Ta₂O₅ (>99%) was purchased from Aladdin Industrial Corporation (USA). The other reagents (analytically pure) were purchased from Guangzhou Chemical Reagent Factory. All reagents were used as received without further purification. Double distilled water was used in all experiments.

A solid state reaction was used to synthesize pristine and Bi-doped NaTaO₃. NaBiO₃·2H₂O and Bi(NO₃)₃·5H₂O were used as the Bi source for Bi doping of the NaTaO₃ samples. A typical solid state reaction is described in the following. A mixture of the starting materials of Ta₂O₅, Na₂CO₃, and Bi source was pressed into pellets, calcined in air at 1000 °C for 10 h and then ground into fine powder. The powder was washed with distilled water, then filtered and dried at 120 °C for use. The molar ratio of Na:Ta:Bi was 1.05:1:0.02 in the starting mixture of Na₂CO₃, Ta₂O₅, and Bi precursor, where the 5% excess Na was to compensate the volatilization loss. The samples were denoted as V-0.02 and III-0.02, where V and III showed the valence state of the Bi source, and 0.02 is from the Bi:Ta ratio.

For comparison, Bi-doped NaTaO₃ with 0.02 molar ratio of Bi to Ta was prepared by the hydrothermal method based on the literature ^[25]. First, the starting materials of Ta₂O₅ and NaBiO₃·2H₂O were dispersed in NaOH aqueous solution (10 mol/L) by strong stirring for 2 h. Then, the slurry was poured into a Teflon-lined autoclave and heated at 180 °C for 48 h. After filtering and washing, the resulting powder was dried at 120 °C to get the final product. In this method, excess Na ion was introduced by a strong base. The sample was denoted as HV-0.02 NaTaO₃.

The crystal structure of the samples was analyzed by an X-ray diffractometer (XRD, Bruker, D8 Advance, Germany) with a Cu K_α radiation source at a scanning rate of 10°/min for all samples. Raman spectra were measured with a Laser Micro-Raman Spectrometer (Renishaw inVia) with a Ar⁺ laser with 633nm excitation. The chemical environment of doped Bi was studied by X-ray photoelectron spectroscopy (XPS, Kratos Axis Ultra XPS, ESCALab250, American; hemispherical electron analyser, 120W Al K_α X-ray source, hv 1486.6 eV). The C 1s reference of 284.1 eV was chosen for calibration. Diffuse reflection spectra (DSR) were recorded on a UV-Vis spectrophotometer (Shimadzu, UV-3150, Japan, equipped with an integrating sphere) to study the optical properties of the samples.

The photocatalytic activity of the samples was examined in a closed gas circulation and evacuation system. Typically, 0.1 g catalyst powder was dispersed in 100 mL methanol aqueous solution (20 vol%) in a Pyrex reaction cell. Pt was photodeposited on the catalyst with 0.3 wt% loading in situ using a H₂PtCl₆·6H₂O precursor under irradiation. The light source was a 300 W Xe lamp (Beijing Changtuo, PLS-SXE-300UV) with an intensity of 200 mW/cm². The light absorption spectra were shown elsewhere ^[27]. The amount of produced hydrogen was analyzed using an online gas chromatograph (Shanghai Tianmei, GC-7890II, China) with a thermal conductivity detector and a molecular sieve 5A column and N₂ carrier.

Figure 1 shows the XRD patterns of the NaTaO₃ and Bi-doped NaTaO₃ samples. The XRD pattern of NaTaO₃ was in agreement with PDF ^#74-2480 (P2/m with a = c = 0.3907 nm, b = 0.3904 nm, and α = γ = 90°, β = 90.1°), and indicated the crystal structure to be the monoclinic phase. For the Bi-doped NaTaO₃ samples, the diffraction patterns were similar to pristine NaTaO₃, and no additional peak was detected. This suggested that the monoclinic phase of NaTaO₃ remained intact and no impurity existed in Bi-doped NaTaO₃. It was reported that the Na/Ta molar ratio of the starting materials affects the site occupancy of Bi at the A- or B-site in the NaTaO₃ lattice, and Bi predominantly occupied the Na-site under a Na-deficient condition ^[24]. In our case, though 5% excess Na ion was introduced in the synthesizing process, the Na:Ta molar ratio in pristine NaTaO₃ was 1:1.9 from the quantitative analysis by XPS. Thus, it was proposed that the Bi ion was located at the A site because of the low Na/Ta ratio.

Fig. 1. XRD patterns of NaTaO3 (1), V-0.02 NaTaO3 (2) and III-0.02 NaTaO3 (3).

Figure 2 shows the Raman spectra of the NaTaO₃ and Bi-NaTaO₃ samples. Two distinct peaks at 578 and 860 cm^-1were observed for both the V-0.02 and III-0.02 samples, which were absent from pristine NaTaO₃. These two peaks corresponded to the Stokes-shifted signals from the transverse and longitudinal optical phonons ^[28]. The appearance of these two peaks indicated that Bi ion doping induced a band gap change in NaTaO₃ ^[29].

Fig. 2. Raman spectra of NaTaO3 (1), V-0.02 NaTaO3 (2) and III-0.02 NaTaO3 (3).

In addition to the energy band structure, Raman spectroscopy of solids is extremely sensitive to the local deviation from the average periodicity, thereby making it a valuable probe of the local structure. The region of 850 to 950 cm^-1 was assigned to the valence vibrations of the Ta-O-Ta bridges that were inactive in the Raman spectra for the centrosymmetric structure ^[30]. The appearance of 860 cm^-1for Bi-doped NaTaO₃ indicated a distorted B-O-B bond from the bond angle of 180°, which has an impact on the photo-induced charge carrier migration in the photocatalyst ^[14].

The chemical environment and oxidation stateof Bi in NaTaO₃ were studied by X-ray photoelectron spectroscopy (XPS). Figure 3 shows XPS spectra of the samples. The Bi 4f_7/2 and 4f_5/2 peaks were at 158 and 164 eV, respectively. For comparison, the XPS spectra of Bi 4f in Bi₂O₃, NaBiO₃, BiTaO₄ and HV-0.02 (synthesized with the hydrothermal method) were also measured to deduce the position and valence state information of the doped Bi ion. The binding energy (BE) of Bi 4f_7/2 for all the samples is summarized in Table 1.

Table 1 Bi 4f7/2 binding energy of Bi2O3, HV-0.02 NaTaO3, NaBiO3, III-0.02 NaTaO3, BiTaO4 and V-0.02 NaTaO3. Sample Binding energy of Bi 4f2/7 (eV) V-0.02 NaTaO3 159.0, 159.5 III-0.02 NaTaO3 158.9 BiTaO4 159.0 NaBiO3 158.8 Bi2O3 158.3 HV-0.02 NaTaO3 158.7

Table 1 Bi 4f7/2 binding energy of Bi2O3, HV-0.02 NaTaO3, NaBiO3, III-0.02 NaTaO3, BiTaO4 and V-0.02 NaTaO3.

Fig. 3. XPS spectra of Bi 4f7/2 and Bi 4f5/2 of Bi2O3, HV-0.02 NaTaO3, NaBiO3, III-0.02 NaTaO3, BiTaO4 and V-0.02 NaTaO3.

In the reference samples, Bi₂O₃ and NaBiO₃ were the precursors of Bi with the +3 and +5 chemical valence, respectively. The BE positions of Bi 4f_7/2 in Bi₂O₃ and NaBiO₃ were 158.3 and 158.8 eV, respectively, and the 0.5 eV chemical shift was ascribed to the oxidation state difference. For BiTaO₃ with the perovskite structure, the Bi ion is located at the A-site with +3 chemical valence. The BE position of BiTaO₃ was 159.0 eV, which was higher than 158.3 eV for Bi³⁺ in Bi₂O₃ and 158.8 eV for Bi⁵⁺ in NaBiO₃. This suggested that the Bi chemical environment exhibited more influence on its BE than its oxidation state, and the A-site Bi substitution for Na can significantly increase the BE.

For HV-0.02 NaTaO₃, the Bi ion is located at the B-site in NaTaO₃ with the +5 chemical valence ^[25]. HV-0.02 NaTaO₃ was synthesized by the hydrothermal method with the strong base of sodium hydroxide, where the low hydrothermal temperature avoided the decomposition of NaBiO₃ to Bi³⁺ and excess sodium ion in the strong base made the Bi⁵⁺ substituted for Ta⁵⁺ at the B-site in NaTaO₃. The position of the BE in HV-0.02 NaTaO₃ was 158.7 eV, which was comparable to the 158.8 eV for Bi⁵⁺ in NaBiO₃. Thus, it was proposed that the B-site Bi substitution for Ta in NaTaO₃ had little impact on the BE position of the Bi ion.

For the III-0.02 NaTaO₃ sample synthesized with the Bi₂O₃ precursor, the position of the BE was 158.9 eV, which was close to 159.0 eV for Bi³⁺ at the A-site in BiTaO₃. Therefore, it was deduced that Bi³⁺ substituted for Na⁺ at the A-site in the III-0.02 sample. For the V-0.02 sample synthesized with the NaBiO₃ precursor, the BE position was higher than for III-0.02 in Fig. 3. Due to the pyrolysis of NaBiO₃, it was reasonable that Bi³⁺ and Bi⁵⁺ would coexist in the V-0.02 sample under the high temperature synthesis condition. Therefore, the XPS curve of V-0.02 was fit by two peaks with the BE positions at 159.0 and 159.5 eV. The lower BE at 159.0 eV was the same as the BE for Bi³⁺ at the A-site in BiTaO₃, therefore, this peak was identified as the Bi³⁺ substitution for Na in the V-0.02 sample. For the other peak at 159.5 eV in V-0.02, three possible assignments should be taken into account. First is the A-site Bi⁵⁺ substitution for Na in V-0.02, second is the B-site Bi⁵⁺ substitution for Ta in V-0.02, and third is the Bi³⁺ in a different chemical environment reported in a previous work ^[31].

In the above paragraph, it was discussed that the BE of Bi⁵⁺ substitution for Ta in HV-0.02 was 158.7 eV, which is far lower than 159.5 eV. So the second possibility of the B-site Bi⁵⁺ substitution for Ta was excluded. Based on the BE difference of Bi₂O₃ and III-0.02 (Bi³⁺ in A-site), and of NaBiO₃ and HV-0.02 (Bi⁵⁺ in B-site), it was proposed that the A-site substitution can more significantly increase the BE of the Bi ion than the B site, and the chemical environment had a more significant influence on the BE than the oxidation state. For Bi³⁺ in a different chemical environment proposed in the literature ^[31], the BE should not be as high as 159.5 eV, considering its +3 oxidation state and that the chemical environment was different from the A-site. That is, the third possibility of Bi³⁺ in a different chemical environment was also excluded. Therefore, we assigned the 159.5 eV peak to the A-site Bi⁵⁺ substitution for Na from an overall consideration of the chemical states. This remains to be further explored in future work.

Thus, we identified that the Bi³⁺ doping occurred at the A-site in the III-0.02 sample, while Bi³⁺ and Bi⁵⁺ co-doping at the A-site occurred in the V-0.02 sample. It was worth mentioning that the Bi⁵⁺ content in V-0.02 was far smaller than Bi³⁺ from the two components of the XPS fitting in Fig. 3. That is, most of NaBiO₃ was pyrolysed to Bi³⁺, and only a small amount of Bi⁵⁺ was doped into the NaTaO₃ cell. The Bi³⁺ and Bi⁵⁺ co-existence in NaTaO₃ induced the production of Na vacancies or interstitial O atoms in the lattice for charge compensation, which will have an impact on the photocatalytic process and interstitial charge migration ^[23].

Diffused reflectance spectra of pristine and Bi-doped NaTaO₃ samples are shown in Fig. 4. The steep edges and intense absorption of NaTaO₃ originated from the transition from the valence band to the conduction band. A shoulder peak on the steep absorption edge of NaTaO₃ was observed in the Bi-doped V-0.02 and III-0.02 samples, which indicated that doped Bi ions extended the absorption edge to a longer wavelength. The valence band (VB) was mainly composed of O 2p orbitals while the conduction band (CB) mainly consists of Ta 5d orbitals for NaTaO₃ ^[18]. This shoulder peak in the absorption spectra can be attributed to the transition from the localized impurity state of the Bi ion dopant. An isolated donor energy state was formed in the forbidden band of NaTaO₃ during Bi doping, from considering the energy structure of BiTaO₄ ^[32].

Fig. 4. UV-Vis diffused reflectance spectra of NaTaO3 (1), V-0.02 NaTaO3 (2) and III-0.02 NaTaO3 (3).

The photocatalytic activity of pristine and Bi-doped NaTaO₃ samples were examined by splitting water into H₂ under simulated solar light. For all samples, H₂ generation was linear with irradiation time as shown in Fig. 5. The highest activity was observed for the V-0.02 sample, and the photocatalytic activity of the III-0.02 sample was comparable with that of NaTaO₃.

Fig. 5. Photocatalytic H2 evolution of NaTaO3 (1), V-0.02 NaTaO3 (2) and III-0.02 NaTaO3 (3).

In the absorption spectra, almost similar absorption profiles were observed for the V-0.02 and III-0.02 samples. The photocatalytic activity distinctly increased with the V-0.02 sample, but remained unchanged with III-0.02. Due to the low Bi ion content in NaTaO₃, the slight extending of light absorption to the visible region contributed little to the photocatalytic activity. In addition to light absorption for charge carrier generation, charge transfer inside the photocatalyst is an important factor that affects the photocatalytic activity. The Raman results indicated that Bi ion doping induced a distorted B-O-B bond from the bond angle of 180°, which was detrimental for charge carrier migration in NaTaO₃. The Bi ion dopant induced the generation of a Na vacancy or an interstitial O in the lattice for charge compensation. These defects served as charge traps that promote the separation of the charge carriers and therefore the photocatalytic activity ^[27]. The separate actions of the positive and negative factors in the III-0.02 sample led to unchanged photocatalytic activity compared to pristine NaTaO₃. For the V-0.02 sample, it was found in the XPS data that both Bi⁵⁺ and Bi³⁺ ions were doped into the A-site. Compared to the III-0.02 sample, there were more defects in the V-0.02 sample that existed as high valence state Bi⁵⁺ ions in the lattice. That is, the separation efficiency of the charge carriers in the V-0.02 sample was higher than III-0.02. Therefore, a higher photocatalytic activity was obtained with V-0.02. Our result indicated that the occupation by a high valence ion at the A-site in the ABO₃ perovskite structure was more favorable for the enhancement of photocatalytic activity.