Enhancing the photocatalytic hydrogen production activity of BiVO4 [110] facets using oxygen vacancies

The activity of the hydrogen evolution reaction (HER) during photoelectrochemical (PEC) water-splitting is limited when using BiVO4 with an exposed [110] facet because the conduction band minimum is below the H+/H2O potential. Here, we enhance the photocatalytic hydrogen production activity through introducing an oxygen vacancy. Our first-principles calculations show that the oxygen vacancy can tune the band edge positions of the [110] facet, originating from an induced internal electric field related to geometry distortion and charge rearrangement. Furthermore, the induced electric field favors photogenerated electron–hole separation and the enhancement of atomic activity. More importantly, oxygen-vacancy-induced electronic states can increase the probability of photogenerated electron transitions, thus improving optical absorption. This study indicates that oxygen-defect engineering is an effective method for improving the photocatalytic activity when using PEC technology.


Introduction
Photoelectrochemical (PEC) water-splitting using solar energy to generate hydrogen is considered to be one of the more promising approaches for renewable energy production. 1,2 Since the initial report of a TiO 2 -based photocatalyst, many semiconductors have been investigated, but the photocatalytic efficiencies are still low and far from being practically applicable because of the following problems: (1) weak visible-light adsorption due to wide band gaps; (2) fast electron-hole pair recombination; (3) low carrier mobility; and (4) band edge positions that do not match the water redox potentials. [3][4][5] For an ideal photocatalyst, its valence band maximum (VBM) should be energetically lower than the O 2 /H 2 O potential and its conduction band minimum (CBM) should be higher than the H + /H 2 O potential; in addition, it should be active towards the H 2 evolution reaction (HER) and O 2 evolution reaction (OER). 6,7 Recently, monoclinic clinobisvanite bismuth scheelite (ms-BiVO 4 ) has attracted extensive attention due to its abundance, strong visible-light adsorption (with a direct band gap of 2.4 eV), and high activity for O 2 evolution. For this photocatalyst, the VBM of BiVO 4 is located at ca. 2.4 V vs. RHE, providing a sufficient overpotential for holes to photo-oxidize water. However, the CBM is below the H + /H 2 O potential, and the excited electrons cannot photo-reduce water. 8,9 Additionally, poor carrier transport properties and rapid electron-hole recombination also limit the PEC performance. 10,11 As a consequence, many measures have been taken, such as doping, morphology control, regulating different exposed facets, heterojunction construction, and surface decoration, to enhance its PEC activity. [12][13][14][15] As is known, many examples of faceted BiVO 4 polyhedra have been synthesized, and each exposed facet exhibits different thermodynamic and photocatalytic behavior; photogenerated electrons and holes can be preferentially separated and accumulated on [010] and [110] facets, where the [010] facet favors proton reduction and the [110] facet favors water oxidation. 16,17 Zhao et al. realized a so-called hydrogen farm project via precisely tuning the (110)/(010) facets, achieving an overall solar-to-chemical efficiency of over 1.9% and a solar-to-hydrogen efficiency exceeding 1.8%. If the photocatalytic activity of a single (110) or (010) facet can be enhanced, the efficiency of photocatalytic hydrogen generation at a (110)/(010) facet heterojunction can also be enhanced. In this case, the oxygen reaction on the (110) facet is a complex reaction, thus, it is important to improve the photocatalytic activity of the (110) facet. 18 Vacancy-defect engineering is a feasible method, utilizing electron redistribution and special chemical properties to enhance the photocatalytic activity. [19][20][21] Recently, experimental studies have shown that oxygen vacancies (O vac ) in a crystal structure could greatly improve the photocatalytic activity. For example, Zhao et al. reported the O vac -boosted photocatalytic nitrogen xation of TiO 2 via providing more active sites with electron redistribution and enhanced electron transport. 22 Our group have shown that surface O vac supported charge separation and transfer, thus improving the OER performance of 3D nanoporous BiVO 4 . 23 Although the use of O vac , which widely exist in metal oxides and are important for the PEC performance, is a promising strategy for enhancing the photocatalytic activity, the mechanism explaining the effects of O vac on photocatalytic water-splitting remains contentious and poorly understood. In this work, we adopt the use of an O vac to improve the photocatalytic hydrogen production activity of the BiVO 4 [110] facet and investigate the mechanism via rst-principles calculations. The calculated results show that the O vac not only upshis the band edge positions to satisfy the requirements of PEC water-splitting but it also assists photogenerated electron-hole separation and optical absorption, as a result of the O vac -induced electric eld.

Computational model and methods
All calculations were based on the Vienna ab initio simulation package (VASP) with density functional theory (DFT), 24,25 and the generalized gradient approximation (GGA) of the Perdew-Burke-Ernzerhof (PBE) functional was adopted. 26,27 As shown in Fig. 1(b) and (c), supercells with a size of 11.9 Â 7.3 Â 23.5Å 3 containing 48 atoms were used to model bilayer BiVO 4 with and without oxygen vacancies, and the models were constructed from the [110] surface of optimized bulk monoclinic BiVO 4 with a 12Å vacuum slab along the z-direction (see Fig. 1(a)). A plane wave cutoff energy of 400 eV and a total change in energy of 1.0 Â 10 À5 eV for geometrical optimization were employed, and the maximum stress was less than 0.01 eVÅ À1 . Monkhorst-Pack kpoint grids of 5 Â 5 Â 1 for geometric optimization and 7 Â 7 Â 1 for electronic structure calculations were sampled as the Brillouin zones.

Results and discussion
As shown in Fig. 1(a), bulk BiVO 4 is a layered monoclinic scheelite-phase structure containing BiO 8 dodecahedra and VO 4 tetrahedra, which are linked via Bi 3+ -O 2À -V 5+ connections and stacked along the main [001] axis direction with an interplanar distance of 2.97Å (close to the experimental value of 2.89Å). 28 Bulk BiVO 4 is a semiconductor, and the calculated band gap is 2.25 eV, basically in accordance with the experimental band gap of 2.40 eV. Additionally, the PBE-calculated lattice constants are a ¼ 5.04Å, b ¼ 5.27Å, and c ¼ 11.89Å, which are consistent with the experimental values of a ¼ 5.09Å, b ¼ 5.20Å, and c ¼ 11.70 A, 29 conrming the reliability of the PBE method.
As shown in Fig. 1(b), the BiVO 4 [110] facet is made up of 7-, 5-, and 4-coordinated Bi, 5-and 4-coordinated V, and 2-and 3coordinated O; it retains semiconductor behavior with a band gap of 2.28 eV, and the VBM mainly consists of O 2p while the CBM is primarily composed of Bi 6p, O 2p, and V 3d (see Fig. 2(a)).
We create an O vac near Bi 7c , which is considered as an active site, as can be judged from the adsorption energy of H 2 O molecules given in the ESI. † 30 As shown in Fig. 1(c Table 1). This change in structure will result in a change in the electronic structure of the [110] facet. As shown in Fig. 2(c) , indicating an increase in the electron concentration and the generation of n-type semiconductor behavior. Furthermore, localized states exist in the band gap, coming from hybridization between V 3d and O 2p states neighbouring the O vac , which can also be seen in the partial density analysis in the inset of Fig. 2(c). These O vacinduced electronic states are conducive to electron transition from the VBM to these states, with a gap of 1.81 eV, and subsequently to the CBM, with a gap of 0.17 eV, beneting the optical absorption.
Comparing the partial charge densities of [110] facets with and without an O vac , as shown in Fig. 2(b) and (d), we nd that charge in the VBM region is mainly concentrated on O atoms and charge in the CBM region is located mainly on V atoms for the pure [110] facet (see Fig. 2(b)). There are obvious changes in the [110] facet with the O vac : the charge in the VBM and CBM regions greatly increases because unmatched electrons and dangling bonds are present, favoring the catalytic behavior. Specially, the charge in the CBM region is distributed more on O atoms but still in the sublayer, the charge in the VBM moves to the surface, and the increased distance between the VBM and CBM is benecial for electron-hole separation. Furthermore, we quantitatively analyse the electric dipole moments of BiVO 4 ; an internal electric eld is present with a magnitude of 13.01 D for the [110] facet with the O vac , which is associated with the greater geometry distortion and charge rearrangement and is stronger than that of the pure [110] facet (7.59 D). As is known, an induced electric eld can effectively improve surface charge separation and change the photoelectrochemical impedance spectroscopy and transient absorption spectroscopy responses. For example, Zhang et al. have shown that tantalum doping induced an electric eld in hematite homojunction nanorods, providing additional driving force to signicantly improve charge separation both in the bulk and at the surface. 32 Hussain et al. have shown that an oxygen-vacancy-induced internal electric eld between [BiO] + and [Br] À had the remarkable capacity to assist effective charge separation and move charge to the surface from the bulk. 33 Fig. 3(a) shows the average electrostatic potentials along the z-direction of the BiVO 4 systems. The work function, dened as the difference between the vacuum level and the Fermi level, is 6.04 eV for the pure [110] facet, which is larger than that of the [110] facet with the O vac (5.87 eV), indicating that charge is more easily transferred to the surface due to the existence of the O vac . Based on the electrostatic potential, the band edge energies (e.g., the VBM and CBM) can be obtained via aligning the eigenvalues to the vacuum level. 34 For the pure [110] facet, the VBM is À6.318 eV and the CBM is À4.038 eV, straddling the oxidation potential but not the H + /H 2 O potential; this means there is a lack of driving force for the HER, limiting the photocatalytic hydrogen generation abilities of BiVO 4 , which is in accordance with the experimental results. 35 Compared to the pure [110] facet, the band edge positions of the [110] facet with the O vac are upshied by 0.161 eV; the VBM position is À6.216 eV and the CBM position is À3.877 eV, straddling the water redox region. The upshi mainly comes from the O vacinduced internal electric eld caused by geometry distortion and charge rearrangement. As we know, the total dipole perpendicular to the surface component (m t ) causes the work function change, that is, DW t ¼ m t /A3 0 , 34,36 where A and 3 0 refer to the surface area of the unit cell and dielectric constant, respectively. Here, the dipole density is À0.428 D nm À2 and, therefore, the resultant work function change is À0.161 eV; based on DV t ¼ ÀDW t , the band edge upshis by 0.161 eV, matching with the calculated energy shi based on the mean electrostatic potential. To clearly describe the change in band gap, we display (Ahn) 1/2 as a function of hn in Fig. 4(b). The intercept of a tangent line to the rst peak with the x-axis relates to the band gap. The intercept is 2.28 eV for the pure [110] facet, relating to the band gap. Compared to the pure [110] facet, two peaks appear for the  [110] facet with the O vac : the intercept of the rst peak relates to a band gap of 2.33 eV and the intercept of the second peak gives a band gap of 2.68 eV. The difference between the intercept of the rst peak and the second small peak is 0. 35 Fig. 2(a)). For the [110] facet with the O vac , a peak appears at 350 nm, relating to an energy of 3.54 eV; the corresponding energy in the PDOS is 1.48 eV, as seen in Fig. 2 Fig. S2 †), and create two and three O vac , forming O vac concentrations of 6.25% and 9.38%, as shown in Fig. S3 in the ESI; † 30 the calculated results show that the O vac site and concentration have a great inuence on the electronic structure and optical adsorption, thus affecting the photocatalytic properties.

Conclusions
Based on electronic structure calculations and band edge alignment analysis, we demonstrate that vacancy-defect engineering is a feasible strategy for improving the photocatalytic water-splitting activity of BiVO 4 . To this end, the O vac plays an important role: (1) the O vac excites the activity of neighbouring atoms due to unmatched electrons and dangling bonds; (2) the O vac -induced internal electric eld is conducive to photogenerated electron-hole separation and can tune the band edges; and (3) the O vac -induced local electronic states favor electron transitions and enhance the optical absorption. As a result, the BiVO 4 [110] facet can become a promising photocatalyst for water-splitting owing to the ideal band gap for enhanced optical absorption, the reduced electron-hole recombination, and the suitable band edges for water redox.

Conflicts of interest
There are no conicts to declare.