Significant Impact of Exposed Facets on the BiVO4 Material Performance for Photocatalytic Water Splitting Reactions

The impact of the four predominant (010), (110), (001), and (121) exposed facets obtained experimentally for monoclinic BiVO4 on its photocatalytic performance for water splitting reactions is investigated on the basis of the hybrid density functional theory including the spin–orbit coupling. Although their electronic structure is similar, their transport and redox properties reveal anisotropic characters based on the crystal orientation and termination. The particular role of each facet in proton reduction was correlated with the surface Bi coordination number and their geometrical distribution. Our work predicts the (001) facet as the only good candidate for both HER and OER, while the (010) facet is a fitting candidate for OER only. The (110) and (121) surfaces are acceptable candidates only for OER but less potential than (001) and (010). These outcomes will efficiently conduct experimentalists for an attentive design of facet-oriented BiVO4 samples toward improving water oxidation and proton reduction.

U pon the report of Fujishima and Honda on TiO 2 water photolysis, 1 photoelectrochemical (PEC) water splitting has been considered as one of the most attractive ways to generate hydrogen using solar energy by maintaining sustainable green environment. 2−4 In order to reduce cost and achieve high conversion order, considerable attempts have been focused on developing an ideal and active semiconductor photocatalyst for O 2 evolution reaction (OER) and H 2 evolution reaction (HER) simultaneously by controlling its fundamental physicochemical properties. In water splitting reactions, the photogenerated charge carriers are situated at the valence band maximum (VBM) and conduction band minimum (CBM) of the photocatalyst. Hence, the wellpronounced delocalized nature of the orbitals associated with the VBM and CBM electronic states inside the bulk is required to ensure good mobility likelihood throughout the crystal structure of the material to the surface. 5−9 Moreover, suitable VBM/CBM positions lying energetically lower than O 2 /H 2 O and higher than H + /H 2 potentials, respectively, are necessary for the driving force of released holes and excited electrons needed to oxidize water and reduce proton. 10 Several ways have been put in to enhance the photocatalytic properties, either by addressing the problem of sluggish transfer kinetics of holes and electrons through the addition of a cocatalyst on top of the electrode surface 11,12 or by forming a heterojunction. 13,14 Lately, facet engineering has become an effective way to tune the semiconductor photoanode characteristics and thus improving its photocatalytic activity. 15−17 A recent study on facet-engineered WO 3 electrode for photoelectrochemical hydrogen generation from natural seawater has given an average rate of 38.18 μmol h −1 . 18 Among numerous visible-light driven semiconductor photocatalysts, BiVO 4 is the frontier in a wide range of applications. 12 The monoclinic crystalline phase was particularly given the highest activity for photocatalytic OER. 19 This result motivated several researchers to focus on the BiVO 4 monoclinic phase in order to determine its photocatalytic behavior. 11,20 As reported, it has an appropriate VBM position for OER oxidizing water with an onset potential around 0.9 V vs the reversible O 2 /H 2 O potential. 21,22 Its CBM position and flat band potential are impartially negative or just positioned at the H 2 /H 2 O level. As invoked in the literature, 18−20 this material has poor carrier transport and slow kinetics transfer at the material−electrolyte interface, which restrict its PEC performance.
Since bismuth vanadate is a common photoanode utilized for water splitting, it is required to determine the role of each exposed facet in photoredox reactions because different facets of BiVO 4 show different thermodynamic and photocatalytic behaviors. 23 Among the low-Miller-index surfaces, (001) tends to be the preferred growth orientation 24 and has also been shown to be the most important facet for photocatalytic oxidation of water for O 2 generation. 25 It has 16 times higher photocurrent density as reported experimentally than the randomly oriented BiVO 4 photoanode. 24 Another dominant surface is (010), which possesses the lowest surface energy 26 and showed good visible photocatalytic degradation activities. 27 Li and co-workers 28 showed in their experiment that the photogenerated charges could be separated between the (010) and (110) facets, where the (110) facet was used for water oxidation and (010) for proton reduction. Cao and coworkers 29 recently reported a first-principle study on the (101) Bi 2 O 4 /(010) BiVO 4 heterojunction, revealing that the (010) BiVO 4 can form a coherent stable interface with the (101) Bi 2 O 4 to improve BiVO 4 utilization of visible light for water splitting with better charge separation. This method of selective photocatalyst deposition yields higher photocatalytic and photoelectrocatalytic OER activities. Li and co-workers 30 also reported a higher PEC activity of (010) than (121) and correlated that with the higher electron mobility of the former one. In summary, selective facets of BiVO 4 yields better charge separation and efficient photochemical water splitting.
Computationally, very few reports have studied the low Miller indexes of BiVO 4 surfaces using PBE. 23,31 Therefore, an accurate investigation of these surfaces is needed to understand the difference in photocatalytic behavior. In our previous works, we examined principle bulk properties of several photocatalytic bismuth-based materials including BiVO 4 . 32−36 We adopted a computational framework based on the density functional theory (DFT) together with the hybrid Hyed− Scuseria−Ernzerhof (HSE06) exchange−correlation functional. 37,38 This method has also proven good accuracy by matching the experimental band gap and band energy positions relative to the water redox potential of a wide number of photocatalytic materials. [5][6][7][8][9][32][33][34][35][36]39,40 Recently, ideal exposed facets for OER and HER of the widely utilized UV light-and visible light-driven photocatalysts, TiO 2 and Ta 3 N 5 , were predicted on the basis of an optoelectronic and redox characteristics combination using this computational scheme and relevant information was proposed for improving materials performance. 41,42 The impact of the four predominant (010), (110), (001), and (121) exposed facets obtained experimentally for monoclinic BiVO 4 on the material performance for photocatalytic water splitting reactions still remained unclear and are addressed in this letter. The objective of this work is to provide this knowledge by performing an accurate and comprehensive DFT-based computational study based on the hybrid HSE06 functional including the spin−orbit coupling (SOC) to take into account the relativistic effects on Bi. Their electronic structure, transport features, and relative band alignment to water splitting limits are investigated. The role of each exposed facet in photoredox process and the most appropriate candidates for HER and OER are given. All details on the computational models and methods used in this work are given in the Supporting Information.
Bulk monoclinic BiVO 4 (C 6 2h point group and C2/c space group) has a layered structure made by Bi 3+ and V 5+ cations in coordination with O 2− and linked to each other by distorted BiO 8 dodecahedra and VO 4 tetrahedra, leading to four and two different oxygen neighbors in each subunit. 43 Our PBE computed lattice constants of 7.23, 11.56, and 5.10 Å reproduced very well the experimental values (7.25, 11.7, and 5.09 Å). 36,43 Four predominant (010), (110), (001), and (121) exposed facets are obtained experimentally in the prepared BiVO 4 samples. 24,25,27,28,30 The (010)  For bulk monoclinic BiVO 4 , our HSE06+SOC computed band gap energy of 2.8 eV is in agreement with the experimental value (2.5 eV) and much more accurate than the PBE+SOC computed one of 1.7 eV. 36 The same effect of SOC on the band gap energy revealing 0.2 eV reduction was obtained at both levels of functional. The CBM is mainly made by V 5+ (3d 0 ) orbitals. The VBM consists of O 2− (2p 6 ) orbitals together with weak Bi 3+ (6s 2 ) orbitals contribution, in agreement with photoemission data. 20,44 The hybridization effect of Bi 3+ (6s 2 )/O 2− (2p 6 ) lone pair states at VBM is attributed to the distortions in BiO 8 dodecahedra, in accordance with the experimental photoemission with X-ray diffraction results. 43 Similar contribution of orbital types at the VBM/CBM states are also obtained for (010)-, (110)-, (001)-, and (121)-oriented BiVO 4 slabs ( Figure 2). Comparing to the bulk material, their corresponding band gap energy is slightly broader by 0.1 or 0.2 eV. The (001) slab gives 2.9 eV, while the (010), (110), and (121) slabs give 3.0 eV. A minimal effect of SOC on the band gap energy was revealed only for (010) and (001) slabs with a reduction of 0.1 eV, as shown in Figure  S2 in the Supporting Information.
The computed partial charge density maps using HSE06+-SOC associated with the VBM and CBM electronic states of the (010)-, (110)-, (001)-, and (121)-oriented BiVO 4 slabs are displayed in Figures 3 and 4. For the (010)-and (001)oriented BiVO 4 slabs, the partial charge densities associated with VBM/CBM electronic states are delocalized over O 2p, Bi 6s, and V 3d orbitals throughout the crystal lattice, leading to good mobility of holes/electrons from bulk to (010) and (001) surfaces ( Figure 3). The partial charge densities corresponding to the CBM electronic state of the (010) slab reveal better orientations of V 3d orbitals than those obtained for (001). This result is expected to lead to easier migration or better mobility of electrons toward (010) than (001). In both cases, the partial charge densities are in line with the projected density of states results shown in Figure 2. For the (110)oriented BiVO 4 slab, the partial charge density associated with   (Figure 4). This result tends to lower the mobility of electrons and holes from bulk to (110) surface. For the (121)-oriented BiVO 4 slab, the partial charge density corresponding to the VBM electronic state is partially distributed over some O 2p orbitals and very few Bi 6s orbitals, forming a disconnected tube-like shape from the bulk to the surface, while the CBM partial electron density is delocalized on V 3d orbitals located in sublayers, forming a tube-like shape parallel to the surface ( Figure 4). As a consequence, this result is expected to lead to lower hole mobility to this surface than to the (010) and (001) surfaces and poor electron mobility to the (121) surface. These predicted results highlight the fundamental origin behind the   The Journal of Physical Chemistry Letters pubs.acs.org/JPCL Letter higher PEC activity of (010) than (121) as reported experimentally and correlated with the higher electron mobility of (010) with respect to (121). 30 The absolute VBM/CBM energies obtained using HSE06+-SOC of BiVO 4 slabs are −2.  Figure 6). However, the VBM levels of (001) and (121) slabs are 1.37 and 1.52 eV lower than the O 2 /H 2 O level, while their respective CBM energy level is 0.3 and 0.25 eV higher than the H + /H 2 level (Figure 4). Due to the incorrect CBM levels of (010) and (110) BiVO 4 slabs with respect to H + /H 2 potential, the excited electrons toward these surfaces after absorption of photons will suffer from absent driving force to reduced proton. On the basis of the correct VBM levels with respect to O 2 /H 2 O potential, these two facets    Figures S3 and S4 in the Supporting Information). The similar phenomenon observed in the redox characteristics between the (001) and (121) slabs can be justified by the appearance of four-coordinated Bi with four-coordinated V and two-coordinated O on both facets. The small downward shift in the energy levels of (110) slab with respect to those of the (010) slab can be justified by the appearance of some fivecoordinated surface Bi. These predicted results highlight the fundamental origin behinds the spatial carrier separation between (010) and (110) facets of BiVO 4 as invoked in experiment. 28 The significant modification in the redox characteristics of (001) and (121) slabs leading to the drastic upward shift in the band energy levels with respect to those obtained for the (010) and (110) slabs, can also be justified by the difference in the surface Bi species coordination number going from six for (010) or six and five for (110) to four for both (001) and (121).
If we merge the redox and carrier transport characteristics together for each surface, our work highlights the (001) facet as the only appropriate candidate for both OER and HER, though the (010) facet is a good candidate for OER only. The two (110) and (121) surfaces are acceptable candidates only for OER but have less potential than (001) and (010). These predicted results explain the fundamental origin behind the reported experimental data showing (001) as the most important and stable facet for photocatalytic oxidation of water for O 2 generation and (010) for good visible photocatalytic degradation activities. 25−27 In water splitting reactions, the two cocatalysts for OER and HER should be anchored over the best facets to boost the holes and electrons transfer kinetics to water. Obtaining appropriate flat band potentials of powder samples with respect to water splitting levels cannot guarantee an active photocatalyst for water splitting reactions due to the anisotropic character of its transport and redox characteristics and their correlation with the type/nature of exposed facet. The obtained results from this work could successfully explain the fundamental reasons why powder BiVO 4 samples have good activity for photocatalytic OER, while relatively poor PEC performance for hydrogen evolution was obtained. 19,20,25−27 Overcoming the H 2 evolution performance limitation in an effective way requires a selective depositions of the HER cocatalyst on the (001) surface. The OER cocatalyst is required to be selectively deposited on the (010) facet for water oxidation to minimize the electron−hole recombination on the photocatalyst surface. The 2-ccordinated O species would be the active site for water oxidation toward OER on the (010) surface. The 4coordinated V together with 4-ccordianted Bi species would be the active sites for proton reduction toward HER on the (001) surface.
In summary, we have investigated the impact of the four predominant (010), (110), (001), and (121) exposed facets obtained experimentally for monoclinic BiVO 4 photocatalyst on its performance for water splitting reactions. Although the electronic structure is similar for these four surfaces, their redox and transport properties revealed an anisotropic nature based on the crystal orientation and termination. The particular role of each facet in proton reduction was correlated with the surface Bi coordination number and their geometrical distribution. After having combined the charge carrier transport and redox characteristics for each surface, our study predicted the (001) facet as the only appropriate candidate for both HER and OER, while the (010) facet is a good candidate only for OER. The (110) and (121) facets were classified as acceptable candidates only for OER but less potential than (001)