperature, O/S exchange penetrates into the bulk phase of the sorbent because of the increase of the solid-state diffusion rate. The desulfurization efficiency and the sulfur capacity improve. At temperatures from 650 to 950 °C, the initial H2S concentrations are lower than the detection limits of GC, and the sulfur capacities increase from 16 to 18 g S/100 g sorbent. Mn/Al sorbents prepared by the co-precipitation method are suitable for high-temperature desulfurization.

As expected, the data listed in Table show that the rate constants increased with the temperature increase. The activation energies of the sorption rate constant and the deactivation rate constant can be calculated by linear regression of the Arrhenius equation. As shown in Figure , the plots of ln k0 and ln kd against 1/T are almost linear. The obtained apparent activation energy (Ea) and deactivation energy (Ed) are 38.18 and 31.05 kJ/mol, respectively. These values are close to those of Cu1Mn9 mixed oxide/SBA-15 sorbents (33.02 and 46.34 kJ/mol) and those of La3Mn97 mixed oxide/KIT-6 sorbents (48.98 and 56.10 kJ/mol) reported by Hong.[26] The obtained Arrhenius formulas of k0 and kd are expressed as followswhere R = 8.314 × 10–3 kJ·mol–1·k–1, and T is the absolute temperature.

Figure 5

Plots of ln k0 and ln kd against 1/T for sample 3; GHSV = 11 942 h–1.

Rate Constants of the Sulfidation–Regeneration Process

Mn/Al sorbents can be regenerated completely with diluted air or steam. In this work, sample 3 was repeatedly sulfided and regenerated at 850 °C for five cycles using diluted air with 10% O2 or using 81% H2O in N2 as the regenerating medium. It was found that DM has a good prediction ability for successive sulfidation.

Figure shows the experimental data and calculated H2S breakthrough curves of successive sulfidations during these tests. The calculated rate constants are listed in Table and the changes of rate constants during the successive sulfidations are shown in Figure . It could be seen that the variation trends of k0 and kd are similar. Using 10% O2 as the regenerating gas, k0 and kd were found to decline during the first three cycles, indicating an initial deactivation of this sorbent. In the next two cycles, k0 and kd are almost stable. The initial deactivation of Mn–Al sorbents has been observed by others for the samples prepared by the wet impregnation method.[33] The main reasons for the initial sorbent decay are proposed to be related to an alumina phase transition, the transformation of the Mn oxides, and the changes in textural properties.

Figure 6

Experimental data and the calculated H2S breakthrough curves in five cycles with (a) 10% O2 or (b) 81% H2O as the regenerating gas; T = 850 °C; GHSV = 11 942 h–1.

Table 5

Rate Parameters Evaluated from the Breakthrough Data for Sample 3 at Successive Sulfidations

cycle number	1	2	3	4	5
kd (min–1)	0.4350	(0.3360)a	(0.1913)	(0.2551)	(0.1987)
		0.4471	0.3981	0.4551	0.4125
k0 (m3·kg–1·min–1)	5.3797	(4.0460)	(2.3290)	(3.0156)	(2.4539)
		5.6526	5.0075	5.5474	4.9599
R2	0.9999	(0.9980)	(0.9889)	(0.9955)	(0.9890)
		0.9947	0.9999	0.9998	0.9998

The data in parentheses are the rate constants for sample 3 regenerated with 10% O2. Others are those for sample 3 regenerated with 81% H2O.

Figure 7

Relationship of the rate constants during five cycles with (a) 10% O2 or (b) 81% H2O as the regenerating gas; T = 850 °C; GHSV = 11 942 h–1.

Using 81% H2O in N2 as the regenerating gas, k0 and kd show no significant reduction. No deactivation was observed after five cycles. The reaction heat of regeneration with O2 and steam at 1100 K is about −530 and −10.19 kJ/mol, respectively.[22] Thus, steam regeneration results in less sintering, which is related to the stabilization of rate constants.

Conclusions

The kinetic behavior for H2S removal over Mn/Al sorbents in a fixed-bed reactor at a high temperature can be evaluated effectively by using the DM. The sorption rate constant (k0), deactivation rate constant (kd), apparent activation energy (Ea), and deactivation energy (Ed) were calculated. The k0 and kd values of the Mn/Al sorbent are much higher than those of pure Mn2O3. The Ea and Ed values for the sorbent with the Mn content (wt %) of 35.4% were 38.18 and 31.05 kJ/mol, respectively. The DM gives a good prediction for the experimental H2S breakthrough data. It can be applied to the kinetic analysis of high-temperature H2S removal over Mn-based sorbents without the requirement of the structural property of sorbents.

Experimental Section

Preparation of Sorbents

Five sorbents with different manganese contents were prepared by the co-precipitation method. 1.27 mol/L Mn(NO3)2 solution and 1 mol/L Al(NO3)3 solution were separately prepared from 50 wt % manganous nitrate solution (AR) and Al(NO3)3·9H2O (AR). For samples 1, 2, 3, and 4, mixed nitrate solutions with different ratios of Mn/Al were neutralized with isovolumetric 10 wt % NH3·H2O. For sample 5, as a reference, pure Mn2O3 was prepared with 1.27 mol/L Mn(NO3)2 solution and isovolumetric 10 wt % NH3·H2O.

During precipitation, the nitrate solution and NH3·H2O were simultaneously added into a reactor with a small amount of water. The pH value was kept in the range of 9–10, and the temperature was 50 °C. The precipitation slurry was aged at 50 °C for 2 h and then filtered and washed with distilled water. The filtration cake was dried, crushed, and sieved to the size of 80–200 mesh and then calcined at 850 °C in air for 6 h.

The Mn contents of the samples 1–5 were analyzed by the ammonium iron(II) sulfate titrimetric method (GB1506-2002-T), which are listed in Table . Each value is the average of three measurements. The absolute difference between the parallel measurements is not more than 1%.

Table 6

Measured Mn Content of Samples 1–5

sample	1	2	3	4	5
Mn content (wt %)	13.7	26.9	35.4	46.5	69.8

Sulfidation–Regeneration Tests

Sulfidation–regeneration tests were carried out in a fixed-bed reactor, which were described in detail in our previous works.[22,34] During the sulfidation stage, the sorbents were sulfided with a gas steam of 1% (15.179 g/m3) H2S in H2 at 450–950 °C. The outlet H2S concentration was measured by an SC2000 gas chromatograph (GC) which was equipped with a thermal conductivity detector and a flame photometric detector. The sulfidation ended when the H2S concentration was close to that of the inlet gas.

The sulfided sorbents were regenerated with diluted air or steam. In the diluted air regeneration, the outlet gas contained SO2 and elemental sulfur. Elemental sulfur was condensed after the reactor. The main product of steam regeneration is H2S. The concentrations of SO2 or H2S were analyzed by GC. The regeneration stages were ended when the concentrations of SO2 or H2S were close to the detection limits of GC, which are 50 and 5 ppm, respectively. The regenerated acceptor was directly used for the sulfidation test in the next cycle.

Each experiment was repeated at least three times in sequence. The average values were reported. The amount of sulfur captured by the sorbents was evaluated by the numerical integration of the breakthrough curves.