Theoretical investigation of H₂S removal on γ-Al₂O₃ surfaces of different hydroxyl coverage

Abstract

The sulfurized processes of H₂S on dehydrated (100) and (110) as well as partially hydrated (110) surfaces of γ-Al₂O₃ were investigated using a periodic density functional theory method. The adsorption configurations of possible intermediates and the potential energy profiles of reaction are depicted. Our results show that H₂S adsorbs preferentially on the Al site along with the S bond, and the adsorption energies are −32.52 and −114.38 kJ mol⁻¹ on the dehydrated (100) and (110) surfaces, respectively. As the reaction temperature of the desulfurization changes, the (110) surface presents different levels of hydroxyl coverage, which affects the adsorption structures of species and reaction energies of dissociation processes. The bonding strengths of H₂S on the partially hydrated (110) surfaces are weaker than on the dehydrated (110) surface. Compared with the 3.0 and 8.9 OH per nm² surfaces, the H₂S has the weakest adsorption energy (−39.85 kJ mol⁻¹) and the highest activation energy (92.06 kJ mol⁻¹) on the 5.9 OH per nm² surface. On the 8.9 OH per nm² surface, the activation energy of the second dissociation step (rate-determining step) for H₂S dissociation is merely 38.32 kJ mol⁻¹. On these involved surfaces, cleavage processes of the two H–S bonds present facile activation energies, which are facilitative to desulfurization.

---

1. Introduction

Hydrogen sulfide (H₂S) is the most common sulfur-containing impurity in combustion of fossil fuels and industrial processes, petroleum/natural gas drilling and refining, coal gasification processes, and biogas generated from anaerobic fermentation of wastes.^1–4 It is a prerequisite to reduce the H₂S content from these streams to a low level as H₂S is extremely malodorous and toxic, being the source of acid rain, causing pipeline corrosion and limiting plant lifetime,^5,6 as well as poisoning most downstream catalysts.⁷

Use of solid metal-oxide sorbents, such as CaO, Fe₂O₃, CuO, ZnO, γ-Al₂O₃ and CeO₂, as candidate desulfurization sorbents has been reported extensively for removal of H₂S.^8–14 Among these metal oxides, γ-alumina (γ-Al₂O₃) has attracted considerable attention because of its highly efficient desulfurization capability.^15,16 γ-Al₂O₃ has been widely used in industrial fields, as a catalyst support or as a catalyst (for the Claus process),¹⁷ with inevitable reaction with H₂S in potential sulfidation environments.^18,19 It would be advantageous to understand the detailed reaction process between γ-Al₂O₃ surface and H₂S. Research on this surface reaction may be helpful to improve performance of desulfurizers. Many attempts have been made to investigate the interaction mechanism of H₂S with γ-Al₂O₃, with adsorption of H₂S on γ-Al₂O₃ surface studied using many instruments.^15,20–24 For example, Travert et al. studied interaction of H₂S on γ-Al₂O₃ surface using infrared (IR) spectroscopy,²⁵ and showed that the adsorption of H₂S irreversibly perturbs the high-frequency region of the IR spectroscopy and results in changes to surface acidity. DeRosset et al.²⁶ estimated that isosteric heats of adsorption of H₂S ranged from −104.6 to −159.0 kJ mol⁻¹, depending on the degree of dehydration of the γ-Al₂O₃. And entropy calculations indicate that mobility of the adsorbed H₂S is highly restricted. Reshetenko et al.¹⁶ carried out heterogeneous decomposition of H₂S on γ-Al₂O₃, and determined the reaction order and effective activation energy to be 2.0 and 72 kJ mol⁻¹, respectively.

Recently, the density functional theory (DFT) method based on quantum chemistry has been proposed as a method for providing some molecular and atomic level information regarding the γ-Al₂O₃ surface, including the positions and behaviors of sulfur components. Arrouvel et al.²⁷ investigated the interaction between H₂S and the γ-Al₂O₃ surface under the usual hydrodesulfurization (HDS) conditions using DFT combined with surface thermochemistry. They pointed out that only rather high temperatures and very low water partial pressure stabilize the sulfidation of the (110) surface, leading to formation of sulfhydryls and hydroxyls. Lo et al.²⁸ also studied adsorption of H₂S on γ-Al₂O₃ surfaces, including both dehydrated and hydrated surfaces. They found that the chemisorption of H₂S on dehydrated surfaces is highly favored, whereas the phenomenon of physisorption is more likely to occur on the hydroxyl surfaces. Furthermore, H₂S adsorption on dehydrated surfaces is more energetically favorable than on hydrated surfaces. Although the adsorption properties of H₂S on γ-Al₂O₃ surfaces have been characterized experimentally and theoretically, a detailed sulfurized mechanism has not been reported to date.

In this study, based on DFT together with a periodic model, we systematically investigate the adsorption energies and geometries of H₂S and resultant species on different γ-Al₂O₃ surfaces, including partially hydrated (110), and dehydrated (100) and (110) surfaces. The reaction processes and potential energy surfaces of H₂S decomposition on the different surfaces are also calculated, which may aid in understanding the reaction state of H₂S under different operating environments and in developing new-style desulfurizers.

2. Computational methods and models

2.1. Calculation methods

All Kohn–Sham DFT calculations were performed using a periodic model and a plane-wave basis set, as implemented in the Vienna ab initio simulation package (VASP).^29,30 The interaction between valence electrons and the core was described by the full-potential projector augmented wave (PAW) method.^31,32 The generalized gradient approximation (GGA) formulation of Perdew, Burke and Enzerhoff (PBE)³³ was employed to treat exchange–correlation energy. Brillouin zone integration was converged with a 3 × 3 × 1 k-point mesh generated by the Monkhorst–Pack algorithm.³⁴ Previous calculations have shown that the 3 × 3 × 1 k-point mesh is sufficient to gain good converged results.^35,36 We studied its reliability (see Table S1 in ESI†) and found that the adsorption energies and activation energy are similar. Therefore, we used the 3 × 3 × 1 k-point mesh. A cutoff energy of 400 eV was sufficient to obtain a satisfactory convergence of the total energy. In the structure optimization and energy calculation, the convergence tolerance was set to 10⁻⁵ eV for electronic self-consistent iteration and the residual forces of free atoms were limited smaller than 0.03 eV Å⁻¹. A Gaussian smearing function with a width of 0.1 eV was utilized to speed up convergence of the total energy. Spin polarization was used in all calculations. The transition states (TSs) and the minimum energy paths (MEPs) were located using the nudged elastic band (NEB) method.^37,38 Eight equally spaced images were linearly interpolated between the reactant and final states. The quasi-Newton algorithm was used to relax the ion positions in all NEB calculations. All reported transition structures were verified to exhibit only one imaginary frequency by frequency calculations.

2.2. Surface models

The crystallographic bulk structure of γ-Al₂O₃ is complex and controversial because of the diversity of metastable phases during the preparation process. As far as we know, three typical γ-Al₂O₃ structures have been reported: the traditional defective spinel structure,³⁹ the Paglia structure,⁴⁰ and the Digne structure.⁴¹ In particular, the Digne structure is widely used in industrial catalysis applications.^42–44 Thus, the Digne structure of γ-Al₂O₃ was chosen for this study. The (110) and (100) surfaces are the main surfaces of γ-Al₂O₃, and contribute to approximately 90% of the overall surface area.⁴⁵ In the medium-high temperature desulfurization environment,^46,47 the two surfaces are shown to have different coverage of surface hydroxyls. According to a report by Digne et al.,⁴⁵ the (100) surface is fully dehydrated above the Claus reaction conditions(∼600 K), and the (110) surface contains 8.9 OH per nm², 5.9 OH per nm², or 3.0 OH per nm² in the temperature range from 600 K to 1150 K. Therefore, the present study focuses on dehydrated (100) and (110) surfaces, and partially hydrated (110) surfaces including 3.0 OH per nm², 5.9 OH per nm², and 8.9 OH per nm² surfaces. The (100) and (110) surfaces were modeled using a p(2 × 1) supercell and a p(1 × 1) supercell, respectively. For the (100) surface, a four-layer slab was employed, where the two surface layers were relaxed and the bottom two layers were kept fixed in their bulk position. For the (110) surface, the slab consisted of five atomic layers, with the bottom two layers frozen to the bulk parameters and the remaining layers allowed to relax. In all calculations, adsorbates were placed on one side of the slab, and a 15 Å thick vacuum spacer was placed in the perpendicular direction to separate the surface slab.

The adsorption energy, E_ad, was defined as

E_ad = E_(ads/slab) − E_(slab) − E_(ads)

where E_(ads/slab) represents the total energy of slab with the adsorbate, E_(slab) represents the energy of the slab, and E_(ads) represents the energy of the free adsorbate. Based on this definition, a negative E_ad value indicates exothermic adsorption, with a greater negative value referring to stronger exothermic interaction between adsorbate and slab.

The reaction energy (ΔE) and the activation energy (E_a) are defined as

ΔE = E_(FS) − E_(R)

E_a = E_(TS) − E_(R)

where E_(FS), E_(R), and E_(TS) are, respectively, the total energies of the final state, of the reactant, and of the transition state in each elementary reaction.

3. Results and discussion

3.1. Bulk γ-Al₂O₃ and gas-phase H₂S and HS in vacuum

In the present study, the bulk structure of γ-Al₂O₃ is monoclinic, and the unit cell contains 8 Al₂O₃ units. The cell volume after structure optimization is 369.13 ų (46.14 ų per Al₂O₃ unit, namely, the cell volume per Al₂O₃ unit), which is about 0.54% smaller than the experimental value (46.39 ų per Al₂O₃ unit).⁴⁸ The corresponding lattice parameters after geometry optimization are a = 5.52 Å, b = 8.33 Å, c = 8.02 Å, and β = 90.62°. The largest deviation of these values is slightly less (1.08%) than the experimental value reported by Krokidis et al.⁴⁹ The relevant results of gas-phase H₂S and HS in vacuum, including bond lengths, bond angles, and vibrational frequencies, are summarized in Table 1. These calculated values are consistent with previous experimental and calculational data.^50–52

Table 1 Geometrical parameters and vibrational frequencies of gas-phase H₂S and HS

	H2S		HS
	Cal^a	Expt^b	Cal^a	Expt^c
a Values in brackets are predicted from ref. 50.b From ref. 51.c From ref. 52.
r(S–H) (Å)	1.349[1.337]	1.328	1.354[1.331]	1.346
θ (deg)	91.6[91.9]	91.6
Vasym (cm^−1)	2663[2673]	2628	2623[2634]	2660
Vsym (cm^−1)	2643[2654]	2615
γbend (cm^−1)	1171[1172]	1183

---

3.2. Adsorption geometries and energies on the γ-Al₂O₃ surfaces

Fig. 1 shows the stable configurations of the dehydrated model γ-Al₂O₃ (100) and (110) surfaces, and the potential adsorption sites, which are used to investigate the interaction with possible adsorbates. For the purposes of discussion, the Al and O atoms in the outmost layer of the slab are labeled I, II, III, IV and A, B, C, D, respectively. On the dehydrated γ-Al₂O₃ (100) surface, Al(I)–Al(III) exposed on the surface are pentacoordinated, whereas Al(IV) is tetracoordinated and in a position below the surface plane which is not available for adsorption.⁵³ All the O atoms are tricoordinated and labeled for the different marks because of different chemical environments. In the case of the dehydrated γ-Al₂O₃ (110) surface, the Al(I) and Al(II) atoms are tetracoordinated but have different chemical environments, and Al(III) is tricoordinated. The O(A) and O(B) atoms are tricoordinated, and the O(C) and O(D) atoms are dicoordinated. The geometries of the partially hydrated (110) surfaces are shown in Fig. 2, including the 3.0, 5.9, and 8.9 OH per nm² surfaces. In particular, the sites on the partially hydrated (110) surfaces share the marks of the dehydrated surface.

Fig. 1 Stable configurations of the dehydrated model γ-Al₂O₃ (100) and (110) surfaces: (a) (100) surface; (b) (110) surface. The I, II, III, and IV labels refer to the Al sites, and A, B, C, and D stand for the O sites (gray, Al; red, O).

Fig. 2 Stable configurations of the model γ-Al₂O₃ (110) surfaces for different hydroxyl coverage: hydrated with θ = 3.0 OH per nm²; hydrated with θ = 5.9 OH per nm²; hydrated with θ = 8.9 OH per nm² (gray, Al; red, O; white, H).

3.2.1. Adsorption of H₂S, HS, S, and H on the dehydrated γ-Al₂O₃ (100) surface. The optimized adsorption structures of H₂S, HS, S and H on the dehydrated γ-Al₂O₃ (100) surface are displayed in Fig. 3, and the corresponding adsorption energies and geometric parameters for all adsorbates are summarized in Table 2. Similar to other metal oxides, H₂S can be adsorbed through the S atom on the metal cation, because of the S lone-pair electrons.^54,55 The H₂S molecule preferentially adsorbs on the Al(III) site, and the plane of the H₂S molecule is nearly parallel to the surface. The bond distance of S–Al(III) is 2.657 Å, and the adsorption energy is −32.52 kJ mol⁻¹, which is in agreement with the result by Ionescu et al. (−37 kJ mol⁻¹).⁵⁶ Meanwhile, the H–S bond lengths are stretched from 1.349 Å in the gas phase to 1.354 and 1.353 Å in the adsorbed state. HS is also favorably absorbed on the Al(III) site with an adsorption energy of −96.48 kJ mol⁻¹, and the S–Al(III) distance is 2.425 Å. These values reveal that HS has stronger bonding capability with the surface compared with H₂S. In the case of S adsorption, the most stable configuration is located at the Al(III)–O(D) bridge site, which is similar to S adsorbed on the “Ce–O bridge” site on the CeO₂ surface.⁵⁷ The O–S and Al–S bond lengths are 1.788 and 2.388 Å, and the corresponding adsorption energy is −256.26 kJ mol⁻¹. Regarding H adsorption, the H atom readily adsorbs on the O site which has two stable adsorption structures, see H(a) and H(b) of Fig. 3. The H atoms in H(a) and H(b) adsorb on O(C) and O(D) sites, respectively. The calculated adsorption energies are −143.49 and −143.55 kJ mol⁻¹ on the O(C) and O(D) sites, and the bond lengths are 0.983 and 0.981 Å, respectively.

Fig. 3 Optimized geometric structures of H₂S, HS, S, and H adsorbed on the dehydrated γ-Al₂O₃ (100) surface. Distances are given in Å (gray, Al; red, O; yellow, S; white, H).

Table 2 Calculated adsorption energies and geometric parameters for all adsorbates on the γ-Al₂O₃ surfaces

Species	Param^a	D100^b	D110^c	3.0 OH^e per nm^2	5.9 OH per nm^2	8.9 OH per nm^2
a Param: parameters.b D100, the dehydrated γ-Al2O3 (100) surface.c D110, the dehydrated γ-Al2O3 (110) surface.d FS denotes the final state.e 3.0 OH per nm^2, 5.9 OH per nm^2 and 8.9 OH per nm^2 represent the different levels of hydroxyl coverage for γ-Al2O3 (110) surface.
H2S	Site	Al III	Al III	Al II	Al I	Al I
	dH–S (Å)	1.354/1.353	1.454/1.355	1.380/1.370	1.396/1.352	1.416/1.362
	dS–Al (Å)	2.657	2.417	2.430	2.586	2.540
	∠HSH (deg)	91.0	95.0	90.6	94.7	94.1
	Ead (kJ mol^−1)	−32.52	−114.38	−92.82	−39.85	−67.57
HS	Site	Al III	Al III	Al I–I	Al I–I	Al I
	dH–S (Å)	1.354	1.354	1.352	1.352	1.366
	dS–Al (Å)	2.425	2.284	2.425/2.477	2.405/2.540	2.383
	Ead (kJ mol^−1)	−96.48	−208.75	−154.79	−158.64	−155.06
S	Site	Al III-OD	Al III-OC/Al I-OB	Al I-OB	Al I-OB	OA
	dS–O (Å)	1.788	1.765/1.741	1.739	1.738	1.812
	dS–Al (Å)	2.388	2.263/2.224	2.235	2.223	—
	Ead (kJ mol^−1)	−256.26	−308.75/−318.59	−318.41	−307.44	−258.12
H	Site	OC/OD	OC	OC	OA	OA
	dS–H (Å)	0.983/0.981	1.091	0.979	0.972	0.971
	Ead (kJ mol^−1)	−143.49/−143.55	−340.80	−269.73	−268.15	−273.33
HS + H	Ead (kJ mol^−1)	−587.55(FS1a)^d	−672.14(FS1c)	−640.40(FS1e)	−619.84(FS1g)	−636.32(FS1i)
S + H	Ead (kJ mol^−1)	−652.14(FS2b)	−798.68(FS2d)	−734.67(FS2f)	−712.82(FS2h)	−687.67(FS2j)

---

3.2.2. Adsorption of H₂S, HS, S, and H on the dehydrated γ-Al₂O₃ (110) surface. The optimized adsorption structure of H₂S is displayed in Fig. 4. In this configuration, the S atom of H₂S locates on the Al(III) site, yielding an adsorption energy of −114.38 kJ mol⁻¹. This adsorption energy is lower than that obtained on the (100) surface by 81.86 kJ mol⁻¹, which reflects the stronger interaction between H₂S and the (110) surface. The bond length of S–Al(III) is 2.417 Å, and two H–S bonds of adsorbed H₂S are elongated to 1.454 and 1.355 Å. The change in HS bond lengths can be attributed to induction of coordinatively unsaturated Al cation for S and O anion for H. Similar to H₂S, the preferential adsorption site for HS is the unsaturated Al(III) site, and the adsorption energy is −208.75 kJ mol⁻¹. For S adsorption, the S atom is adsorbed on the Al(III)–O(C) or Al(I)–O(B) site to form two stable configurations [S(a) and S(b)], with adsorption energies of −308.75 and −318.59 kJ mol⁻¹, respectively. The corresponding S–O and S–Al bonds are 1.765 and 2.263 Å, 1.741 and 2.224 Å, respectively. Regarding H adsorption, the optimized adsorption structure is that the H atom sits on the O(C) site with an adsorption energy of −340.80 kJ mol⁻¹ and a bond distance of 1.091 Å. As seen from Table 2, the bonding strengths of these species on the (110) surface are stronger than those on the (100) surface.

Fig. 4 Optimized geometric structures of H₂S, HS, S, and H adsorbed on the dehydrated γ-Al₂O₃ (110) surface. Distances are given in Å (gray, Al; red, O; yellow, S; white, H).

3.2.3. Adsorption of H₂S, HS, S, and H on the partially hydrated γ-Al₂O₃ (110) surfaces. The most stable adsorption configurations for H₂S, HS, S, and H on the partially hydrated surfaces are shown in Fig. 5, and the corresponding adsorption energies and geometric parameters are summarized in Table 2. For H₂S, the H₂S still occupies the Al site with an S bond on the surface. The S–Al distances are 2.430, 2.586, and 2.540 Å, and the adsorption energies are −92.82, –39.85, and −67.57 kJ mol⁻¹ on the 3.0, 5.9, and 8.9 OH per nm² surfaces, respectively. The H–S bond lengths of H₂S on these surfaces are longer than those of the free H₂S, indicating that the H₂S is activated. The calculated values of adsorption energies show that the bonding strength of H₂S on the dehydrated surface is smaller than on the dehydrated surface (−114.38 kJ mol⁻¹). Clearly, surface water has an influence on the chemical environment of intrinsic adsorption sites and changes the stable adsorption configurations of H₂S on these surfaces. It is of note that the adsorption energy on the 8.9 OH per nm² surface is as much as 27.72 kJ mol⁻¹ lower than that on the 5.9 OH per nm² surface (−67.57 vs. −39.85 kJ mol⁻¹), which is consistent with surface multimolecular adsorption of H₂S.⁵⁸

Fig. 5 Optimized geometric structures of H₂S, HS, S, and H adsorbed on the partially hydrated γ-Al₂O₃ (110) surfaces. Distances are given in Å (gray, Al; red, O; yellow, S; white, H).

In terms of HS adsorption on the 3.0 OH per nm² surface, the HS steadily absorbs on two adjacent Al(I) atoms via the bridge bond mode (see Fig. 5 HS), and the corresponding adsorption energy is −154.79 kJ mol⁻¹. Similar to HS adsorption on the 3.0 OH per nm² surface, the S atom of HS is in direct contact with both Al(I) sites on the 5.9 OH per nm² surface. However, the bridge site occupied by the HS favorably absorbs on the single Al(I) site and its H atom lies toward the O(A) site on the 8.9 OH per nm² surface.

For S adsorption on the 3.0 OH per nm² surface, the S atom binds to the surface Al(I) and O(B) atoms via the bridge bond mode forming a stable structure, as shown in Fig. 5, with absorption energy of −318.41 kJ mol⁻¹. Likewise, the S atom that binds with the Al(I)–O(B) bridge site is still stable on the 5.9 OH per nm² surface with adsorption energies of −307.44 kJ mol⁻¹, and the bond lengths of S–Al(I) and S–O(B) are 2.223 and 1.738 Å, respectively. In contrast with the 3.0 OH per nm² and 5.9 OH per nm² surfaces, the S atom preferentially occupies the O(A) site with a S bond on the 8.9 OH per nm² surface with an absorption energy of −258.12 kJ mol⁻¹, which is 57.03 kJ mol⁻¹ lower than that on the Al(I)–O(B) bridge site. The adsorbed surface hydroxyl inhibits S adsorption on the Al(I)–O(B) bridge site.

For H adsorption, the optimal adsorption sites are O(C) on the 3.0 OH per nm² surface, and O(A) on the 5.9 and 8.9 OH per nm² surfaces, as illustrated in Fig. 5, and the corresponding adsorption energies are −269.73, −268.15, and −273.33 kJ mol⁻¹, respectively. These absorption energy values are approximately 70 kJ mol⁻¹ higher compared with the dehydrated surface. This indicates that the bonding strength of the H adsorption is weaker than on the dehydrated surface.

3.3. Reaction mechanism of H₂S/γ-Al₂O₃ interactions

The probable potential energy profiles of H₂S interacting with the γ-Al₂O₃ surfaces are shown in Fig. 6, and the calculated values of reaction energy and activation energy for each elementary step are given. Fig. 6(a) shows the first dehydrogenation process (H₂S → HS + H), and the second dehydrogenation step (HS → H + S) is displayed in Fig. 6(b). The relative structures on this potential energy profiles are depicted in Fig. 7–9, which include final states (FSs) and TSs on the different γ-Al₂O₃ surfaces. In addition, the coadsorption energies of FSs are given in the Table 2. In particular, the reaction of H₂S on hydrated (110) surfaces will never involve desorption of H₂O prior to S–H activation because of the high adsorption energy of H₂O, similar to the CH₄ dissociation on the hydrated γ-Al₂O₃ surfaces.⁴⁴

Fig. 6 Calculated probable potential energy profiles for dissociation of H₂S (a) and HS (b) on dehydrated and partially hydrated γ-Al₂O₃ surfaces. (D100 surface, the dehydrated γ-Al₂O₃ (100) surface; D110 surface, the dehydrated γ-Al₂O₃ (110) surface).

Fig. 7 Calculated transition states (TSs) and corresponding final states (FSs) for dissociation of H₂S and HS on the dehydrated γ-Al₂O₃ (100) surface. Distances are given in Å (gray, Al; red, O; yellow, S; white, H).

Fig. 8 Calculated transition states (TSs) and corresponding final states (FSs) for dissociation of H₂S and HS on the dehydrated γ-Al₂O₃ (110) surface. Distances are given in Å (gray, Al; red, O; yellow, S; white, H).

Fig. 9 Calculated transition states (TSs) and corresponding final states (FSs) for dissociation of H₂S and HS on partially hydrated γ-Al₂O₃ (110) surfaces. Distances are given in Å (gray, Al; red, O; yellow, S; white, H).

3.3.1. Dehydrated γ-Al₂O₃ (100) surface. In the first dissociation step of H₂S, the most stable adsorption configuration of H₂S is selected as the initial state (IS) (see H₂S in Fig. 3). During the dissociation process (H₂S → HS + H), the H–S bond is broken via TS1_a leading to HS + H(FS1_a) with a small activation energy of 13.22 kJ mol⁻¹. In contrast, a reaction barrier of 4.82 kJ mol⁻¹ must be overcome on the CeO₂ (111) surface,⁵⁹ and a much higher energy barrier of 52.40 kJ mol⁻¹ surmounted on the Cu₂O (111) surface.⁶⁰ As presented in Fig. 7, the breaking H–S bond in TS1_a is 1.510 Å, which approximately elongates by 0.156 Å compared with that in the IS. The Al–S bond is shortened from 2.657 Å in IS to 2.462 Å in TS1_a. In FS1_a, the H atom absorbs on the O(C) site with a bond distance of 1.016 Å, and HS occupies the Al(III) site, which is similar to the adsorption mode of single HS. The dehydrogenation process and coadsorption structure (FS1_a) are similar to H₂O dissociation on the γ-Al₂O₃ (100) surface.⁶¹ The adsorption energy of the coadsorption structure is much lower than the sum of the individual fragments on the surface, which has been attributed to charge transfer between the adsorbates through the support, consistent with the results of Christiansen et al.⁶¹ and Huang et al.⁶² In this step, the energy released is close to 17.49 kJ mol⁻¹.

In the second dissociation step, the HS can undergo the dehydrogenation process (HS → H + S) forming H + S(FS2_b) by overcoming an activation energy of 51.60 kJ mol⁻¹ in TS2_b, as shown in Fig. 6(b). This calculated value of activation energy is as much as 38.38 kJ mol⁻¹ higher than that of the first dissociation step. This result indicates that this dissociation step is slightly difficult from the first dissociation step, and the second step is the rate-determining step in the dissociation reaction. By this reaction, the Al–S and H–S bond distances change from 2.425 and 1.354 Å in the HS to 2.322 and 1.511 Å in the TS2_b, finally reaching 2.272 and 1.987 Å in the dissociation state (FS2_b). In FS2_b, the H and S atoms occupy the O(C) and Al(III) sites, respectively, with a coadsorption energy of −652.14 kJ mol⁻¹.

3.3.2. Dehydrated γ-Al₂O₃ (110) surface. In the IS, the H–S bond of H₂S can be completely activated showing elongation (1.454 Å vs. 1.349 Å in the gas-phase; see H₂S in Fig. 4). Thus, the configurations of HS + H (FS1_c) can be formed via TS1_c overcoming a 4.31 kJ mol⁻¹ activation energy. The reaction energy of this step is −20.21 kJ mol⁻¹. In FS1_c, the HS still occupies the Al(III) site and the H atom attaches to the O(B) site with bond distances of 1.028 Å. The breaking H–S and forming H–O bonds are 1.540 and 1.424 Å in TS1_c, respectively. The first dissociation step occurs easily because of the low activation energy.

Further dehydrogenation can take place via TS2_d to form surface atomic S and another H atom. In TS2_d, the H–S bond is stretched to 1.626 Å from 1.354 Å in HS (see Fig. 4). The activation energy is 77.31 kJ mol⁻¹, which increases by 73.00 kJ mol⁻¹ compared with the parameter of TS1_c. After TS2_d, the dissociative H atom attaches to the O(C) site, and the S atom diffuses to the Al(II)–Al(III) bridge site from the Al(III) site, forming the structure of FS2_d which is similar to the dissociative adsorption geometry of HCN on the γ-Al₂O₃ (110) surface.⁶³ By comparing the stable adsorption sites of the species between the coadsorption structure (see Fig. 8 FS2_d) and the single optimal adsorption structure (see Fig. 4 S(a)), it is found that the most stable coadsorption site is not the superposition of every single optimal adsorption site. The detailed structural parameters are depicted in Fig. 8.

3.3.3. Partially hydrated γ-Al₂O₃ (110) surfaces. To further understand the effect of hydrated surfaces on H₂S removal, we also calculated the dissociation of H₂S on 3.0 OH per nm², 5.9 OH per nm², and 8.9 OH per nm² surfaces. The corresponding structures of TSs and FSs along the reaction paths are depicted in Fig. 9.

For H₂S dissociation on the 3.0 OH per nm² surface, based on the most stable adsorption structure of H₂S (see Fig. 5 (3.0 OH per nm² surface) H₂S) in which the H–S bond is elongated (1.380 Å vs. 1.349 Å in the gas-phase) toward the surface O(C) site, the reaction starts from approaching the H to the O(C) site. The H–S bond distances change from 1.380 Å in the IS to 1.543 Å in the TS1_e, finally reaching 1.989 Å in the FS1_e. In FS1_e, the produced O–H bond is 1.049 Å. The calculated activation energy is 6.73 kJ mol⁻¹, which is similar to that of the dehydrated surface (4.31 kJ mol⁻¹). Meanwhile, the reaction energies show little difference between these two surfaces (−10.03 vs. −20.21 kJ mol⁻¹). As depicted in Fig. 6(a), the activation energies of the first dehydrogenation on the 5.9 and 8.9 OH per nm² surfaces are 29.19 and 5.23 kJ mol⁻¹, and the corresponding reaction energies are −42.44 and −31.20 kJ mol⁻¹, respectively. It is clear that this step occurs easily, both kinetically and thermodynamically, on these three surfaces.

Immediately following, the second dissociation step was investigated, and the potential energy profile is depicted in Fig. 6(b). The HS (see Fig. 5 (3.0 OH per nm² surface)) can break the H–S bond to produce FS2_f via transition states TS2_f with activation energy of 62.81 kJ mol⁻¹. In TS2_f, the breaking H–S and forming H–O bonds are 1.485 and 1.506 Å, respectively. After TS2_f, this H–O bond length reduces further and reaches 0.993 Å in FS2_f. This process is 38.93 kJ mol⁻¹ exothermic. In HS dissociation on the 5.9 OH per nm² surface, the H atom is abstracted from HS via TS2_h to form H + S(FS2_h) with an activation energy of 92.06 kJ mol⁻¹ and an exothermicity of 13.22 kJ mol⁻¹. On the 8.9 OH per nm² surface, the calculated activation energy (TS2_j) for H–S bond scission from HS (see Fig. 5 (8.9 OH per nm² surface)) to produce FS2_j is 38.32 kJ mol⁻¹, with a reaction energy of 8.35 kJ mol⁻¹. In TS2_j, the H–S bond distance is 1.773 Å, which is stretched 0.407 Å compared with the one of HS (1.366 Å). Detailed structural parameters are shown in Fig. 9.

From the above, it has been shown that the bonding strengths of sulfur-containing species on the dehydrated (110) surface are stronger than those on the dehydrated (100) surface. As the coverage of surface hydroxyl on the (110) surface, the bonding strengths of H₂S on these surfaces are ranked in the following order: H₂S (adsorbed on D110) > H₂S (adsorbed on 3.0 OH per nm² surface) > H₂S (adsorbed on 8.9 OH per nm² surface) > H₂S (adsorbed on 5.9 OH per nm² surface) > H₂S (adsorbed on D100). The adsorption energies of HS on the three hydrated surfaces are nearly equal, and the bonding strengths are in the order: HS (adsorbed on D110) > HS (adsorbed on 5.9 OH per nm² surface) ≈ HS (adsorbed on 8.9 OH per nm² surface) ≈ HS (adsorbed on 3.0 OH per nm² surface) > HS (adsorbed on D100). Comparing the activation energies of the rate-determining step (HS → H + S), it is found that the value on the 5.9 OH per nm² surface is highest and on the 8.9 OH per nm² surface is smallest. It is noted that the adsorption energy and dissociative activation energy of H₂S have non-linear relationships on these three partially hydrated (110) surfaces. On the 3.0 OH per nm² surface, the energy level of the Al(II) site is stronger than that of Al(I), therefore, the Al(II) site benefits the adsorption and activation of H₂S.^45,64 On the 8.9 OH per nm² and 5.9 OH per nm² surfaces, the adsorption energy difference of H₂S is caused by the hydrogen bonds because of the same adsorption site (Al(I)). The H–S bond lengths of H₂S are 1.362 and 1.416 Å on the 8.9 OH per nm² surface, and the H–S bond lengths of H₂S are 1.352 and 1.396 Å on the 5.9 OH per nm² surface. This shows that the influence of hydrogen bonds on the 8.9 OH per nm² surface is larger than on the 5.9 OH per nm² surface. Therefore, the adsorption stability of H₂S on the 8.9 OH per nm² surface is larger than on the 5.9 OH per nm² surface. The bond lengths of H₂S on the 8.9 OH per nm² surface are longer than on the 5.9 OH per nm² surface, indicating that H₂S dissociation is easier on the 8.9 OH per nm² surface than on the 5.9 OH per nm² surface. The result is similar to CH₄ dissociation on the hydrated γ-Al₂O₃ (110) surfaces.⁴⁴ It seems reasonable that the reaction occurs at 600 K, which is in accordance with the Claus process. The highest activation energy is also only 92.06 kJ mol⁻¹, which is consistent with the result by Bishara et al. (76 kJ mol⁻¹).⁶⁵ Therefore, H₂S can easily dissociate into S species on the involved γ-Al₂O₃ surfaces.

4. Conclusions

In this paper, the interactions of H₂S with γ-Al₂O₃ surfaces, including partially hydrated (110), and dehydrated (100) and (110) surfaces, were investigated using DFT. Possible adsorption structures and reaction pathways for H₂S dissociation were identified. The reported data show that H₂S and HS prefer to adsorb on the Al site, and S and H atoms preferentially locate on the Al–O bridge and O sites, respectively. The bonding strengths of these species on the dehydrated (100) surface are weaker than on the dehydrated (110) surface, which is in good agreement with previous experiments. Because of the surface active sites occupied, the bonding strengths of H₂S on hydrated (110) surfaces are smaller than on the corresponding dehydrated surfaces.

In this dehydrogenation reaction, the activation energy of the second dissociation step becomes higher compared with that of the first step. Therefore, the second step could be the rate-determining step for H₂S dissociation on these surfaces. Comparing the dehydrated (110) surfaces, the presence of surface OH groups has an impact on activation energies and reaction energies in H₂S dissociation reactions. The 8.9 OH per nm² surface creates the lowest activation energy for dissociating H₂S, with a positive effect on desulfurization, which is in accordance with the Claus reaction occurring at 600 K. Apparently reasonable hydroxyl coverage is beneficial to removal of H₂S. Of course, compared with all the activation energies on these involved surfaces, the highest activation energy is merely 92.06 kJ mol⁻¹ on the 5.9 OH per nm² surface. The results show that H₂S decomposition is facile on these involved γ-Al₂O₃ surfaces, both thermodynamically and kinetically. It was also proven that the γ-Al₂O₃ desulfurizer is highly efficient for removal of H₂S.

Acknowledgements

The authors gratefully acknowledge the financial support of this study by the National Natural Science Foundation of China (21406154), Natural Science Foundation of Shanxi (2013021007-5), and Special/Youth Foundation of Taiyuan University of Technology (2012L041 and 2013T092).

References

1. X. Li, Y. Wang, Y. Lei and Z. Gu, RSC Adv., 2012, 2, 2302–2307.
2. S. Vallejos, T. Stoycheva, F. E. Annanouch, E. Llobet, P. Umek, E. Figueras, C. Canè, I. Gràcia and C. Blackman, RSC Adv., 2014, 4, 1489–1495.
3. Y. S. Hong, Z. Zhang, Z. Cai, X. Zhao and B. Liu, Energy Fuels, 2014, 28, 6012–6018.
4. J. J. Chen, W. W. Li, H. Q. Yu and X. L. Li, AIChE J., 2013, 59, 3824–3833.
5. R. He, F.-F. Xia, J. Wang, C.-L. Pan and C.-R. Fang, J. Hazard. Mater., 2011, 186, 773–778.
6. K. Guo, J. Wen, Y. Zhao, Y. Wang, Z. Zhang, Z. Li and Z. Qian, Environ. Sci. Technol., 2014, 48, 6844–6849.
7. J. Butt, Activation, deactivation, and poisoning of catalysts, Elsevier, 2012.
8. M. Husmann, C. Hochenauer, X. Meng, W. d. Jong and T. Kienberger, Energy Fuels, 2014, 28, 2523–2534.
9. X. Ren, L. Chang, F. Li and K. Xie, Fuel, 2010, 89, 883–887.
10. F. Yazdanbakhsh, M. Bläsing, J. A. Sawada, S. Rezaei, M. Müller, S. Baumann and S. M. Kuznicki, Ind. Eng. Chem. Res., 2014, 53, 11734–11739.
11. L. Neveux, D. Chiche, J. Perez-Pellitero, L. Favergeon, A.-S. Gay and M. Pijolat, Phys. Chem. Chem. Phys., 2013, 15, 1532–1545.
12. G. Buelna and Y. Lin, Sep. Purif. Technol., 2004, 39, 167–179.
13. C. Apesteguia, S. Trevizan, T. Garetto, J. P. de los Reyes and J. Parera, React. Kinet. Catal. Lett., 1982, 20, 1–6.
14. B. Guo, L. Chang and K. Xie, Ind. Eng. Chem. Res., 2014, 53, 8874–8880.
15. O. Saur, T. Chevreau, J. Lamotte, J. Travert and J.-C. Lavalley, J. Chem. Soc., Faraday Trans. 1, 1981, 77, 427–437.
16. T. Reshetenko, S. Khairulin, Z. Ismagilov and V. Kuznetsov, Int. J. Hydrogen Energy, 2002, 27, 387–394.
17. W. C. Content, Petroleum Technology, John Wiley & Sons, Inc, New Jersey, 2007.
18. A. R. Ferreira, M. J. Martins, E. Konstantinova, R. B. Capaz, W. F. Souza, S. S. X. Chiaro and A. A. Leitao, J. Solid State Chem., 2011, 184, 1105–1111.
19. P. S. Prasad, J. W. Bae, S.-H. Kang, Y.-J. Lee and K.-W. Jun, Fuel Process. Technol., 2008, 89, 1281–1286.
20. I. Desyatov, E. Paukshtis and A. Mashkina, React. Kinet. Catal. Lett., 1990, 41, 85–88.
21. A. Datta and R. G. Cavell, J. Phys. Chem., 1985, 89, 450–454.
22. Y. Okamoto, M. Ohhara, A. Maezawa, T. Imanaka and S. Teranishi, J. Phys. Chem., 1986, 90, 2396–2407.
23. R. Glass and R. Ross, J. Phys. Chem., 1973, 77, 2576–2578.
24. T. Slager and C. Amberg, Can. J. Chem., 1972, 50, 3416–3423.
25. A. Travert, O. Manoilova, A. Tsyganenko, F. Maugé and J. Lavalley, J. Phys. Chem. B, 2002, 106, 1350–1362.
26. A. DeRosset, C. Finstrom and C. Adams, J. Catal., 1962, 1, 235–243.
27. C. Arrouvel, H. Toulhoat, M. Breysse and P. Raybaud, J. Catal., 2004, 226, 260–272.
28. J. M. Lo, T. Ziegler and P. D. Clark, J. Phys. Chem. C, 2011, 115, 1899–1910.
29. G. Kresse and J. Furthmüller, Phys. Rev. B: Condens. Matter Mater. Phys., 1996, 54, 11169.
30. G. Kresse and J. Furthmüller, Comput. Mater. Sci., 1996, 6, 15–50.
31. P. E. Blöchl, Phys. Rev. B: Condens. Matter Mater. Phys., 1994, 50, 17953.
32. G. Kresse and D. Joubert, Phys. Rev. B: Condens. Matter Mater. Phys., 1999, 59, 1758.
33. J. P. Perdew, K. Burke and M. Ernzerhof, Phys. Rev. Lett., 1996, 77, 3865.
34. H. J. Monkhorst and J. D. Pack, Phys. Rev. B: Condens. Matter Mater. Phys., 1976, 13, 5188.
35. J. Joubert, A. Salameh, V. Krakoviack, F. Delbecq, P. Sautet, C. Copéret and J. M. Basset, J. Phys. Chem. B, 2006, 110, 23944–23950.
36. P. Hirunsit, K. Faungnawakij, S. Namuangruk and C. Luadthong, Appl. Catal., A, 2013, 460, 99–105.
37. G. Henkelman, B. P. Uberuaga and H. Jónsson, J. Chem. Phys., 2000, 113, 9901–9904.
38. G. Mills, H. Jónsson and G. K. Schenter, Surf. Sci., 1995, 324, 305–337.
39. H. Knözinger and P. Ratnasamy, Catal. Rev.: Sci. Eng., 1978, 17, 31–70.
40. G. Paglia, C. Buckley, A. Rohl, B. Hunter, R. Hart, J. Hanna and L. Byrne, Phys. Rev. B: Condens. Matter Mater. Phys., 2003, 68, 144110.
41. M. Digne, P. Sautet, P. Raybaud, P. Euzen and H. Toulhoat, J. Catal., 2002, 211, 1–5.
42. A. Salameh, J. Joubert, A. Baudouin, W. Lukens, F. Delbecq, P. Sautet, J. M. Basset and C. Coperet, Angew. Chem., Int. Ed., 2007, 46, 3870–3873.
43. X. Tan, X. Ren, J. Li and X. Wang, RSC Adv., 2013, 3, 19551–19559.
44. R. Wischert, P. Laurent, C. Copéret, F. o. Delbecq and P. Sautet, J. Am. Chem. Soc., 2012, 134, 14430–14449.
45. M. Digne, P. Sautet, P. Raybaud, P. Euzen and H. Toulhoat, J. Catal., 2004, 226, 54–68.
46. Z. Wang and M. Flytzani-Stephanopoulos, Energy Fuels, 2005, 19, 2089–2097.
47. Y. I. Yoon, M. W. Kim, Y. S. Yoon and S. H. Kim, Chem. Eng. Sci., 2003, 58, 2079–2087.
48. S. Wilson, J. Solid State Chem., 1979, 30, 247–255.
49. X. Krokidis, P. Raybaud, A.-E. Gobichon, B. Rebours, P. Euzen and H. Toulhoat, J. Phys. Chem. B, 2001, 105, 5121–5130.
50. S.-F. Peng and J.-J. Ho, J. Phys. Chem. C, 2010, 114, 19489–19495.
51. G. Herzberg, Molecular spectra and molecular structure, vol. 3: Electronic spectra and electronic structure of polyatomic molecules, Van Nostrand Reinhold, New York, 1966, p. 1.
52. K. Huber and G. Herzberg, Constants of diatomic molecules, vol. IV of Molecular spectra and molecular structure, Van Nostrand Reinhold, New York, 1979.
53. Z. Zuo, W. Huang, P. Han, Z. Gao and Z. Li, Appl. Catal., A, 2011, 408, 130–136.
54. J. A. Rodriguez and A. Maiti, J. Phys. Chem. B, 2000, 104, 3630–3638.
55. X. Chu, Z. Lu, Y. Zhang and Z. Yang, Int. J. Hydrogen Energy, 2013, 38, 8974–8979.
56. A. Ionescu, A. Allouche, J.-P. Aycard, M. Rajzmann and F. Hutschka, J. Phys. Chem. B, 2002, 106, 9359–9366.
57. H.-T. Chen, Y. Choi, M. Liu and M. Lin, J. Phys. Chem. C, 2007, 111, 11117–11122.
58. C. Ren, X. Wang, Y. Miao, L. Yi, X. Jin and Y. Tan, J. Mol. Struct.: THEOCHEM, 2010, 949, 96–100.
59. D. Marrocchelli and B. Yildiz, J. Phys. Chem. C, 2012, 116, 2411–2424.
60. R. Zhang, H. Liu, J. Li, L. Ling and B. Wang, Appl. Surf. Sci., 2012, 258, 9932–9943.
61. M. A. Christiansen, G. Mpourmpakis and D. G. Vlachos, ACS Catal., 2013, 3, 1965–1975.
62. W.-F. Huang, H.-T. Chen and M. Lin, J. Phys. Chem. C, 2009, 113, 20411–20420.
63. S. Kim, D. C. Sorescu and J. T. Yates, J. Phys. Chem. C, 2007, 111, 5416–5425.
64. G. Feng, C.-F. Huo, C.-M. Deng, L. Huang, Y.-W. Li, J. Wang and H. Jiao, J. Mol. Catal. A: Chem., 2009, 304, 58–64.
65. A. Bishara, O. Salman, N. Khraishi and A. Marafi, Int. J. Hydrogen Energy, 1987, 12, 679–685.

---

Footnote

† Electronic supplementary information (ESI) available: The differences of adsorption energies and activation energy using different k-point mesh. See DOI: 10.1039/c5ra05443e

---