1. Introduction The major characteristics of the structural chemistry of metal borates are the abundant and diverse connectivity of borate ionic framework, where metal cations are imbedded. Most borates are insulators or semiconductors because the borate ionic framework cannot conduct electrons or holes efficiently, therefore optical properties, i.e. second harmonic-generation and photoluminescent properties, are well explored for metal borates [1–3]. In literature, there are only a few particular metal borates showing substantial photocatalytic properties. For example, InBO3 can photocatalyze the degradation of 4-chlorophenol [4]; CuB2O4 and Cu3B2O6 (by loading Pt as cocatalyst) can catalyze water reduction or oxidation under visible light irradiation [5]; Bi4B2O9 and Bi2O2[BO2(OH)] show observable visible light activities for methylene blue degradation [6]; and the nonlinear optical material K3B6O10Br shows a good catalytic activity in UV-induced de-chlorination of chlorophenols [7]. It is commonly agreed that photocatalytic water splitting is a green process to convert the solar energy into chemical energy. The ongoing research in this field is very active. People are seeking for materials capable of catalyzing the overall water splitting without sacrificial reagents, which, however, is particularly difficult because of the high recombination rate of the photogenerated e− and h+ before they could react with substrates [8,9]. There are of course several strategies to atleast partially restrain the charge recombination. One is to shorten the pathway for the electrons moving to the surface by making catalysts in nanosize or mesoporous form. For example, a gallium borate (Ga4B2O9) cannot photocatalyze the water reduction in its bulk form even with the assistance of CH3OH in aqueous solution, while it became highly efficient to photocatalyze overall water splitting without cocatalysts, when it was prepared by sol–gel method and crystallized in nanostripes [10]. On the other hand, it is usually helpful to suppress the charge recombination by loading cocatalysts, such as Pt, Pd, RuOx, Ag, and Au [11,12]. Extensive efforts have been applied on this aspect to improve the catalytic activities of well-known photocatalysts like TiO2 and g-C3N4 [13–20]. Very recently, an open-framework gallium borate Ga9B18O33(OH)15·H3B3O6·H3BO3 (Ga-PKU-1) has been reported as a UV-light photocatalyst for water splitting, however, requiring the assistance of sacrificial regents [21]. This boron-rich compound possesses a GaO6-based three-dimensional (3D) framework (see Fig. S1 in Supplementary information, SI), and the unoccupied 4s orbitals of Ga contribute to the bottom of the conducting band. Ga-PKU-1 can be prepared by hydrothermal method and were single crystals in micrometer (see Fig. S2 in SI). It is a natural idea to apply the cocatalyst loading to enhance its photocatalytic activity. Here we performed a systematic investigation on this aspect, where it indeed exhibits a greatly improved activity and is even capable of overall water splitting under UV-light irradiation. 2. Experimental sections The preparation of Ga-PKU-1 was performed according to the therapy in literature [21]. First, β-Ga2O3 was dissolved in concentrated HNO3 at 180 °C for 10 h in a closed system. The resultant solution evaporates to nearly dry by just opening to air. Thereafter, H3BO3 (3.0 g) was charged and the system was sealed again and maintained at 220 °C for Catalysis Communications 71 (2015) 17–20 ⁎ Corresponding author. E-mail address: taoyang@cqu.edu.cn (T. Yang). http://dx.doi.org/10.1016/j.catcom.2015.08.009 1566-7367/© 2015 Elsevier B.V. All rights reserved. Contents lists available at ScienceDirect Catalysis Communications journal homepage: www.elsevier.com/locate/catcom another 5 days. Finally, the white powder product was washed by deionized water for several times. Powder X-ray diffraction (XRD) was performed to confirm the phase purity. As a representative, the powder XRD pattern of as-prepared Ga-PKU-1 was refined using the cell lattice reported in literature (see Fig. S3 in SI) [21], no impurity refection could be found. In addition, all samples used in current study, including monometallic and bimetallic modified catalysts, were characterized by powder XRD after catalytic reactions and none of them show any degeneration, indicating the high stability of Ga-PKU-1 (see Figs. S4– S6 in SI). Powder XRD data were collected on a PANalytical X'pert diffractometer equipped with a PIXcel 1D detector (Cu Kα radiation). The operation voltage and current are 40 kV and 40 mA, respectively. Photocatalytic activities were tested on a gas-closed circulation system equipped with a vacuum line (CEL-SPH2N system), a reaction vessel and a gas sampling port that is directly connected to a gas chromatograph (Shanghai Techcomp-GC7900, TCD detector, molecular sieve 5A). In a typical run, 50 mg of a catalyst was dispersed by a magnetic stirrer in 50 mL pure water or 20 vol.% methanol aqueous solution in a 150 mL Pyrex glass reactor with a quartz cover. The solution has been kept stirred, and a 5 °C recycling water bath was applied to keep the reaction vessel cool. The light irradiation source was generated by a 500W high-pressure mercury lamp (CEL-M500, Beijing AuLight). The spectrum of the lamp was provided in Supplementary information. The H2 and O2 evolution was analyzed by an online gas chromatograph, separately. For detecting so-generated H2 and O2, the gas carrier was selected to be N2 and He, respectively, due to the sensitivity of the TCD detector. The detailed gas production curves against time are presented in Figs. S7–S10 in SI. The noble metal or metal oxide co-catalyst loading was performed in a photo-deposition method using the above setup. For example, 0.1 g of catalyst together with 1 wt.% Ru aqueous solution (RuCl3) was charged into 50 mL 40 vol.% methanol aqueous solution under magnetic stirring. Then a UV incident beam was irradiated to this mixture for 3 h. Then, the powder sample was collected and washed by deionized water. For other metal loading, the used sources are Ni(NO3)2·6H2O, AgNO3, H2PtCl6·6H2O, HAuCl4·4H2O and CoCl2·6H2O, respectively. It is assumed that most cocatalyst ions in aqueous solution were successfully loaded. The accurate amount of the loading cocatalyst is in fact difficult to determine. For example, we performed semi-quantitative elemental analyses on NiOx and CoOx-loaded samples by inductively coupled plasma-atomic emission spectrometry (ICP-AES) on a Leeman Profilespec. The results show that the content of NiOx and CoOx was roughly 0.2 wt.% and 0.1 wt.%, respectively. Nevertheless, the photo-deposition method was commonly used by most researchers and it is proved to be effective, though the accurate quantitative of cocatalyst cannot be easily determined. 3. Results and discussion In literature, the structure of Ga-PKU-1 was described as an openframework metal borate, where the GaO6 octahedra share the common edges to form a 3D continuous backbone and the triangular BO3 species are grafted on the surface [21]. The theoretical calculations suggest that the top of the valence band (VB) is composed of O 2p orbitals and the bottom of the conduction band (CB) is contributed by both the Ga 4s and B 2p orbitals. The band structure is as-expected and suggests a wide gap characteristic. Accordingly, when irradiated by UV-light photons, the VB electrons, which are located at the O 2p orbitals, can be promoted to excited states, either Ga 4s or B 2p unoccupied orbitals. Benefiting from the uninterrupted 3D Ga-O ionic framework, the photon-excited electrons and holes can itinerate to the surface of the particle and catalyze the water reduction. The preliminary results in literature proved the above scheme by carrying out both two half reactions, i.e. the water reduction and oxidation, with the assistance of CH3OH and AgNO3, respectively [21]. However the activity is not high because the sample prepared by solvothermal method is in fact small single crystals in micrometer (see Fig. S2 in SI) and the recombination problem of e− and h+ is severe. As shown in Fig. 1, the additive of 1 wt.% cocatalysts, such as CoOx, RuOx, Pt, Au, NiOx and Ag, could significantly increase the H2 evolution rate from 3.3 to 47.9 μmol/h/g (in 20 vol.% methanol aqueous solution). It is commonly agreed that a close contact can facilitate the charge (either e− or h+) immigration from the catalyst to the corresponding cocatalyst, resulting in the extended lifetime of charge carriers [22,23]. In addition, these loaded cocatalyst nanoparticles on the surface behave as catalytic sites, and probably offer strong binding sites with substrates [11]. As shown in Fig. 1, the loading of cocatalysts in most cases leads to the enhancement of photocatalytic activity in one order. Here we need to mention that CoOx and RuOx are well-known cocatalyst for O2 evolution. CoOx and RuOx particles can behave as collectors for h+ and accelerate its consumption by reacting with sacrificial reagents. As a consequence, it is beneficial to extend the lifetime of e−, and therefore increase the H2 evolution rate. More importantly, the representative samples loaded with 1 wt.% RuOx, Ag and Pt eventually endow GaPKU-1 crystallites observable ability of the pure water splitting (see Fig. 2). Note that the commonly used P25 (TiO2 nanoparticles) without cocatalyst shows no activity of pure water splitting when applying the same experimental conditions. Usually, Pt particles offer the water reduction sites and the RuOx cocatalyst is capable of water oxidation, which means that photogenerated e− and h+ prefers to move to Pt and RuOx particles, respectively [24,25]. We selected Pt and RuOx for the bimetallic loading onto Ga-PKU-1 and changed the loading content for each component. As shown in Fig. 3, it shows an optimized activity for 1 wt.% RuOx–1 wt.% Pt loaded sample and the observed H2 evolution rate is as high as 28.4 μmol/h/g in pure water. It is also interesting to observe that the sum of the activities for the samples with monometallic loaded (9.2 μmol/h/g for 1 wt.% RuOx and 14.2 μmol/h/g for 1 wt.% for Pt) is smaller than 28.4 μmol/h/g. This is a hint that there exists a synergetic effect when loading an appropriate amount of RuOx and Pt simultaneously. As is known, RuOx and Pt cocatalysts can prompt the O2 and H2 generation, and their acceleration effects of consuming e− and h+ show a good match when loading 1 wt.% RuOx and 1 wt.% Pt in our case. Either increasing or decreasing the content of Pt or RuOx would lead to a decrease of the activity. The combinations of RuOx–Ag and Pt–Ag were also tested but without substantial increasing of the activity (see Fig. 3). In addition, the H2 and O2 production rates were determined for the optimized sample (see Fig. 4), which is very

Fig. 1. H2 evolution rates under UV-light irradiation for Ga-PKU-1, and those loaded with 1 wt.% CoOx, RuOx, Pt, Au, NiOx, and Ag. Photocatalytic conditions: 50 mg of photocatalyst, 50 mL of 20 vol.% methanol aqueous solution.

Fig. 2. H2 evolution rates under UV-light irradiation for 1 wt.% RuOx, Ag, and Pt loaded GaPKU-1. Photocatalytic conditions: 50 mg of photocatalyst, 50 mL pure water.

close to the stoichiometric ratio of 2:1. It is a strong proof that the water reduction and oxidation can occur simultaneously on the surface of the crystallites and it is therefore an important progress during the improvement of the photocatalytic activity for Ga-PKU-1. To maintain the activity and be stable is extremely important for a photocatalyst. Of course, people can use sacrificial reagents to partially solve the photo-corrosion problem. For example, metal sulfide photocatalysts would be self-oxidized by photon-excited h+, which can be prohibited by adding S2−/SO3 2− in aqueous solution [26–28]. For these catalysts aiming at overall water splitting, it is a prerequisite to be stable in pure water under the incident beam irradiation. As shown in Fig. 5, the photocatalytic activity shows about 10% decrease after 3 cycles of reaction, while the recovered powder sample from photocatalytic water splitting experiments after 14 h irradiation shows a high stability checked by powder XRD. Thus, the decreased activity is very likely due to the mechanical loss of the cocatalyst on the surface. In such a case, more efforts are needed to further improve the conditions of the cocatalysts loading, such as post-annealing treatments

Fig. 3. H2 evolution rates under UV-light irradiation for bimetallic loaded samples. Photocatalytic conditions: 50 mg of photocatalyst, 50 mL pure wate

Fig. 4. H2 and O2 evolution under UV-light irradiation for 1 wt.% RuOx–1 wt.% Pt loaded GaPKU-1. Photocatalytic conditions: 50 mg of photocatalyst, 50 mL pure water.

Overall, the post-treatments on Ga-PKU-1 with various cocatalysts loading significantly improve the photocatalytic water splitting ability, from barely seen activity of the host to be capable of pure water splitting in stoichiometry. In literature, CuB2O4 and Cu3B2O6 only show very limited photocatalytic water reduction or oxidation ability with the assistance of sacrificial reagents and Pt cocatalyst [5]. Another interesting

Fig. 5. (a) Time-dependence of H2 evolution for 1 wt.% RuOx–1 wt.% Pt loaded Ga-PKU-1 in pure water system. After each cycle, the system was evacuated; (b) powder XRD comparison for as-prepared and recycled samples

gallium borate, Ga4B2O9, shows no activity in bulk form (crystallites in micrometer), but it possesses a very high activity in nano-stripes even without any cocatalyst (118 μmol/h/g for H2 evolution and 58 μmol/h/ g for O2 evolution in pure water) [10]. In such a case, we believe that the presumably-existed nano-form of Ga-PKU-1 might have a higher activity than Ga4B2O9 nano-stripes. Further effects are needed to the synthetic chemistry of Ga-PKU-1. 4. Conclusions In conclusion, we developed the photocatalytic activity of Ga-PKU-1 by loading 1 wt.% metal or metal oxide cocatalyst (CoOx, RuOx, Pt, Au, NiOx and Ag). The water reduction activity increases from 3.3 up to 47.9 μmol/h/g in 20 vol.% methanol aqueous solution. It can also photocatalyze the pure water splitting when loading Pt, RuOx, and Ag. More importantly, bimetallic loading of RuOx and Pt probably shows a synergetic effect of the activity enhancement and the overall water splitting activity is 28.4 μmol/h/g for H2 and 14.5 μmol/h/g for O2, respectively, which is close to the stoichiometric ratio. Generally, the photocatalytic activity of metal borates has not been extensively studied, our work on the open-framework Ga-PKU-1 suggests the potential functionality in photocatalysis for those borates with metal–oxygen ionic backbones. Acknowledgment This work was financially supported by the Natural Science Foundation of China (Grants 91222106, 21171178) and the Natural Science Foundation Project of Chongqing (Grants 2012jjA0438, 2014jcyjA50036). Appendix A. Supplementary data Crystal structure of Ga-PKU-1 and a photograph of single crystals, powder XRD patterns for all used photocatalysts and a representative figure of Le Bail fitting, gas evolution curves at different conditions and for various catalysts, UV–vis absorption spectra. This material is available free of charge via the Internet at http://www.elsevier.com. Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j.catcom.2015.08.009