1. Introduction With the increasing human activities, fossil fuels, which have caused many serious environmental problems, are difficult to maintain the growing energy demand. Therefore, many research groups have already focused on the clean and renewable energy, such as wind, solar, biomass and hydrogen. In particular, hydrogen, generated by photocatalytic water splitting, has attracted considerable attention as an alternative solution for energy crisis and environmental pollution. The photoelectrochemical hydrogen production was first reported in 1972, in which TiO2 was used as the photoanode material [1]. Since then, photocatalytic water splitting for hydrogen production using TiO2 has been extensively studied due to its high photocatalytic activity, stability, non-corrosion and nontoxicity. However, several drawbacks hinder its widespread applications, such as the recombination of photoexcited electron–hole pairs and the large electronic band gap (anatase, 3.2 eV) [2]. One of the effective methods to overcome the drawbacks is doping nonmetal elements into TiO2, e.g. N, C, S and F [3–7], among which, nitrogen doping shows obvious advantages [8,9] due to the small ionic radius of nitrogen, unique properties of high thermal stability and low recombination centers. Previous studies [3,10,11] have also indicated that narrow optical band gap and high photocatalytic activity can be obtained by doping N into TiO2. So far, many techniques have been developed to synthesize N-doped TiO2, including sol–gel [12,13], hydrothermal method [14], oxidation of titanium nitride [15], pulsed laser deposition [16] and sputtering [17–21], among which TiO2 films deposited by sputtering exhibit uniform composition [22] and facile control of film thickness [23]. In addition, many methods have been proposed to enhance the water splitting by use of TiO2-based materials, such as constructing photoelectrochemical cells (PEC, consist of a single-crystal TiO2 photoanode and a Pt cathode) [1] and p/n-photoelectrolysis cells [24], adjusting pH value [25], adopting particulate photocatalytic system [26], employing NaOH coating [27] and adding cocatalyst (e.g. Pt, Au) [28]. However, there are few reports on using TiO2 films deposited by sputtering to generate hydrogen without the assistance of metal cathode, bias or loading noble metal. In this work, N-doped TiO2 films were prepared by RF magnetron sputtering in a mixture of O2, N2 and Ar. The intrinsic properties of the films were investigated by XRD, SEM, AFM, XPS and UV–vis spectrophotometer. Hydrogen production was examined by immersing the as-deposited films into 10% aqueous methanol solution. It is found thatthe photocatalytic activity of N-doped TiO2 film in water splitting was far higher than that of undoped TiO2 film and even Degussa P25 powders.                 

2.  2. Experimental N-doped TiO2 films were deposited on quartz glass substrates (2 × 4 cm) by RF reactive magnetron sputtering. These substrates were beforehand cleaned, sequentially with acetone, alcohol and de-ionized water, respectively for 20 min in the ultrasonic bath. The working gas was a mixture of N2 (99.999% pure), O2 (99.99% pure) and Ar (99.999% pure). The target was a 2-inch-diameter Ti metal plate (99.999% pure). The target-substrate distance was fixed at 70 mm and the base vacuum was 5.8 × 10−3 Pa. Before deposition, the substrate temperature was maintained at 400 ◦C for 2 h to degas and the target was pre-sputtered for 30 min by argon plasma. Afterwards, working gas was introduced into the chamber through three mass flow controllers and flow rate fixed at 5.0sccm (standard cubic centimeter per minute), 3.9sccm, 40sccm for N2, O2 and Ar respectively. The total working pressure was set at 0.3 Pa and the RF power was fixed at 130W. The whole deposition process was conducted at 450 ◦C for 4 h, during which the substrate rotated around its axis at 8 rotations per minute. Pure TiO2 film was also deposited under the same condition without N2 as reference. The crystalline structure of the as-deposited N-doped TiO2 films was identified by X-ray diffraction (XRD, Model D/Max 2550V, Rigaku, Japan). The thickness and morphology were determined by field emission scanning electron microscopy (FE-SEM, JSM6700F, JEOL, Japan). The surface roughness was estimated by atomic force microscopy (AFM, Nanoscope, NS3A-02, Veeco, USA). The transmittance spectra of the films were measured by a UV–vis spectrophotometer (Lambda 900, PerkinElmer, USA). The chemical compositions and valence-band spectra were determinated by X-ray photoelectron spectroscopy (XPS, ESCALAB 250, Thermo Scientific, USA). The hydrogen production was examined by the photocatalytic water splitting testing system (CEL-SPH2N, AULTT, China), irradiated by a 300W Xe lamp (CEL-HXF300, AULTT, China). The total output of the lamp was 50W and only 2.6W of incoming irradiance was obtained below 390 nm. The photocatalytic reaction was carried out in a quartz cell containing 100 mL aqueous methanol solution (CH3OH: H2O = 1:10 v/v). Before reaction,the as-deposited film was immersed in the aqueous solution and then oxygen was removed by a mechanical pump. The film-lamp distance was fixed at 12 cm. The amount of H2 produced was analyzed by a gas chromatograph (SP7800, AULTT, China) using N2 as carrier gas. 

3. 3. Results and discussion 3.1. Crystalline structure and morphology Fig. 1 shows the XRD patterns of N-doped TiO2 film. Diffraction peaks observed at 2 = 25.28◦, 36.95◦, 37.88◦, 38.55◦, 48.05◦, 54.09◦, 54.88◦, 62.67◦, and 68.76◦ correspond well with (1 0 1), (1 0 3),(0 0 4),(1 1 2),(2 0 0),(1 0 5),(2 1 1),(2 0 4), and (1 1 6) planes of anatase phase of TiO2 (JCPDS No. 21-1272), indicating that no other phase (e.g., rutile or brookite) is observed and N-doping has little effect on the crystal structure. Nevertheless, it can be seen that N-doping can greatly influence the growth orientation of anatase TiO2 particles. As shown in Fig. 1, the intensities of the (0 0 4), (1 1 2), (2 0 0), and especially (2 1 1) peaks become stronger while (1 0 1) peak become weaker for the N-doped TiO2 film, compared with those of the undoped TiO2 film. For example, the ratios of I(2 1 1) to I(1 0 1) [I(2 1 1) and I(1 0 1) are the intensities of (2 1 1) and (1 0 1) peaks, respectively] for the N-doped and undoped TiO2

   

   Fig. 1. XRD patterns of pure TiO2 and N-doped TiO2 films prepared by RF reactive magnetron sputtering

   
   

films are 100% and 33%. As is well-known, the surface energies of (0 0 1), (1 1 2), (1 0 0), and (2 1 1) planes are all higher than that of (1 0 1)plane [29]. Therefore,highphotocatalytic activity of N-doped TiO2 film is expected, due to its large percentage of exposed (0 0 4), (1 1 2), (2 0 0), and (2 1 1) facets. Fig. 2a illustrates the surface morphology of the N-doped TiO2 film, which exhibits a rough appearance of interconnected TiO2 nodules. Columnar structure is clearly observed from the cross-sectional image, as shown in Fig. 2b. The width of columnar nanograins is ∼40 nm, and the thickness of the film is 900 nm ± 20 nm. The root-mean-square (RMS) roughness of the film is 20.4 nm, measured by AFM (see Fig. 2c)