Nowadays, human society relies on fossil fuels for energy. However, the depletion of fossil resources and increasing environmental concerns about the emission of large amounts of CO₂ during the combustion of fossil fuels have emphasized the need to develop renewable and clean energy resources. Therefore, great efforts have been made to reduce the use of fossil fuels that increase the concentration of CO₂ in the atmosphere. Photocatalytic CO₂ reduction to produce valuable organic fuels, such as methane, formic acid, and methanol, is of particular interest because of its environmental friendliness and ability to mimic natural photosynthesis to transform solar energy into chemical energy ^{[1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11]}. Among the semiconductors, TiO₂ has been studied worldwide and proven to be the most promising due to its biological and chemical inertness, strong oxidation power, and low cost ^{[12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25]}. However, the practical application of TiO₂ is limited by low visible light absorption ability and fast electron-hole recombination, which significantly reduce the photoconversion efficiency. Recently, it has been reported that the photocatalytic properties of TiO₂ greatly depend on its structure and morphology ^{[26, 27]}. Therefore, tailoring the TiO₂ morphology, crystal architecture, and surface and textural properties is effective strategies to modify and improve its photocatalysis for practical applications ^[28].

The design of the TiO₂ morphology with specific efforts to control exposed crystal faces is receiving a lot of attention. Crystal face engineering is an efficient way to improve photocatalytic performance. Since the first successful preparation of single crystal anatase TiO₂ sheets with 47% (001) faces by Yang et al. ^[29], many methods have been developed to fabricate anatase TiO₂ dominated by the (001) face, including hydrothermal, solvothermal, and high temperature gas phase reactions. Han et al. ^[30] demonstrated that nano-sized titania sheets with 89% (001) faces can be synthesised by an aqueous synthesis route. The formation of mainly (001) crystal faces was achieved by using a capping agent, which facilitated the formation of these crystal faces by lowering their free energy and making them more stable ^{[30, 31, 32]}. A report by Selloni ^[33] indicated that there was a strong preferential interaction between fluorine and the (001) faces of anatase TiO₂ crystals. Meanwhile, the formation of anatase (001) faces largely depended on the presence of plentiful fluorine under acidic conditions. This was attributed to the strong coordination between Ti⁴⁺ and F-, which was enhanced by the acid and which preserved the TiO₂ (001) faces ^{[33, 34]}. It was reported that fluorine enhanced the crystallization and promoted the growth of the anatase phase of TiO₂ nanosheets with mainly (001) faces for higher photocatalytic activity ^[34].

Recently, it was revealed that the co-existence of exposed (001) and (101) faces on anatase TiO₂ had specific synergetic effects, which showed a higher photocatalytic activity than that of TiO₂ with mainly (001) or (101) faces. This is because the difference in the energy levels of the (001) and (101) faces drives the electrons and holes to diffuse to different crystal faces ^[34], resulting in faster charge separation rates. Further, Yu et al. ^[34] proposed a novel “surface heterojunction” concept based on density functional theory (DFT) calculations, which explained the superior photocatalytic activity of TiO₂ with both (001) and (101) faces. Here, we report the fabrication of visible light responsive N-doped TiO­₂ microsheets (N-TiO₂ MS) with exposed (101) and (001) faces that formed surface heterojunctions by a simple hydrothermal method. Compared with commercial TiO­₂ (P25) and the precursor TiN, the N-doped TiO­₂ sample exhibited a higher photocatalytic CO₂ reduction activity. The surface heterojunction formed by the (001) and (101) faces enhances electron-hole separation, which was shown by electrochemical impedance spectroscopy (EIS) analysis. This work provides new insight into the design and preparation of new highly efficient visible light responsive photocatalytic materials.

The TiO₂ sample was synthesised by the hydrothermal method. TiN (0.9 g) was mixed with 1 mL HF, 3 ml HCL, and 26 mL deionized water under constant stirring for 1 h. The mixture was transferred into a dry Teflon-lined stainless steel autoclave and heated at 200 °C for 18 h. The resulting suspension was cooled to room temperature. Then, the sample was washed three times with deionized water and ethanol. Finally, the sample was dried in an oven at 80°C for 10 h and labelled N-TiO₂ MS. For comparison, the precursor TiN was treated under the same condition without HF and HCl. In addition, another N-doped TiO₂ nanoparticle (N-TiO₂ NP) sample was prepared using tetrabutylorthotitanate as precursor with 1 wt% NH₃·H₂O solution according to a previous study ^[35] for visible light comparison.

The surface morphology of the as-prepared samples was characterized by a JSM-6510 microscope (JEOL, Tokyo, Japan) at an accelerating voltage of 20 kV. The crystalline structure of the samples was analyzed by an XRD diffractometer (Rigaku, Japan) with Cu K_α radiation (λ = 0.15418 nm) at a scanning rate of 0.04°/s. The Brunauer-Emmett-Teller (BET) specific surface areas (S_BET) of the samples were determined by nitrogen adsorption data obtained on a Micromeritics ASAP 3020 nitrogen adsorption apparatus (USA). All the samples were degassed at 180 °C prior to the nitrogen adsorption measurement. A desorption isotherm was used to determine the pore size distribution by the Barret-Joyner-Halender (BJH) method, by using a cylindrical pore model ^[36]. The nitrogen adsorption volume at the relative pressure (p/p₀) of 0.994 was used to calculate the pore volume and average pore size. The XPS measurement was performed in an ultrahigh vacuum VGESCALAB 210 electron spectrometer equipped with a multichannel detector using Mg K_α (1253.6 eV) radiation (operated at 200 W) of a twin anode in the constant analyzer energy mode with a pass energy of 30 eV. All the binding energy was referenced to the C 1s peak at 284.8 eV of surface adventitious carbon. Survey scans and detailed scans of the Ti 2p, C 1s, O 1s, F 1s, and N 1s photoelectron peaks were recorded for the samples. Optical absorption by the samples was characterized by a UV-visible spectrophotometer (UV-2550, Shimadzu, Japan). BaSO₄ was used as a reflectance standard in the UV-visible diffuse reflectance experiment.

The electrochemical behavior of the samples was determined by EIS. EIS was performed in a three electrode system (CH1660C instruments, CHI, China) using Ag/AgCl as the reference electrode, Pt wire as the counter electrode, and FTO as the working electrode. Na₂SO₄ aqueous solution (0.5 mol/L) was used as the electrolyte. The samples were coated on a 2 cm x 1.5 cm fluorine-tin oxide (FTO) glass electrode and dried at room temperature for 24 h. The Nyquist plot was used out to analyze the measured results.

The photocatalytic reduction of CO₂ was carried out in a 200 mL capacity homemade Pyrex reactor at room temperature and atmospheric pressure described in our recent reports ^{[37, 38]}. In a typical photocatalytic CO₂ reduction experiment, 0.1 g of the sample was suspended in 20 mL of deionized water in the reactor and ultra-sonicated and dried in an oven at 80 °C to form a thin film at the bottom of the Pyrex reactor. Prior to illumination, the reactor was tightly sealed and fastened using a rubber septum. Then, ultra-pure nitrogen gas was blown through the system for 30 min to remove air and put the reaction system into an anaerobic condition. A chemical reaction between 0.12 g NaHCO₃ (added into the reactor before sealing) and 0.3 mL HCl aqueoussolution(4 mol/L), which was added into the reactor by a syringe, was used to produce CO₂ and H₂O vapour in situ. Then, the reactor was irradiated for 2 h with a Xe arc lamp with a power of 300 W without (as a UV-visible full spectrum light source) and with a cut filter of 400 nm as a visible light source, which was placed 10 cm above the photocatalytic reactor. After the photocatalytic CO₂ reduction reaction, 1 mL of mixed gas was taken out from the reactor and analyzed by a gas chromatograph (GC2014C, Shimadzu, Japan) equipped with a flame ionization detector (FID) and methanizer. Products were analyzed on the basis of the retention time of the standard gaseous sample. The amount of methanol produced after the reaction was quantified as a major product of the reaction.

The phase structure of the TiO₂ sample is important for determining its photocatalytic activity. Figure 1 shows the XRD patterns of the different samples. The XRD pattern of the N-TiO₂ MS sample showed the formation of pure anatase (JCPDS No. 21-1272) with no residual TiN phase. Further analysis of the pattern indicated that the intensity of the diffraction peak of the N-TiO₂ MS sample was more intense and sharper than that of P25, suggesting the formation of well-crystallized TiO₂. This result is in good agreement with previous reports that surface fluorination and an acidic synthesis condition improved the growth and crystallization of TiO₂ crystallites ^[39].

Fig. 1. XRD patterns of TiN, as-prepared N-TiO2 MS, and P25 samples.

According to the Wulff construction, the high surface energy (001) faces grow and diminish rapidly, and thus a conventional anatase TiO₂ grows into a unique ocatahedral bipyramid shape, and 90% of the exposed faces are the thermodynamically stable (101) face ^[29]. The morphology of our samples was studied by SEM. As shown in Fig. 2, N-TiO₂ MS exhibited a characteristic microsheet morphology with a high percentage of the (001) face. It has an average side length of 3.5 μm and thickness of 0.3 μm, indicating the formation of larger crystallites and an enhanced crystallization in comparison to P25. The percentage of the exposed (001) faces on as-prepared N-TiO₂ MS was determined to be 65% from the FESEM image (Fig. 2(c)) using Eqs. (1)-(3) below ^[40].This is significantly higher than that on natural TiO₂ (around 10%).Note that this exposed percentage of (001) faces (65%) is very close to that reported recently (58%) for the optimized formation of a surface heterojunction ^[34]. This is an indication that a properly exposed ratio of (001) and (101) faces was achieved. Therefore, a good photocatalytic activity can be expected.

Fig. 2. SEM images of as-prepared N-TiO2 MS (a) and TiN (b), and FESEM image of N-TiO2 MS showing (001) and (101) faces (c).

The percentage of exposed (001) faces was calculated from Eqs. (1)-(3) ^[40]:

where l and d are the average length and average thickness, respectively, of the sheets measured from SEM images, S₀₀₁ is the area of the (001) face, S₁₀₁ is the area of the (101) face, P is the percentage of exposed the (001) face, and θ is the value for the angle between the (001) and (101) faces of anatase TiO₂.

It is well-known that a high surface area and large pore structure are beneficial for improving photocatalytic activity. This is because of more surface active sites for the adsorption of reactant molecules and ease of transport of reactant molecules and products through the interconnected pore network ^{[41, 42]}. The specific surface area and pore structure of the as-prepared N-doped TiO₂ samples were investigated by nitrogen adsorption isotherms. The results are shown in Table 1 and Fig. 3. The isotherm of the as-prepared N-TiO₂ MS sample was type IV of the BDDT (Brunauer-Deming-Deming-Teller) classification, indicating the presence of mesopores. Moreover, the isotherm with an H3 hysteresis loop suggested that there were slit-shaped pores formed within aggregates of plate-like particles ^{[43, 44, 45]}. The pore size distribution determined from the desorption branch of the nitrogen isotherm by the BJH (Barrett-Joyner-Halenda) method is shown in the inset of Fig. 3. This indicated a wide pore diameter distribution in a range of 10-100 nm centered at 40 nm for the as-prepared N-TiO₂ MS sample. This mesoporous structure would be useful as the transport pathway during the photocatalytic reaction by allowing the rapid diffusion of reactant molecules in the pores.

Table 1 Physical properties and estimated band gap energy of the samples. Sample (001) face (%) S BET (m 2/g) Average pore size (nm) Pore volume (cm 3/g) Band gap energy (eV) TiN - 1.6 6.2 < 0.012 - N-TiO 2 MS 65 5.0 7.8 < 0.010 3.2

Table 1 Physical properties and estimated band gap energy of the samples.

Fig. 3. Nitrogen adsorption isotherms of TiN and as-prepared N-TiO2 MS. The inset shows the pore size distribution curves.

The XPS measurements were carried out to investigate the chemical state and surface chemical composition of the samples. Figure 4(a) shows the XPS survey spectra of the as-prepared N-TiO₂ MS and precursor TiN samples. Sharp photoelectron peaks appeared at the binding energy of 458 (Ti 2p), 531 (O 1s), 684 (F 1s), and 285 eV (C 1s). The C 1s peak resulted from residual carbon on the sample and adventitious hydrocarbon from the XPS instrument itself.

Fig. 4. XPS survey spectra (a) of as-prepared N-TiO2 MS and TiN, and high resolution XPS spectra of Ti 2p (b), F 1s (c), and N 1s (d) of as-prepared N-TiO2 MS.

Two characteristic peaks at the binding energy of 458.8 and 464.5 eV (Fig. 4(b)) were observed for Ti 2p_3/2 and Ti 2p_1/2, respectively, indicating that the oxidation state of the Ti element in the as-prepared N-TiO₂ MS sample was the same as that of bulk TiO₂. The high resolution spectrum of O 1s showed a core level peak at 530 eV, which corresponded to the lattice oxygen associated with Ti-O-Ti linkages in TiO₂, and a shoulder peak at 531.4 eV, which came from the hydroxyl group (-OH) of adsorbed water molecules (not shown here). An important feature was that one characteristic high resolution F 1s XPS peak at the binding energy of 684.5 eV appeared for the as-prepared N-TiO₂ MS sample, as shown in Fig. 4(c). This peak indicated that F ions were adsorbed on the surface of the TiO₂. More significantly, the high resolution N 1s XPS peak of as-prepared N-TiO₂ MS, shown in Fig. 4(d), showd one wide N 1s peak centered at 395 eV, which was attributed to nitrogen doping in the TiO₂ microsheets with a moderate doping content (surface atomic concentration of 1.98% from XPS) derived from the TiN precursor. Usually, N doping is very hard to achieve in highly crystalline TiO₂ microsheets, yet, it is crucial for visible light absorption.

The UV-visible spectra provide insightful information into the interactions of the photocatalytic materials (i.e., TiO₂ samples) with photons of different energy.Figure 5 shows the comparison of the optical absorption spectra of TiN, as-prepared N-TiO₂ MS, and reference P25. The TiN precursor showed strong visible light absorption with an obvious decrease around 500 nm. In contrast, the reference P25 exhibited almost no detectable visible light absorption above 400 nm. Notably, as-prepared N-TiO₂ MS showed an obvious red shift in the absorption edge as compared to the reference P25. Moreover, strong absorption in the whole visible light spectrum was observed for as-prepared N-TiO₂ MS, which is consistent with its dark colour inherited from the TiN precursor. Therefore, it was not surprising that the prepared N-TiO₂ MS sample exhibited visible light photocatalytic activity.

Fig. 5. UV-visible diffuse reflection spectra of TiN, as-prepared N-TiO2 MS, and P25.

Many studies have demonstrated that charge recombination versus separation and transport is an important factor in the performance of a photocatalyst. In order to develop a photocatalyst with a high photoconversion efficiency, a reduced e^-/h⁺ recombination rate and enhanced electron transport are desirable. In this study, EIS was carried out to investigate the interfacial charge transport process. Figure 6 depicts the Nyquist plot of the three samples, where the radius of each arc correlated with the charge transporat ability of the corresponding photocatalyst. Generally, a larger radius of the arc indicates a lower ability to transport charges ^{[46, 47]}. It can be seen that as-prepared N-TiO₂ MS showed a much smaller arc than that for the TiN precursor or reference P25 samples. Several reasons account for the higher separation and transport efficicency of charge carriers in N-TiO₂ MS. First, the exposure of (001) and (101) faces with a proper ratio on N-TiO₂ MS faciliated the formation of surface heterojunctions and significantly enhance charge separation and transport. Second, the highly crystalline feature of N-TiO₂ MS favored the separation and transport of charge carriers. Finally, the F^- ions on the surface of TiO₂ were responsible for a reduction of the recombination rate of photogenerated electrons and holes ^[48].

Fig. 6. Nyquist plots of the TiN, as-prepared N-TiO2 MS, and P25 samples.

The photocatalytic CO₂ reduction activity of the samples was investigated under both full spectrum light irradiation and visible light (with 400 nm cutoff filter) irradiation. The detected product over the N-TiO₂ MS sample is shown in Fig. 7. The control experiment showed that no hydrocarbon compound was detected in the absence of either photocatalyst or light irradiation, suggesting that the resulting hydrocarbon compounds were produced by a photocatalytic reaction on the photocatalyst. In particular, under both full spectrum light irradiation and visible light irradiation, CO₂ was mainly transformed into methanol on the N-TiO₂ MS sample during the photocatalytic reduction reaction (Fig. 7(a) and (b)). In contrast, no product was detected for reference P25 under visible light irradiation, and the dominant product was methane for reference P25 under full spectrum light irradiation.

Fig. 7. GC chromatograms for the photocatalytic CO2 reduction over TiN, N-TiO2 MS, and P25 under full spectrum light irradiation (a) and visible light irradiation (b) for 2 h.

Figure 8 shows a comparison of the photocatalytic methanol production rates by the TiN precursor, N-TiO₂ MS, reference P25, and N-TiO₂ NP samples as photocatalyst. TiN showed no UV or visible light photocatalytic activity because of its very high charge recombination rates (based on the EIS analysis). P25 also showed no methanol production under visible light irradiation because it has no visible light absorption ability (Fig. 5). In contrast, N-TiO₂ MS and N-TiO₂ NP exhibited obvious photocatalytic activity for methanol production under visible light irradiation because of the nitrogen doping. Also, the UV activity under full spectrum light irradiation for producing methanol was considerable on N-TiO₂ MS relative to reference P25 and N-TiO₂ NP (Fig. 8). The high photocatalytic activity of the N-TiO₂ MS sample was attributed to the exposed 65% (001) and 35% (101) faces on the TiO₂, which helped build surface heterojunctions on the TiO₂ microsheet surface that favored charge separation and transport. In particular, the photogenerated electron migrated through the surface heterojunction to (101) faces for the photoreduction process, while the photogenerated hole was transferred through the surface heterojunction to (001) faces for a photooxidation process. As a result, the electron-hole recombination rate was greatly reduced. Moreover, the existence of the F^- ions on the TiO₂ sample acted as a mediator of interfacial charge transfer, which significantly reduced the electron-hole recombination. In summary, the N-TiO₂ MS sample exhibited superior visible light photocatalytic performance due to the synergistic effect of the N doping, surface heterojunction, and surface fluorination.

Fig. 8. Comparison of CH3OH production rates over TiN, N-TiO2 MS, P25, and N-TiO2 NP under full spectrum light irradiation (1) and visible light irradiation (2) for 2 h.

N-doped anatase TiO₂ microsheets with both exposed (101) and (001) faces that formed surface heterojunctions were fabricated by a simple hydrothermal method using TiN as precursor in the presence of HF as capping agent and HCl for pH control. The prepared TiO₂ sample exhibited 65% (001) and 35% (101) exposed faces, high crystallinity, a large surface area, and desirable wide spectrum optical response. The prepared TiO₂ sample was chemically modified by N doping and surface fluorination. As a result, the prepared N-doped TiO₂ sample showed significantly higher photocatalytic activity than TiN and P25 under visible light. This was due to the synergistic effect of the surface heterojunction, N doping, and surface fluorination. In particular, the surface heterojunctions improved electron-hole separation. N doping favored visible light absorption. Surface adsorbed F^- ions act as a mediator for interfacial charge separation. The surface fluorinated N-doped TiO₂ with exposed (001) and (101) faces has potential applications in gas sensors, photocatalysts, solar cells, biomedical engineering, and photonic and optoelectronic devices.