Monoclinic Ga₂O₃ (100) surface as a robust photocatalyst for water-splitting

Abstract

The β-Ga₂O₃ (100) surface, with or without defects, as a robust photocatalyst for water decomposition was studied on the basis of density functional theory (DFT). The surface defects considered, herein, were oxygen vacancies and doping with higher chalcogens, such as S, Se and Te. Narrowed bandgaps of the defective surfaces, leading to a high utilization of solar energy with respect to pure Ga₂O₃, were observed. By optimizing the geometrical structures of the initial molecular adsorption states (IS), the transition states (TS) and the final dissociative adsorption states (FS), the reaction activation energy and the adsorption energy of each species in the reaction pathway were obtained. Water acts as a Lewis base and provides electrons to the surfaces. The presence of water on the surfaces more likely preferred the molecular modes. The reaction results demonstrate that the surface is robust for water decomposition, where the defects, both vacancies and doping with high chalcogens, have no evident influence on the reaction parameters. The reaction pathway can be improved by vacancies or Se doping. These findings for water decomposition on Ga₂O₃ (100) surfaces can be used in synthesis of photocatalysts and for understanding the interactions across the reaction pathway.

---

Introduction

Hydrogen is considered the ultimate clean energy source for solving the global energy and environmental issues. There are several ways of producing hydrogen, such as from fossil fuels, natural gas by steam reforming,^1–3 electrolysis of water,⁴ reforming of biomass,⁵ and photocatalytic water-splitting.^6–8 Producing hydrogen from water using natural energy, such as sunlight, is considered one of the most promising technologies for solving these problems. Semiconductors with a d¹⁰ electron configuration have attracted considerable attention for superior photocatalytic activities, mainly due to their conduction bands being formed by hybridized sp orbits with a large dispersion, which makes them able to generate photoexcited electrons with large mobility.^9–11 As a representative of such d¹⁰ materials, gallium oxide (Ga₂O₃) exhibits high activity in water-splitting and in the degradation of organic pollutants. Among the five polymorphs of gallium oxide,¹² the β phase with a monoclinic structure was reported to be commonly formed under ordinary conditions. This material has drawn much attention for its potential application in optoelectronic devices, gas sensors, spintronic devices, etc.^13–18 In addition, several theoretical simulations have been conducted to explore its structural,^19,20 electronic,²¹ and optical properties.²²

Recently, β-Ga₂O₃ has also been particularly noted because of its high photocatalytic activity for the water-splitting reaction.²³ The most stable low index surface of β-Ga₂O₃ is the (100)-B type, as has been proven by theoretical calculations and experiments.^20,24 The absorption behaviors of materials such as CO₂,²⁵ CH₄,²⁶ H₂,^27–29 formate,² and CH₃OH,³⁰ on Ga₂O₃ have been investigated using DFT methods, with most of these simulations based on the (100) termination surface. However, the large bandgap of β-Ga₂O₃ (about 4.8 eV) leads to a low utilization of the solar spectrum because the material is only active in the ultraviolet region. Moreover, vacancies and doping on non-metal atoms of the catalysts are feasible considerations to improve the photo-adsorption properties by influencing the electronic structure of the photocatalysts. It was found that N doping can lead to a p-type electrical conductivity³² and enable tunable light emission.³³ The study also showed that the non-metal doping of Ga₂O₃ could provide an optical bandgap in the visible region.³¹

In the published literature about photocatalytic water-splitting, the interactions between water and the surfaces of catalysts have been a very important subject.^14,34–36 Considering photocatalysts based on Ga₂O₃ (100) surfaces, several researchers have paid great attentions to their catalytic efficiency and optical properties through carrying out experiments involving the surfaces.^37–40 Water adsorption on low index α-Ga₂O₃ surfaces, with high activation but low stability, was proven by theoretical simulations.⁴¹ Moreover, for the most stable phase of β-Ga₂O₃, the favorable reaction path for water-splitting on a perfect surface and on a NiO_x-cluster-modified surface were studied, and it was shown that the β-Ga₂O₃ (100) surface after loading NiO_x had a significant enhancement of activity for generating oxygen.^42,43 To the best of our knowledge, there is no report in literature on the impact of the water-splitting reaction with chalcogenide doping. Because the partial decomposition (H₂O → H + OH) of water is easier than the complete decomposition (H₂O → 2H + O), the surface hydroxyl radical is more critical for photocatalysis. Hence, in this study, we only considered the partial decomposition of water.

In this study, we performed DFT calculations to investigate the water adsorption and decomposition on β-Ga₂O₃ (100) surfaces. Introducing a non-metal vacancy and chalcogen doping were considered to ameliorate the solar utilization and reaction path. The main purpose of this study was to comprehend the interactions between water and the photocatalytic surfaces in order to supply theoretical guidance to efficiently synthesize Ga₂O₃ catalysts.

Computational details and models

All the DFT calculations were performed using the Vienna Ab-initio Simulation Package (VASP) code.^44,45 The projector-augmented wave (PAW) method was applied to describe the electron–ion interactions.^46,47 The exchange–correlation potentials were described through the Predew–Burke–Ernzerhof (PBE) function within the generalized gradient approximation (GGA) formalism.⁴⁶ A plane wave basis set with a cutoff energy of 420 eV and a 5 × 3 × 1 k-points grid, determined by the Monkhorst–Pack method⁴⁸ for all the surfaces were found to give good converged results. The atomic structures were relaxed using the conjugate gradient algorithm as implemented in the VASP code until the forces on all the unconstrained atoms were smaller than 0.01 eV Å⁻¹. Dipole correction calculations were performed to compensate for the long-range dipolar interactions with periodic images across the vacuum region. Transition states along the reaction pathway were tested using the CI-NEB (climbing image nudged elastic band) method^49,50 and the number of images inserted between the IS and FS were approximately equal to three times the number obtained by the running script (dist.pl) in the VTST script packages.⁵¹

The formation energies of X-doped defect surfaces, ΔE^X_f, and the adsorption energies of H₂O on all the substrate surfaces, , are defined as follows:

ΔE^X_f = E^Total_X-doped − E^Total_pefect + μ_O − μ_X,

where E^Total_X-doped, E^Total_pefect, and E^Total_Sub represent the total energy of the surfaces with and without dopants, and with and without H₂O absorption, respectively. The quantity μ_X (X = O, S, Se, Te and H₂O) is the chemical potential of the X atom or molecule obtained from the energy calculation for the X atom or molecule in a cubic cell with a side length of 15 Å. According to the above definitions, positive values of the absorption energy correspond to an endothermic process, whereas negative values indicate that the adsorption is thermodynamically favorable.

The average difference in electron density for water adsorption on the β-Ga₂O₃ (100) surfaces along the z direction is defined as follows:

where ρ_water+Surf.(x, y, z), ρ_Surf.(x, y, z) and ρ_water(x, y, z) are the charge density at the (x, y, z) points in the lattice of the water adsorbed surface, the surface and the water molecule, and the decomposition products lattice, respectively. Consequently, positive values indicate an increase in charge density at this position, whereas negative values indicate a decrease in charge density.

The unit cell of β-Ga₂O₃ contains 8 Ga and 12 O atoms with a base centered monoclinic structure (Fig. 1(a)). Gallium atoms can be divided into two categories: Ga₁, which is tetrahedrally coordinated, and Ga₂, which has six O neighbors in a distorted octahedral arrangement. Furthermore, the oxygen atoms can be distinguished by O₁, O₂ and O₃. O₁, shared by one octahedron and two tetrahedron gallium atoms, is not on the same plane of a Ga triangle, whereas O₂, shared by two octahedrons and one tetrahedron, lies in the plane of a Ga triangle. In addition, O₃ has four Ga neighbors tetrahedrally arranged. A vacuum layer of 15 Å between successive slabs along the [001] direction was used to separate the two heterostructures in order to avoid spurious interactions due to the nonlocal nature of the correlation energy.⁵² The perfect (100) surface is shown in Fig. 1(b) and (c). On the pure surface, 4-fold-coordinated gallium (Ga₁), 5-fold-coordinated gallium (Ga₂), 3-fold-coordinated oxygen (O₁) and 3-fold-coordinated (O₃) were exposed, among which Ga₂ and O₃ were unsaturated. In total, hundred and twenty atoms with twenty-four molecular units in the slabs were distributed in six atomic layers. The bottom two layers were fixed in their bulk positions in all the calculations, whereas the top four layers, together with the adsorbents, were allowed to relax.

Fig. 1 The models of the β-Ga₂O₃ unit cell and the perfect (100) surface: (a) the bulk β-Ga₂O₃ unit cell crystal, (b) a cross view of the β-Ga₂O₃ (100) surface, (c) top view of the (100) surface. In all of this text, the color of the gallium (Ga) oxygen (O), sulfur (S), selenium (Se), tellurium (Te) and hydrogen (H) atoms are identified by gray, red, yellow, dark goldenrod, dark orange, and white, respectively.

Results and discussion

1. Electronic properties of the β-Ga₂O₃ (100) surface with an oxygen vacancy or doped with higher chalcogens (S, Se and Te)

β-Ga₂O₃ has a C2/m symmetry with four formula units per crystallographic cell, and is characterized by four lattice parameters, namely a, b, c and β. The optimized lattice parameters for β-Ga₂O₃ are a = 12.26 Å, b = 3.05 Å, c = 5.82 Å, and β = 103.65°, and agree well with experimental values and other DFT values.^23,31,40 The density of states (DOS) of the bulk β-Ga₂O₃, shown in Fig. 3(a), was calculated to obtain the structure of the electronic states for the material. The valance band (VB) was mainly contributed by the oxygen 2p orbits, where the O₃ component was the smallest and O₂ the largest. In the conductive band (CB), the gallium s character was the main component and the 3s orbits of the Ga₂ atoms were slightly larger than those of the Ga₁. The energy bandgap (E_b) in this calculation was 2.29 eV (less than the experimental value, which is within 4.6–4.8 eV), which can be explained by the GGA-PBE pseudopotential chosen in the DFT calculations, which underestimates the bandgap by 2–3 eV, and overestimates the lattice parameters. The agreement with the other DFT values was excellent.

Four types of positions for both vacancies and doping were considered in the initial configurations, which were categorized by the atoms on the surfaces. The most stable structures of the defect surfaces are shown in Fig. 2, which are all located at O₃ sites, not only for vacancies but also for the doping with S, Se and Te. The calculated formation energy of the V_O3 surface is 2.909 eV, which is higher than those reported earlier.^14,15 This may be due to the different choice of the chemical potential of the oxygen atoms in this study compared to the other reports in the literature. A single value of atomic energy was used in this study, which was half the energy of molecular oxygen described in the previous literature. For S-, Se-, and Te-doped surfaces, the formation energies are listed in Table 1. The S-doped surface had the lowest energy and the Te-doped surface had the highest energy because the large radius of the Te atom hinders the doping. There were only a few areas of damage in the construction of the surfaces, which demonstrated that the defects were reasonable. For V_O3 (Fig. 2(a and b)), the most distorted part after optimization was around the Ga₁ site. Because of the relatively large atomic radius of S, Se and Te compared to O (Te 1.44 Å, Se 1.15 Å, S 1.01 Å, O 0.60 Å), the Ga–X (X = S, Se and Te) bond lengths were elongated, leading to the dopants being above the surfaces by 0.75 Å, 0.97 Å and 1.15 Å, respectively.

Fig. 2 Structures of the defective surfaces. (a) Across view of the β-Ga₂O₃ (100) surface with O₃ vacancy (V_O3), (b) top view of V_O3, (c) across view of S-doped β-Ga₂O₃ (100) (S-doped) surface, (d) top view of S-doped surface, (e) across view of Se-doped β-Ga₂O₃ (100) (Se-doped) surface, (f) top view of Se-doped surface, (g) across view of Te-doped β-Ga₂O₃ (100) (Te-doped) surface, (h) top view of Te-doped surface.

Table 1 Data of the formation energy (in eV), charge transfer (in |e|) and heights (in Å) of the dopants compared to the perfect surface

	ΔE^Xf	eGa1	eGa2	eO1	eO3	echalcogen	ΔeSurf.	Δh
Bulk		−1.837	−1.884	1.240	1.243
Perfect		−1.810	−1.859	1.207	1.213		−7.497
VO3	2.909	−1.046	−1.644	1.194	1.201		−7.429
S-doped	3.307	−1.651	−1.739	1.221	1.217	0.797	−7.514	0.745
Se-doped	4.005	−1.594	−1.697	1.221	1.217	0.630	−7.522	0.968
Te-doped	4.849	−1.515	−1.619	1.222	1.220	0.390	−7.553	1.150

---

The DOS of all the surfaces are shown in Fig. 3. For the perfect (100) surface, the total DOS for the structure was almost the same as for the DOS of the bulk; however, the bandgap was slightly smaller than that for the unit cell, as shown in Fig. 3(b). An extra peak appeared in the DOS for V_O3 (seen in Fig. 3(c)) between CBM and VBM, which is called the “impurity state”. The peak comes from the three dangling bonds left behind after O₃ leaves. The impurity state is mainly contributed by the 4s and 4p orbits of Ga near the defect site, with the Ga₁ atom having the greatest contribution, as indicated in Fig. 3(d). The band gaps for the S-doped surface, Se-doped surface and Te-doped surface were 1.84 eV, 1.66 eV and 1.49 eV, respectively. The VBM edges of the doped systems were shifted upwards, and were mainly composed of S 3p, Se 4p and Te 5p orbits, which are shown in Fig. 3(d), (f) and (j), whereas the CBM remained almost unchanged. This was due to the higher energy of S 3p, Se 4p and Te 5p compared to the O 2p orbits, in which this trend is consistent with the other results.³¹ The narrowing band gap indicates that the utilization ratio of solar energy for water-splitting by this type of photocatalysts, if they could be used, would be significantly improved.

Fig. 3 DOS of the unit cell and (100) surfaces. (a) Bulk, (b) pure (100) surface, (c) V_O3 surface, (d) LDOS of Ga₁ atom, which is the nearest of vacancy on the V_O3 surface, (e) S-doped surface, (f) LDOS of S in the S-doped surface, (g) Se-doped surface, (h) LDOS of Se in the Se-doped surface, (i) Te-doped surface, (j) LDOS of Te in the Te-doped surface.

Charge transfer was calculated by Bader charge analysis^53–55 to further understand the distribution of surface atoms, which are labeled in Table 1. Although the absolute values of the Bader charges are not the exact values for the electrons transferred, the relative values can clearly show us the distribution of charges. In this article, negative values indicate electron loss, whereas positive values represent a gain in electrons. Ga₁, Ga₂, O₁, and O₃, respectively, denote the atom types defined in the previous paragraph. However, for the deficient surfaces, they represent the atom's proximity to the defects, because the symmetries are destroyed by the defects.

The Δe_Surf. in Table 1 is defined as the sum of charge transfers for all the atoms on the surfaces. For all the surfaces, the Δe_Surf. had negative values, which indicates that the surfaces had lost charge and could easily gain electrons. Surfaces with a charge loss or gain will more easily absorb small molecules.⁴⁷ Comparing all the surfaces together, the Δe_Surf. for the V_O3 surface was the lowest (7.429 |e|), whereas the Te-doped surface lost the most charge (7.553 |e|), indicating that the Te-doped surface had the highest adsorption ability.

The charge transfer of Ga₁ and Ga₂ to the V_O3 surface was changed the most because of the bond breaking. Compared with the pure surface, the charge transfer of Ga₁ and Ga₂ was reduced by 0.791 |e| and 0.242 |e|, respectively, leading to the least surface charge loss, with a value of 7.429|e|. When S, Se or Te was doped, the transferred charges of S, Se and Te of the defective surface were all smaller than that for O₃ (0.797–1.213 |e|, 0.630–1.213 |e|, and 0.390–1.213 |e|, respectively), which manifested that the S, Se and Te atoms got fewer electrons than O. Therefore, the strengths of the Lewis bases for S, Se and Te are weaker than that for O and this will lower the bond strength for adsorbing the H atoms from the water molecules.

2. Water adsorption and decomposition on Ga₂O₃ (100) surfaces

For all the surfaces, the Bader charges of Ga atoms had negative values, indicating that these atoms all carried positive charges, which indicates that these sites would become strong Lewis acids and could absorb the H₂O molecules through the O_w.⁵⁶ On the contrary, surface oxygen atoms showing positive values, which carried negative charges, would be strong Lewis bases for adsorbing the H atoms of water molecules. Ga₂ lost more electrons than Ga₁, whereas O₃ gained more electrons than O₁, which can be seen in Table 1, with these trends also being consistent with the other results.²⁰

For the water molecule, the charge transfer of O_w, as shown in Table 2, was 1.142 |e| and for the H atom, it was −0.571 |e|. After the combination of two hydrogen atoms, which respectively lost one electron, the O_w formed two pairs of bonding electrons by combination with two single 2p orbits of the O atoms. The pairs of electrons in the 2s and 2p orbits of the O_w are called lone pair electrons, which can act as Lewis bases by their loss in a reaction. On the contrary, the two H atoms will act as Lewis acids due to the bonding electron pairs tending toward the O atom, forming empty orbits. Adsorption sites between the H₂O molecule with the Ga₂O₃ (100) surfaces were considered with Lewis acids or bases of both the H₂O molecules and the structures of the surfaces. Seven initial fundamental models are shown in Fig. 4. The water molecule acts as a Lewis base, and the O_w–Ga bond is the main driving force in water adsorption, as shown in Fig. 4(c)–(f). In addition, the molecule acts as a Lewis acid and the H–O bond plays an important role, as shown in Fig. 4(b) and (g). The O_w–Ga bond and H–O bond cooperate (Fig. 4(a)) because H₂O can act as both a Lewis base and a Lewis acid.

Table 2 Data of the adsorption energy (in eV), the structures (lengths in Å and angles in deg.) and charge transfer (in |e|) for the water molecule and its adsorption and decomposition on perfect and defective surfaces

		Eads.	Ga–Ow	Ow–H1	Ow–H2	∠HOH	∠W–S	eGa-ads.	eO-ads.	eOw	eH1	eH2O/HOH
H2O molecule				0.972	0.972	104.477				1.142	−0.571
Perfect	IS	−0.480	2.176	0.982	0.982	104.569	6.704	−1.872	1.214	1.211	−0.639	−0.068
	TS	−0.266	1.939	1.331	0.976	108.927	32.948
	FS	−0.271	1.927	1.377	0.975	110.605	33.006	−1.863	1.235	1.193	−0.638	−0.039
VO3	IS	−0.459	2.184	0.981	0.981	104.691	11.244	−1.877	1.225	1.231	−0.645	−0.059
	TS	−0.277	1.984	1.194	0.978	108.859	31.627
	FS	−0.332	1.919	1.414	0.976	110.018	32.085	−1.850	1.271	1.203	−0.636	−0.038
S-doped	IS	−0.484	2.388	0.975	0.974	105.899	25.843	−1.761	1.226	1.216	−0.641	−0.050
	TS	−0.251	2.211	1.088	0.974	108.055	51.126
	FS	−0.255	1.938	1.549	0.974	110.666	44.463	−1.764	1.297	1.220	−0.660	−0.042
Se-doped	IS	−0.470	2.199	0.981	0.979	105.230	17.755	−1.872	1.235	1.229	−0.628	−0.061
	TS	−0.268	2.188	1.017	0.977	113.546	24.535
	FS	−0.309	1.930	1.369	0.975	110.471	32.273	−1.869	1.272	1.213	−0.640	−0.038
Te-doped	IS	−0.502	2.191	0.986	0.981	104.603	7.061	−1.868	1.232	1.271	−0.679	−0.059
	TS	−0.285	2.003	1.117	0.979	108.410	57.325
	FS	−0.333	1.918	1.429	0.975	111.407	32.602	−1.855	1.281	1.245	−0.640	−0.046

---

Fig. 4 All the possible models of a water molecule adsorbed on the Ga₂O₃ (100) surfaces. (a) Ga–O bridge site, (b) O–O bridge site, (c) Ga–Ga bridge site, (d–f) Ga top sites, (g) O top site.

2.1 Water adsorption and decomposition on a perfect (100) surface. As many as 24 types of initial adsorption structures were modeled according to Fig. 4 and the superficial atomic types. The most stable structure after being fully relaxed is shown in Fig. 5(a) for IS, which was optimized from the structure shown in Fig. 4(d). The O_w atom lies on top of the Ga₂ atom and the two H atoms move toward the O₃. That is accounted for by the Ga₂ sites having a stronger acidity compared to Ga₁. The lengths of O_w–Ga₂ and O_w–H and the angle of H–O_w–H (∠HOH) are given in Table 2. The plane of the water molecule is approximately parallel to the surface, with an angle (∠W–S) of 6.704°. There are good agreements with the other DFT values for water adsorption on the β-Ga₂O₃ (100) surface.⁴³ The adsorption energy of water on the perfect surface was −0.480 eV. Compared with the water molecule, the bond length of O_w–H is elongated about 0.1 Å and the angle of H–O_w–H is increased by 0.2°. This structural distortion of the adsorbed H₂O compared with its gas-phase structure indicates that the adsorbed H₂O is activated.

Fig. 5 (a) Adsorption configurations and decomposition reaction pathway for water on the perfect Ga₂O₃ (100) surface. (b) Average difference in electron density for water adsorption on the perfect Ga₂O₃ (100) surface along the z direction. The horizontal dashed line with the blue color in (b) denotes the position of the surface.

According to the Lewis acid–base theory, the OH radical will be adsorbed onto the top site of Ga atoms, becoming a surface-adsorbed hydroxyl radical, and another product H atom is absorbed onto the O atoms of the surface.⁴⁶ According to the most stable adsorption structure of adsorption, the possible original structures for the decomposition states could be structured with the OH vertically put on top of Ga_ads., O_w next to Ga₂ and the H atom from the water placed on top of O₁ or O₃ near the Ga₂ atom. The decomposition structure after being fully relaxed is shown in Fig. 5(a) for FS, with an adsorption energy of −0.271 eV. The O_w atom lies on top of the Ga₂ atom and the two H atoms move toward the O₃. The angle between OH + H fragment and the surface is 33.006°. The detailed structural parameters of the water adsorption and decomposition on the intact surfaces are shown in Table 2.

The configuration of the transition state between these two states was also calculated, and is shown in Fig. 5(a) for TS. The structure of TS is similar to the dissociative state, i.e., it belongs to the “late barrier” and its adsorption energy is −0.266 eV. The decomposition pathway of water on the perfect surface is shown in Fig. 5. By the abovementioned results, the decomposition reaction on the perfect Ga₂O₃ (100) surface is shown to be an endothermic reaction with a reaction energy of 0.209 eV. Water molecules need an energy of 0.214 eV from the initial molecular state to accomplish the decomposition reaction, but the reverse reaction barrier is very small, with a value of 0.005 eV. These data demonstrate that the molecular adsorption state is the favorable state for water on the perfect Ga₂O₃ (100) surface. That is to say, the water is mainly in the form of a molecule and it is very difficult to dissociate it on the perfect surface.

The Ga atoms below the O_w atoms were defined as “Ga_ads.”, whereas the O atom on the surfaces connected to the decomposition products of H atoms were called “O_ads.”. H₁ is defined as the H atom which decomposes to form H₂O because of its inequitable position.

The results of the charge transfers are summarized in Table 2. Charge transfers for free water molecules are also given in Table 2 as a reference. In both molecular adsorption and decomposition on the surface, the electrons transfer from the adsorption water to the surface, indicating that H₂O acts as a Lewis base and provides electrons to the surface.

The average difference in charge density for water adsorption on the perfect Ga₂O₃ (100) surface along the z direction, as shown in Fig. 5(b), was calculated to further understand the charge redistribution in this process. For water molecule adsorption, the average charge density between the water and the surface is increased, indicating that a covalent bond is formed. However, the average charge density of the region between the water decomposition products is mainly negative and the electrons significantly concentrate in the position of the O_w atom, which indicates that there is an ionic bonding interaction between the surface and the main decomposition product of OH.

2.2 Water adsorption and decomposition on the oxygen vacancy β-Ga₂O₃ (100) surface. The symmetries of the surfaces are destroyed by the existence of defects. Up to 32 types of initial adsorption structures were modeled for the V_O3 surface. The initial adsorption sites considered, herein, are described in Fig. S1.† We also take into account the adsorption energies for when water adsorbs upon or near the defect sites in Table S1.† The stable structure, with an adsorption energy of −0.459 eV, is shown in Fig. 6 (IS), which also was optimized from the structure shown in Fig. 4(d). The O_w of the H₂O molecule lies on top of the Ga₂ atoms and the two H are placed toward the O₃ atoms. The stable site chosen by water is neither the vacancy nor the farthest position from the vacancy, which is similar to the perfect surfaces. The plane of the water molecule has an angle of 11.244° to the surface. The other structural parameters are shown in Table 2. Compared with the water molecule, the bond length of O_w–H is elongated by about 0.1 Å and the angle of H–O_w–H is increased by 0.02°, indicating an active water molecule.

Fig. 6 (a) Adsorption configurations and decomposition reaction pathway of water on the V_O3 surface. (b) Average difference in electron density for water adsorption on the V_O3 surface along the z direction.

The dissociative adsorption state is shown in Fig. 6(a) for FS, in which one of the decomposition products of OH come to the bridge between Ga_ads. and O_ads. with H moving toward O_ads., and another product is an H atom adsorbed onto the O₃ on the surface to form another surface-terminated OH radical. The adsorption energy of the dissociative state was −0.032 eV and the angle between the plane of the OH + H fragment and the surface was 32.085°.

The transition state, shown in Fig. 6 (TS), is similar to the decomposition state and belongs to the “later transition state”, with an adsorption energy of −0.277 eV. The reaction barrier for the water decomposition reaction was 0.182 eV, whereas its reverse reaction barrier was 0.055 eV. These results demonstrate that it is more reasonable for a water molecule to be adsorbed on the V_O3 surface, but not in the dissociative state. The barrier for the reverse reaction in this pathway is higher than for water dissociation on the perfect surface.

In both the water molecule adsorption and decomposition on the V_O3 surface, the electrons transfer from the adsorption water to the surface, as can be seen in Table 2, indicating that H₂O acts as a Lewis base and provides electrons to the surface. Fig. 6(b) shows the average difference in electron density for water adsorption on the V_O3 surface along the z direction. The molecular adsorption state is similar to water adsorption on the perfect surface, with an increase in the average charge density between the water and the surface, which demonstrates that a covalent bond was formed. The average charge density of the region between the water decomposition product OH and the surface, as shown in Fig. 6(b), has a negative value, but a small positive range (close to 1.25 Å), and the electrons are significantly concentrated around the position of the O_w and upon the surface, which indicates that the interactions between the surface and the main decomposition product of OH are ionic bonding, whereas covalent bonding only plays a very weak role.

2.3 Water adsorption and decomposition on chalcogenide-doped surfaces. When water molecules adsorb on chalcogenide-doped surfaces, the most stable structures are obtained from 33 types of initial adsorption structures and these are shown in Fig. 7(a) for IS, Fig. 8(a) for IS, and Fig. 9(a) for IS, respectively. The initial adsorption sites considered here are modeled by Fig. 4 and S1.† For the S-doped surface, the O_w is almost occupied upon the Ga₂ atom, which is the nearest of the S atoms, and the two H atoms move toward O₁ and O₃, which optimizes the structure seen in Fig. 4(d). However, for Se- and Te-doped surfaces, the position of Ga_ads. is the same as the V_O3 surface, which is neither the nearest nor the furthest from the defect site, and two H atoms move toward two different directions, as shown in Fig. 8(a) for IS and Fig. 9(a) for IS. The high adsorption height for water on the S-doped surface can be explained by the neighbors of the S atoms on the surface, which have bigger radical than O atoms and fewer electrons. Nevertheless, for Se- and Te-doped surfaces, the Ga_ads. moves away from the dopants. The water molecule on the Te-doped surface is the most stable case among the three. The angle between the plane of the water molecular and the chalcogenide-doped surfaces were 25.843°, 17.755°, and 7.061°. With increasing the atomic number of the dopants, the bond lengths of O_w–H are elongated and the angle of H–O_w–H is increased. Therefore, surfaces doped with chalcogenides have a higher activity for water adsorption than a perfect surface.

Fig. 7 (a) Adsorption configurations and decomposition reaction pathway of water on the S-doped surface. (b) Average difference in electron density for water adsorption on the S-doped surface along the z direction.

Fig. 8 (a) Adsorption configurations and decomposition reaction pathway of water on the Se-doped surface. (b) Average difference in electron density for water adsorption on the Se-doped surface along the z direction.

Fig. 9 (a) Adsorption configurations and decomposition reaction pathway of water on the Te-doped surface. (b) Average difference in electron density for water adsorption on the Te-doped surface along the z direction.

The decomposition configurations on the chalcogenide-doped surfaces are shown in Fig. 7(a) for FS, Fig. 8(a) for FS, and Fig. 9(a) for FS. For water composed on the S-doped surface, the H₁ atom connects with O₃ and another following, with O_w pointing to the defect site. When water is decomposed on the Se-doped surface, the H₁ connects with O₁ and another follows, with O_w moving toward another O₁ site near Ga_ads., which connects with O_w. The O_w atom lies on top of the Ga₂ atom and the two H atoms move toward the O₃ in the case of Te doping. Other structural parameters of water decomposition on the chalcogenide-doped surfaces of Ga₂O₃ (100) are exhibited in Table 2.

Fig. 7(a) for TS, Fig. 8(a) for TS, and Fig. 9(a) for TS show the transition states, and also the reaction pathway. All the structures of the TS are similar to the dissociative states, which also belongs to the “late barrier”, and the adsorption energies for the S-, Se-, and Te-doped surfaces are −0.251 eV, −0.268 eV and −0.285 eV, respectively. In line with increasing the atomic number of the dopants, the reaction barriers for the water decomposition reaction respectively are 0.223 eV, 0.202 eV and 0.217 eV. However, the reverse reaction barriers are 0.004 eV, 0.041 eV, and 0.048 eV respectively, which are much smaller than the reaction barriers. Therefore, the water is also mainly in the form of a molecule and it is difficult for it to exist on the chalcogenide-doped surfaces.

The charge transfer for water molecule adsorption and decomposition (Table 2) has negative values, indicating that H₂O acts as a Lewis base and provides electrons to the surfaces. The average difference in electron density for water adsorption on chalcogenide-doped Ga₂O₃ (100) surfaces along the z direction, respectively, can be seen in Fig. 7(b), 8(b) and 9(b). In the case of molecular adsorption, the electrons are mainly gathered between the surfaces and the position of the O_w atoms, indicating that the interactions between the surfaces and the water molecules are mainly covalent bonds. The interactions between the surface and the decomposition fragments are chiefly ionic bonding, with a few covalent bonds, because the average charge density between them has mainly negative values, as was shown before.

Conclusion

In order to clarify the interactions between water and photocatalytic surfaces, a comprehensive DFT study was implemented for water adsorption and decomposition on the β-Ga₂O₃ (100) surfaces with oxygen vacancy and by chalcogen doping. Based on the formation energy, it was easy to obtain the O vacancy and S-doped surfaces, but difficult for Se and Te doping due to their larger atomic radii. For surfaces with an oxygen vacancy and high-chalcogen doping, the bandgap reduces with respect to the perfect surface, denoting that the utilization ratio of the solar energy for water-splitting by this type of photocatalysts, if it could be used, would be significantly improved. H₂O acted as a Lewis base for both adsorption and decomposition. In all cases, the transition states all belonged to the “later transition state”, whereas the adsorption energy of water molecule adsorption was lower than for the decomposition state. Therefore, the present states of water on the surfaces are more likely to favor water molecules. The reaction energies were in order V_O3 < Se-doped < Te-doped < perfect surface < S-doped. The reverse reaction barriers decreased in the order V_O3, Te-doped, Se-doped, perfect and S-doped. Consequently, it was more beneficial for water-splitting when the surface had a vacancy and with Se doping. The defects, both vacancy and chalcogenide-doped, had no evident effect on the reaction parameters. Furthermore, the surface with a robust non-defective-sensitive structure could be used as an anti-poisoning photocatalyst. Our study has important implications for understanding the decomposition of water on β-Ga₂O₃ (100) surfaces with an oxygen vacancy or chalcogenide doping. Furthermore, it can provide theoretical guidance to efficiently synthesize Ga₂O₃ photocatalysts.

Acknowledgements

This study was financially supported by the National Natural Science Foundation of China (Grant no. 61674053, 11504092, U1404109 and 11504334) and by the High Performance Computing Center of Henan Normal University.

References

1. R. H. Crabtree, ChemInform, 2010, 26, 987–1007.
2. M. Calatayud, S. E. Collins, M. A. Baltanás and A. L. Bonivardi, Phys. Chem. Chem. Phys., 2009, 11, 1397.
3. S. U. M. Khan, M. Al-Shahry and W. B. Ingler, Science, 2003, 297, 2243–2245.
4. J. Rossmeisl, Z. Qu, H. Zhu, G. J. Kroes and J. K. Nørskov, J. Electroanal. Chem., 2007, 607, 83–89.
5. R. D. Cortright, R. R. Davda and J. A. Dumesic, Nature, 2002, 418, 964–967.
6. A. L. Linsebigler, G. Lu and J. T. Yates, Chem. Rev., 1995, 95, 735–758.
7. Z. Zou, J. Ye, K. Sayama and H. Arakawa, Nature, 2002, 414, 625–627.
8. K. Nakata and A. Fujishima, J. Photochem. Photobiol., C, 2012, 13, 169–189.
9. H. Kadowaki, et al., Chem. Lett., 2007, 36, 424–425.
10. Z. Lu, et al., Phys. Chem. Chem. Phys., 2014, 16, 12488–12494.
11. J. Sato, N. Saito and H. Nishiyama, J. Photochem. Photobiol., A, 2002, 148, 85–89.
12. H. He, et al., Phys. Rev. B: Condens. Matter Mater. Phys., 2006, 74, 195123.
13. H. Peelaers and C. G. Van de Walle, Phys. Status Solidi B, 2015, 252, 828–832.
14. J. B. Varley, et al., Appl. Phys. Lett., 2010, 97, 142106.
15. T. Zacherle, et al., Phys. Rev. B: Condens. Matter Mater. Phys., 2013, 87, 235206.
16. F. Zhang, et al., J. Cryst. Growth, 2014, 387, 96–100.
17. Y. Chen, et al., J. Mater. Sci.: Mater. Electron., 2015, 26, 3231–3235.
18. W. Seiler, et al., Thin Solid Films, 2010, 589, 556–562.
19. S. Yoshioka, H. Hayashi, A. Kuwabara, F. Oba, K. Matsunaga and I. Tanaka, J. Phys.: Condens. Matter, 2007, 19, 1351–1356.
20. V. M. Bermudez, Chem. Phys., 2006, 323, 193–203.
21. K. Yamaguchi, Solid State Commun., 2004, 131, 739–744.
22. A. B. Rahane, M. D. Deshpande and S. Chakraborty, J. Phys. Chem. A, 2012, 116, 10559–10565.
23. X. Wang, et al., Angew. Chem., Int. Ed., 2012, 51, 13089–13092.
24. T. Liu, et al., J. Mater. Chem. A, 2015, 3, 10309–10319.
25. Y. Pan, C. Liu, D. Mei and Q. Ge, Langmuir, 2010, 26, 5551–5558.
26. D. Kohl, T. Ochs and W. Geyer, Sens. Actuators, B, 1999, 59, 140–145.
27. Y. Yang and P. Zhang, Phys. Lett. A, 2010, 374, 4169–4173.
28. Y. Pan, D. Mei, C. Liu and Q. Ge, J. Phys. Chem. C, 2011, 115, 10140–10146.
29. E. A. Gonzalez, P. V. Jasen, A. Juan, S. E. Collins, M. A. Baltanás and A. L. Bonivardi, Surf. Sci., 2005, 575, 171–180.
30. M. M. Branda, E. C. Sebastián, J. C. Norberto, A. B. Miguel and L. B. Adrian, J. Phys. Chem. B, 2006, 110, 11847–11853.
31. W. Guo, et al., Phys. Chem. Chem. Phys., 2015, 17, 5817–5825.
32. L. Liu, M. Li, D. Yu, J. Zhang, H. Zhang, C. Qian and Z. Yang, Appl. Phys. A, 2010, 98, 831.
33. J. L. F. Da Silva, A. Walsh and A. Wei, Phys. Rev. Lett., 2008, 100, 256401.
34. K. Maeda and K. Domen, J. Phys. Chem. Lett., 2010, 1, 2655–2661.
35. Y. Wang, N. Li, P. Duan, X. Sun, B. Chu and Q. He, J. Nanomater., 2015, 2015, 191793.
36. L. S. Reddy, et al., Nanoscale Res. Lett., 2015, 10, 364.
37. C. Li, et al., Energy Environ. Sci., 2015, 8, 1493–1500.
38. H. Hidaka, Photochem. Photobiol. Sci., 2015, 14, 919–928.
39. X. Chai, et al., Sci. China: Chem., 2015, 58, 532–538.
40. K. Maeda, et al., J. Phys. Chem. B, 2005, 109, 20504–20510.
41. X. Zhou, E. J. M. Hensen, R. A. van Santen and C. Li, Chem.–Eur. J., 2014, 20, 6915.
42. X. Zhou, H. Dong and A. M. Ren, Int. J. Hydrogen Energy, 2016, 41, 5670.
43. X. Zhou, H. Dong and A. M. Ren, Phys. Chem. Chem. Phys., 2016, 18, 11111.
44. G. Kresse and J. Furthmüller, Comput. Mater. Sci., 1996, 6, 15–50.
45. P. E. Blöchl, Phys. Rev. B: Condens. Matter Mater. Phys., 1994, 50, 17953–17979.
46. G. Kresse and D. Joubert, Phys. Rev. B: Condens. Matter Mater. Phys., 1999, 59, 1758–1775.
47. J. P. Perdew, K. Burke and M. Ernzerhof, Phys. Rev. Lett., 1996, 77, 3865–3868.
48. H. J. Monkhorst and J. D. Pack, Phys. Rev. B: Solid State, 1976, 13, 5188–5192.
49. G. Henkelman, B. P. Uberuaga and H. A. Jónsson, J. Chem. Phys., 2000, 113, 9901–9904.
50. G. Henkelman and H. Jónsson, J. Chem. Phys., 2000, 113, 9978–9985.
51. D. Sheppard, P. Xiao, W. Chemelewski, D. Johnson and G. Henkelman, J. Chem. Phys., 2016, 136, 074103.
52. A. Bermudez, F. Jelezko, M. B. Plenio and A. Retzker, Phys. Rev. Lett., 2011, 107, 3745.
53. M. Yu and D. R. Trinkle, J. Chem. Phys., 2010, 134, 498–503.
54. W. Tang, E. Sanville and G. Henkelman, J. Phys.: Condens. Matter, 2009, 21, 84204–84210.
55. E. Sanville, S. D. Kenny, R. Smith and G. Henkelman, J. Comput. Chem., 2007, 28, 899–908.
56. J. Chen, et al., Langmuir, 2013, 29, 7025–7037.

---

Footnote

† Electronic supplementary information (ESI) available. See DOI: 10.1039/c6ra24798a

---