Regulating protonation paths for enhanced photocatalytic CO₂ methanation by coupling Pt sites on WO_2.9/TiO₂

Abstract

CO₂ methanation via photocatalysis with water vapor is a sustainable technique for reducing CO₂ emission but is challenged by the high energy barrier associated with the initial adsorption, activation and protonation of CO₂ molecules. In this work, a substoichiometric WO_2.9 thin film with strong Lewis acidity was coated on TiO₂ microspheres, followed by the deposition of Pt cocatalysts on WO_2.9 with controlled Pt single atoms and clusters (Pt–WO_2.9/TiO₂). The methane production rate reached 10.74 μmol h⁻¹ g⁻¹ with a selectivity of 99.8%, which was ∼40 times higher than that of bare TiO₂ (0.27 μmol h⁻¹ g⁻¹). The high methane production rate was attributed to the synergy of Pt sites on the WO_2.9/TiO₂ heterojunction, where the Pt clusters facilitated water dissociation, thereby providing H* through hydrogen spillover on the surface, and the presence of a substoichiometric WO_2.9 surface further enhanced the spillover process. The high density of active H* promoted the protonation pathway for CO₂ activation (CO₂ → COOH⁺ → *COOH), which improved the adsorption of the essential intermediate *CO on Pt single atoms and displayed a significantly reduced energy barrier for the protonation reaction of C1 intermediates, resulting in a mixed reaction pathway. This work provides new insights into a mechanism to regulate the reaction path to facilitate efficient photocatalytic CO₂ methanation.

---

1. Introduction

Methane is the main component of natural gas and is widely used as a clean energy source or raw material for bulk production of chemicals.¹ In industry, CH₄ is synthesized using CO₂ and H₂ as raw materials via the Sabatier reaction,² but the carbon emission and energy consumption of the process need to be improved. With the increasing demand for natural gas in the coming decades,³ CH₄ synthesis via a green and sustainable approach is highly promising, with photocatalysis emerging as a method to achieve CO₂ methanation from solar energy. However, owing to the high thermodynamical stability of CO₂, it is challenging to achieve high methane yield via photocatalysis.⁴ Moreover, the conversion from CO₂ to CH₄ is an eight-electron reduction reaction that requires a high density of protons and electrons, resulting in relatively slow reaction kinetics.⁵ In addition, the competing reaction for CO₂ reduction is the recombination of protons to form hydrogen. Therefore, photocatalysts that can accumulate photogenerated electrons and promote the migration of protons to carbon intermediates may enhance methane production.⁶

The substoichiometric tungsten oxides, such as WO_2.72, W₂₄O₆₈ and WO_2.9, are stable in air,⁷ and the light absorption of WO_3−x can be extended to the near-infrared region owing to the defect states below the conduction band minimum (CBM).^8,9 The V_O (oxygen vacancy) and unsaturated metal coordination sites pair can be used as Lewis acid sites to adsorb and activate CO₂ molecules by reducing the activation energy barrier of CO₂ molecules.¹⁰ In addition, W(VI) sites in WO_3−x can be used as electron reservoirs by reducing them to W(V).¹¹ The tungsten oxide with oxygen deficiency (V_O) as substoichiometric tungsten oxides (WO_3−x) are introduced to construct a heterojunction with TiO₂ to improve the charge transfer kinetics of titania and increase the electron density on the surface. Titania is the most investigated semiconductor in photocatalysis due to its high abundance, high chemical stability and resistivity, but it is limited by its overall low reaction rate. In this work, mesoporous TiO₂ microspheres were used as a support for substoichiometric tungsten oxide loading. The mesoporous structure and high surface area can provide abundant exposed active sites for surface reduction reactions and facilitate the construction of heterojunctions with intimate interfacial contact. The closely contacted anatase/rutile phase junctions, combined with highly crystalline anatase and rutile nanoparticles, may synergistically promote charge separation and transport.^12,13

For methane production, noble metal cocatalysts can effectively enhance photocatalytic activity because of their electron transfer ability.^14–16 However, hydrogen evolution reaction, which is a competing reaction in CO₂ reduction, can also be accelerated by noble metals as well.¹⁷ Previous studies have shown that the coordination environment,^16,18 size and distribution¹⁹ of Pt can effectively inhibit hydrogen evolution. It was reported in the 1960s that hydrogen spillover from noble metals to the oxide support occurred at the interface between the metal and the support.²⁰ For example, it has recently been demonstrated that Pt nanoparticles (Pt NPs) dissociate H₂O into absorbed H atoms (*H) and spillover from Pt NPs to the CO₂ adsorption sites on the support to produce methane.¹⁶

Moreover, the activation of CO₂ during photoreduction involves two different pathways: activation by electrons (CO₂ → ·CO₂⁻) and protonation (CO₂ → COOH⁺). Recently, experiments have demonstrated that CO₂ activation via protonation on semiconductor photocatalysts, forming an O=C=O–H⁺ intermediate,²¹ can significantly decrease the reaction energy barrier when such an intermediate undergoes further multiple proton-electron coupling reactions. Therefore, it is essential to design photocatalysts that favor the protonation of CO₂. In this work, substoichiometric WO_2.9 was fabricated on TiO₂ microspheres for photocatalytic CO₂ methanation, where the deposition of Pt cocatalysts and substoichiometric WO_2.9 with good distribution can be expected to control the hydrogen spillover and improve the methane production rate (Scheme 1).

Scheme 1 Schematic of the synthesis of TiO₂ microspheres, WO_2.9/TiO₂, and Pt–WO_2.9/TiO₂ photocatalyst.

2. Experimental

2.1 Chemicals and materials

Pluronic F127 ((H(OC₂H₄)_x(OC₃H₃)_y(O₂H₄)_zOH), M_w = 15 000 g mol⁻¹), H₂PtCl₆·6H₂O, WCl₆ (99.9%), tetrahydrofuran (THF, 99.9%), sodium bicarbonate (99%) and tetrabutyltitanate (TBOT, 99%) were purchased from Adamas. Acetic acid (AcOH, 99%), absolute ethyl alcohol (C₂H₅OH, 99.8%), and hydrochloric acid (HCl, 37 wt%) were purchased from Chron in Chengdu. Ultrapure water (resistivity: 17.38 MΩ cm) was used in the synthetic process. All reagents were used as received without further purification.

2.2 Synthesis of the catalysts

Synthesis of TiO₂ microspheres. In a typical procedure, 1 mL of HOAc and HCl (37 wt%) was dropped in 15 mL of THF. 0.75 g F127 yellow powder was slowly introduced into the above solution. The mixture was vigorously stirred to form a clear and transparent solution. Subsequently, 1.7 mL of TBOT was dropped under vigorous stirring to form a clear yellow solution. The yellow solution was then left in a drying oven to evaporate the THF solvents at 40 °C for 20 h and at 80 °C for another 10 h. The obtained precipitate was calcined in a tubular furnace under N₂ flow at 400 °C for 3 h. Finally, the white powder was obtained by calcination in a muffle furnace at 400 °C for 3 h.

Synthesis of WO_2.9/TiO₂ microspheres. In the N₂ atmosphere, 17.1 mg WCl₆ and 100 mg TiO₂ microspheres were dispersed and dissolved in 40 mL of anhydrous ethanol, and the mixture was continuously stirred for 3 hours in an ice-water bath. After that, it was poured into a 50 mL hydrothermal kettle and kept in an oven at 180 °C for 12 hours. After natural cooling to room temperature, the samples were washed three times with deionized water, dried in a vacuum oven at 60 °C, and calcined at 400 °C in a 5% Ar/H₂ atmosphere for 2 hours.

Synthesis of Pt–WO_2.9/TiO₂ microspheres. 100 mg WO_2.9/TiO₂ microspheres were dispersed in 20 mL DI water, 1 mg chloroplatinic acid was added and sonicated in N₂ atmosphere and ice water bath for 40 minutes. After freezing in the refrigerator (−20 °C) for 16 hours, keep it frozen and irradiate it with 275 nm ultraviolet light for 1 h in an ice/water bath. After the natural melting of the ice and the temperature reaching room temperature, the powders were centrifuged, washed three times with DI water and dried overnight at 80 °C in a vacuum oven.

2.3 Material characterization

The morphology of all samples was observed on a field emission scanning electron microscope (FESEM) (JEOL, JSM-7500, Japan). Transmission electron microscope (TEM) images were acquired with a Japan JEOL-F200 at an acceleration voltage of 200 kV. Samples were dispersed in ethanol, and 10 μL suspension was dropped on an ultra-thin carbon film supported by a 300-mesh copper TEM grid and dried for 1 h under an IR light. Powder X-ray diffraction (XRD) patterns were recorded on a Rigaku MiniFlex600 instrument equipped with a Cu Kα radiation (λ = 0.154178 nm) in the 2θ range of 10–80° with a step size of 0.01°. The lattice spacing d was calculated using Bragg's law: 2d sin θ = nλ. The Rietveld refinement line of TiO₂ was obtained using Maud software, in which the crystal structure models of anatase (9008213.cif) and rutile (9004141.cif) were introduced, respectively, and background and scale parameters, basic phase parameters, microstructure parameters, crystal structure parameters and all parameters for texture were adjusted in turn to calculate the ratio of rutile and anatase in TiO₂ microspheres. X-ray photoelectron spectroscopy (XPS) was performed on a Thermo Scientific K-Alpha instrument with a monochromatic Al Kα X-ray source (12 kV, 6 mA). The vacuum in the analysis room is 5.0 × 10⁻⁷ mbar. The binding energies of all the spectra were calibrated to the binding energy of C 1s at 284.6 eV.

2.4 Photocatalytic reduction of CO₂

The photocatalytic CO₂ reduction experiments were performed using a solid–vapor reaction in a Pyrex glass batch reactor with a window on top for light illumination (Labsolar 6A, Beijing Perfect light, China). The reaction temperature was controlled at approx. 10 °C by circulating cooling reagents at 5 °C. A 300 W Xe-lamp equipped with an AM 1.5 filter was placed on top and the optical power density was 100 mW cm⁻². 20 mg sample was dispersed in 1 mL of DI water and dipped on a porous ceramic substrate (Beijing Perfect Light Technology Co., Ltd) with a 30 mm diameter, then dried at 80 °C for 30 minutes. After drying, the substrate loaded with the catalyst was placed on the quartz support in the center of the reactor, 10 cm above the bottom of the reactor. The reactor was filled with 50 mL DI water at the bottom. After the reactor was sealed, it was continuously vacuum-treated for 10 minutes and purged three times with high-purity CO₂ (99.999%) gas, where CO₂ gas was bubbled through the water at the bottom of the reactor and continuously flowed until the pressure inside the reactor reached ∼50 KPa. During irradiation, the products in the gas phase at different reaction times were analyzed by an online gas chromatograph (GC2030, SHIMADZU, Japan) with a barrier discharge ionization detector (BID) using high-purity helium (99.999%) as the carrier gas and Carboxen@1010 PLOT (30 m × 0.53 mm) as a chromatographic column. In blank experiments, in which there was no light source or photocatalyst, high-purity N₂ was used instead of CO₂.

The selectivity of CO₂ reduction (S_CO2) and CH₄ (S_CH4) are calculated using the following equations:

2.5 Photoelectrochemical measurements

The photoelectrochemical measurements of TiO₂, Pt–TiO₂, WO_2.9/TiO₂, and Pt–WO_2.9/TiO₂ were performed in a three-electrode quartz electrolytic cell using an electrochemical workstation (Autolab PGSTAT 204A Instruments by Metrohm, Utrecht, Netherlands). The counter electrode was Ag/AgCl and the reference electrode was a platinum sheet (1 × 1 cm). The working electrode was prepared by the spin-coating method. 5 mg of the samples and 20 μL Nafion solution were placed into 0.5 mL of ethanol and sonicated for 1 h to obtain a uniform suspension, which was then uniformly coated on the glassy carbon electrode and dried in a vacuum oven at 80 °C. 0.5 M Na₂SO₄ solution was used as the electrolyte.

The transient photocurrent spectra were tested at a bias of 0.5 V with an on/off period of 40 s for each transient. The EIS measurements were carried out in the frequency range of 0.01–10⁵ Hz at a bias of 0.5 V_OCP. Mott–Schottky (M–S) curves were obtained at frequencies of 600 Hz, 800 Hz, 1000 Hz and 1200 Hz in the potential range of −1.5–1.2 V.

2.6 DRIFTS measurements

DRIFTS measurements were conducted using a Thermo Scientific Nicolet FTIR spectrometer equipped with a liquid nitrogen-cooled HgCdTe (MCT) detector. The spectra were obtained by applying an average of 128 scans with a spectral resolution of 2 cm⁻¹. The final spectra were obtained after removing the background spectra.

DRIFTS for CO Adsorption. The chamber was purged with Ar (99.999%) at room temperature (RT) for 30 min to eliminate air after loading the samples and then the background spectrum was recorded. Subsequently, 8% CO in Ar was purged until the adsorption equilibrium was reached. The CO was then stopped by purging Ar into the chamber to remove the CO in the gas phase.

DRIFTS for in situ CO₂ reaction. The reaction chamber was purged with N₂ (99.999%) at 100 °C for 30 min to remove moisture and air from the sample and then cooled to RT. After recording the background spectrum in N₂, CO₂ (99.999%) and H₂O vapor via bubbling of CO₂ gas through water were introduced into the reaction chamber until adsorption equilibrium was reached. The chamber was then sealed and the spectra were recorded at certain time intervals under the illumination of a 300 W Xe lamp with an AM 1.5 filter.

2.7 DFT calculations

All the DFT calculations were performed using the Vienna ab initio simulation package (VASP)²² with the projector augmented wave (PAW) method.²³ The exchange-functional is treated using the generalized gradient approximation (GGA) with Perdew–Burke–Ernzerhof (PBE)²⁴ functional. The energy cutoff for the plane-wave basis expansion was set to 450 eV. Partial occupancies of the Kohn–Sham orbitals were allowed using the Gaussian smearing method with a width of 0.2 eV. The WO_2.9 (010) and anatase TiO₂ (101) surfaces were constructed, where the Pt cluster with 10 Pt atoms was established loading on the surface. In addition, single atom Pt was constructed by replacing the W and Ti atoms on WO_2.9 (010) and anatase TiO₂ (101) surfaces, respectively. The k-point of 2 × 1 × 1 and 1 × 1 × 1 was used for the calculations on WO_2.9 (010) and anatase TiO₂ (101) surfaces, respectively. The self-consistent calculations applied a convergence energy threshold of 10⁻⁵ eV, and the force convergency was set to 0.05 eV Å⁻¹.

The free energy corrections were calculated at 298 K using the following equation:

ΔG = ΔE + ΔG_ZPE + ΔG_U − TΔS

where ΔE, ΔG_ZPE, ΔG_U, and ΔS refer to the DFT-calculated energy change, the correction from zero-point energy, the correction from inner energy and the correction from entropy.²⁵

3. Results and discussion

3.1 Morphology and structure analysis

In brief, TiO₂ microspheres were synthesized via solvent evaporation-driven oriented assembly using tetrabutyltitanate precursors.¹² A typical scanning electron microscopy (SEM) image of TiO₂ (Fig. 1a) exhibits a regular spheric shape with a diameter of 1.57 ± 0.10 μm, consisting of anatase (JCPDS 21-1272) and rutile (JCPDS 21-1276), with an R (R = I_A/I_R) of 23.6%, as determined by Rietveld refinement ((Fig. 1b, Table S1 and Fig. S1†). The homojunction between anatase and rutile in the microspheres allows for improved charge-carrier separation efficiency, higher reduction potential of electrons (E_CB of rutile ≤0.6 eV) and light absorption ability (E_g of rutile: ∼3.0 eV).²⁶

Fig. 1 SEM image and statistical analysis on particle size of (a) TiO₂ microspheres and (b) Pt–WO_2.9/TiO₂; (c) HRTEM image (inset: TEM image); (d) AC-HAADF-STEM image and (e) EDS elemental mapping of Pt–WO_2.9/TiO₂; CO-DRIFTS for (f) Pt–WO_2.9/TiO₂ and (g) Pt–TiO₂, WO_2.9/TiO₂.

A substoichiometric tungsten oxide film was loaded onto TiO₂ microspheres via a hydrothermal method using tungsten hexachloride (WCl₆) as a metal precursor in deoxygenated anhydrous ethanol. The morphology of tungsten oxide depends on the concentration of WCl₆ in solution, but with TiO₂ microspheres as support, the morphology of tungsten oxide is directed by the support, where W⁶⁺ ions adsorb on the oxide sites, form nuclei and grow as a film (Fig. S2†).

The XRD spectrum (Fig. S1b†) for WO_2.9/TiO₂ shows no corresponding diffraction peaks for WO_2.9, which may be attributed to the high dispersion of WO_2.9 on the surface of TiO₂ microspheres. For reference, pure tungsten oxide in a dark blue color was synthesized using the same hydrothermal method. The XRD spectrum of the tungsten oxide (Fig. S3†) shows obvious diffraction peaks at 23.6° and 48.3°, which can be ascribed to the (010) and (020) facets in the WO_2.9 phase (JCPDS 05-0386). The Raman spectroscopy (Fig. S4†) revealed characteristic Raman active modes with symmetries E_g at 443 cm⁻¹ and A_1g at 609 cm⁻¹ for rutile TiO₂, and E_g at 144 cm⁻¹ for anatase. The broad peak at about 230 cm⁻¹ is aroused from the multiple phonons scattering. The attenuation of the peak at 144 cm⁻¹ may be due to the change in the anatase phase in the TiO₂ microspheres. The coating amount of WO_2.9 was optimized based on the catalytic performance to be 10 wt% (Fig. S5a†).

The deposition of Pt was performed using photochemistry under restricted conditions.²⁷ The ice lattice naturally confines the dispersion of precursors, nucleus formation and crystal size. The density, size, dispersion and allocation of Pt atoms can be fine-regulated by the density of photoelectrons,²⁸ and the Pt precursor concentration. In this work, Pt cocatalysts were deposited as single atoms and nanoclusters on WO_2.9 film, reduced by photogenerated electrons under irradiation of ultraviolet light at λ = 275 nm. For reference, Pt cocatalysts were deposited on TiO₂ microspheres via the same photochemistry method and annotated as Pt–TiO₂. The amount of the Pt precursor was optimized based on the catalytic performance to be 1 wt% (Fig. S5b†).

The diameter of TiO₂ microspheres in Pt–WO_2.9/TiO₂ is 1.61 ± 0.11 μm, where the increase of 40 nm was attributed to the coating of WO_2.9 film (Fig. 1b). The lattice fringes at the edge of the microsphere were analyzed to be 0.377 nm, 0.370 nm and 0.364 nm from the HRTEM image (Fig. 1c and S6a†), corresponding to the (010), (302) and (106) crystal planes of WO_2.9, respectively.

The atomic dispersion of Pt was analyzed using aberration-corrected high-angle annular dark-field scanning TEM (AC HAADF-STEM) imaging (Fig. 1d and S6b†), which clearly resolves isolated Pt single atoms (marked by orange circles) and uniformly dispersed Pt clusters (marked by red circles) composed of several Pt atoms on WO_2.9, with no evidence of Pt nanoparticles exceeding 1 nm in size. The compositions of the elements were analyzed with a uniform distribution of Pt and W ((Fig. 1e). To further clarify the distribution of Pt, CO-adsorption diffuse reflectance infrared Fourier transform spectroscopy (CO-DRIFTS) measurements for Pt–WO_2.9/TiO₂ are performed (Fig. 1f). First CO coverage saturation is reached with two strong IR peaks for gas phase CO. Then, gas phase CO and physically adsorbed CO were removed, which leaves chemical adsorption of CO with two sets of absorption bands centered at 2103 cm⁻¹ and a broad peak at 2070 cm⁻¹. The former is ascribed to linearly adsorbed CO on single-atom Pt^δ+,²⁹ and the latter is assigned to linearly adsorbed CO on Pt⁰ in Pt particles.³⁰

The CO-DRIFTS spectrum of WO_2.9/TiO₂ exhibits no clear bands for adsorbed CO (Fig. 1g), indicating that the CO adsorption bands observed for Pt–WO_2.9/TiO₂ originate from the CO adsorbed on the Pt species. For Pt–TiO₂, two sets of CO absorption bands are observed, with one centered at 2119 cm⁻¹ and another broad band centered at 2070 cm⁻¹, where the former peak at 2119 cm⁻¹ is attributed to CO adsorption on single Pt^δ+ atoms. Compared with CO on Pt–WO_2.9/TiO₂ (2103 cm⁻¹), there is a blue shift in the CO adsorption peak on single Pt^δ+ sites, indicating that the back donation from WO_2.9 to Pt^δ+ is stronger than that from TiO₂. To evaluate the stability of Pt on TiO₂ and WO_2.9/TiO₂, the photocatalysts were pretreated in Ar/H₂ at 100 °C and CO-DRIFTS spectra were collected after the treatment (Fig. S7†).

X-ray photoelectron spectroscopy (XPS) was used to analyze the chemical species of the photocatalysts. The XPS survey spectra of TiO₂, Pt–TiO₂, WO_2.9/TiO₂ and Pt–WO_2.9/TiO₂ demonstrate that no impurity elements were present in the as-synthesized samples, except for Pt, W, O, Ti and C (Fig. S8†). The XPS Ti 2p spectrum of TiO₂ shows four symmetric peaks after deconvolution of the spectrum, with a pair of binding energy at 458.4 eV and 464.1 eV for Ti 2p_3/2 and Ti 2p_1/2 of Ti⁴⁺, and another pair at 456.9 eV and 462.6 eV representing Ti 2p_3/2 and Ti 2p_1/2 for Ti³⁺, with an area ratio of 36.5% for Ti³⁺ (Fig. 2a).

Fig. 2 High-resolution XPS spectra of (a) Ti 2p; (b) W 4f; (c) Pt 4d5/2; and (d) O 1s.

With the coating of WO_2.9, for the Ti 2p spectra of WO_2.9/TiO₂ and Pt–WO_2.9/TiO₂, a doublet peak at 458.4 eV for Ti 2p_3/2 and 464.1 eV for Ti 2p_1/2 of Ti⁴⁺ can be detected, where the shoulder peak of Ti³⁺ vanishes, which may be attributed to the oxidation during hydrothermal reaction in solution. During the formation of WO_2.9, electrons are preferentially extracted from Ti³⁺-associated oxygen vacancies in TiO₂, thereby oxidizing Ti³⁺ to Ti⁴⁺.

For Pt–TiO₂, the shoulder peak for Ti³⁺ vanished, possibly because of Pt deposition on the vacancy sites. However, the binding energy of Ti⁴⁺ shifts to lower energy (∼0.30 eV), suggesting either the defects partially exist, or the electrons transfer from Pt to TiO₂ at the interface, which may induce upward surface band bending of TiO₂.

In the W 4f XPS spectrum of WO_2.9/TiO₂, two doublets can be deconvoluted-peaks with a binding energy of 35.2 eV and 37.3 eV attributed to W⁶⁺, and another doublet at 33.9 eV and 36.0 eV attributed to W⁵⁺ (Fig. 2b). With Pt deposition, the peak of W⁶⁺ for Pt–WO_2.9/TiO₂ shifts to higher binding energy (∼0.13 eV), indicating a decrease in the electron density of the W species, probably due to the transfer to the Pt sites. The shoulder peak for W⁵⁺ exists after the photodeposition of Pt, with an area ratio of 13.7%, which is close to the ratio for W⁵⁺ in WO_2.9/TiO₂ (13.0%) (Table S2†). Compared with the change in Ti 2p for Pt–TiO₂ after deposition of Pt, the defect sites in WO_2.9 remain for Pt–WO_2.9/TiO₂, suggesting a stable structure of substoichiometric WO_2.9.

The Pt 4d_5/2 spectrum of Pt–WO_2.9/TiO₂ was resolved into three components, Pt⁴⁺ at 316.8 eV, Pt²⁺ at 314.4 eV, and Pt⁰ at 311.6 eV (Fig. 2c). Compared with the Pt 4d_5/2 spectrum for Pt–TiO₂, the binding energies of Pt²⁺ show a shift to lower binding energy (∼0.20 eV, 314.6 eV), indicating an overall increase in the electron density of Pt on Pt–WO_2.9/TiO₂. For both samples, the Pt²⁺ species is dominant (∼55%), but the concentration of Pt⁰ species in Pt–TiO₂ (22.6%) is higher than that in Pt–WO_2.9/TiO₂ (15.7%) (Table S2†).

The O 1s XPS spectrum (Fig. 2d) of TiO₂ shows three components at 527.8 eV, 529.6 eV, and 531.7 eV, possibly corresponding to electron-rich oxygen (O_R < 529.0 eV), lattice O (O_L, Ti–O), and hydroxyl O (O_C, –OH), respectively.³¹ With the coating of WO_2.9, for the O 1s spectra of WO_2.9/TiO₂ and Pt–WO_2.9/TiO₂, the peak related to electron-rich oxygen disappeared, which may be attributed to the oxidation during the hydrothermal reaction in solution. An additional peak at ∼531.2 eV appears, which is characteristic of the oxygen-deficient tungsten oxide (O_V). Compared with the O 1s spectra for TiO₂ and WO_2.9/TiO₂, the O 1s spectra for Pt–TiO₂ and Pt–WO_2.9/TiO₂ show a shift to lower binding energy, which may be due to photoreduction and the interaction between the surface and deposited Pt.

The Mott–Schottky spectra show that the flat band (E_fb) potential of TiO₂ and WO_2.9 is −0.85 and 0.10 V vs. Ag/AgCl (about −0.65 and 0.30 V vs. NHE, E_NHE (V) = E_Ag/AgCl + 0.197), respectively. (Fig. 3a) Furthermore, for n-type semiconductors, the conduction band minimum (CBM) is typically ∼0.2 V more negative than the flat-band potential. Consequently, the CBM positions of TiO₂ and WO_2.9 were measured to be −0.85 V and 0.10 V (vs. NHE), respectively. The optical band gap was obtained from Tauc plots according to the following equation:

(αhν)ⁿ = A(hν − E_g)

where α is the absorption coefficient, h is Planck's constant and n is the vibration frequency (n = 2 for TiO₂, n = 1/2 for WO_2.9). The band gap (E_g) of TiO₂ and WO_2.9 was thus calculated to be 3.10 and 2.36 eV (ref. 32 and 33) (Fig. 3b), respectively. The band structures of TiO₂ and WO_2.9 are thus proposed to be a type II heterojunction.

Fig. 3 (a) Mott–Schottky spectra and (b) Tauc plots of TiO₂ and WO_2.9, inset: proposed band structures of TiO₂ and WO_2.9. High-resolution XPS spectra of (c) Ti 2p, and (d) W 4f of WO_2.9/TiO₂ before and after light irradiation.

In situ XPS Ti 2p and W 4f spectra under illumination (Fig. 3c and d) demonstrate that the Ti 2p peaks shifted +0.36 eV to higher binding energy, and the W 4f peaks shifted −0.18 eV, accompanied by an increase in W⁵⁺ content (13.7 → 24.0%). These observations confirm the directional electron transfer from TiO₂ to WO_2.9, consistent with a type II heterojunction mechanism.

3.2 Photocatalytic CO₂ reduction and photoelectrochemical performance

The above XPS analysis demonstrates the electronic interactions between WO_2.9 and Pt, and between WO_2.9 and TiO₂, which may contribute to the photocatalytic methane production. The photocatalytic CO₂ reduction performance of the Pt–WO_2.9/TiO₂ composites was evaluated and measured, leading to the production of CO, CH₄, and H₂ in the gas phase. Each experiment was repeated three times for the reference samples as TiO₂, Pt–TiO₂, and WO_2.9/TiO₂, to ensure reproducibility. The comparison of CH₄, H₂ and CO yield over reaction time under illumination is shown in Fig. S10.†

For TiO₂, the yield of CH₄ after 6 hours is 1.61 μmol g⁻¹ (Fig. 4a), which increased to 3.68 μmol g⁻¹ after coating with WO_2.9 for WO_2.9/TiO₂. For Pt–WO_2.9/TiO₂, the yield of CH₄ reaches 64.41 μmol g⁻¹, which is 17.5 times higher than that of WO_2.9/TiO₂. However, for Pt–TiO₂, the CH₄ yield is 37.63 μmol g⁻¹, which is about half of that for Pt–WO_2.9/TiO₂.

Fig. 4 Photocatalytic production yield (a) and selectivity (b) of TiO₂, WO_2.9/TiO₂, Pt–WO_2.9/TiO₂ and Pt–TiO₂ in a 6 h reaction time. (c) Photocurrent transient spectra and (d) electrochemical impedance spectra (EIS) of TiO₂, Pt–TiO₂, WO_2.9/TiO₂, and Pt–WO_2.9/TiO₂.

Compared with the methane production rate reported in the literature, the value reported in this work is among the top values (Table S4†). In the reference experiments, there was no detection of the gas product, e.g. CH₄ or CO, in the experiments without CO₂ (using substituted N₂), without light irradiation, or as a photocatalyst (Fig. S11†).

In addition, the H₂ yield is 3.34 μmol g⁻¹ for TiO₂ and 0.80 μmol g⁻¹ for WO_2.9/TiO₂, suggesting that the WO_2.9/TiO₂ heterojunction significantly improved the overall CO₂ conversion rate and suppressed the competing reaction for hydrogen evolution. Moreover, the hydrogen yield is 41.12 μmol g⁻¹ for Pt–TiO₂, which is 4.8 times higher than that for Pt–WO_2.9/TiO₂ (8.47 μmol g⁻¹), indicating that the deposition of Pt promotes the hydrogen evolution reaction, but with WO_2.9, hydrogen evolution can be effectively inhibited.

The selectivity of CO₂ reduction (S_CO2) and CH₄ (S_CH4) are calculated based on the yield, where S_CO2 refers to the selectivity for electron consumption CO₂ reduction (see Fig. S12†) in competition with hydrogen evolution, and S_CH4 refers to the methane selectivity among CH₄ and CO (Fig. 4b). The S_CO2 is 67.1% for TiO₂ microspheres, which increases significantly to 95.0% with the loading of WO_2.9 and reaches a remarkably high value of 96.8% for Pt–WO_2.9/TiO₂. However, S_CO2 for Pt–TiO₂ decreases to 78.6%, which suggests that the presence of WO_2.9 facilitates the electron transfer to CO₂ reduction reaction.

The stability of Pt–WO_2.9/TiO₂ was evaluated by performing six consecutive reaction cycles (6 hours for each). The average methane production rate in the fifth cycle was 73.3% of the first cycle (Fig. S13†), showing a slight decrease in the activity, which may stem from partial deactivation of the Pt active sites or surface contamination by intermediates.

To understand the impact of Pt sites on the CO₂ reduction reaction, Pt NP–WO_2.9/TiO₂ was prepared by annealing Pt–WO_2.9/TiO₂ in N₂ to convert Pt to nanoparticles. The methane production rate of Pt NP–WO_2.9/TiO₂ significantly decreased to 5.25 μmol g⁻¹ h⁻¹, and the hydrogen production rate increased to 19.19 μmol g⁻¹ h⁻¹ (Fig. S14†), that gives a S_CO2 of 52.4%. These results indicate that the competitive reaction of hydrogen evolution is favoured on Pt nanoparticles.

The S_CH4 among CH₄ and CO for TiO₂ was 94.9%, that is like S_CH4 for WO_2.9/TiO₂ (96.2%). In the presence of Pt, the S_CH4 is 99.8% for Pt–WO_2.9/TiO₂, 99.7% for Pt–TiO₂, and 99.9% for Pt NP–WO_2.9/TiO₂, which suggests that Pt can significantly improve the methane selectivity, possibly contributing to the proton production and spillover during the reaction.

The photocurrent density of Pt–WO_2.9/TiO₂ (1.77 μA cm⁻²) is nearly 3 times as compared to TiO₂ (0.56 μA cm⁻²), demonstrating a significant improvement in the separation and transfer of photogenerated charge carriers (Fig. 4c). The electrochemical impedance spectroscopy (EIS) spectra show that the charge transfer resistance of Pt–WO_2.9/TiO₂ is 15.31 kOhm, which is lower than the R_ct of Pt–TiO₂ (27.92 kOhm, Table S5†). The lower charge transfer resistance favoured the transport and separation of photogenerated electron–hole pairs (Fig. 4d). The coating of WO_2.9 on TiO₂ not only extends the absorption edge but improves the separation efficiency of photogenerated carriers.

After five cycles of experiments, the post-reaction photocatalyst was analysed. A slight decrease in photocurrent density (Fig. S15a†) and an increase in charge transfer resistance (Fig. S15b†) are in line with a decrease in photocatalytic activity, possibly due to partial deactivation of the Pt active sites.

3.3 CO₂ photoreduction mechanism

Except for the charge transfer kinetics, the surface reaction rate also determines the overall reaction rate. The reaction intermediates during the photocatalytic methane production were thus detected by in situ DRIFTS measurements (Fig. 5 and S16†). After the adsorption equilibrium of CO₂ and water vapor, the DRIFTS for Pt–WO_2.9/TiO₂ (Fig. 5a, bottom spectrum) show the peaks at 1273 cm⁻¹, 1372 cm⁻¹, 1424 cm⁻¹, 1700 cm⁻¹, that could be attributed to the features of the *COOH intermediates, that is, OH deformation, C–O stretch, symmetric stretch and C=O stretch, respectively. The signals appeared at 1067 cm⁻¹, 1464 cm⁻¹, 1150 cm⁻¹ can be assigned to *CHO, *CH₂O and *OCH₃, respectively. Furthermore, the *CO absorption band at ∼2077 cm⁻¹ can be detected, indicating abundant CO intermediates were generated on the surfaces after the adsorption of CO₂ and the *CHO intermediate originated from *CO. The bands at 2987 cm⁻¹ to 2898 cm⁻¹ can be attributed to the –CH₃/–CH₂ stretch and the wide band at 1637 cm⁻¹ can also be observed, which is attributed to physically adsorbed H₂O. The results suggest that the dissociation of CO₂, H₂O and proton reactions occur on the surface after the adsorption of CO₂ and H₂O in the dark.

Fig. 5 In situ DRIFTS spectra recorded during the photocatalytic CO₂ reduction of (a) Pt–WO_2.9/TiO₂ and (b) TiO₂. (c) Proposed pathways for CO₂ methanation.

After illumination of the Xe lamp with AM 1.5 filter, the DRIFTS feature for *CO at ∼2077 cm⁻¹ and the bands at 2987 to 2898 cm⁻¹ for –CH₃/–CH₂ stretching almost vanishes, suggesting that the intermediates were consumed during the reaction. The intensity for the *CHO at 1067 cm⁻¹, *OCH₃ at 1150 cm⁻¹ and *CH₂O at 1464 cm⁻¹ decrease during the 40 min illumination duration. However, the intensity of the signals for the *COOH intermediate features increased over the illumination time, that is OH deformation at 1273 cm⁻¹, symmetric stretch at 1424 cm⁻¹, and C=O stretch at 1700 cm⁻¹. Based on the above analysis, the pathway of CO₂ methanation on Pt–WO_2.9/TiO₂ photocatalyst is a mixed pathway (Fig. 5c).

For the DRIFTS of TiO₂ (Fig. 5b), after the adsorption of CO₂, the peaks of monodentate carbonates at 1518 cm⁻¹ and 1437 cm⁻¹ (m-CO₃²⁻), bidentate carbonates at 1587 cm⁻¹, and 1327 cm⁻¹ (b-CO₃²⁻) can be detected, as well as *CO at 2077 cm⁻¹, physically adsorbed H₂O at 1638 cm⁻¹ and –CH₃/–CH₂ at 2987 to 2898 cm⁻¹. After illumination of AM1.5 for 40 min, the signal of the IR peaks related to b-CO₃²⁻ and *CO intermediates completely vanishes, but the intensity for the bands at 1518 and 1437 cm⁻¹ for m-CO₃²⁻ increases over illumination time. Moreover, the intensity of the characteristic peak at 1700 cm⁻¹ of *COOH (C=O) hardly changes with illumination. The above phenomena indicate that light excitation enhances the adsorption and activation ability of TiO₂ for CO₂ gas (CO₂ to adsorbed carbonate species), but the protonation process of the intermediates is limited, resulting in low catalytic efficiency. In addition, CO₂ methanation on TiO₂ occurs via the carbene pathway. The formation of the *CO intermediate needs to overcome a high energy barrier to form the *C intermediate, so CH₄ production through the mixed pathway is favoured instead of the carbene pathway.³⁴ Hence, the CH₄ yield and selectivity of Pt–WO_2.9/TiO₂ are higher than those of TiO₂, consistent with the yield and reaction rate detected in the CO₂ photoreduction experiments.

The DRIFTS spectra of Pt–TiO₂ and WO_2.9/TiO₂ were acquired as a reference (Fig. S16†). Compared to TiO₂, the peaks of b-CO₃²⁻ disappear and the intensity of peaks related to m-CO₃²⁻ significantly decreased in the DRIFTS spectra of WO_2.9/TiO₂ acquired after adsorption and under illumination, and the intensity of the peak related to *COOH (C=O) significantly increased with the extension of illumination time (Fig. S16a†). The *COOH species is a key intermediate in the photoreduction of CO₂ to CO or CH₄, and its formation and adsorption behavior significantly affect the yield and selectivity of the CO₂ reduction reaction relative to the hydrogen evolution reaction.

For the DRIFTS of Pt–TiO₂ in the dark and under illumination (Fig. S16b†), the intensity of the peaks related to *COOH (C=O) also increased. The peak corresponding to *CH₃O at 1150 cm⁻¹ appears after adsorption of CO₂ and the intensity decreased over time under illumination, demonstrating that the Pt sites regulate the CO₂ methanation pathway from carbene to mixed pathway.

Based on the above discussion and analysis, it is clear that the most significant adsorption peak of the *COOH species appears for Pt–WO_2.9/TiO₂, but no observation of the adsorption peak corresponding to the carbonate species before and after illumination suggests that the synergy between WO_2.9 and Pt sites can promote the conversion of CO₂ into the *COOH intermediate. The absence of carbonate intermediates indicates that the reaction followed a proton-induced pathway for CO₂ reduction (CO₂ → COOH⁺ → *COOH).

In addition to the photoelectric performance and spectroscopic analysis, it can be determined that the formation of WO_2.9/TiO₂ heterojunction and the loading of Pt sites on the WO_2.9 film significantly improve the light absorption and achieve a higher separation efficiency of photogenerated carriers. Furthermore, the strong Lewis acidity of the WO_2.9 film, the electron-rich nature of the Pt sites and the hydrogen spillover channel from Pt clusters (Pt NPs) to Pt single atoms (Pt₁) optimize the adsorption and activation processes of CO₂, significantly promoting the generation of key intermediates for CO₂ methanation, such as *COOH and *OCH₃. Consequently, the highest CH₄ production and selectivity in Pt–WO_2.9/TiO₂ are simultaneously achieved.

Hydrogen spillover occurs between noble metal particles and oxide supports (particularly reducible support), which allows the migration of active hydrogen atoms from metal particles (where a hydrogen atom is generated) to the support.^35,36 In the hydrogen evolution reaction, the hydrogen spillover effect can significantly enhance reaction rates.^37–39 Therefore, the rational design of spillover pathways on the catalyst can effectively improve CO₂ hydrogenation rates by suppressing competitive HER.^16,40

To further evaluate the hydrogen spillover assisted by WO_2.9, density functional theory (DFT) calculations were performed using Pt/TiO₂ and Pt–WO_2.9/TiO₂. Fig. 6a shows that on the Pt–WO_2.9 surface, H* preferentially absorbs at the Pt/WO_2.9 interface with ΔG_H values of −0.17 eV (site 1) and −0.08 eV (site 2), where the negative values indicate spontaneous hydrogen adsorption. The ΔG_H at the W top site (site 3, between Pt NPs and Pt₁) was 0.15 eV, resulting in a thermodynamic energy barrier from site 2 to site 3 of 0.23 eV. On the surface of Pt–TiO₂, the values of ΔG_H on site 1 and site 2 were −0.25 eV and −0.50 eV, respectively, indicating spontaneous hydrogen bonding at the interface between the Pt cluster and TiO₂. The ΔG_H for H* adsorption on the Ti top (site 3) was 0.11 eV, leading to a considerably higher thermodynamic energy barrier (0.61 eV) from site 2 to site 3, compared to the Pt–WO_2.9/TiO₂, which may hinder hydrogen spillover from site 2 to site 3 on the surface of TiO₂.

Fig. 6 Calculated free energy diagram for (a) hydrogen spillover on Pt/WO_2.9 in comparison to Pt/TiO₂ and (b) CO₂ to *CHO or CO on Pt₁/WO_2.9/TiO₂ and on WO_2.9. The insets are stick-ball models, showing the stable adsorption configuration of intermediates on the active sites.

Based on the results of CO₂ photoreduction experiments, the highest H₂ production rate (19.19 μmol g⁻¹ h⁻¹) is observed for Pt NPs on WO_2.9/TiO₂ without Pt₁, suggesting that protons may accumulate on Pt NPs and recombine to form hydrogen gas, which is a competitive reaction process for methane production. With Pt₁ and Pt clusters (Pt–WO_2.9/TiO₂), H₂ production rate decreased significantly (1.42 μmol g⁻¹ h⁻¹), and the methane production rate greatly increased (10.74 μmol g⁻¹ h⁻¹). In addition, for 2Pt–WO_2.9/TiO₂ (with nominal 2.0 wt% Pt precursor), the hydrogen evolution rate was nearly two times higher than that of Pt–WO_2.9/TiO₂ (Fig. S5b†). Therefore, Pt SAs optimize CO₂ adsorption/activation, and Pt clusters enhance H₂O dissociation, which synergistically determines the overall activity and selectivity for CO₂ methanation.

For the sample without WO_2.9 (Pt–TiO₂), the methane production rate is 6.27 μmol g⁻¹ h⁻¹ and the H₂ production rate is 6.85 μmol g⁻¹ h⁻¹, which may be due to the limited spillover of hydrogen. The experimental results and DFT calculations demonstrate that during the photocatalytic reduction reaction, protons at the Pt clusters transfer from the interface between the Pt clusters and the oxide support to the surface of the oxide, then desorb and participate in the reaction with CO₂ molecules on adjacent Pt₁, where the spillover is favoured on the WO_2.9, in comparison to TiO₂. The synergistic effect of Pt clusters and Pt₁ and the promotion of WO_2.9 on hydrogen spillover led to high efficiency for CO₂ methanation and inhibited HER activity.

To further demonstrate the role of the synergistic Pt sites, calculations on the Gibbs free energy pathways of CO₂ reduction from adsorbed CO₂ to *CHO or CO were performed using Pt–WO_2.9/TiO₂ and bare WO_2.9 (Fig. 6b). On the surface of bare WO_2.9, ΔG_H value of CO₂ to *COOH was 0.66 eV, which decreased to 0.14 eV for Pt–WO_2.9/TiO₂. The lower energy barrier on the surface of Pt–WO_2.9/TiO₂ can be ascribed to the electronic structure of Pt₁ favouring the adsorption compared to that of W. Moreover, due to the hydrogen spillover to the O atom adjacent to Pt₁, an OH bond forms near Pt₁, which can significantly weaken the original Pt–O bond (Fig. S17†). Consequently, when a C₁ intermediate adsorbs and forms a chemical bond with Pt₁, the Pt–O bond strengths further decrease and may break, leading to stable adsorption of the C1 intermediate on the Pt₁ site (inset in Fig. 6b).

In addition, on Pt–WO_2.9/TiO₂, the energy barrier from *CO to *CHO was 0.44 eV, which was significantly lower than the energy barrier for *CO desorption (1.55 eV). Therefore, *CO tends to further protonation reaction rather than desorption of CO on Pt–WO_2.9/TiO₂. However, in WO_2.9, the energy barrier from *CO to desorbed CO molecules was 0.40 eV, which was lower than the energy barrier from *CO to *CHO (0.58 eV). Therefore, *CO intermediates tend to desorb from the WO_2.9 surface to free CO molecules, which is in line with the experimental results, in which the CO production rate is higher for WO_2.9/TiO₂ without Pt presence on the surface.

For Pt–WO_2.9/TiO₂, CH₄ selectivity reaches almost 100% with highest production rate under simulated solar light. Hydrogen spillover from Pt clusters to the vicinity of Pt₁ sites not only regulates the coordination environment of Pt₁ sites for CO₂ adsorption and activation but also provides protons for CO₂ methanation.

4. Conclusion

In this work, CO₂ molecules were converted to methane using a Pt–WO_2.9/TiO₂ photocatalyst, achieving a high CH₄ selectivity of 99.8% and an average CH₄ production rate of 10.74 μmol g⁻¹ h⁻¹ under simulated solar light. Dissociated water on the Pt clusters provides active H* for CO₂ methanation through the hydrogen spillover channel, while the Pt single atoms act as the adsorption and activation sites for CO₂. The strong Lewis acidity of WO_2.9 enhances hydrogen spillover, effectively inhibiting the competitive hydrogen evolution activity. Moreover, the hydrogen spillover influences the coordination of Pt₁ for the adsorption of C₁ intermediates and reduces the energy barrier for further protonation. The yield of CH₄ is thus enhanced by promoting the activation of CO₂ molecules and the CH₄ production pathways. Therefore, this work provides insights into the atomic design of cost-effective photocatalysts for CO₂ methanation.

Data availability

The data supporting this article have been included as part of the ESI.†

Conflicts of interest

There are no conflicts to declare.

Acknowledgements

This work was financially supported by the National Natural Science Foundation of China (No. 22108179, X. Z., 22308234, C. Z., 22176135 C. L.) and the Postdoctoral Fellowship Program of CPSF (GZC20231775, C. Z.). The authors would like to thank Dr. Yingming Zhu at the Institute of New Energy and Low-Carbon Technology for UV-vis measurements, and Yanping Huang at the Center of Engineering Experimental Teaching, School of Chemical Engineering, Sichuan University.

References

1. X. Meng, X. Cui, N. P. Rajan, L. Yu, D. Deng and X. Bao, Chem, 2019, 5, 2296–2325.
2. M. Tommasi, S. N. Degerli, G. Ramis and I. Rossetti, Chem. Eng. Res. Des., 2024, 201, 457–482.
3. U. Ulmer, T. Dingle, P. N. Duchesne, R. H. Morris, A. Tavasoli, T. Wood and G. A. Ozin, Nat. Commun., 2019, 10, 3169.
4. T. Zhang, X. Han, N. T. Nguyen, L. Yang and X. Zhou, Chin. J. Catal., 2022, 43, 2500–2529.
5. X. Li, J. Yu, M. Jaroniec and X. Chen, Chem. Rev., 2019, 119, 3962–4179.
6. J. Fu, K. Jiang, X. Qiu, J. Yu and M. Liu, Mater. Today, 2020, 32, 222–243.
7. T. Paik, M. Cargnello, T. R. Gordon, S. Zhang, H. Yun, J. D. Lee, H. Y. Woo, S. J. Oh, C. R. Kagan, P. Fornasiero and C. B. Murray, ACS Energy Lett., 2018, 3, 1904–1910.
8. K. Manthiram and A. P. Alivisatos, J. Am. Chem. Soc., 2012, 134, 3995–3998.
9. H. Khan, M. G. Rigamonti and D. C. Boffito, Appl. Catal., B, 2019, 252, 77–85.
10. C. Lu, J. Li, J. Yan, B. Li, B. Huang and Z. Lou, Appl. Mater. Today., 2020, 20, 100744.
11. F. Riboni, L. G. Bettini, D. W. Bahnemann and E. Selli, Catal. Today, 2013, 209, 28–34.
12. Y. Liu, R. Che, G. Chen, J. Fan, Z. Sun, Z. Wu, M. Wang, B. Li, J. Wei, Y. Wei, G. Wang, G. Guan, A. A. Elzatahry, A. A. Bagabas, A. M. Al-Enizi, Y. Deng, H. Peng and D. Zhao, Sci. Adv., 2015, 1, e1500166.
13. L. Yang, J. Du, J. Deng, N. H. M. Sulaiman, X. Feng, C. Liu and X. Zhou, Small, 2024, 20, 2307007.
14. J. Ran, M. Jaroniec and S.-Z. Qiao, Adv. Mater., 2018, 30, 1704649.
15. C. Dong, C. Lian, S. Hu, Z. Deng, J. Gong, M. Li, H. Liu, M. Xing and J. Zhang, Nat. Commun., 2018, 9, 1252.
16. Y. Wang, J. Zhao, Y. Li and C. Wang, Appl. Catal., B, 2018, 226, 544–553.
17. S. Xie, Y. Wang, Q. Zhang, W. Deng and Y. Wang, ACS Catal., 2014, 4, 3644–3653.
18. H. Wang, L. Zhang, Y. Zhou, S. Qiao, X. Liu and W. Wang, Appl. Catal., B, 2020, 263, 118331.
19. X. Shi, Y. Huang, Y. Bo, D. Duan, Z. Wang, J. Cao, G. Zhu, W. Ho, L. Wang, T. Huang and Y. Xiong, Angew. Chem., Int. Ed., 2022, 61, e202203063.
20. L. Kuai, Z. Chen, S. Liu, E. Kan, N. Yu, Y. Ren, C. Fang, X. Li, Y. Li and B. Geng, Nat. Commun., 2020, 11, 48.
21. S. Yin, Y. Zhou, Z. Liu, H. Wang, X. Zhao, Z. Zhu, Y. Yan and P. Huo, Nat. Commun., 2024, 15, 437.
22. G. Kresse and J. Furthmüller, Comput. Mater. Sci., 1996, 6, 15–50.
23. P. E. Blöchl, Phys. Rev. B: Condens. Matter Mater. Phys., 1994, 50, 17953–17979.
24. J. P. Perdew, J. A. Chevary, S. H. Vosko, K. A. Jackson, M. R. Pederson, D. J. Singh and C. Fiolhais, Phys. Rev. B: Condens. Matter Mater. Phys., 1992, 46, 6671–6687.
25. C. Guo, X. Tian, X. Fu, G. Qin, J. Long, H. Li, H. Jing, Y. Zhou and J. Xiao, ACS Catal., 2022, 12, 6781–6793.
26. C. Gao, T. Wei, Y. Zhang, X. Song, Y. Huan, H. Liu, M. Zhao, J. Yu and X. Chen, Adv. Mater., 2019, 31, 1806596.
27. H. Wei, K. Huang, D. Wang, R. Zhang, B. Ge, J. Ma, B. Wen, S. Zhang, Q. Li, M. Lei, C. Zhang, J. Irawan, L.-M. Liu and H. Wu, Nat. Commun., 2017, 8, 1490.
28. A. Kubiak, Sci. Rep., 2024, 14, 13827.
29. W. Zhang, H. Wang, J. Jiang, Z. Sui, Y. Zhu, D. Chen and X. Zhou, ACS Catal., 2020, 10, 12932–12942.
30. K. Ding, A. Gulec, A. M. Johnson, N. M. Schweitzer, G. D. Stucky, L. D. Marks and P. C. Stair, Science, 2015, 350, 189–192.
31. M. Kan, C. Yang, Q. Wang, Q. Zhang, Y. Yan, K. Liu, A. Guan and G. Zheng, Adv. Energy Mater., 2022, 12, 2201134.
32. L. Li, X. Dai, C. Cheng, F. Chen, S. M. Wabaidur, W. Wang and Y. Hu, Chem. Eng. J., 2025, 507, 160101.
33. L. Li, H. Liu, C. Cheng, X. Dai, F. Chen, J. Ning, W. Wang and Y. Hu, ACS Catal., 2024, 14, 10204–10213.
34. Ş. Neaţu, J. A. Maciá-Agulló, P. Concepción and H. Garcia, J. Am. Chem. Soc., 2014, 136, 15969–15976.
35. W. C. Conner, Jr. and J. L. Falconer, Chem. Rev., 1995, 95, 759–788.
36. M. D. Marcinkowski, A. D. Jewell, M. Stamatakis, M. B. Boucher, E. A. Lewis, C. J. Murphy, G. Kyriakou and E. C. H. Sykes, Nat. Mater., 2013, 12, 523–528.
37. Z. Gu, M. Li, C. Chen, X. Zhang, C. Luo, Y. Yin, R. Su, S. Zhang, Y. Shen, Y. Fu, W. Zhang and F. Huo, Nat. Commun., 2023, 14, 5836.
38. J. Li, J. Hu, M. Zhang, W. Gou, S. Zhang, Z. Chen, Y. Qu and Y. Ma, Nat. Commun., 2021, 12, 3502.
39. J. Li, H.-X. Liu, W. Gou, M. Zhang, Z. Xia, S. Zhang, C.-R. Chang, Y. Ma and Y. Qu, Energy Environ. Sci., 2019, 12, 2298–2304.
40. P. Liu, Z. Huang, X. Gao, X. Hong, J. Zhu, G. Wang, Y. Wu, J. Zeng and X. Zheng, Adv. Mater., 2022, 34, 2200057.

---

Footnote

† Electronic supplementary information (ESI) available. See DOI: https://doi.org/10.1039/d5cy00167f

---