Effect of Rh valence state and doping concentration on the structure and photocatalytic H₂ evolution in (Nb,Rh) codoped TiO₂ nanorods†

Jiquan Huang *^ab, Ting Lv ^a, Qiufeng Huang ^a, Zhonghua Deng ^a, Jian Chen ^b, Zhuguang Liu ^b and Guo Wang *^ab
aFujian Science & Technology Innovation Laboratory for Optoelectronic Information of China, Fuzhou, 350108, P. R. China. E-mail: hjq@fjirsm.ac.cn; guowang@fjirsm.ac.cn
^bKey Laboratory of Optoelectronic Materials Chemistry and Physics, Fujian Institute of Research on the Structure of Matter, Chinese Academy of Sciences, Fuzhou, 350002, P. R. China

First published on 21st October 2020

---

---

Introduction

The increasing demand for global energy and environmental protection has accelerated the search for renewable energy sources as an alternative to fossil fuels. As an energy carrier with high energy capacity and ultra-low environmental load, H₂ has been considered the next-generation fuel, which gives rise to wide concern about its promotion to industrial production.^1–3 An ideal way to produce H₂ is the direct conversion of solar energy into chemical energy through photocatalytic water splitting.^4–6 However, the solar-to-hydrogen (STH) conversion efficiency is so limited that it could not be applied for large-scale applications.⁷ It is known that the STH conversion efficiency depends mainly on the capacity for sunlight absorption and the corresponding quantum yield of the photocatalysts used in the photocatalytic water splitting system.^7–9 For decades, hundreds of semiconductors have been developed and evaluated as potential photocatalyst materials, and their morphologies and bandgap structures have been widely engineered to optimize solar light harvesting, electron/hole separation, and long-term stability.^1,7,10–13 Yet still, semiconductors with H₂ production performance exceeding that of TiO₂ have rarely been reported.

By virtue of its inexpensive, nontoxic, stable and durable qualities, TiO₂ exhibits outstanding UV photocatalytic performance.^14–16 Great progress has been made in developing TiO₂-based semiconductors for photocatalytic water splitting. One-dimensional (1D) anatase TiO₂ nanostructures (such as nanotubes, nanorods, and nanowires) are of particular interest due to their high surface area, abundant surface active sites, efficient electron/hole separation, and the great oxidizing ability of photogenerated holes.^15–18 For example, a high STH conversion efficiency of 1.05% was achieved in a cross-linked TiO₂ nanowire photoanode,¹⁹ and a photon-to-current conversion efficiency of nearly 100% at 330 nm was achieved in Pd-sensitized TiO₂ nanotube arrays.¹⁵ Very recently, we reported the fabrication of dendritic nanorod arrays with an H₂ evolution rate of 44.7 mmol g⁻¹ m⁻² under simulated sunlight and an apparent quantum yield of 63.8% at 365 nm.²⁰ However, the STH conversion efficiency of TiO₂ is still too low to meet the target value for possible commercial applications, since a wide bandgap of 3.2 eV (for anatase) prevents it from harvesting visible light (note: UV and visible light account for 5% and 52% of the solar spectrum, respectively).^15–17 Thus, it is of great importance to broaden the light absorption from UV to the visible-light region.

Many strategies have been explored to break through the limitations of the wide bandgap, one of which is doping.²¹ Doping of appropriate element(s) into the TiO₂ crystal lattice creates donor or acceptor levels in the forbidden gap.²² Many metal and non-metal elements, such as Fe, Zr, Cr, Al, V, Rh, Ru, Mo, C, N, F, B, Si, and P, have been proved to be effective dopants for enhancing visible-light absorption.^21–25 However, it is found that the expected significant improvements in STH conversion efficiency were not achieved, and even the photocatalytic activity under UV light declined drastically, due to an increase in electron/hole recombination.²² Obviously, both the limited absorption of visible light and the fast recombination of photoinduced electron/hole pairs must be addressed to improve the STH conversion efficiency. In this regard, one notable discovery is that a trivalent Rh(III) dopant with a d⁶ electron configuration can bring about an in-gap donor level without introducing additional recombination centres.²⁶ The electron transition from the Rh(III) donor level to the conduction band (CB) makes Rh-doped TiO₂ a highly active visible-light-driven photocatalyst.^16,26–29 Furthermore, Rh(III) doping can even promote electron–hole separation while increasing the photo-induced carrier density.¹⁶ However, in Rh-doped TiO₂ (and titanates such as SrTiO₃), Rh ions prefer to exist in the form of Rh(IV) rather than Rh(III) for charge neutralization.²⁷ Rh(IV) has an open shell t_2g⁵ configuration, which leads to paramagnetic dopant centres acting as trapping sites.²⁶ Therefore, additional compensating dopants such as Sb(V), Nb(V), and Ta(V) should be presented to suppress the oxidation of Rh(III) to Rh(IV). Recently, we reported the fabrication of an (Nb,Rh) codoped TiO₂ semiconductor (Ti_0.996Nb_0.002Rh_0.002O₂ nanorods) by a two-step method.¹⁶ This photocatalyst displays ultra-high activity whether under ultraviolet or visible light. Its H₂ production rate is far superior to that of undoped TiO₂ nanorods, (Sb,Rh) codoped rutile TiO₂, or Pt-loaded P25, which implies that (Nb,Rh) codoping into TiO₂ (and titanate) is rather efficient for the development of a visible-light-responsive photocatalyst with a high quantum yield and high STH conversion efficiency.

In the present work, we fabricated a series of anatase phase (Nb,Rh) codoped TiO₂ nanorods to gain insight into the maximum improvement in photocatalytic performance. The factors that control visible-light absorption, carrier transfer and recombination, and the H₂ evolution rate were revealed by regulating the dopant rations and doping concentrations, as well as tailoring the morphology and crystal phase. It is expected to provide a useful reference for evaluating the potential industrial application of (Nb,Rh) codoped TiO₂-based photocatalysts in this study.

Experimental section

Synthesis of photocatalysts

Nb and/or Rh doped TiO₂ photocatalysts were synthesized by a two-step method, which was described in our previous paper.¹⁶ The first step was to prepare protonated titanate nanotubes by a sol–gel–hydrothermal reaction. Typically, Ti(OC₄H₉)₄, NbCl₅, and RhCl₃·3H₂O were dissolved in ethanol (containing 0.1 vol% HCl as stabilizer) to form 1 mM, 0.1 M, and 0.02 M solutions, respectively. NbCl₅ (and/or RhCl₃·3H₂O) solution was added dropwise to Ti(OC₄H₉)₄ solution under magnetic stirring. The volume ratio of these solutions was adjusted in accordance with the chemical composition of Ti_1−x–yNb_xRh_yO₂ (see Table 1). Then 3 mL of ultrapure water was dripped slowly into 50 mL of the above mixed solution under vigorous stirring to form a homogeneous sol. The sol was left in a water bath (set at 50 °C) and stirred mildly until a transparent gel came into being. After aging for 5 h, the gel was added into 100 mL of 2 M NaOH solution, and stirred vigorously for 20 h to obtain a well-dispersed slurry. The solid content was separated from the slurry by vacuum filtration, and then re-dispersed in 70 mL of 8 M NaOH aqueous solution by stirring for 1 h. The specimen was transferred into a 100 mL autoclave and then heated at 180 °C in an oven for 24 h. After the hydrothermal reaction, the precipitate was collected and washed with ultrapure water until pH = 9, and then with diluted HNO₃ solution (pH = 3) until pH = 3. The obtained product was protonated titanate nanotubes (see Fig. S1 in the ESI†). The titanate nanotubes were washed repeatedly with ultrapure water and ethanol to remove the absorbed nitric acid and water, followed by drying in air at 70 °C. Subsequently, Nb and/or Rh doped TiO₂ photocatalysts were obtained by calcining the dried protonated titanate products in air for 2 h, and their morphology, crystallinity and crystal phase could be tailored by changing the calcined temperature from 400 °C to 900 °C.

---

Characterization

The morphology and elemental distribution of the materials were observed by transmission electron microscopy (Tecnai G2 F20, FEI, USA). The crystal structure was determined by an X-ray diffractometer (Miniflex 600, Rigaku, Japan). The surface chemical structure and band alignment were analysed by X-ray photoelectron spectroscopy (ESCALAB 250xi, Thermo Scientific, USA), and all the spectra were calibrated with reference to the C 1s peak at 284.8 eV. Solid-state optical diffuse reflectance spectra were measured with a UV-Vis spectrophotometer (Lambda 900, PerkinElmer, USA) with BaSO₄ as a standard, and were converted to absorption spectra by using the Kubelka–Munk function. The PL (emission) spectra were measured with an Edinburgh Instruments FLS980 spectrometer, and the excitation wavelength was 380 nm.

Electrochemical measurements

Electrochemical measurements were carried out on a CHI660E electrochemical working station with a standard three-electrode cell. Pt, Ag/AgCl, and Na₂SO₄ solution (0.5 M), were used as the counter electrode, reference electrode, and electrolyte, respectively. The working electrode was fabricated by depositing the photocatalyst on an ITO glass. The procedure was as follows: first, 5 mg of photocatalyst, 10 μL of naphthol solution (5 wt%), 40 μL of ethanol, and 200 μL of ultrapure water were mixed in a quartz tube, and then ultrasonically scattered for 2 h. Second, 40 μL of the mixed sample was deposited onto an ITO conductive glass (the deposition area was 1 × 1 cm²) and evaporated to dry. The electrochemical measurements were performed in a square electrolytic cell containing 125 mL of electrolyte, and the light source was a 300 W xenon-lamp. The photocurrent with ON/OFF cycles was measured with a cut-off filter (λ ≥ 420 nm) and with a bias potential of 0.4 V. Electrochemical impedance spectroscopy (EIS) plots were recorded in a frequency range from 10 to 800 kHz. Mott–Schottky plots were collected at frequencies of 1000, 2000, and 3000 Hz.

Photocatalytic hydrogen generation test

The hydrogen generation experiments took place in an online photocatalytic activity evaluation system (CEL-SPH2N, CEAuLight, China) equipped with a top-irradiation quartz reactor. A 100 mg sample was dispersed in 100 mL of 10 vol% methanol solution in the reactor. After being bubbled with N₂ for 0.5 h, the reactor was evacuated by a pump and then irradiated by a 300 W xenon-lamp. In the course of the experiment, the suspension was magnetically stirred and the temperature was maintained at 6 °C. The generated H₂ was detected by appended online chromatography.

Results and discussion

It is known that hydrothermally derived protonated titanate will be converted into TiO₂ upon calcination.² Accordingly, (Nb,Rh)-codoped TiO₂ will be obtained if Nb and Rh are pre-doped into the titanate. Fig. 1 shows the XRD patterns of pristine TiO₂ (TNR) and a representative (Nb,Rh)-codoped sample Ti_0.998Nb_0.006Rh_0.006O₂ (N6R6TNR; the chemical composition and nomenclature of the samples are listed in Table 1). It was found that titanate had been completely converted to anatase TiO₂ at 400 °C. As the temperature increased, the crystallinity of the TiO₂-based product continually improved, and the crystal phase gradually transformed from anatase to rutile. By calcining at 700 °C, rutile appeared as a minor phase in N6R6TNR, while the pure anatase phase persisted in TNR. Similarly, at 800 °C, TNR had a higher anatase/rutile ratio than N6R6TNR. It is therefore suggested that Nb/Rh codoping can accelerate the phase transformation from anatase to rutile (A–R). Such acceleration is more pronounced when TiO₂ is doped with Rh alone. As shown in Fig. S2,† the rutile phase was observed in R6TNR at a low calcination temperature of 500 °C. It is widely accepted that the A–R transformation depends greatly on oxygen vacancy and stress (or lattice distortion).^30–32 In principle, when Nb⁵⁺ enters into a TiO₂ crystal substituting for Ti⁴⁺, the charge compensation is considered to decrease the concentration of oxygen vacancies and consequently suppress atomic transport in the anatase structure, leading to retardation of the A–R transformation.^30,33 In contrast for Rh-doped TiO₂, the valence states of substitutional rhodium ions consist of a dominant +4 and a minor +3 (vide infra). The incorporation of Rh³⁺ should promote the formation of oxygen vacancies and thereby facilitate the phase transformation.³³ For (Nb,Rh)-codoped TiO₂, such as N6R6TNR, the substitution of equimolar Nb⁵⁺ and Rh³⁺ for Ti⁴⁺ does not significantly affect the defect chemistry of TiO₂. However, their larger effective ionic radii (0.665 Å for Rh³⁺, 0.64 Å for Nb⁵⁺, and 0.605 Å for Ti⁴⁺ in a 6-coordinate environment) will expand the unit cell and increase the stress. Such structural deformation is also conducive to the A–R transformation.³² The cell expansion induced by doping can be demonstrated by the shift of diffraction peaks in the XRD patterns, as shown in Fig. S2 and S3.†

The evolution of the morphology with calcination temperature is presented in Fig. 2 and Fig. S1.† It is found that the phase transformation from titanate to TiO₂ under calcination is accompanied by the collapse of nanotubes. At 500 °C, nanotubes had completely transformed into nanorods with diameters of about 7 nm and lengths of tens to hundreds of nanometers (Fig. 2a and b). The HRTEM images (Fig. 2b) revealed two sets of lattice fringes with the same spacing of 0.35 nm and an angle of ∼82°, which correspond to the (101) and (011) planes of anatase phase. Fig. S1g† shows the HRTEM image of the end face of a nanorod. High-density dislocations were observed. Since the nanorods are formed by the inward collapse and reconstruction of nanotubes (Fig. S1a–g†), the generation of large-scale crystal defects is inevitable. During the high-temperature calcination, although some of the crystal defects will be eliminated and the crystallinity will be improved, the migration of high-concentration dislocations will also lead to the fracture of the nanorods. At a higher calcination temperature of 700 °C, the length of most nanorods was found to be shorter than 70 nm, and quasi-spherical and elongated nanoparticles were also found (Fig. 2c). The HRTEM image (Fig. 2d) reveals the coexistence of anatase and rutile. (101) diffraction planes of the anatase phase were clearly displayed in a rod-like nanocrystal. In addition, two sets of lattices with spacings of 0.23 and 0.32 nm were observed in another nanoparticle, which can be ascribed to the (200) and (110) planes of the rutile phase, respectively. As the calcination temperature was further raised to 900 °C, only quasi-spherical (or elongated) particles were observed (Fig. 2e). Fig. 2f shows the EDX elemental mapping of N15R15TNR, where the homogeneous distribution of Ti, Nb and Rh elements highlighted the formation of (Nb,Rh)-codoped TiO₂ without local aggregation or second phases.

Fig. 3 shows the effect of calcination temperature on the photocatalytic activity. When the calcination temperature was in the range of 400–700 °C, the hydrogen evolution rate of N6R6TNR was at least two orders of magnitude higher than that of TNR, indicating that (Nb,Rh)-codoping is an effective strategy to improve the photocatalytic activity of TiO₂. For both samples, the photocatalytic activity reached a maximum at 500 °C, and fell sharply when the temperature was higher than 600 °C. Generally, the performance of a photocatalyst depends largely on crystallinity, particle size and shape, specific surface area, and crystal structure (e.g., anatase, rutile and brookite phases for TiO₂).³⁴ For example, (i) high crystallinity and small particle size are often required for inhibiting the bulk recombination of photogenerated carriers; (ii) a higher surface area means more surface active site for redox reactions; (iii) compared with zero-dimensional (0D) nanoparticles, 1D nanostructures exhibit advantages such as enhanced light absorption and excellent charge transport;^15–18 (iv) among the three allotropic crystalline forms of TiO₂ (anatase, rutile, and brookite), the photocatalytic activity of the anatase phase is generally considered to be the highest. In this study, compared with samples obtained by calcining at higher temperature, the nanotubes obtained at 400 °C possessed the highest specific surface area. However, their crystallinity was also the lowest, which led to a lower hydrogen generation rate than that of nanorods obtained at 500 °C. As the temperature was further increased (T > 500 °C), the hydrogen generation rate decreased, even though the crystallinity increased. This indicated that the contribution of crystallinity was no longer dominant, and the importance of other factors (such as morphology, surface area, and A–R transformation) increased. As more and more nanorods broke into 0D nanoparticles at T > 600 °C (Fig. 2 and S1†), the hydrogen generation rate decreased sharply, indicating that the photocatalytic performance of 1D nanorods is far superior to that of 0D nanoparticles. It is noticed that, for N6R6TNR, the hydrogen generation rate decreased from 8035 μmol g⁻¹ h⁻¹ at 600 °C (anatase) to 51 μmol g⁻¹ h⁻¹ at 900 °C (rutile). Similarly, we also synthesized Ti_0.99Nb_0.005Rh_0.005O₂ 0D nanoparticles by a simple sol–gel–calcination method. The average particle size increased slightly from 15 nm at 600 °C (anatase) to 20 nm at 900 °C (rutile), while the corresponding hydrogen generation rate decreased obviously from 440 μmol g⁻¹ h⁻¹ (600 °C) to 9 μmol g⁻¹ h⁻¹ (900 °C). Obviously, the A–R transformation has a significant impact on the photocatalytic performance. The anatase phase exhibits much higher photocatalytic activity than the rutile phase. Therefore, the preparation of Rh³⁺-doped rutile TiO₂ through high-temperature calcination is not a good choice for achieving excellent performance. In contrast, the synthesis of anatase TiO₂-based 1D nanostructures with high crystallinity is meaningful, but is also very difficult to achieve. In recent years, we have made many attempts to prepare high-quality TiO₂-based nanotubes with little success. In this study, under the balance of various factors, the sample calcined at 500 °C exhibited the best performance for hydrogen evolution. Therefore, in the following studies, the calcination temperature was set to 500 °C.

As mentioned above, the photocatalytic performance of TiO₂ can be greatly improved by (Nb,Rh) codoping. Undoubtedly, this improvement is related not only to the calcination temperature, but also to the doping concentration of both Nb and Rh. In order to optimize the performance, under the premise that the calcination temperature has been fixed (i.e., 500 °C), our strategy is to determine the preferred Nb/Rh mol ratio before determining the total doping concentration.

Fig. 4 shows the solid state UV-Vis absorption spectra of TiO₂-based photocatalysts with different Nb/Rh ratios. Both TNR and N6TNR exhibited an absorbance edge at ∼385 nm and hardly absorb visible light, indicating that Nb⁵⁺ doping cannot contribute to the visible photocatalytic activity of TiO₂. For N6R3TNR, N6R6TNR, and N3R6TNR, intensive absorption from 385 to 530 nm was observed, which can be attributed to the electronic excitation from the in-gap Rh³⁺ donor level to the CB.^16,35 For R6TNR, the presence of a broad absorption peak with wavelength greater than 500 nm can be assigned to the electronic transition from the valence band (VB) to Rh⁴⁺ unoccupied levels.^16,35,36 Additionally, the significant absorption between 400 and 500 nm implied the possible existence of Rh³⁺ in this sample. By comparing the absorption spectra between 500 and 600 nm (corresponding to the Rh⁴⁺ level), it is inferred that the Rh⁴⁺ content (relative to Rh³⁺) was increasing in the order of R6TNR > N3R6TNR > N6R6TNR > N6R3TNR, which can be further confirmed by XPS detection.

Fig. 5a shows the Rh 3d core level spectra of several doped TiO₂ samples with different Nb/Rh ratios. The difference in the outer electron cloud density of Rh ions leads to a lower binding energy for Rh³⁺ than for Rh⁴⁺. For Rh³⁺ and Rh⁴⁺, the 3d_5/2 peak was located at 307.7 and 309.3 eV, and the 3d_3/2 peak was located at 312.5 and 314.1 eV, respectively. For N6R3TNR, two symmetric peaks were found at 307.7 and 312.5 eV, indicating the sole existence of Rh³⁺. For N3R6TNR, both Rh 3d_5/2 and Rh 3d_3/2 peaks can be deconvoluted into two comparable components, indicating the coexistence of Rh³⁺ and Rh⁴⁺ with a ratio close to 1. For N6R6TNR, in addition to Rh³⁺, a trace of Rh⁴⁺ was also detected. In reverse, for R6TNR, the Rh 3d_5/2 profile can be fitted with a dominant Rh⁴⁺ peak and a secondary Rh³⁺ peak, implying that it is not a completely isovalent substitution of Rh⁴⁺ for Ti⁴⁺ in Rh-doped TiO₂. This may be due to the fact that the stable oxidation state of Rh is +3. The change in valence state of Rh with Nb/Rh ratio will affect the charge balance, and thus alter the electronic structure of the doped TiO₂. Fig. 5b shows the VB spectra. It is found that the Fermi level (E_F) was located at about 2.78, 2.98, 2.84, 2.73, 2.25, and 1.97 eV above the valence band maximum (VBM) for TNR, N6TNR, N6R3TNR, N6R6TNR, N3R6TNR, and R6TNR, respectively. In other words, Nb-doping will shift the E_F towards the CB while Rh-doping will shift the E_F towards the VB. This is because Nb⁵⁺ and Rh³⁺ act as donor and acceptor impurities in TiO₂, respectively. The existence of Nb⁵⁺ in N6TNR makes it more n-type while Rh³⁺ in R6TNR makes it more p-type (exhibited as the downward shift of the E_F). Based on the aforementioned XPS and UV-Vis observations, a schematic band structure diagram was proposed, as shown in Fig. S4.† For all the samples, the E_F was located above the impurity (Rh⁴⁺ and/or Rh³⁺) levels, indicating their n-type semiconductor nature. This may be due to the slight doping and the resultant negligible charge imbalance.

It is generally accepted that, in Rh-doped TiO₂, Rh(III) is the active species, while Rh(IV) doping introduces additional trapping sites and thus its formation should be avoided. As shown in Fig. 5 and Fig. S4,† Rh(IV) coexists with Rh(III) in the samples with Rh/Nb ≥ 1, and only Rh(III) exists in N6R3TNR. It is thereby anticipated that N6R3TNR may exhibit the highest photocatalytic activity among these samples. But surprisingly, the hydrogen evolution rate of N6R3TNR was only 2681 μmol g⁻¹ h⁻¹, which was much lower than that of N6R6TNR (9194 μmol g⁻¹ h⁻¹) or N3R6TNR (4170 μmol g⁻¹ h⁻¹), as shown in Table 1. One possible explanation is that the Rh content of N6R3TNR is lower than that of the other two samples. However, it is also noticed that the hydrogen evolution rate of N3R3TNR (which has the same Rh doping concentration as N6R3TNR) was about three times that of N6R3TNR. Similar phenomena have also been reported in previous work.^35,37 For example, the hydrogen evolution rate of (La,Rh) codoped SrTiO₃ (with a constant Rh doping level of 4 mol%) decreases drastically with an increase in the La doping concentration from 4 to 10 mol%.³⁵ To explore the origin of the enhanced performance, EIS measurement was carried out. Fig. 6 and Fig. S6† show the Nyquist plots for (Nb,Rh) co-doped TiO₂. As shown in Fig. 6, for samples N6R3TNR, N6R6TNR, and N3R6TNR, N6R3TNR showed the largest semicircles, revealing its highest electron transfer resistance.^38,39 In contrast, the smallest semicircular diameter of N3R6TNR indicated that the charge transfer in this sample was easier and more efficient than for the other two samples. Fig. S5† shows the Mott–Schottky plots of these three samples. Linear behaviour was observed when the applied bias ranged from 0 to 0.8 V. The positive slopes indicate that these samples were n-type semiconductors, which was in consistent with the aforementioned analysis (see the proposed band structure diagram in Fig. S4†). The slope for N6R3TNR was larger than that for N6R6TNR or N3R6TNR. It is known that the slope of the Mott–Schottky plot is inversely proportional to the donor density.^39,40 Therefore, the larger slope of N6R3TNR suggested its lower donor density. According to the Nyquist and Mott–Schottky plots, it is possible to conclude that N6R3TNR possesses higher electron transfer resistance and lower carrier density. On the other hand, the presence of Rh(IV) in N6R6TNR and N3R6TNR induces an increase in the conductivity and carrier density, which allows more efficient charge separation, and allows more electrons and holes (per unit time) to reach the semiconductor surface for photocatalytic water splitting. It is therefore deduced that the existence of a certain amount of Rh(IV) may be beneficial to the improvement of photocatalytic performance. Meanwhile, it should be emphasized again that the Rh(IV) unoccupied level plays the role of a carrier recombination centre. As discussed above, the content of Rh(IV) in N3R6TNR and R6TNR was far higher than that in N3R3TNR and N6R6TNR. However, the hydrogen evolution rate of N3R6TNR (4170 μmol g⁻¹ h⁻¹) was much lower than that of N3R3TNR (6788 μmol g⁻¹ h⁻¹) or N6R6TNR (9194 μmol g⁻¹ h⁻¹), and the hydrogen evolution rate of R6TNR (622 μmol g⁻¹ h⁻¹) was even just about 1/15 of N6R6TNR (see Table 1). Obviously, the concentration of Rh(IV) must be limited to a low level by controlling the Nb/Rh ratio. According to Table 1, the optimal Nb/Rh ratio for achieving excellent photocatalytic activity is around 1. It should be pointed out that the valence state of Rh in a TiO₂ lattice depends not only on the Nb/Rh ratio, but also on the preparation methods and conditions (such as calcination temperature and atmosphere). It is thus of little significance to accurately determine the optimal Nb/Rh ratio.

With the determination of calcination (500 °C) and Nb/Rh ratio (1:1), the dependence of photocatalytic activity on the doping concentration was studied. Fig. 7a shows the hydrogen evolution performance of Ti_1−2xNb_xRh_xO₂ nanorods from methanol solution under the full arc irradiation of a Xe lamp. The amount of produced H₂ increased almost linearly with irradiation time in the 10 h test, indicating the stability of the (Nb,Rh) codoped TiO₂ photocatalysts. The stable hydrogen evolution rate of the samples is presented in Table 1. Obviously, the photocatalytic activity can be greatly improved by slight doping. The hydrogen evolution rate of N3R3TNR (x = 0.003) was 6788 μmol g⁻¹ h⁻¹, which was about 80 times higher than that of undoped TiO₂. An ultrahigh evolution rate of 9194 μmol g⁻¹ h⁻¹ was achieved by further increasing the doping concentration to x = 0.006 (N6R6TNR). Thereafter, the photocatalytic performance decreased with increasing doping concentration, but remained above 1400 μmol g⁻¹ h⁻¹ even if the doping concentration reached x = 0.02. Fig. S7† shows the visible-light-driven hydrogen generation over Ti_1−2xNb_xRh_xO₂ photocatalysts (λ ≥ 420 nm). N6R6TNR presented the highest hydrogen evolution rate (75 μmol g⁻¹ h⁻¹) among these samples. A further increase in doping concentration led to a progressive downturn in the visible-light activity. The change in photocatalytic activity with doping concentration is related to the recombination of photogenerated electron–hole pairs. As shown in Fig. S8,† the intensity of the PL emission was markedly suppressed with slight doping (x < 0.009). However, the PL intensity increased with a further increase in the doping concentration (x > 0.009). In general, a weaker PL intensity is indicative of a lower recombination rate of photogenerated carriers.^41,42 The outstanding H₂ evolution performance of N3R3TNR, N6R6TNR, and N9R9TNR correlated well with their much lower electron–hole recombination. Fig. 7b shows transient photocurrent responses under visible-light irradiation (λ ≥ 420 nm). The photocurrent densities of N3R3TNR, N6R6TNR, and N9R9TNR were close, and significantly higher than those of other samples. In other words, the generation and migration rate of electron/hole pairs reached a maximum when the doping concentration fell within the range 0.003 ≤ x ≤ 0.009, which is in accord with the photocatalytic activity. Such a low optimal doping concentration is of great importance for cost saving due to the high price of rhodium.

Conclusions

In summary, we synthesized (Nb,Rh) codoped anatase TiO₂ nanorods through the solid-state transformation of titanate nanotubes derived from sol–gel–hydrothermal reaction, and investigated the effect of dopant ratio, doping concentration, and calcination temperature on the structural and photocatalytic properties. Upon calcination at a low temperature of about 400 °C, protonated titanate nanotubes are transformed into anatase TiO₂ nanotubes with low crystallinity, which are further transformed into anatase nanorods and then into rutile nanoparticles with high crystallinity as the temperature increases. The competition of various factors, such as crystallinity, particle size and shape, and crystal phase, makes anatase nanorods obtained at 500 °C become an attractive photocatalyst with the best hydrogen evolution performance. In Rh-doped and (Nb,Rh) codoped TiO₂, both Rh(III) and Rh(IV) construct impurity levels in the forbidden band of TiO₂, resulting in intensive absorption of visible light with different wavelengths. The electronic excitation from the Rh(III) donor level to the CB contributes to visible-light activity and high photocatalytic performance, while the role of Rh(IV) is more complicated. As we know, Rh(IV) acts as electron/hole trapping sites, and thus the hydrogen evolution rate of Rh(IV)-doped TiO₂ is much lower than that of Rh(III)-doped TiO₂. On the other hand, a small amount of Rh(IV)-doping will increase the carrier density and electron transfer conductivity, which is beneficial to improving photocatalytic activity. By controlling the Nb/Rh ratio, the oxidation state of Rh can be regulated to optimize the photocatalytic performance of (Nb,Rh) codoped TiO₂. It is found that the optimal Nb/Rh ratio is around 1. For TiO₂ nanorods codoped with equimolar Nb and Rh (i.e., Ti_1−2xNb_xRh_xO₂), both a large photocurrent density and a high photocatalytic hydrogen evolution rate are realized when the doping concentration is in the rage 0.003 ≤ x ≤ 0.009, demonstrating that (Nb,Rh) codoping is an effective strategy to boost visible-light activity and to improve the hydrogen production efficiency of TiO₂-based photocatalysts.

Conflicts of interest

The authors declare no conflicts of interest.

Acknowledgements

This research was supported by the National Key R&D Program of China (2017YFB0403200) and the Strategic Priority Research Program of Chinese Academy of Sciences (XDA22010301).

References

1. X. B. Chen, S. H. Shen, L. J. Guo and S. S. Mao, Chem. Rev., 2010, 110, 6503–6570 CrossRef CAS .
2. J. Q. Huang, Y. B. Jiang, G. J. Li, C. B. Xue and W. Guo, Renewable Energy, 2017, 111, 410–415 CrossRef CAS .
3. M. Sheikhzadeh, S. Hejazi, S. Mohajernia, O. Tomanec, M. Mokhtar, A. Alshehri, S. Sanjabi, R. Zboril and P. Schmuki, ChemCatChem, 2019, 11, 6258–6262 CrossRef CAS .
4. M. M. Zhu, C. Zhu, D. Wu, X. Wang, H. B. Wang, J. Guo, H. Huang, C. F. Shi, Y. Liu and Z. H. Kang, Nanoscale, 2019, 11, 15984–15990 RSC .
5. M. R. Gholipour, C. T. Dinh, F. Beland and T. O. Do, Nanoscale, 2015, 7, 8187–8208 RSC .
6. L. Ma, Y. L. Che, D. J. Yang, H. X. Li, S. J. Ding, L. Xiong, P. L. Qin and X. B. Chen, Nanoscale, 2020, 12, 4383–4392 RSC .
7. S. S. Chen, Y. Qi, C. Li, K. Domen and F. X. Zhang, Joule, 2018, 2, 2260–2288 CrossRef CAS .
8. Z. Z. Cheng, F. M. Wang, T. A. Shifa, C. Jiang, Q. L. Liu and J. He, Small, 2017, 13, 1702163 CrossRef .
9. X. Q. Sun and X. X. Xu, Appl. Catal., B, 2017, 210, 149–159 CrossRef CAS .
10. S. Martha, P. C. Sahoo and K. M. Parida, RSC Adv., 2015, 5, 61535–61553 RSC .
11. F. E. Osterloh, Chem. Soc. Rev., 2013, 42, 2294–2320 RSC .
12. L. Yuan, C. Han, M. Q. Yang and Y. J. Xu, Int. Rev. Phys. Chem., 2016, 35, 1–36 Search PubMed .
13. J. X. Luo, J. G. Yu, M. Jaroniec, S. Wageh and A. A. Al-Ghamdi, Adv. Mater., 2017, 29, 1601694 CrossRef .
14. S. Bellamkonda, N. Thangavel, H. Y. Hafeez, B. Neppolian and G. R. Rao, Catal. Today, 2019, 321–322, 120–127 CrossRef CAS .
15. J. G. Yu, Y. Hai and B. Cheng, J. Phys. Chem. C, 2011, 115, 4953–4958 CrossRef CAS .
16. J. Q. Huang, G. J. Li, Z. F. Zhou, Y. B. Jiang, Q. Q. Hu, C. B. Xue and W. Guo, Chem. Eng. J., 2018, 337, 282–289 CrossRef CAS .
17. M. Z. Ge, Q. S. Li, C. Y. Cao, J. Y. Huang, S. H. Li, S. N. Zhang, Z. Chen, K. Q. Zhang, S. S. Al-Deyab and Y. K. Lai, Adv. Sci., 2017, 4, 1600152 CrossRef .
18. G. W. An, L. K. Dhandole, H. Park, H. S. Bae, M. A. Mahadik and J. S. Jang, ChemCatChem, 2019, 11, 5606–5614 CrossRef .
19. M. Liu, N. D. L. Snapp and H. Park, Chem. Sci., 2011, 2, 80–87 RSC .
20. G. J. Li, J. Q. Huang, C. B. Xue, J. Chen, Z. H. Deng, Q. F. Huang, Z. G. Liu, C. Gong, W. Guo and R. Cao, Cryst. Growth Des., 2019, 19, 3584–3591 CrossRef CAS .
21. A. E. R. Mohamed and S. Rohani, Energy Environ. Sci., 2011, 4, 1065–1086 RSC .
22. M. Y. Wang, J. Loccozia, L. Sun, C. J. Lin and Z. Q. Lin, Energy Environ. Sci., 2014, 7, 2182–2202 RSC .
23. H. T. Gao, B. Lu, W. C. Liu, C. H. Ding, F. F. Liu and G. J. Liu, J. Comput. Theor. Nanosci., 2012, 9, 2198–2207 CrossRef CAS .
24. R. T. Yunarti, M. Lee, Y. J. Hwang, J. W. Choi, D. J. Suh, J. Lee, I. W. Kim and J. M. Ha, Catal. Commun., 2014, 50, 54–58 CrossRef CAS .
25. S. C. Sun, P. Gao, Y. R. Yang, P. P. Yang, Y. J. Chen and Y. B. Wang, ACS Appl. Mater. Interfaces, 2016, 8, 18126–18131 CrossRef CAS .
26. F. E. Oropeza and R. G. Egdell, Chem. Phys. Lett., 2011, 515, 249–253 CrossRef CAS .
27. R. Konta, T. Ishii, H. Kato and A. Kudo, J. Phys. Chem. B, 2004, 108, 8992–8995 CrossRef CAS .
28. J. Kuncewicz and B. Ohtani, RSC Adv., 2016, 6, 77201–77211 RSC .
29. S. G. Kim, L. K. Dhandole, Y. S. Seo, H. S. Chung, W. S. Chae, M. Cho and J. S. Jang, Appl. Catal., A, 2018, 564, 43–55 CrossRef CAS .
30. J. Arbiol, J. Cerda, G. Dezanneau, A. Cirera, F. Peiro, A. Cornet and J. R. Morante, J. Appl. Phys., 2002, 92, 853–861 CrossRef CAS .
31. D. R. Zhang, X. Z. Jin and J. H. Li, Mater. Chem. Phys., 2016, 176, 68–74 CrossRef CAS .
32. M. A. Barakat, G. Hayes and S. I. Shah, J. Nanosci. Nanotechnol., 2005, 5, 759–765 CrossRef CAS .
33. J. P. Xu, S. B. Shi, L. Li, J. F. Wang, L. Y. Lv, F. M. Fang and Y. W. Du, J. Phys. Chem. Solids, 2009, 70, 511–515 CrossRef CAS .
34. A. Kudo and Y. Miseki, Chem. Soc. Rev., 2009, 38, 253–278 RSC .
35. D. H. K. Murthy, H. Matsuzaki, Q. Wabgm, T. Suzuki, K. Seki, T. Hisatomi, T. Yamada, A. Kudo, K. Domen and A. Furube, Sustainable Energy Fuels, 2019, 3, 208–218 RSC .
36. J. Kuncewicz and B. Ohtani, Chem. Commun., 2015, 51, 298–301 RSC .
37. T. Takata and K. Domen, J. Phys. Chem. C, 2009, 113, 19386–19388 CrossRef CAS .
38. C. Y. Huang, R. T. Guo, W. G. Pan, J. Y. Tang, W. G. Zhou, X. Y. Liu, H. Qin and P. Y. Jia, Appl. Surf. Sci., 2019, 464, 534–543 CrossRef CAS .
39. X. Bi, G. H. Du, D. F. Sun, M. Zhang, Y. Yu, Q. M. Su, S. K. Ding and B. S. Xu, Appl. Surf. Sci., 2020, 511, 145617 CrossRef CAS .
40. A. Wolcott, W. A. Smith, T. R. Kuykendall, Y. P. Zhao and J. Z. Zhang, Adv. Funct. Mater., 2009, 19, 1849–1856 CrossRef CAS .
41. K. Wang, Y. Li, J. Li and G. K. Zhang, Appl. Catal., B, 2020, 263, 117730 CrossRef CAS .
42. L. S. Jiang, J. Li, K. Wang, G. K. Zhang, Y. Li and X. Y. Wu, Appl. Catal., B, 2020, 260, 118181 CrossRef CAS .

---

Footnote

† Electronic supplementary information (ESI) available: XRD patterns of TiO2-based photocatalysts with different Nb/Rh ratio and doping concentration, TEM and HRTEM images of selected samples calcined at different temperature, Nyquist and Mott–Schottky plots. See DOI: 10.1039/d0nr05695b

This journal is © The Royal Society of Chemistry 2020