Topochemical synthesis of mesoporous TiO₂ co-doped with nitrogen and fluorine for improved photocatalytic O₂ evolution under visible light

Abstract

Mesoporous rutile TiO₂ photocatalysts co-doped with N and F were synthesized via a topotactic ammonolysis approach using mesostructured TiO₂ as a precursor. The co-substitution of N and F into the rutile lattice led to substantial modulation of the Ti electronic structure and to extension of visible-light absorption, accompanied by local distortion of TiO₆ octahedra. Systematic characterization revealed that the balance between dopant incorporation and structural integrity of the mesoporous framework played a decisive role in determining photocatalytic performance for half-cell O₂ evolution. Although higher-temperature ammonolysis promoted N incorporation and enhanced visible-light absorption, it also compromised the mesostructure, reducing the overall activity. The highest O₂ evolution rate under visible-light irradiation was achieved with the sample prepared under optimal conditions that maintained both phase-pure rutile and mesoporosity. These results highlight the importance of controlling both chemical doping and structural features when designing high-performance non-metal-doped oxide photocatalysts for solar-driven water oxidation.

---

New concepts

We report a topotactic ammonolysis strategy to synthesize mesoporous rutile TiO₂ co-doped with nitrogen and fluorine, achieving simultaneous control over crystal phase, lattice electronic structure, and mesostructural integrity. While non-metal doping of TiO₂ has long been explored to extend visible-light activity, the challenge of preserving a high-surface-area mesoporous framework during high-temperature treatments remains largely unsolved. Our approach uniquely balances lattice substitution with mesoporous preservation, enabling co-modulation of bandgap and defect chemistry without sacrificing the hierarchical pore network essential for mass transport. This synergy allows efficient separation and transfer of photogenerated charge carriers, yielding significantly enhanced visible-light-driven O₂ evolution. The work provides a conceptual advance by demonstrating that photocatalytic performance in doped oxides is not dictated solely by light-harvesting ability, but by a fine interplay between electronic tuning and nanostructural stability. This insight is broadly applicable to the design of next-generation non-metal-doped oxides for solar energy conversion, electrocatalysis, and environmental remediation, where multifunctional structural control at the nanoscale is required.

Introduction

Photocatalytic water splitting is a promising strategy for converting abundant solar energy into clean H₂ fuel in an environmentally friendly manner.^1–4 Among the two half-reactions, water oxidation remains a major challenge because of its complex four-electron transfer process, which often limits the overall efficiency of the system.^5,6 To overcome this bottleneck, researchers have directed extensive efforts toward developing visible-light-responsive photocatalysts with enhanced charge separation and surface reactivity.¹

Mixed-anion compounds such as oxysulfides, oxyhalides, and oxynitrides have attracted attention because of their unique electronic structures and properties originating from the incorporation of additional anionic species.^7–11 Among these compounds, oxynitrides have emerged as promising photocatalysts capable of absorbing visible light, primarily because of the contribution of N 2p orbitals to the valence band. However, N doping in oxides such as TiO₂ can induce O vacancies, which accelerate charge recombination.¹² To address this problem, researchers have proposed co-doping with F (via the substitution 2O²⁻ ⇌ N³⁻ + F⁻) to suppress recombination by balancing the local charge and maintaining the anion framework.^13–15 Indeed, N,F-co-doped TiO₂ has shown enhanced activity toward photocatalytic water oxidation.^14,16

Mesoporous materials often exhibit improved photocatalytic performance because of their short charge carrier diffusion lengths and high surface areas, which facilitate surface chemical reactions.^17,18 For example, mesoporous TiO₂ has been reported to effectively photocatalyze the decomposition of organic dye molecules such as rhodamine B and methylene blue, as well the photoreduction of CO₂ to methane.^19–21 Mesoporous TiO₂ was first synthesized via a modified sol–gel method in the presence of phosphate surfactants by Antonelli and Ying in 1995.²² Since then, considerable efforts have been devoted to developing mesoporous-structured TiO₂. Such materials are particularly interesting because of their ability to interact with atoms, ions, molecules, and nanoparticles not only on the surface but also throughout the bulk.²³ Nevertheless, developing a new template-free route to synthesize highly organized mesoporous TiO₂ spheres remains challenging because of the limited availability of appropriate precursors, poor controllability of the reactivity of transition-metal precursors, and structural collapse during crystallization and during grain growth upon heat treatment.^24,25

This study combines two effective strategies for designing efficient water oxidation photocatalysts: (1) the formation of active, crystalline mesoporous TiO₂ via a microwave-assisted, template-free solvothermal method and (2) bandgap narrowing through N and F co-substitution. Although N,F-co-doped mesoporous TiO₂ (meso-TiO₂:N,F) has been widely investigated for use in applications such as the photodegradation of organic substances, its application in visible-light-driven water oxidation has not yet been reported. To the best of our knowledge, this work is the first demonstration of such a material being applied to photocatalytic water oxidation under visible light.

In the present work, meso-TiO₂:N,F is synthesized by a topochemical reaction—a powerful approach previously shown to be effective for obtaining various mixed-anion phases.^26–29 However, the potential of topochemical methods for synthesizing mesoporous mixed-anion materials has not yet been sufficiently elucidated, nor have their photocatalytic properties. In particular, this study focuses on clarifying the correlation between the mesoporous structure and the O₂ evolution rate from aqueous AgNO₃ by systematically comparing meso-TiO₂, meso-TiO₂:N, and meso-TiO₂:N,F. Our results demonstrate that moderate N doping both preserves the mesoporous structure and promotes photocatalytic water oxidation under visible light.

Experimental

Materials and reagents

Rutile TiO₂ (Catalysis Society of Japan Reference Catalyst, JRC-TIO-6), rutile TiO₂ (Wako Pure Chemicals, 99.0%), titanium(IV) tetrabutoxide (TBOT, Wako Pure Chemicals, 95.0%), hexane (Wako Guaranteed Reagent, >96.0%), (NH₄)₂TiF₆ (Wako Pure Chemicals, 95.0%), AgNO₃ (Wako Pure Chemicals, 99.8%), La₂O₃ (TCI, 99.9%), HCl (Kanto Chemicals, 35–37%), ethanol (Wako Guaranteed Reagent, >99.5%), acetone (Wako 1st Grade, >99.0%) were used as received.

Synthesis of meso-TiO₂

TBOT (2.5 mL) was added to a mixture of hexane (25 mL) and HCl (2.5 mL) in a pressure-tight vessel, and the mixture was heated at 423 K for 4 h using a Mars 6 microwave reaction system (CEM Japan). The resultant precipitate was collected by filtration, washed several times with ethanol and distilled water, and then dried in an oven at 353 K.

An agate mortar and pestle were used to mix the as-prepared TiO₂ with (NH₄)₂TiF₆ in different molar ratios in the presence of acetone. For convenience, the molar concentration of (NH₄)₂TiF₆ in the mixture is denoted as C (mol%), calculated using the equation

The resultant precursor was loaded onto a Ni plate and placed at the center of an alumina tube reactor to prevent contamination from the reactor walls. After the reactor was purged with N₂ gas, the sample was heated at 673 K for 15 h under a flow of dry NH₃ (heating ramp: 10 K min⁻¹, flow rate: 100 mL min⁻¹) unless otherwise stated.

Synthesis of nano-TiO₂:N,F

Nano-TiO₂:N,F was synthesized under the optimized conditions reported previously.¹⁴ Briefly, JRC-TIO-6 and (NH₄)₂TiF₆ were mixed in a molar ratio of 95 to 5 (i.e., C = 5) in the presence of acetone using an agate mortar and pestle. The nitridation conditions were identical to those used for the preparation of meso-TiO₂:N,F.

Synthesis of bulk-TiO₂:N,F

Bulk-TiO₂:N,F was synthesized via a previously reported method.¹⁶ Commercial rutile TiO₂ and (NH₄)₂TiF₆ were mixed in a molar ratio of 85 to 15 (i.e., C = 15) in the presence of methanol using an agate mortar and pestle. The resultant precursor was loaded onto a Ni plate and placed at the center of an alumina tube reactor. After the reactor was purged with N₂ gas, the sample was heated at 773 K for 1 h under a dry NH₃ flow (ramp: 10 K min⁻¹, flow rate: 50 mL min⁻¹).

General characterization

X-ray diffraction (XRD) patterns were recorded on a MiniFlex600 (Rigaku) powder diffractometer. Raman spectra were acquired using a NRS-4500 spectrometer with a 532 nm laser (JASCO). UV-vis diffuse-reflectance spectra were obtained using a V-670 (JASCO) spectrometer equipped with an integration sphere. The specific surface area was calculated with the BELMaster software based on the Brunauer–Emmett–Teller (BET) theory. The surface morphology was observed by field-emission scanning electron microscopy (FE-SEM; S-5500; Hitachi High-Tech), and the amounts of N and F were determined using an elemental analyzer (Elemental Analyzer-1; JM10, JM11; J-Science). Transmission electron microscopy (TEM) and reflection electron microscopy (REM) images were acquired using a field-emission transmission electron microscope (JEM-2100F; JEOL) and a scanning electron microscope (JCM-7000; JEOL), respectively. The cross section of the sample was prepared using a cross-section polisher (CP; SM-09010; JEOL) with an Ar-ion beam. Elemental analysis, SEM, TEM, REM, and CP were conducted at the Core Facility Center, Institute of Science Tokyo. Ti K-edge X-ray absorption fine structure (XAFS) measurements were performed at beamline 9A of the Photon Factory (High Energy Accelerator Research Organization, Tsukuba, Japan). Time-resolved microwave conductivity (TRMC) measurements were carried out for powdered samples prepared on adhesive tape on a quartz substrate. All of the experiments were conducted at room temperature in air. X-band microwave radiation with a frequency of ∼9 GHz was used. The third harmonic (355 nm) of a Nd:YAG laser (Continuum, Surelite II, 5–8 ns pulse duration, 10 Hz) was used for carrier excitation (incident photon density I₀ = 9.1 × 10¹⁵ photons cm⁻² pulse⁻¹). The photoconductivity (Δσ = ΔP_r/(AP_r), where A is the sensitivity coefficient, P_r is the reflected microwave power, and ΔP_r is the change in P_r during the irradiation) was converted into the product of the quantum yield (φ) and the sum of the charge carrier mobilities Σμ (= μ⁺ + μ⁻) using the relational formula φΣμ = Δσ(eI₀F_light)⁻¹, where e and F_light are the electron charge and a correction coefficient, respectively.

Electrochemical impedance measurements

Meso-TiO₂ and meso-TiO₂:N,F (C = 15) electrodes were prepared via electrophoretic deposition onto conductive glass substrates coated with a fluorine-doped tin oxide (FTO) layer. Two FTO glasse plates were immersed in parallel (1.6 cm apart) into an acetone dispersion containing 50 mg of meso-TiO₂ or meso-TiO₂:N,F powder and 10 mg of I₂ in 45 mL of solution. A DC voltage of 30 V was applied for 15 s using a power supply (GW Instek PSW 80–13.5), resulting in uniform photocatalyst coatings on the FTO surfaces. The coated area was fixed at 1.5 cm × 3.5 cm. After deposition, the electrodes were dried under vacuum overnight.

Electrochemical impedance measurements were performed at room temperature using a potentiostat (HZ-Pro, Hokuto Denko) and a three-electrode electrochemical cell made of Pyrex glass. The working electrodes were meso-TiO₂/FTO and meso-TiO₂:N,F (C = 15)/FTO. An Ag/AgCl (saturated KCl) electrode and a Pt wire were used as the reference and counter electrodes, respectively. The electrolyte was an aqueous 0.1 M Na₂SO₄ solution whose pH was adjusted by the addition of H₂SO₄ or NaOH solution as needed. Prior to the start of the measurement, the solution was purged with Ar gas for 30 min to remove dissolved O₂. Mott–Schottky (MS) plots were recorded at a frequency of 100 Hz. Cyclic voltammetry (CV) was performed for 30 cycles over the potential range from −0.63 to +0.98 V (vs. reversible hydrogen electrode (RHE)).

Photocatalytic reactions

Photocatalytic reactions were conducted in a top-irradiation-type Pyrex reaction vessel connected to a closed gas circulation system.³⁰ In each experiment, 50 mg of meso-TiO₂:N,F and 200 mg of La₂O₃ (pre-calcined at 1273 K for 2 h to serve as a pH adjuster) were dispersed in 140 mL of a 10 mM AgNO₃ aqueous solution. The system was evacuated several times to remove air, and a small amount of Ar gas was introduced prior to irradiation. A 300 W Xe lamp was used as the light source (typically operated at 20 A), and illumination was conducted through a water filter and an L42 cutoff filter (λ > 400 nm) in combination with a CM-1 cold mirror. Evolved gases were analyzed using an on-line gas chromatograph (GC-8A, Shimadzu) equipped with a thermal conductivity detector and a molecular sieve 5A column; Ar was used as the carrier gas. During irradiation, the reaction temperature was maintained at room temperature by cooling water circulated around the reaction vessel.

The apparent quantum yield (AQY) for the water oxidation reaction was measured using the same experimental setup, and was calculated using eqn (2):

AQY (%) = A × R/I × 100 (2)

where A, R, and I represent a coefficient based on the reaction (A = 4 for O₂ evolution), the O₂ evolution rate, and the rate of incident photons, respectively.^31,32 The total number of incident photons (∼0.40 mW) was measured using a calibrated Si photodiode.

Results and discussion

X-ray diffraction and elemental analyses

The as-prepared meso-TiO₂ was found to crystallize in a single rutile phase, and no phase transition was observed during the nitridation process at 673 K to yield the C = 0 sample, as confirmed by XRD analysis (Fig. 1(A)). With increasing concentration of (NH₄)₂TiF₆ (i.e., increasing C value), impurities such as anatase TiO₂ and TiOF₂ appeared, particularly at high C values (C = 25 and 50), consistent with previous reports.^13,16 The absence of impurity phases at lower C values suggests that the impurities are caused by the direct pyrolysis of (NH₄)₂TiF₆ that had not reacted with meso-TiO₂.³³

Fig. 1 (A) XRD patterns and (B) Raman spectra for meso-TiO₂ and meso-TiO₂:N,F obtained by nitriding mixtures of meso-TiO₂ and (NH₄)₂TiF₆ with different C values.

Raman spectroscopy

Raman spectroscopy was employed to investigate the local structure of the synthesized materials. The Raman spectrum of meso-TiO₂ exhibits a weak peak at 235 cm⁻¹ (multi-photon process) and two intense peaks at 446 and 612 cm⁻¹, corresponding to the E_g- and A_1g-modes, respectively. These features are consistent with rutile-phase TiO₂ prepared by the hydrothermal method.^34–36 Peak broadening and slight frequency shifts became more pronounced with increasing C value, except for C = 50, which can be attributed to variations in particle size and/or slight non-stoichiometry induced by oxygen vacancies.³⁷ In particular, the E_g peak, sensitive to local inhomogeneity and lattice distortion around dopants, underwent a successive shift toward lower frequencies with increasing the C value, indicating strong influence of N doping into the rutile TiO₂ lattice. For C = 50, additional peaks at 155 and 517 cm⁻¹ were observed, which are assigned to anatase TiO₂ and TiOF₂ impurities, respectively.^34,38 This observation is consistent with the XRD result (Fig. 1(B)).

Elemental analysis was conducted to confirm the presence of N and F in the doped meso-TiO₂ samples. Compared with the C = 0 sample, the samples with higher C values exhibited greater substitution of N³⁻ for O²⁻ in the TiO₂ lattice (for C ≥ 5), indicating that F compensates for the charge imbalance caused by N doping and thus enables the incorporation of a greater amount of N (Table 1). As expected, the F content increased with the addition of more (NH₄)₂TiF₆. Interestingly, the N content showed a volcanic-type trend; increasing the C value from 25 to 50 led to a decrease in the N concentration. This observation suggests that excessive (NH₄)₂TiF₆ may lead to decomposition or structural disruption that hinders further N doping.

Table 1 Results of elemental analysis and N₂ adsorption/desorption measurements for meso-TiO₂:N,F prepared with different C values

Entry	C	Amount^a/wt%		Specific surface area^b/m^2 g^−1	Total pore volume^c/cm^3 g^−1
		N	F
a Determined by elemental analysis.b Determined by nitrogen adsorption/desorption measurements at liquid nitrogen temperature according to the BET method.c Determined by nitrogen adsorption/desorption measurements at p/p^0 ∼1.d Precursor oxide.
1^d	—	—	—	43	0.061
2	0	0.50	—	21	0.13
3	5	0.52	1.3	18	0.065
4	15	0.81	2.0	14	0.042
5	25	1.1	3.2	10	0.035
6	50	0.96	5.5	4	0.011

---

N₂ adsorption–desorption isotherms

N₂ adsorption–desorption experiments were performed for meso-TiO₂, meso-TiO₂:N (C = 0), and meso-TiO₂:N,F (C ≥ 5) (Fig. 2). The isotherms for meso-TiO₂ exhibited a Type IV profile with an H3 hysteretic loop, in which the lower limit of the desorption branch corresponds to the cavitation-induced region.³⁹ This type of loop is typically observed in mesoporous materials composed of non-rigid aggregates of flaky particles, as will be discussed later. With increasing C values during the nitridation process, the H3-type hysteresis gradually disappeared, accompanied by a notable decrease in the specific area and the total pore volume (Table 1). These results indicate that a greater degree of nitridation, while beneficial for introducing N and F into the meso-TiO₂ lattice and thereby enhancing visible-light absorption, simultaneously leads to degradation of the mesoporous structure.

Fig. 2 N₂ adsorption–desorption isotherms of (A) as-prepared meso-TiO₂ and (B)–(F) meso-TiO₂:N,F with different C values.

Electron microscopy

The microwave-assisted solvothermal method allowed for the formation of monodispersed TiO₂ particles, although some irregularly shaped particles were also observed (Fig. 3(A)). Some of the particles are spherical, with an average diameter of ∼3 µm (Fig. 3(B)). They were composed of nanoparticles ranging from several tens of nanometers in size (Fig. 3(C)). The aggregation of these nanoparticles likely contributed to the observed porous structure, as supported by the N₂ adsorption–desorption isotherms (Fig. 2).

Fig. 3 SEM images of (A)–(C) as-prepared meso-TiO₂ and (D)–(F) meso-TiO₂:N,F with C = 15.

Precise control over the mesostructure and morphology in TiO₂ synthesis remains challenging because of the rapid hydrolysis and peptization of Ti precursors.⁴⁰ Although various synthetic strategies such as sol–gel, template-assisted, and hydrothermal methods have been developed, they often involve time-consuming procedures and harsh reaction conditions.^41–46 Moreover, mesoporous TiO₂ materials synthesized via self-assembly approaches typically consist of anatase nanocrystals because the formation of the rutile phase generally requires higher crystallization temperatures that can collapse the mesostructure as a result of thermal instability.⁴⁷

In our microwave-assisted solvothermal method, the use of an organic solvent and HCl effectively moderated the hydrolysis of the Ti precursor, allowing a balance between rapid hydrolysis and controlled self-assembly while also suppressing excessive crystal growth. Microwave heating further reduced the temperature required for rutile phase formation and enabled the rapid construction of mesoporous TiO₂ microspheres with a well-defined rutile phase. This synthesis was achieved via micelle-assisted assembly in a sealed autoclave within a remarkably short reaction time.

After the nitridation with (NH₄)₂TiF₆ (C = 15), the spherical morphology of meso-TiO₂ was largely retained, although some surface roughening and particle growth were observed (Fig. 3(D) and (E)). These changes are attributed to the growth of smaller primary particles upon N/F co-substitution (Fig. 3(F)), which also contributed to the reduced pore volume discussed in the previous section. The thermal ammonolysis of meso-TiO₂ with (NH₄)₂TiF₆ was accompanied by inevitable particle growth and the formation of anatase and TiOF₂ impurities, the latter becoming more prominent at higher initial concentrations of (NH₄)₂TiF₆.

TEM images of the meso-TiO₂ and the C = 15 sample (Fig. 4) revealed distinct lattice fringes with spacings of 0.26–0.27 nm, consistent with the (101) facet of rutile TiO₂. Given the similar ionic radii of N³⁻, F⁻, and O²⁻, the preservation of the lattice spacing is reasonable. These results demonstrate successful topochemical conversion of meso-TiO₂ into TiO₂:N,F (here, C = 15) while retaining crystallinity, although partial degradation of the original mesoporous structure was observed.

Fig. 4 TEM images of (A) the as-prepared meso-TiO₂ and (B) the C = 15 sample.

Optical properties and bandgap structures

UV-vis diffuse-reflectance spectroscopy (DRS) measurements were performed on meso-TiO₂:N (C = 0) and meso-TiO₂:N,F (C ≥ 5) samples to investigate their optical properties (Fig. 5). The spectra of both the meso-TiO₂:N (C = 0) and meso-TiO₂:N,F (C = 5) samples exhibited a new absorption band in the visible range (400–700 nm), in addition to the original absorption edge of meso-TiO₂ at ∼400 nm. The N,F co-doped samples exhibited substantially stronger visible-light absorption than the N-doped sample, primarily because of enhanced N incorporation facilitated by F co-doping. The absorption extending beyond 600 nm is attributable to reduced Ti species, which may form under high-temperature ammonolysis with NH₃ gas acting as a reductant.^16,48

Fig. 5 UV-vis diffuse-reflectance spectra of as-prepared meso-TiO₂ and meso-TiO₂:N,F with different C values.

As reported by Asahi et al., N doping introduces N 2p orbitals above the O 2p valence band, effectively narrowing the bandgap.⁴⁹ F doping, by contrast, has been reported to primarily modify the electronic structure around the conduction-band edge of TiO₂.⁵⁰ Therefore, the co-substitution of N and F is expected to synergistically narrow the bandgap of meso-TiO₂, enhancing its visible-light absorption.

Because rutile TiO₂ is considered an indirect-bandgap semiconductor, we estimated the bandgaps of meso-TiO₂ and meso-TiO₂:N,F (C = 15) from Tauc plots using the square root of the Kubelka–Munk function (Fig. S1).^30,51,52 The bandgap decreased substantially from 3.03 eV (meso-TiO₂) to 2.25 eV (C = 15), consistent with the green coloration of the doped sample (Fig. S2).

The electrochemical properties were further examined by CV and MS analyses. The C = 15 sample displayed stable redox behavior under both anodic and cathodic polarization across multiple CV cycles (Fig. S3), indicating good electrochemical stability. The band edge potential could be estimated from electrochemical impedance measurements. MS plots showed a positive slope, confirming the n-type semiconductor character of the as-prepared materials (Fig. S4). The flat-band potentials were estimated to be −0.86 V and −0.62 V vs. Ag/AgCl at pH 6.4 for meso-TiO₂ and the C = 15 sample, respectively.

On the basis of previous reports,^16,53 we assumed a 0.2 V difference between the conduction-band potential and the flat-band potential. Under this assumption, in conjunction with the bandgaps and the flat-band potentials estimated from DRS and MS plots, we can depict the bandgap structure (Fig. S5). The conduction-band potential for TiO₂:N,F is approximately the same as that for undoped rutile TiO₂.¹⁶ In the present study, we confirmed this trend for the C = 15 sample, supporting the validity of this assumption.

X-ray absorption spectroscopy

The effect of co-substitution of N and F into rutile TiO₂ on the electronic state of Ti was also confirmed by Ti-K edge X-ray absorption near edge structure (XANES) spectra. The pre-edge structure at the Ti-K edge for TiO₂ features three peaks: A1, A2, and A3 (Fig. 6(A)). The A₁ peak is assigned to the quadrupole transition from 1s to t_2g of the local Ti absorber. The A₂ peak originates from the overlap of two transitions, which are the quadrupole transition from 1s to e_g of the local Ti absorber and the dipole transition from 1s to t_2g of neighboring Ti atoms. The A₃ peak is attributed to the dipole transition from 1s to e_g of the neighboring Ti atoms.^54,55 The XANES spectra of the meso-TiO₂:N,F samples resemble that of rutile TiO₂. However, the relative intensity of A2 increased with increasing C value (Fig. 6(B)), indicating increased distortion of the octahedral TiO₆ unit.⁵⁶ In addition, the post-edge features also provide information on the local structure of Ti. The main peak in the post-edge region resembles that for rutile TiO₂; however, the edge crest of the white line peak at ∼4984 eV is substantially broadened with increasing C value, indicating the loss of long- and medium-range order (Fig. 6(C)).⁵⁷

Fig. 6 Ti K-edge XANES spectra of as-prepared meso-TiO₂ and meso-TiO₂:N,F with different C values. (A) Entire XANES spectrum and (B) pre-edge and (C) white-line regions.

Photogenerated charge carrier dynamics

The photogenerated charge carrier dynamics of the as-prepared meso-TiO₂:N,F samples were evaluated by time-resolved microwave conductivity (TRMC) measurements. TRMC is a powerful technique for probing the dynamic behavior of charge carriers in semiconductor materials, with numerous examples being reported in mixed-anion compounds.^58–61 Although the literature includes TRMC studies for TiO₂,^62–64 no such studies have yet been reported for TiO₂:N,F.

Fig. 7 shows the time-dependent TRMC signals for meso-TiO₂ and meso-TiO₂:N,F samples with different C values. The corresponding values of φΣμ_max and τ_1/2 are summarized in Table S1. In general, aliovalent substitution of an element into a host material causes a charge imbalance, which promotes charge carrier recombination.^12,65 The reduced φΣμ_max (i.e., photoconductivity) for the C = 0 sample compared with that for pristine meso-TiO₂ is attributed to the detrimental effects of N³⁻/O²⁻ substitution on carrier mobility. Interestingly, this attenuation in photoconductivity was not observed in the C = 15 sample, suggesting that an optimal level of N/F co-doping does not exacerbate charge recombination. Instead, it may enhance carrier transport dynamics relative to those in the undoped meso-TiO₂. However, increasing the C value to 25 led to a noticeable decrease in photoconductivity, likely due to the introduction of additional recombination centers associated with excessive doping. The TRMC profile for the C = 50 sample appears more complex, which is attributable to the higher concentration of impurity phases, including anatase TiO₂ (or TiO₂:N,F) and TiOF₂ (Fig. 1).

Fig. 7 TRMC transient decay profiles of as-prepared meso-TiO₂ and meso-TiO₂:N,F samples with different C values.

Photocatalytic water oxidation

A volcano-like trend was observed between the O₂ generation rate and the C value (Fig. 8(A)), with the highest photocatalytic activity being obtained at C = 15, which gave an AQY of 5.2% at 365 nm. The time course of O₂ evolution over the optimal meso-TiO₂:N,F composition is shown in Fig. 8(B). The full time-course data for the other specimens are compared with those for C = 15 in Fig. S6. As the reaction proceeds, the reaction rate gradually decreases from 10.6 µmol h⁻¹ during the first hour to 3.6 µmol h⁻¹ after 10 h. This decrease is attributed to the reduction of Ag⁺ ions in the solution to metallic Ag on the surface of the photocatalyst; the metallic Ag not only occupies the active sites but also interferes with light absorption, ultimately leading to a decrease in the reaction rate.⁶⁶ The optimized meso-TiO₂:N,F exhibited greater activity than previously reported nano-TiO₂:N,F and bulk-TiO₂:N,F (Fig. S7).

Fig. 8 Photocatalytic activity of meso-TiO₂:N,F with different C values for O₂ evolution under visible-light irradiation in AgNO₃ aqueous solution. Reaction conditions: photocatalyst, 50 mg; La₂O₃, 200 mg; 10 mM AgNO₃ aqueous solution, 140 mL; light source, 300 W Xe lamp fitted with a CM-1 mirror and an L42 cutoff filter (λ > 400 nm).

Factors influencing the photocatalytic activity

The results of the photocatalytic reactions indicated that the photocatalytic activity of meso-TiO₂:N,F was strongly dependent on the C value in the synthesis and that the highest activity was obtained with the C = 15 sample. Here, morphology (or porosity), visible-light absorption capability, and photoconductivity are factors considered to affect the photocatalytic activity of meso-TiO₂:N,F. A porous structure may be beneficial for charge carrier migration from the bulk to the surface and for mass transport of reactants. Materials with more pronounced visible-light absorption can, in principle, generate more photogenerated charge carriers, which may participate in surface chemical reactions, depending on the photoconductivity of the material.

Meso-TiO₂ did not show activity under visible light because it did not exhibit visible-light absorption (Fig. 5). The C = 0 and 5 samples exhibited more pronounced visible-light absorption than meso-TiO₂ because of the N/F incorporation. Therefore, meso-TiO₂ doped with N or co-doped with N/F at C = 0 or 5 showed photocatalytic activity under visible-light irradiation, although the photoconductivity was lower than that of meso-TiO₂ (Fig. 7 and 8). The decrease in photoconductivity is not favorable for improving photocatalytic activity. Nevertheless, the fact that the C = 0 and 5 samples showed photocatalytic activity under visible-light irradiation strongly suggests that the visible-light absorption capacity is a major factor influencing the photocatalytic activity. However, the visible-light absorption of the C = 0 and 5 samples is not sufficient because of their lower N content (Table 1 and Fig. 5), which may lead to lower photocatalytic activity than for the C = 15 sample, although the mesoporous structure could be maintained.

Although the C = 25 sample had the highest N content among the synthesized meso-TiO₂:N,F samples and a morphological character similar to that for the C = 15 sample, its O₂ evolution rate was lower than that for the C = 15 sample. One possible explanation for the decrease in activity is that the excess N/F doping increased the density of electron–hole recombination centers.⁴⁸ This idea is consistent with the TRMC measurement results, which indicated that the photoconductivity was attenuated from C = 15 to C = 25 (Fig. 7). A correlation between photocatalytic O₂ evolution rate and photoconductivity has been reported previously. For example, the photocatalytic O₂ evolution rate over the oxyhalides Bi₄TaO₈Cl and PbBiO₂Cl has been reported to be correlated with the product of the maximum TRMC signal (φΣμ_max) and the half-lifetime of charge carriers (τ_1/2).^59,60

For rutile TiO₂, the rate of O₂ evolution from aqueous AgNO₃ tends to increase with increasing particle size of rutile TiO₂ (i.e., with increasing crystallinity of rutile TiO₂).^67,68 The higher activity of the bulk TiO₂:N,F compared with that of the nano-TiO₂:N,F (Fig. S7) is subject to this trend, although the weaker visible-light absorption of nano-TiO₂:N,F may contribute to the lower activity to a certain extent (Fig. S8). Interestingly, the meso-TiO₂:N,F exhibited even higher activity than the bulk TiO₂:N,F. Although the specific surface areas of the meso-TiO₂:N,F and the bulk-TiO₂:N,F are similar, their morphological characteristics differ.¹⁶ This observation again suggests that the mesoporous structure may contribute to the higher activity of the TiO₂:N,F.

Meanwhile, a comparison of the photoconductivity and half-lifetime of the C = 15 sample with those of bulk-TiO₂:N,F is intriguing. The photoconductivity (φΣμ_max) of the C = 15 sample was lower than that of bulk TiO₂:N,F, whereas its half-lifetime was much longer. As a result, the product of φΣμ_max and τ_1/2 for the C = 15 sample was greater than that for bulk-TiO₂:N,F, which may be another explanation for the higher activity of the C = 15 sample. Notably, N and F co-substitution led to a pronounced attenuation of the TRMC signals in both nano- and bulk-TiO₂, whereas an enhancement was observed exclusively in the meso-TiO₂ (Fig. S9). This result suggests that co-doping of N and F introduces additional defects into the nano and bulk forms, thereby hindering the transport of photogenerated carriers. By contrast, the unique structure of the mesoporous TiO₂ presumably promotes efficient transport of photogenerated charge carriers within the nanochannels, leading to improved charge carrier mobility.

The importance of retaining a mesoporous structure was further underscored using another set of meso-TiO₂:N,F samples synthesized with a fixed C value of 15 but subjected to different ammonolysis temperatures. Thermal ammonolysis at 623 K failed to yield single-phase rutile TiO₂:N,F, and temperatures above 673 K were required to achieve phase purity (Fig. S10). However, N₂ adsorption–desorption measurements revealed that such a high-temperature treatment led to collapse of the mesoporous structure (Fig. S11), despite the concomitant enhancement in visible-light absorption (Fig. S12). Consequently, the sample that simultaneously retained the mesoporous texture derived from the meso-TiO₂ precursor and exhibited no impurity phases showed the highest photocatalytic activity (Fig. S13). These results clearly demonstrate that maximizing the O₂ evolution activity in rutile TiO₂:N,F requires a delicate balance between enhancing visible-light absorption and preserving the mesoporous architecture.

Conclusions

We successfully synthesized mesoporous rutile TiO₂:N,F photocatalysts via a topotactic ammonolysis approach using mesoporous rutile-type TiO₂ (meso-TiO₂) as a precursor. Notably, the synthesis of rutile-type meso-TiO₂ without a template—previously considered difficult—was achieved under mild conditions (low temperature and short reaction time) with the aid of microwaves, representing a significant advance in inorganic synthesis.^69,70 Co-substitution of N and F into the rutile lattice modulated the electronic structure of Ti and extended visible-light absorption while simultaneously inducing local distortion of the TiO₆ octahedra. Systematic studies revealed that the visible-light-driven O₂ evolution activity was governed by a delicate balance between doping-induced light absorption and retention of the mesoporous framework. Although high-temperature ammonolysis enhanced visible-light absorption by promoting N incorporation, it also disrupted the mesostructure, resulting in decreased photocatalytic performance. The best-performing sample was obtained under conditions that allowed for both phase-pure rutile formation and preservation of the mesoporous texture, enabling efficient separation of photogenerated charge carriers. These findings underscore the critical importance of structural control in designing non-metal-doped oxide photocatalysts for solar energy conversion. While the optimized meso-TiO₂:N,F does not yet match the activity of state-of-the-art photocatalysts such as BiVO₄, further improvement is expected through the introduction of suitable O₂ evolution cocatalysts. Moreover, the high surface area of TiO₂-based mesostructures suggests broader potential as adsorbents or catalyst supports for capturing harmful gases such as NO_x and volatile organic compounds.^71,72 Ongoing work in our group is directed toward these applications.

Author contributions

S. L. and K. M. designed the project and wrote a draft of the manuscript. S. L. conducted most of the experiments and analyses with X. Z. R. N., F. I. and A. S. performed the time-resolved microwave conductivity measurements. M. O. provided comments on experiments. All of the authors made contributions to the preparation of the manuscript.

Conflicts of interest

There are no conflicts to declare.

Data availability

The data supporting this article have been included as part of the supplementary information (SI). Supplementary information: additional characterization and reaction data. See DOI: https://doi.org/10.1039/d5nh00584a.

Acknowledgements

This work was supported by Grant-in-Aid for Scientific Research (A) (JP25H00899). It was also partially supported by Grant-in-Aids for Transformative Research Areas (A) “Supra-ceramics” (JP22H05142, JP22H05148 and JP 23H04626).

Notes and references

1. S. Sun, Y.-C. Zhang, G. Shen, Y. Wang, X. Liu, Z. Duan, L. Pan, X. Zhang and J.-J. Zou, Appl. Catal., B, 2019, 243, 253–261.
2. Z. Wang, C. Li and K. Domen, Chem. Soc. Rev., 2019, 48, 2109–2125.
3. J. Xing, W. Q. Fang, H. J. Zhao and H. G. Yang, Chem. – Asian J., 2012, 7, 642–657.
4. J. Yan, J. Liu, Y. Ji, M. Batmunkh, D. Li, X. Liu, X. Cao, Y. Li, S. Liu and T. Ma, ACS Catal., 2020, 10, 8742–8750.
5. G. Ge, M. Liu, C. Liu, W. Zhou, D. Wang, L. Liu and J. Ye, J. Mater. Chem. A, 2019, 7, 9222–9229.
6. P. Li, L. Kong, J. Liu, J. Yan and S. Liu, ACS Sustainable Chem. Eng., 2019, 7, 17941–17949.
7. M. M. S. Karim, A. M. Ganose, L. Pieters, W. W. W. Leung, J. Wade, L. Zhang, D. O. Scanlon and R. G. Palgrave, Chem. Mater., 2019, 31, 9430–9444.
8. Q. Wang, M. Nakabayashi, T. Hisatomi, S. Sun, S. Akiyama, Z. Wang, Z. Pan, X. Xiao, T. Watanabe, T. Yamada, N. Shibata, T. Takata and K. Domen, Nat. Mater., 2019, 18, 827–832.
9. H. Fujito, H. Kunioku, D. Kato, H. Suzuki, M. Higashi, H. Kageyama and R. Abe, J. Am. Chem. Soc., 2016, 138, 2082–2085.
10. S. Wei and X. Xu, Appl. Catal., B, 2018, 228, 10–18.
11. M. Hojamberdiev, R. Vargas, Z. C. Kadirova, K. Kato, H. Sena, A. G. Krasnov, A. Yamakata, K. Teshima and M. Lerch, ACS Catal., 2022, 12, 1403–1414.
12. A. Nakada, S. Nishioka, J. J. M. Vequizo, K. Muraoka, T. Kanazawa, A. Yamakata, S. Nozawa, H. Kumagai, S.-I. Adachi, O. Ishitani and K. Maeda, J. Mater. Chem. A, 2017, 5, 11710–11719.
13. K. Nukumizu, J. Nunoshige, T. Takata, J. N. Kondo, M. Hara, H. Kobayashi and K. Domen, Chem. Lett., 2003, 32, 196–197.
14. A. Miyoshi, K. Kato, T. Yokoi, J. J. Wiesfeld, K. Nakajima, A. Yamakara and K. Maeda, J. Mater. Chem. A, 2020, 8, 11996–12002.
15. H. A. Seibel Ii, P. Karen, T. R. Wagner and P. M. Woodward, J. Mater. Chem., 2009, 19, 471–477.
16. A. Miyoshi, J. J. M. Vequizo, S. Nishioka, Y. Kato, M. Yamamoto, S. Yamashita, T. Yokoi, A. Iwase, S. Nozawa, A. Yamakata, T. Yoshida, K. Kimoto, A. Kudo and K. Maeda, Sustainable Energy Fuels, 2018, 2, 2025–2035.
17. Y. Takahara, J. N. Kondo, T. Takata, D. Lu and K. Domen, Chem. Mater., 2001, 13, 1194–1199.
18. X. Wang, K. Maeda, X. Chen, K. Takanabe, K. Domen, Y. Hou, X. Fu and M. Antonietti, J. Am. Chem. Soc., 2009, 131, 1680–1681.
19. J. B. Joo, I. Lee, M. Dahl, G. D. Moon, F. Zaera and Y. Yin, Adv. Funct. Mater., 2013, 23, 4246–4254.
20. D. S. Kim, S. J. Han and S. Y. Kwak, J. Colloid Interface Sci., 2007, 316, 85–91.
21. T. Wang, X. Meng, P. Li, S. Ouyang, K. Chang, G. Liu, Z. Mei and J. Ye, Nano Energy, 2014, 9, 50–60.
22. D. M. Antonelli and J. Y. Ying, Angew. Chem., Int. Ed., 2003, 34, 2014–2017.
23. W. Li, Z. Wu, J. Wang, A. A. Elzatahry and D. Zhao, Chem. Mater., 2013, 26, 287–298.
24. J. H. Pan, X. S. Zhao and W. I. Lee, Chem. Eng. J., 2011, 170, 363–380.
25. Y. Zhang, G. Li, Y. Wu, Y. Luo and L. Zhang, J. Phys. Chem. B, 2005, 109, 5478–5481.
26. Y. Kobayashi, O. J. Hernandez, T. Sakaguchi, T. Yajima, T. Roisnel, Y. Tsujimoto, M. Morita, Y. Noda, Y. Mogami, A. Kitada, M. Ohkura, S. Hosokawa, Z. Li, K. Hayashi, Y. Kusano, J. Kim, N. Tsuji, A. Fujiwara, Y. Matsushita, K. Yoshimura, K. Takegoshi, M. Inoue, M. Takano and H. Kageyama, Nat. Mater., 2012, 11, 507–511.
27. F. D. Romero, A. Leach, J. S. Moller, F. Foronda, S. J. Blundell and M. A. Hayward, Angew. Chem., Int. Ed., 2014, 53, 7556–7559.
28. T. Oshima, T. Ichibha, K. Oqmhula, K. Hibino, H. Mogi, S. Yamashita, K. Fujii, Y. Miseki, K. Hongo, D. Lu, R. Maezono, K. Sayama, M. Yashima, K. Kimoto, H. Kato, M. Kakihana, H. Kageyama and K. Maeda, Angew. Chem., Int. Ed., 2020, 59, 9736–9743.
29. Y. Tang, K. Kato, T. Oshima, H. Mogi, A. Miyoshi, K. Fujii, K. I. Yanagisawa, K. Kimoto, A. Yamakata, M. Yashima and K. Maeda, Inorg. Chem., 2020, 59, 11122–11128.
30. S. Nishioka, F. E. Osterloh, X. Wang, T. E. Mallouk and K. Maeda, Nat. Rev. Methods Primers, 2023, 3, 42.
31. K. Maeda, T. Maeda, C. Suppaso, S. Nishioka, Y. Kamakura, S. Yasuda and T. Yokoi, Sustainable Energy Fuels, 2024, 8, 36–42.
32. C. Wang, H. Zhang, W. Luo, T. Sun and Y. Xu, Angew. Chem., Int. Ed., 2021, 60, 25381–25390.
33. D. Chen, Z. Jiang, J. Geng, J. Zhu and D. Yang, J. Nanopart. Res., 2008, 11, 303–313.
34. J. Yang, S. Mei, J. M. Ferreira, P. Norby and S. Quaresma, J. Colloid Interface Sci., 2005, 283, 102–106.
35. C. T. Cherian, M. V. Reddy, T. Magdaleno, C.-H. Sow, K. V. Ramanujachary, G. V. S. Rao and B. V. R. Chowdari, CrystEngComm, 2012, 14, 978–986.
36. H. L. Ma, J. Y. Yang, Y. Dai, Y. B. Zhang, B. Lu and G. H. Ma, Appl. Surf. Sci., 2007, 253, 7497–7500.
37. Y. Huo, Y. Jin, J. Zhu and H. Li, Appl. Catal., B, 2009, 89, 543–550.
38. S. Guan, Q. Fan, Z. Shen, Y. Zhao, Y. Sun and Z. Shi, J. Mater. Chem. A, 2021, 9, 5720–5729.
39. M. S. Mel’gunov, Adsorption, 2022, 29, 199–208.
40. W. Zhou and H. Fu, ChemCatChem, 2013, 5, 885–894.
41. R. O. Da Silva, R. H. Goncalves, D. G. Stroppa, A. J. Ramirez and E. R. Leite, Nanoscale, 2011, 3, 1910–1916.
42. P. A. R. Bleta and L. Lorenzato, J. Phys. Chem. C, 2010, 114, 2039–2048.
43. J. B. Joo, Q. Zhang, I. Lee, M. Dahl, F. Zaera and Y. Yin, Adv. Funct. Mater., 2011, 22, 166–174.
44. D. V. Bavykin, V. N. Parmon, A. A. Lapkin and F. C. Walsh, J. Mater. Chem., 2004, 14, 2581–2587.
45. J. Yu, G. Wang, B. Cheng and M. Zhou, Appl. Catal., B, 2007, 69, 171–180.
46. Z. Wang, D. Li, X. Zhang, H. Yu, K. Deng, Y. Xu, H. Wang and L. Wang, ACS Appl. Nano Mater., 2024, 7, 9868–9873.
47. K. Lan, R. Wang, W. Zhang, Z. Zhao, A. Elzatahry, X. Zhang, Y. Liu, D. Al-Dhayan, Y. Xia and D. Zhao, Chem, 2018, 4, 2436–2450.
48. Y. W. Irie and K. Hashimoto, J. Phys. Chem. B, 2003, 107, 5483–5486.
49. T. M. Asahi, T. Ohwaki, K. Aoki and Y. Taga, Science, 2001, 293, 269–271.
50. T. Yamaki, T. Umebayashi, T. Sumita, S. Yamamoto, M. Maekawa, A. Kawasuso and H. Itoh, Nucl. Instrum. Methods Phys. Res., Sect. B, 2003, 206, 254–258.
51. Y. Wu, M. Xing, B. Tian, J. Zhang and F. Chen, Chem. Eng. J., 2010, 162, 710–717.
52. S. Suzuki, A. Iwase and A. Kudo, Catal. Sci. Technol., 2020, 10, 4912–4916.
53. Y. Matsumoto, J. Solid State Chem., 1996, 126, 227–234.
54. L. X. Chen, T. Rajh, Z. Wang and M. C. Thurnauer, J. Phys. Chem. B, 1997, 101, 10688–10697.
55. T. Uozumi, A. Kotani and J. C. Parlebas, J. Electron Spectrosc. Relat. Phenom., 2004, 137–140, 623–627.
56. S. Nozawa, T. Iwazumi, H. Osawa and T. Uozumi, Appl. Phys. Express, 2013, 6, 061501.
57. D. R. M. Schwartz, D. R. Mullins, W. Yan, H. Zhu, S. Dai and S. H. Overbury, J. Phys. Chem. C, 2007, 111, 17322–17332.
58. K. Aihara, R. Mizuochi, M. Okazaki, S. Nishioka, S. Yasuda, T. Yokoi, F. Ishiwari, A. Saeki, M. Inada and K. Maeda, ACS Mater. Lett., 2023, 5, 2355–2360.
59. H. Suzuki, M. Higashi, H. Kunioku, R. Abe and A. Saeki, ACS Energy Lett., 2019, 4, 1572–1578.
60. H. Suzuki, S. Kanno, M. Hada, R. Abe and A. Saeki, Chem. Mater., 2020, 32, 4166–4173.
61. H. Tsuchikado, S. Anabuki, O. Cretu, Y. Kinoshita, M. Hattori, Y. Shiroma, D. Fan, M. Okazaki, T. Soma, F. Ishiwari, S. Nozawa, T. Yokoi, M. Hara, K. Kimoto, A. Yamakata, A. Saeki and K. Maeda, ACS Appl. Energy Mater., 2025, 8, 3541–3552.
62. H. H. Scot, T. Martin, W. Choi and M. R. Hoffmann, J. Chem. Soc., Faraday Trans., 1994, 90, 3315–3322.
63. S. Nakajima and R. Katoh, J. Mater. Chem. A, 2015, 3, 15466–15472.
64. A. Saeki, Y. Yasutani, H. Oga and S. Seki, J. Phys. Chem. C, 2014, 118, 22561–22572.
65. K. D. Tsuyoshi and T. Takata, J. Phys. Chem. C, 2009, 113, 19386–19388.
66. T. Hisatomi, C. Katayama, Y. Moriya, T. Minegishi, M. Katayama, H. Nishiyama, T. Yamada and K. Domen, Energy Environ. Sci., 2013, 6, 3595–3599.
67. H. Kominami, S.-Y. Murakami, J.-I. Kato, Y. Kera and B. Ohtani, J. Phys. Chem. B, 2002, 106, 10501–10507.
68. O.-O. Prieto-Mahaney, N. Murakami, R. Abe and B. Ohtani, Chem. Lett., 2009, 38, 238–239.
69. J. Li, R. Li, W. Wang, K. Lan and D. Zhao, Adv. Mater., 2024, 36, e2311460.
70. A. Wang, Y. Ma and D. Zhao, ACS Nano, 2024, 18, 22829–22854.
71. H. Ma, Y. Li, C. Wang, Y. Li and X. Zhang, Catalysts, 2025, 15, 103.
72. S. Bagheri, N. Muhd Julkapli and S. Bee Abd Hamid, Sci. World J., 2014, 2014, 727496.

---