Simultaneous catalytic oxidation of elemental mercury and arsine over CeO2(111) surface: a density functional theory study

Ceria (CeO2)–based materials are potential catalysts for the removal of the Hg0 and AsH3 present in reducing atmospheres. However, theoretical studies investigating the Hg0 and AsH3 removal capacity of ceria remain limited. In this study, the adsorption behavior and mechanistic pathways for the catalytic oxidation of Hg0 and AsH3 on the CeO2(111) surface, including the calculation of optimized adsorption configurations and energies, were investigated using density functional theory calculations. The results suggest that Hg0 and AsH3 are favorably adsorbed on the CeO2(111) surface, whereas CO is not, which is crucial for selective removal when CO is a desirable gas component. Furthermore, AsH3 is adsorbed more favorably than Hg0. In addition, the calculations revealed that the Hg atom is initially adsorbed on the surface and then oxidized by lattice oxygen to form HgO. Concerning AsH3 decomposition, the stepwise dehydrogenation of AsH3 followed by bonding with lattice O atoms to form the As-O bond seems the most plausible. Finally, the adsorbed As-O bond is further forms elemental As and As2O3. Therefore, CeO2 can adsorb and remove Hg0 and AsH3, making it a promising catalyst for the simultaneous catalytic oxidation of Hg0 and AsH3 in strongly reducing off-gas.


Introduction
Yellow phosphorus is an important raw material. The production of each ton of yellow phosphorus produces about 2500-3000 m 3 of off-gas, and the CO content of this gas is high (80-95%), making it an important C1 chemical raw material. C1 chemical industry refers to the organic chemical industry process that syntheses chemical products or liquid fuels with the material containing one carbon atom as raw material, such as CO, CO 2 , CH 4 , CH 3 OH, and HCHO. In order to improve the quality of subsequent C1 chemical products, it is necessary to carry out deep purification of tail gas. In addition, to CO, pollutants, including H 2 S, elemental mercury (Hg 0 ), and arsine (AsH 3 1 the phosphorus off-gas, which should be purified [1]. Atmospheric heavy metal pollution seriously threatens human health and environmental safety, and the phosphorus industry is a major contributor to this, especially in southwest China. Therefore, laws and regulations have been developed to prevent and control pollution [2]. To comply with these increasingly stringent requirements, several purification and resource utilization technologies have been developed to treat yellow phosphorus off-gas, and the removal of Hg 0 and AsH 3 has received particular attention because of their high toxicity, volatility, and persistence in the environment [3,4]. In addition to the health risks, the presence of Hg 0 and AsH 3 in yellow phosphorus off-gas limits its use as a raw material in other processes [5]. Therefore, Hg 0 and AsH 3 removal technology is urgently needed. Mercury in the gas phase mainly exists in three forms: particulate-bound mercury (Hg p ), elemental mercury (Hg 0 ), and oxidized mercury (Hg 2+ ) [6]. Hg 2+ is easily soluble in water, and Hg p remains in the atmosphere for an extremely short time. Therefore, both of these species can be captured by electrostatic precipitators, fabric filters, and wet scrubbers [7][8][9]. However, Hg 0 is difficult to dissolve in water or adsorb on solid surfaces because of its high vapor pressure and low water solubility [10]. In contrast, arsenic can be dispersed in the gas phase as oxidized (As 2 O 5 , As 2 O 3 , and arsenate) and reduced (AsH 3 ) species [5]. In yellow phosphorus off-gas, which is a highly reducing mixture, mercury mainly exists as Hg 0 , whereas arsenic mainly exists as AsH 3 [11]. Therefore, Hg 0 and AsH 3 removal has become a research focus.
Currently, the main methods of Hg 0 and AsH 3 removal involve adsorption and catalytic oxidation [12]. In particular, noble metal catalysts, such as Au [13], Ag [14], Pd [15,16], Pt [17], Ru [18], and Ir [19], have excellent catalytic activity. These materials have, thus, been used as adsorbents and catalysts to remove Hg 0 and AsH 3 . However, the high cost of noble metals has limited their commercialization and application. Fortunately, noble metal-modified carbon materials, which are significantly cheaper than pure noble metals, also show catalytic effects [19,20]. In addition, single or binary metal oxides, especially transition metal oxides, other modified carbon materials, zeolites, and molecular sieves, have been reported to show high Hg 0 and AsH 3 removal efficiencies [21][22][23][24]. However, because of the high reactivity of CO, most catalysts will also react with this useful molecule. Furthermore, the use of separate equipment or catalysts to remove Hg 0 and AsH 3 would result in significant capital investment for equipment and increase operating costs and process complexity. Therefore, selective catalysts for Hg 0 and AsH 3 removal from reducing gas mixtures are required.
The key to developing selective catalysts for Hg 0 and AsH 3 removal is finding a material that lowers the energy barriers of the oxidizing reactions. Ceria is a promising catalytic oxidation material with low cost, no toxicity, and a large oxygen storage capacity. Crucially, CeO 2 contains the Ce 3+ /Ce 4+ redox couple, which can vary from CeO 2 to Ce 2 O 3 and vice versa under oxidizing and reducing atmospheres, respectively [25,26]. To date, there have been many studies on the application of ceria catalysts. Li et al. and Fan et al. showed that CeO 2 has superior catalytic performance for the oxidation of Hg 0 , having 90% oxidation efficiency in simulated flue gas [25,27]. Xie et al. [28] also used ceria to achieve a high AsH 3 removal efficiency. Zhao et al. reported the modification of a commercial selective catalytic oxidation (SCR) catalyst with a series of metal oxides and found that the CeO 2 -modified SCR catalyst showed the highest Hg 0 oxidation capacity [29]. In addition, in our previous work, we found that CeO 2 can oxidize Hg 0 and AsH 3 in a reducing atmosphere [5,28]. Specifically, the removal efficiency was better than 90% at a low reaction temperature of 150 °C. X-ray photoelectron spectroscopy (XPS) measurements showed that Hg 0 and AsH 3 are converted to HgO, elemental As, and As 2 O 3 on the catalyst surface, suggesting that Hg 0 and AsH 3 are oxidized to these species by the lattice oxygen on the surface of the catalyst, and the mechanism of heterogeneous Hg 0 and AsH 3 oxidation should follow the Mars-Maessen mechanism [5].
Nevertheless, few theoretical investigations of the mechanism of Hg 0 and AsH 3 oxidation over ceria have been carried out, and, to date, there have been no reports of quantum chemistry studies concerning the reaction pathway over the CeO 2 (111) surface. In this study, based on our previous work, we systematically and comprehensively studied the adsorption of Hg, CO, and AsH 3 on the CeO 2 (111) surface using density functional theory (DFT) to assess the feasibility of the simultaneous catalytic oxidation of Hg 0 and AsH 3 over CeO 2 for the treatment of yellow phosphorus off-gas. As a result, the reaction pathways and energy barriers were studied and assessed. The objective of this work is to provide theoretical guidance for the development of catalysts for the simultaneous oxidation of Hg 0 and AsH 3 in reducing gas mixtures.

Computational methods
All calculations in this work were carried out by the Dmol3 program included in the Materials Studio based on density function theory (DFT) [30,31]. The exchange-correction potential was modeled using the generalized gradient approximation (GGA) type Perdew-Burke-Ernzerhof (PBE) functional [32,33]. The inner and outer electrons were described by the projector augmented wave (PAW) function [34]. The energy of the system was minimized without any symmetry constraints.
A 10-Å-thick vacuum layer was used to prevent mirror interactions between ceria layers in the supercell, and cutoff energy of 450 eV was used throughout the calculations. Furthermore, a 3 × 3 × 1 Monkhorst Pack k-point grid was used during geometry optimization. The width for Gaussian smearing was 0.2 eV, and the geometric convergence criteria for the energy change tolerance and maximum force tolerance were 10 −5 eV/atom and 0.05 eV/Å, respectively. These values were chosen to enable accurate but practical calculations because of the complex surface structure. Transition states were confirmed by the presence of a single imaginary frequency.
The adsorption energy (E ads ) describes the adsorption strength between the catalyst surface and corresponding gases, as given by Eq. (1) [35,36].
Here, E total , E 1 , and E 2 represent the adsorption energies of the whole adsorption system on the CeO 2 (111) surface, pure catalyst, and isolated free molecules, respectively. If the value of E ads is negative, the adsorption process is favorable, and when the absolute value of E ads exceeds 0.5 eV, the adsorption process results in stable chemical adsorption [37].

Atomic model of CeO 2 (111) surface
CeO 2 has a cubic fluorite crystal structure [38] containing four Ce 4+ ions and eight O 2− ions in the unit cell. Usually, adsorption occurs at the crystal surface rather than in the crystal bulk, so we studied the adsorption behavior on one crystal surface, which is not only more practical, but also reduced the workload. There are three low-index surfaces of crystalline CeO 2 : (111), (110), and (100) [39]. The most stable surface in CeO 2 is the (111) surface, which is formed of a continuous O-Ce-O network interlayer structure. Therefore, this surface was selected as a representative model to study. The CeO 2 supercell structure is shown in Fig. 1a and   Fig. 2. By calculating the adsorption parameters at these four active sites, we can know the adsorption characteristics of pollutants on the catalyst surface, the specific calculation and model are shown in each chapter. A 2 × 2 × 1 supercell structure was used to study the adsorption of Hg 0 , AsH 3 , and CO on the CeO 2 (111) surface.
As mentioned in the "Materials and methods" section, a 10-Å-thick vacuum region was used to eliminate unphysical interactions between adjacent slabs [40].

Hg 0 , AsH 3 , and CO adsorption on the CeO 2 (111) surface
Homogeneous oxidation efficiency is typically low (nearly zero), so we studied the interactions between gases and catalyst surface [41]. Adsorption is the first step in the catalytic site is the most stable site for Hg 0 adsorption. The distance between Hg and Ce in this adsorption mode is 3.241 Å.
Next, we studied the adsorption of AsH 3 on the CeO 2 (111) surface to understand the competitive adsorption of Hg 0 and AsH 3 . The adsorption configurations of AsH 3 on the O top, Ce top, O-O bridge, and Ce-O bridge sites were investigated, and the adsorption energies for these sites were found to be − 1.213, − 1.391, − 1.271, and − 1.313 eV, respectively. The optimized configurations for AsH 3 adsorption on the CeO 2 (111) surface are shown in Fig. 4, and the corresponding adsorption energies and structural parameters are listed in Table 2. The bond length of arsine molecular looks short and results of a sigma-hole bonding interaction [42]. The adsorption of AsH 3 is energetically more favorable than that of Hg 0 , and the most stable binding site for AsH 3 adsorption was the Ce top site. In this configuration, the distance between As and the Ce atom is 2.772 Å.
Finally, the adsorption of CO on the CeO 2 (111) surface was studied (Fig. 5), and the adsorption energies and structural parameters are listed in Table 3  Therefore, in terms of the adsorption energies, on the CeO 2 (111) surface, AsH 3 adsorption is most favorable, followed by Hg 0 adsorption; in contrast, CO adsorption is unfavorable. These results are consistent with our experimental observations that CeO 2 can adsorb and remove Hg 0 and AsH 3 in a strongly reducing atmosphere [5].

Mechanism of Hg 0 catalytic oxidation over the CeO 2 (111) surface
In our previous experimental study, we found that HgO is formed on the catalyst surface, suggesting that lattice oxygen in the catalyst plays an important role in the heterogeneous oxidation process; crucially, the consumed lattice oxygen is replenished by gas-phase oxygen [5]. Therefore, we next discuss the possible Hg 0 catalytic oxidation reaction based on the Mars-Maessen mechanism. The proposed reaction pathway is shown in Fig. 6 for the following steps: reactant → intermediate 1 (IM1) → transition state 1 (TS1) → product. All energy barriers are relative to the reactant. The optimized geometries of the intermediates, transition states, and products in the reaction pathway are shown in Fig. 7. In this model, Hg is initially adsorbed on the surface and then oxidized by the lattice oxygen to form HgO. The TS1 transition state had a single imaginary frequency of − 72.45 cm −1 , and the energy barrier to form HgO  was calculated to be 0.21 eV. Based on the first-principles calculation, we calculate that the activation energy of Hg to generate HgO on the CeO 2 surface is 0.09 eV (T = 0 K). According to Gibbs free energy formula, the free energy of generating HgO is negative at 423 K. When the temperature is 423 K, the reaction can proceed spontaneously under the drive of temperature. This result is consistent with our previous experimental observations.

Mechanism of AsH 3 catalytic oxidation over the CeO 2 (111) surface
Based on our previous studies [5,28], AsH 3 is absorbed on the ceria surface, decomposed, and then, oxidized by lattice oxygen to form elemental As and As 2 O 3 . Furthermore, there are two possible pathways for AsH 3 decomposition on the CeO 2 (111) surface. The combination of AsH 3 with O atoms of the CeO 2 (111) surface followed by dehydrogenation or gradual dehydrogenation and the formation of bonds with O atoms from the CeO 2 (111) surface. Two possible AsH 3 molecule decomposition pathways are proposed: pathway I ( rea cta n t → I M2 → TS2 → IM3 → TS3 → IM 4 → T S4 → IM5 → IM6) and p ath way II (r eacta nt → IM2 → TS5 → IM7 → TS 6 → IM8 → TS7 → IM9 → IM10), as shown in Fig. 8 with relative energy barriers.
In the case of pathway I, the optimized geometries of the intermediates and transition states for AsH 3 decomposition on the CeO 2 (111) surface are shown in Fig. 9. In , which has an energy barrier of 0.47 eV. Finally, the third As-H bond is broken in a spontaneous process (IM5 → IM6) to yield the As-O bond bound to the surface, as well as three H atoms. Thus, one AsH 3 molecule is decomposed by stepwise breakage to one As-O bond on the CeO 2 (111) surface. Figure 10 presents the optimized geometries of the intermediates and transition states for AsH 3 decomposition on CeO 2 (111) surface via pathway II. First, AsH 3 is adsorbed on the surface, and then, the first As-H bond is broken to yield an AsH 2 + H intermediate (IM2 → TS5 → IM7; imaginary frequency: − 34.67 cm −1 ), which has an energy barrier of 0.72 eV. Next, the second As-H bond is broken to form the AsH + 2H intermediate (IM7 → TS6 → IM8; imaginary frequency: − 79.34 cm −1 ), which has an energy Fig. 11 Reaction pathway and energy barriers for the conversion of As-O bonds to products over CeO 2 (111) Fig. 12 Optimized geometries of IM, TS, and products for the conversion of As-O bonds to the final products on CeO 2 (111) surface barrier of 0.93 eV. And then, As + 3H is formed as the third As-H bond is broken (IM8 → TS7 → IM9; imaginary frequency: − 47.92 cm −1 ). This step has an energy barrier of 0.30 eV. Finally, the adsorbed As forms a bond with surface O spontaneously (IM9 → IM10) in a barrierless process, yielding an As-O bond and three H atoms. Comparing the two possible reaction pathways, pathway II has the lowest energy barrier. Hence, the stepwise dehydrogenation of AsH 3 followed by the formation of a bond with an O atom of the CeO 2 (111) surface seems the most plausible mechanism.
The length of the As-O bond in the As-O bond formed from AsH 3 is 1.86 Å, which is similar to that in As 2 O 3 , which may be the final product. Therefore, to determine the final products of the catalytic oxidation of AsH 3 , we carried out calculations on three As-O bonds on the CeO 2 (111) surface. This resulted in the formation of elemental As and As 2 O 3 . The proposed reaction pathway (IM10 → TS8 → IM11 → TS9 → IM12 → product) is shown in Fig. 11 with relative energy barriers. First, two As-O bonds overcome the energy barrier to generate a new As-O bond, to form As-O + As-O-As-O intermediate (IM10 → TS8 → IM11; imaginary frequency: − 74.24 cm −1 ), this process has an energy barrier of 0.18 eV. Next, a new As-O bond is formed to yield an As-O-As-O-As-O intermediate (IM11 → TS9 → IM12; imaginary frequency: − 93.23 cm −1 ) in a process having an energy barrier of 0.67 eV. Finally, a further As-O bond is broken spontaneously, and IM12 is converted to the products: elemental As and As 2 O 3 . The optimized geometries are shown in Fig. 12. The calculation results are consistent with our experimental observations [5]. Thus, overall, AsH 3 is dehydrogenated stepwise and then forms bonds with O on the CeO 2 (111) surface to yield elemental As and As 2 O 3 .

Conclusions
In this work, we systematically studied the adsorption and catalytic oxidation of Hg 0 and AsH 3 on the CeO 2 (111) surface. Comparing the adsorption energies and the adsorption configurations of Hg 0 , AsH 3 , and CO, we found that the most stable binding sites for Hg 0 and AsH 3 are the Ce-top sites, having adsorption energies of − 0.121 and − 1.391 eV, respectively. In addition, the adsorption energy for CO was positive. Thus, AsH 3 adsorption is most favorable, followed by that of Hg 0 , whereas CO adsorption on CeO 2 (111) is unfavorable. This result suggests that CeO 2 could selectively adsorb and remove Hg 0 and AsH 3 in strongly reducing gas mixtures, such as yellow phosphorous off-gas. In addition, the reaction mechanisms were investigated using DFT calculations. The catalytic oxidation of Hg 0 follows the Mars-Maessen mechanism: Hg atoms are first adsorbed on the surface and then oxidized by the lattice oxygen to form HgO. In contrast, AsH 3 is gradually dehydrogenated and forms bonds with an O atom of CeO 2 (111) surface to yield the As-O bond. Finally, the As-O bonds are converted to elemental As and As 2 O 3 . The results of this study provide an understanding of Hg 0 and AsH 3 adsorption and catalytic oxidation on the CeO 2 surface, and we conclude that ceria is a promising catalyst for the removal Hg 0 and AsH 3 from strongly reducing off-gases.