1. Introduction
SO
2 is an important raw material that produces bulk chemicals, and it is also a major air pollutant, which kills lives [
1], erodes metals [
2,
3,
4], and poisons catalysts [
5,
6,
7]. The interaction of SO
2 with metal is drawing researchers’ attention because it plays an important role in the environment and industry [
8].
One of the main sources of SO
2 is the burning of fossil fuels. Desulfurization is a widely used strategy to decrease SO
2 emissions from fossil fuels [
9,
10]. The S Zorb process of SINOPEC Corp., as a typical reactive adsorption desulfurization (RADS), shows its distinguished advantages [
11,
12,
13,
14,
15]. It is carried out in a circulating fluidized bed reactor using a sorbent with Ni and ZnO as the main active phases. In the reactor, sulfur is transferred to the sorbent under a H
2 atmosphere to form ZnS. The sorbent with a particular sulfur loading is then regenerated using air to convert ZnS back to ZnO with SO
2. Although numerous efforts on revealing the chemistry of desulfurization reaction have been made [
16,
17,
18], the behavior of nickel in the SO
2-rich gas during regeneration is still unclear. A deep understanding of the SO
2 reaction on metal surfaces would provide insights that would help develop more efficient catalysts and processes.
Despite the sulfur-removing strategies, SO
2 still remains a question for catalyst scientists. Using density functional theory (DFT), Benjamin D. Gould et al. studied the effect of different sulfides on the Pt of a fuel cell electrode and found that SO
2 had a poisoning ability similar to H
2S and COS [
19]. This is due to the strong adsorption energy of S on Pt’s surface [
20,
21], which is a similar case for sulfides deactivating Ni [
21]. The three-way automotive catalysts could also be poisoned by SO
2 in the tail gas [
22].
Based on those issues, understanding SO
2 adsorption and conversion over a metal surface could shed light on developing new catalysts. A few groups have devoted themselves to studying the adsorption of SO
2 on different transition metal surfaces using first-principles calculations, such as Ni [
23,
24,
25], Cu [
26,
27], Pt [
28,
29,
30,
31], and others [
32]. Xi Lin reported 20 different configurations of SO
2 on the Pt (111) surface and figured out the most energetically stable configurations [
28]. Markus Happel et al. performed a combined experiment and density functional study on the adsorption of SO
2 on clean and oxygen-precovered Pt(111), and also found that there were parallel or perpendicular geometries [
30]. M. J. Ungerer et al. compared the interaction of SO
2 with the Pt’s different surfaces and concluded that the order of adsorption energy was (001) > (011) > (111) [
31]. Yoshiko Sakai et al. used cluster models with 4 and 15 Ni atoms and found that the most stable configuration was a molecular plane nearly parallel to the surface, with all the S and O atoms on bridge sites [
23]. M.J. Harrison et al. used a slab model and concluded that the parallel–hollow geometry was the more preferred adsorption site on Ni [
24]. Xin Wei et al. noticed that the existence of atomic O strengthened the adsorption capacity of SO
2 on a pristine Ni(111) surface [
25].
The conversion of SO
2 was also studied. Chen-Hao Yeh studied the oxidation of S on Pt(111), Ni(111), and the core–shell structures [
33]. They considered four absorbed structures and found that S could be oxidized with either atomic O or O
2 on a Ni surface to form SO, continuing to be oxidized to SO
2 of the parallel bridge orientation. The core–shell structure could change both the adsorption energy and the reaction barrier for SO
2 formation. H. N. Sharma et al. used DFT calculations coupled with microkinetic modeling to examine the oxidation of SO
x to SO
4 on Pd(111) and Pt(111) surfaces, in which three different orientations were considered [
34]. Natasha M. Galea studied removing sulfur from Ni by O
2 using a
p(2 × 2) three-layer model and found that atomic sulfur could be removed up to an initial coverage of 50% at high temperatures [
35].
In this study, we used DFT to investigate the adsorption of SO2 on the Ni(111) surface with different configurations and studied the decomposition pathways of the three most stable ones. The effect of doping metals, namely Cu/Co/Pd/Rh, on whether a second metal would improve the tolerance of SO2 relative to pure Ni was further modeled.
3. Methods
Density functional theory (DFT) was performed on the CASTEP (Cambridge Sequential Total Energy Package) code implemented in Material Studio of Accelrys Inc., Cambridge, UK [
40]. The electron exchange correlation energy was modeled with ultrasoft pseudopotentials and Perdew–Burke–Ernzerh (PBE) exchange–correlation functional based on the generalized gradient approximation (GGA). A wave function energy cutoff of 400 eV was used according to the literature [
35,
41]. A Fermi smearing of 0.1 eV was utilized. The spin polarization calculation was used when Ni was included in the model. The convergence criteria for the structure optimization and energy calculation were set to (a) an SCF tolerance of 1.0 × 10
−6 eV/atom, (b) an energy tolerance of 1.0 × 10
−5 eV/atom, (c) a maximum force tolerance of 0.03 eV/Å, and (d) a maximum displacement tolerance of 1.0 × 10
−3/Å. For the adsorption of O and S on different metals, 2 × 2 supercell of a four-layer slab and the 6 × 6 × 1 k-point Monkhorst-Pack mesh were used. The SO
2 adsorption, transition structures, and products of the reactions over the pristine and doped Ni(111) surface were modeled using periodic 3 × 3 supercell of a four-layer slab, and the 5 × 5 × 1 k-point Monkhorst-Pack mesh was used. For both cases, the slabs were separated with a vacuum spacing of 15 Å to minimize interactions between the slabs. All the metals have a face-centered cubic lattice. The calculated lattice parameters of the pure metals are 3.54 Å for bulk Ni for both 2 × 2 and 3 × 3 supercells, which are in satisfactory agreement with experimental observations of 3.52 Å. The lattice parameters of Co, Cu, Rh, and Pd were 3.46, 3.64, 3.89, and 3.93 Å, respectively. The corresponding experimental values were 3.54, 3.61, 3.80, and 3.89 Å, respectively. All of the calculated results were within 3% of the measured values. During geometry optimization, the bottom two layers were fixed, whereas the top two layers were allowed to relax.
The dissociation pathway was sequential abstraction of O atom from SO
2. The transition states (TSs) were searched using the complete LST/QST method [
42], and the convergence criterion of root-mean-square forces on atoms tolerance of was set to 0.05 eV/Å. The adsorption energy ΔE
ads, reaction energy (ΔE
r), and activation barrier E
a was defined as follows:
where E
A-S is the energy of the slab together with the adsorbate, and E
adsorbate, E
slab, E
FS, E
TS, and E
IS are the total energy of the free adsorbate, bare slab, final state (FS), transition state (TS), and initial state (IS).