Photocatalysis can induce different redox reactions, including pollutant degradation, water splitting [1], and CO₂ reduction [2, 3], and has attracted increasing attention for its promising energy and environmental applications. All these reactions are driven by the photogenerated electrons and holes surviving the recombination process after the photo-excitation of electrons from valence bands to conduction bands in a semiconductor photocatalyst [4, 5]. It is well established that both the number and energetics of the photogenerated electrons and holes intrinsically affect photocatalysis efficiency. In many important reactions involving water oxidation, the holes induce oxidative half-reactions, which is considered a rate-determining step. Taking photocatalytic water splitting as an example, the reduction of protons to hydrogen molecules is a one-electron process, while the oxidation of water to produce oxygen is a four-electron process [6-9]. Increasing the number and/or energetics of photogenerated holes is always favorable for oxygen evolution and overall water splitting [10].

Currently, most studies focus on improving the participation of photogenerated holes in water oxidation by introducing defects [11], doping [12], surface fluorination [13] and constructing various junctions [10]. These methods fall within the scope of narrowing the bandgap, which elevates the valence band maximum (VBM), and promoting surface transfer. Among these options, doping with the appropriate heteroatoms represents a powerful way of modifying the band structures of photocatalysts to change the energetics of the charge carriers. One typical example on this aspect is that introducing interstitial boron with a concentration gradient in anatase or rutile TiO₂ microspheres can induce the downward band bending, and thus increase the energetics of photogenerated holes [14, 15]. Consequently, TiO₂ with a concentration gradient of boron with its maximum at the surface shows much higher photocatalytic water oxidation ability than surface boron-free TiO₂. However, the lower VBM of surface-layer containing boron compared to that of the boron-free internal region of TiO₂ restricts the transport of holes from the bulk to the surface. To solve this problem, it is necessary to realize homogeneous doping of interstitial boron in TiO₂ because this can modify the band structure of the whole material, as demonstrated in layered materials such as cesium titanates [16, 17] and carbon nitride polymers [18, 19]. Unlike layered materials with interlayer galleries that facilitate the diffusion of dopant species from the surface to the bulk for homogeneous doping, TiO₂ without natural diffusion pathways for dopants is always difficult to be doped homogeneously.

Using unique compounds containing dopants as precursors of metal oxides represents an effective method for preparing doped metal oxides, as demonstrated by the hydrolysis of metal borides to prepare boron-doped metal oxides under hydrothermal conditions [14, 15]. A vapor-phase hydrothermal route for treating TaB₂ was also developed to construct TaB₂/Ta₂O₅ core/shell particles because of the local space-confined hydrolysis of TaB₂ and subsequent heterogeneous nucleation and growth of Ta₂O₅ on TaB₂ [20]. Moreover, the good interface contact between TaB₂ and Ta₂O₅ led to boron doping of Ta₂O₅ at the interface, which inspired us to conduct homogeneous boron doping in metal oxides supported by metal borides. In this study, a modified vapor-phase hydrothermal method in a moist argon atmosphere was developed to prepare a boron homogeneously doped TiO₂ shell supported by a TiB₂ core. The spatial distribution of the boron dopant together with the chemical state of TiO₂ was investigated. A correlation between the greatly improved photocatalytic water oxidation of boron doped TiB₂/TiO₂ and the modified band structures was established. The results in this study demonstrate the significance of controlling the homogeneous distribution of heteroatoms in modifying band structures of photocatalysts to increase their activity.

500 mg of TiB₂ (Sigma Aldrich, 98% metal basis) was used as the precursor in a ceramic container and was heated in an argon atmosphere with a flow rate of 50 mL/min bubbling through room temperature deionized water at 500 ℃ for 12 h. The ramping rate was 2 ℃/min, and the sample was naturally cooled down to room temperature after the heating duration. The resultant sample was washed with deionized water at least five times to remove dissolvable byproducts and then dried in air at 90 ℃ overnight.

X-ray diffraction (XRD) patterns of the samples were obtained on a Rigaku diffractometer using Cu K_α irradiation (λ = 1.54056 Å).The morphology and microstructures of the samples were studied by scanning (SEM, Nova NanoSEM 430) and transmission electron microscopy (TEM, FEI Tecnai-F20). Raman spectra (632.8 nm) were collected with LabRam HR 800. The chemical states of the samples were determined by X-ray photoelectron spectroscopy (XPS, Thermo Escalab 250, using a monochromatic Al K_α X-ray source). All binding energies were referred to the C 1s peak (284.8 eV) that arises from adventitious carbon. Thermogravimetry-differential scanning calorimetry (TG-DSC) analysis was conducted on a Netzsch 449C Jupier with a flow rate of 50 mL/min of air at a ramping rate of 10 ℃/min. The optical absorption spectra were recorded on a UV-visible-infrared diffuse reflectance spectrophotometer (Jasco V-770).

Photocatalytic oxygen generation measurements were conducted in a commercial on-line automatic testing system for photocatalytic reactions (Perfectlight Sci & Tech Corp Ltd., Labsolar-6A). The reaction container of the system had a volume of 250 mL. 100 mg of the photocatalyst was suspended in a solution of 0.85 g AgNO₃ in 100 mL water as the electron acceptor. The light source used was a 300 W Xe lamp (Perfectlight Sci & Tech Corp Ltd., PLS-SXE-300UV). The gas produced in the reaction was analyzed by gas chromatography (Shimadzu, GC2014).

To determine an appropriate temperature for the hydrolysis of TiB₂ crystals in a moist argon flow, the thermal oxidation process of TiB₂ crystals in an oxygen atmosphere as a function of temperature was monitored by thermal analysis to simulate water molecule induced hydrolysis. Based on the fact that the O₂ induced oxidation of TiB₂ produced TiO₂ and B₂O₃, the mass of materials increased with a temperature increase. The TG-DSC curves in Fig. 1(a) indicate that the oxidation process starting at temperatures above 480 ℃ can be divided into three distinct stages. Stage-Ⅰ at the low temperature region between 480 and 520 ℃ is reasonably assigned to the oxidation of the surface layer of TiB₂ crystals. The formation of products, including titania and boron oxide from Stage-Ⅰ, probably hinders the access of O₂ to the internal regions of the TiB₂ crystals. This results in Stage-Ⅱ having a lowered oxidation rate between 520 and 620 ℃. A further increase in the temperature leads to the appearance of dominant Stage-Ⅲ beyond 620 ℃ because of the greatly enhanced oxygen diffusion ability through the external product layer. According to the characteristics of the three oxidation stages, it is reasonable to induce the oxidation of TiB₂ crystals at Stage-Ⅰ to obtain the desired TiB₂/TiO₂ core/shell particles. The oxidation of TiB₂ in a moist argon atmosphere in this study was therefore conducted at 500 ℃.

Fig. 1(b) shows TGA-DSC curves of the resultant TiB₂/TiO₂ core/shell particles in an oxygen atmosphere. Unlike the three distinct oxidation stages of pristine TiB₂, the oxidation of TiB₂/TiO₂ core/shell particles only experiences one stage that is similar to Stage-Ⅲ of the pristine TiB₂. This result is consistent with the structural characteristics of the TiB₂/TiO₂ core/shell particles. In addition, based on the mass increase of TiB₂ and TiB₂/TiO₂ in an oxygen atmosphere, the percentage of TiO₂ in the TiB₂/TiO₂ core/shell particles is estimated to be around 30%.

XRD patterns (Fig. 2(a)) and Raman spectra (Fig. 2(b)) of the TiB₂/TiO₂ core/shell particles together with those of the pristine TiB₂ crystals were recorded to analyze crystal structure evolution. XRD peaks of TiB₂ after the thermal oxidation in a moist argon atmosphere retained similar features. An obvious new peak at around 25.3° is attributed to the diffraction of (101) place of anatase TiO₂. Two weak peaks at 36° and 41° confirm the existence of rutile TiO₂. The coexistence of the anatase and TiO₂ phases was also confirmed by Raman spectra. Due to the partial overlap of the major 27.4° peak of rutile TiO₂ with the 27.6° peak of TiB₂, it was difficult to directly determine the ratio of anatase to rutile based on the Reference Intensity Ratio (RIR) method. To solve this problem, the overlapped peak at around 27° is fitted with two peaks at 27.45° and 27.62° that are assigned to rutile (110) and TiB₂ (001) planes, respectively, as shown in Fig. 2(c). Based on the fitting results, the percentages of anatase and rutile in TiO₂ were determined to be around 65% and 35% respectively, by the RIR method.

Morphology and microstructure of pristine TiB₂ andTiB₂/TiO₂ core/shell particles were investigated by SEM and TEM. Fig. 3(a) shows the typical morphology of pristine TiB₂ crystals with the particle size of several microns and a smooth surface. After thermal oxidation in a moist argon atmosphere, the resultant TiB₂/TiO₂ core/shell particles (Fig. 3(b)) have increased particle size distribution and surface roughness, with the formation of a shell consisting of TiO₂ particles. According to the TEM image in Fig. 3(c), the thickness of the TiO₂ shell completely surrounding TiB₂ core is estimated to be around 400 nm. The dense polycrystalline TiO₂ shell consists of the mixture of anatase and rutile nanoparticles as indicated by Fig. 3(d). This is consistent with XRD results.

The compositions of the TiO₂ shell and its chemical states were investigated by XPS, as shown in Fig. 4. The shell consists of Ti, O and B compositions. The XPS spectrum of O 1s has three distinct peaks at 530.5, 532.7 and 534.1 eV, which originates from lattice oxygen, hydroxyl groups and oxygen bonded to the interstitial boron, respectively [14, 15, 21, 22]. Besides the major signal of Ti⁴⁺ in XPS spectrum of Ti 2p, a minor signal centered at 457.3 eV that is typically attributed to Ti³⁺ was formed due to the presence of interstitial boron. Fig. 4(c) shows the sputtering time dependent XPS signal of B 1s. The atomic ratio of boron to titanium determined from the pristine surface is 3:10. The value of 192.0 eV is a typical binding energy for the core electrons of interstitial boron in TiO₂. The less changed intensity of the XPS B 1s peak with the increase of sputtering time suggests a homogeneous distribution of the interstitial boron in the TiO₂ shell. It is known that the interstitial boron tends to diffuse from the bulk towards the surface and finally concentrates in the surface layer upon thermal treatment [11]. The homogeneous distribution of interstitial boron in the TiO₂ shell supported on the TiB₂ core is largely contributed by the sufficient supply of boron from the core.

As demonstrated in previous studies [11], introducing interstitial boron can cause the downward shift of the VBM of TiO₂ to enhance photocatalytic oxygen evolution because of the increased oxidative power of photogenerated holes. A comparison of the XPS valence band spectra of boron homogeneously doped TiO₂ shell on the TiB₂/TiO₂ core/shell particles, commercial rutile TiO₂ and anatase TiO₂ in Fig. 5(a) reveals that the VBM of the TiB₂/TiO₂ core/shell particles is lowered by around 0.3 eV than that of anatase or rutile TiO₂. This downward shift caused by the homogeneous boron doping is comparable to the shift caused by a gradient doping of interstitial boron in TiO₂ [14, 15].

The UV-visible absorption spectrum of the TiB₂/TiO₂ samplein Fig. 5(b) shows a typical intrinsic absorption band before 400 nm and additional absorption bands beyond 400 nm. Similar absorption features were also observed in other metallic/semiconducting core/shell particles [20, 23]. Based on the results of the XPS valence band spectra and UV-visible absorption spectrum, the band alignments of the TiB₂/TiO₂ sample refers to un-doped TiO₂and is proposed in Fig. 5(c). The homogeneous doping of interstitial boron in the shell of TiB₂/TiO₂ core/shell particles leads to the downward shift of both valence band and conduction band edges, with no bandgap changes. This is due to interstitial boron doping, which brings some localized states close to the conduction band minimum [24]. Compared to the band bending caused by the gradient doping of boron, the advantages of this band alignment for photocatalytic water oxidation include not only the lowering of the VBM for the generation of the holes with higher energetics, but also the facilitate transport of the holes from the bulk to the surface.

Fig. 6(a) compares the photocatalytic water oxidation activity of the TiB₂/TiO₂ sample with three reference samples (commercial anatase TiO₂, rutile TiO₂ with a particle size of several hundreds of nanometers, and P25 TiO₂ with a particle size of around 20 nm, with a mixed phase of anatase to rutile in a ratio of 4:1) in the presence of AgNO₃ as sacrificial agent. The TiB₂/TiO₂ sample shows a much superior activity in producing oxygen from photocatalytic water oxidation when compared to the three reference samples, though the former has a much larger particle size and smaller specific surface area. The oxygen evolution rate of the TiB₂/TiO₂ sample is 154 μmol/h, which is around 5 times higher than that of the benchmark P25 TiO₂ photocatalyst. Considering the mixed phases of anatase and rutile that are favorable for the charge separation in these two samples, the superiority of the TiB₂/TiO₂ sample is largely attributed to the lowered VBM of the homogeneous boron doped TiO₂ shell. The stability test in Fig. 6(b) shows the good stability of the TiB₂/TiO₂ sample as a photocatalyst. In addition, it should be pointed out that the TiB₂/TiO₂ sample has no activity under visible light irradiation. This observation can be understood as the localized states below the conduction band usually not being unable to produce charge carriers with sufficiently high energetics as observed in oxygen-deficient TiO₂ [25].

At high temperatures, hydrolysis of TiB₂ in a wetting argon atmosphere was used to grow TiB₂/homogeneous boron doped TiO₂core/shell particles. The resultant TiO₂ shell has the mixed phases of anatase and rutile. Because of the homogeneous boron doping in the shell, the downward shift of the valence band edge of TiO₂ is realized to increase the oxidative power of photogenerated holes. Compared to three relevant reference samples (anatase, rutile, anatase/rutile TiO₂), the resultant core/shell particles show greatly improved photocatalytic water oxidation activity to produce oxygen under UV-visible light irradiation. The results obtained in this study demonstrate the significance of modulating the redox power of photogenerated carriers in improving photocatalytic activity.

This work was supported by the National Science Foundation of China (51702327, 51521091), the Major Basic Research Program, Ministry of Science and Technology of China (2014CB239401), the Key Research Program of Frontier Sciences CAS (QYZDB-SSW-JSC039). Y. Q. Yang and G. Liu are grateful for the awards of the IMR SYNL-T.S. Kê Research Fellowship and the Newton Advanced Fellowship, respectively.