Photocatalytic water splitting using nanosized semiconductors is an ideal technology for converting the free and abundant solar energy into chemical energy [1-5]. Owing to their unique crystal and energy band structures, Nb-based semiconductors consisting of corner-sharing octahedral NbO₆, which are beneficial for the transfer of photogenerated electrons and holes, have attracted much attention in various photocatalysis applications [6-10]. Importantly, the conduction band (CB) levels consist of an Nb 4d orbit, which is much more negative than a Ti 3d orbit, giving the photogenerated electrons a strong reducing capability [11-13]. Although many Nb-based semiconductors that function efficiently under UV light have been developed, only a few visible light responsive Nb-based photocatalysts have been reported for photocatalytic water splitting, and further improvement of the photocatalytic water splitting activity is required [5, 14-17]. Therefore, it is highly desirable to search for a novel Nb-based semiconductor nanocatalyst with improved visible-light-driven photocatalytic water splitting activity for practical applications.

Sn₂Nb₂O₇ with a pyrochlore structure has attracted particular research interest in visible-light-driven photocatalytic water splitting owing to its narrow band gap (approximately 2.4 eV), and proper CB and valence band (VB) potentials [18-20]. This material can absorb visible light because Sn^Ⅱ with a 5s² electronic configuration contributes to the formation of a hybrid orbital with O 2p as a VB. However, the reported bulk Sn₂Nb₂O₇ powders (bulk SNO) obtained by the high-temperature solid state reaction method have shown unsatisfactory photocatalytic activities owing to its large-sized crystals with low specific surface areas [21, 22]. Therefore, it is still a challenge to develop a facile yet effective method for preparing nanosized Sn₂Nb₂O₇ with enhanced photocatalytic activity for future applications.

Herein, we have developed a facile two-step hydrothermal method to prepare Sn₂Nb₂O₇ nanocrystals (SNO NCs) with an average size of approximately 20 nm. The photocatalytic activity of SNO NCs is evaluated by photocatalytic H₂ production from water using lactic acid as a sacrificial agent under visible light irradiation. SNO NCs show enhanced photocatalytic H₂ production activity when compared with bulk SNO powders owing to its relatively high dispersity of nanosized particles and larger specific surface area.

SNO NCs were synthesized by the facile two-step hydrothermal method shown in Scheme 1. First, a clear solution of soluble potassium niobate K₈Nb₆O₁₉·10H₂O (KNO) was obtained after hydrothermal treatment of Nb₂O₅ powder (0.5 g) in 45 mL KOH solution (3 mol/L) at 200 ℃ for 12 h. Then, ethanol was added into the above KNO clear solution (V_ethanol:V_{KNO clear solution} = 1:2) to obtain a white precipitant, which reacted with SnCl₂ to produce light yellow SNO precursors by ultrasonication for 5 min under room temperature. Finally, the yellow SNO NCs were obtained by further hydrothermal treatment of the light yellow SNO precursors dispersed into a pure water solution (30 mL) at 200 ℃ for 12 h.

The morphology of the SNO NCs was characterized by scanning electron microscopy (SEM) and transmission electron microscopy (TEM). As shown in Fig. 1(a) and (b), SNO NCs with an average size of approximately 20 nm in a large scale were observed. From the high resolution transmission electron microscopy (HRTEM) image of a single nanoparticle (Fig. 1(c)), the spacing between the adjacent fringes was measured to be 0.61 nm, which corresponded to the (111) plane of cubic Sn₂Nb₂O₇. The successful fabrication of SNO NCs was further confirmed by X-ray diffraction (XRD) analysis. As demonstrated in Fig. 1(d), all of the diffraction peaks could be exactly indexed to the pure phase of Sn₂Nb₂O₇ (JCPDS 23-0593). No impurity peaks were detected. The calculated crystal size of SNO NCs was approximately 20 nm, according to the Scherrer Equation based on the peak at 2θ = 29.2°, which indicated that pure SNO NCs could be successfully fabricated through a facile two-step hydrothermal process.

The band gap is known to play an important role in the efficient capture of solar energy in semiconductors. The band structures of the as-prepared SNO NCs and the bulk SNO powder were examined by UV-Vis diffuse reflectance spectroscopy. As shown in Fig. 2(a), both the SNO NCs and bulk SNO powder exhibited an obvious photoabsorption in the visible light region. Compared with the bulk SNO powder, the absorption edge of the SNO NCs exhibited an obvious blue-shift, which was ascribed to the nanosized effect [23-25]. The band gaps of the SNO NCs and the bulk SNO powder, calculated by a plot of (ahv)² versus photo energy, were 2.52 and 2.38 eV, respectively (Fig. 2(b)). Obviously, SNO NCs have a wider energy band gap compared with that of the bulk SNO powder. To detect the detailed band structures of SNO NCs, valence band X-ray photoelectron spectra (VB XPS) were recorded to determine the relative VB maximum. Compared with the bulk SNO powder, an obvious downshift for the VB edge of SNO NCs was observed, as shown in Fig. 2(c). Based on the VB XPS results and Fig. 2(b), the schematic band structures for SNO NCs and the bulk SNO powder are shown in Fig. 2(d). The CB and VB edges were calculated at −0.27 and 2.25 V for SNO NCS, and −0.73 and 1.65 V for the bulk SNO powder, respectively.

The specific surface area of photocatalysts is also crucial for their photocatalytic performance. N₂ adsorption-desorption measurements were performed to determine the specific surface areas of the as-obtained samples. Fig. 3 shows a typical Ⅳ isotherm with an H3-type hysteresis loop, which suggested the co-existence of mesopores and macropores in SNO NCs. The specific surface areas of SNO NCs was measured to be 52.2 m²/g, which was much larger than that of the bulk SNO particles (approximately 2.3 m²/g). It has been extensively reported that photocatalysts with a high specific surface area beneficial for photocatalysis [26-28]. Therefore, SNO NCs with a large surface area could exhibit a high visible-light-driven photocatalytic H₂ production activity.

The photocatalytic activity of SNO NCs was evaluated by photocatalytic H₂ production from lactic acid aqueous solutions under visible light irradiation (λ > 400 nm). The visible-light-driven photocatalytic H₂ evolution rate of SNO NCs without loading cocatalyst was 2.1 µmol/(h·g), which was approximately three times higher than that of the bulk SNO powder (0.65 µmol/(h·g)). The relatively lower photocatalytic H₂ evolution rates of both the samples were ascribed to the quick recombination of the photogenerated electrons and holes [20]. When Pt nanoparticles (NPs) as cocatalyst were loaded on the surface of the samples by an in-situ photodeposition method, both samples showed an improved photocatalytic H₂ production activity. As shown in Fig. 4(a), the photocatalytic H₂ evolution rate of SNO NCs with 0.3 wt% loading of Pt was 82 µmol/(h·g) under visible light irradiation (λ > 400 nm), which was 5.5 times higher than that of the bulk SNO powder (14.8 µmol/(h·g)). The higher photocatalytic H₂ production activity of SNO NCs could be ascribed to its high dispersity of nanosized particles, larger specific surface area and more positive VB potential. First, the decrease in the particle size of SNO NCs caused a dramatically decreased migration distance of the photogenerated electrons and holes, which allowed photogenerated charge carriers to move rapidly to the surface to take part in the photocatalytic reaction [29-31]. Second, the specific surface area of SNO NCs was 52.2 m²/g, which was much higher than that of bulk SNO particles (2.3 m²/g). It is well-known that a high surface area could increase the surface active sites and charge transfer of a semiconductor photocatalyst, which would result in the improvement of the photocatalytic activity [32, 33]. Third, the more positive VB potential of SNO NCs could largely expedite the oxidation capability of holes (Fig. 2(c)), leading to an enhanced separation efficiency of photoinduced electron-hole pairs and thus improved photocatalytic H₂-production activity [5, 23]. All these favorable factors co-contribute to the exceptionally high photocatalytic performance of SNO NCs.

The proposed photocatalytic reaction mechanism of SNO NCs under visible light irradiation can be illustrated as follows. Under visible light irradiation, the VB electrons of SNO NCs are excited to the CB, simultaneously leaving behind positive holes in the VB. Normally, these charge carriers quickly recombine and only a fraction of the electrons and holes participate in the photocatalytic reaction, which results in a poor photocatalytic activity. However, when SNO NCs were modified by Pt NPs, these photogenerated electrons on the CB of SNO NCs tend to transfer to Pt NPs, which leads to electron-hole separation and thus a higher photocatalytic H₂ production activity.

To demonstrate the applicability of SNO NCs in photocatalysis, a 24-h recycling experiment with intermittent evacuation every 6 h was performed under visible light irradiation (λ > 400 nm). After four consecutive cycles, the SNO NCs maintained an almost constant photocatalytic H₂ production rate without noticeable catalyst deactivation (Fig. 4(b)). The satisfying stability makes SNO NCs a good candidate for photocatalytic water splitting under visible light irradiation.

In conclusion, SNO NCs with an average size of approximately 20 nm were successfully prepared on a large scale through a two-step hydrothermal route. SNO NCs showed an improved visible-light photocatalytic H₂-production activity compared with the bulk SNO powder. The higher photocatalytic activity of SNO NCs was mainly attributed to its relatively high dispersity of nanosized particles and larger specific surface area when compared with the bulk powder. This work provides a novel synthesis method to develop nanosized Nb-based semiconductor photocatalysts with high photoactivity for solar energy conversion.