Tailoring atomically dispersed Fe-induced oxygen vacancies for highly efficient gas-phase photocatalytic CO₂ reduction and NO removal with diminished noxious byproducts

Abstract

Single-atom-supported metal oxides have attracted extensive interest in energy catalysis, offering a promising avenue for mitigating greenhouse gas emissions and environmental pollution. This study presents a facile synthesis of single-atom Fe-modified Bi₂WO₆ photocatalysts. By carefully tuning the Fe ratios, the 1.5Fe-Bi₂WO₆ sample demonstrates exceptional photocatalytic efficiency in CO₂ to CO reduction (36.78 μmol g⁻¹). Additionally, an outstanding NO removal performance is also achieved through this photocatalyst with an impressively low conversion of toxic NO₂ at just 0.37%. The reaction intermediates and mechanisms governing the photocatalytic reduction of CO₂ into CO are elucidated using in situ DRIFTS and in situ XAS techniques. Regarding NO removal, the introduction of Fe single-atoms, along with induced oxygen vacancies, plays a pivotal role in facilitating the transformation of NO and NO₂ into nitrate by stabilizing NO and NO₂ species. Mechanistic insights into photocatalytic NO oxidation are garnered through scavenger trapping and EPR experiments employing DMPO. This study emphasizes single-atom-supported metal oxide's potential in sustainable chemistry and air purification, providing a promising solution for urgent environmental challenges.

---

1. Introduction

Carbon dioxide (CO₂) and nitrogen oxides (NO_x) are harmful atmospheric pollutants. CO₂ intensifies global warming and climate change, while NO_x contributes to air pollution and health problems.^1–3 Photocatalysis has emerged as a promising solution, harnessing sunlight to transform CO₂ and diminish NO_x concentrations into valuable compounds.^4–6 Extensive research efforts have been committed to developing and constructing diverse semiconductor systems for such purposes. Yet, when it comes to the photocatalytic oxidation of NO, it inevitably leads to the production of an intermediate compound, NO₂, which is even more hazardous than the initial reactant. As a consequence, if catalytic reactions lead to the predominance of NO₂, it could potentially compromise the overall effectiveness of nitric oxide (NO) removal.⁷ To refrain from such a situation, tailoring active sites is essential in designing the photocatalytic system.

Zhang et al. first proposed the terminology of the heterogeneous single-atom (SA) catalyst (Pt₁/FeO_x) for CO oxidation.⁸ This innovative approach involves catalysts composed of isolated or incorporated metal atoms with unique electronic structures and unsaturated coordination.^9–12 However, the high surface energy of single atoms will lead to undesired aggregation; hence, the interaction between single atoms and support materials also plays a crucial role in stabilizing single atoms. Among the several types of supported SAs, metal oxide-supported single-atom catalysts have received much attention because of the specific and variable properties observed in metal oxides. The stability of single atoms within the lattice matrix is a key factor, as they tend to maintain thermodynamic stability due to their robust interactions with the surrounding lattice atoms.^13,14 Determining the exact anchor location is quite a challenge, especially when metal oxide supports possess fewer surface defects, and thus their interaction with metal species is weak. In this study, we selected bismuth tungstate (Bi₂WO₆), a member of the Aurivillius oxides family, and an n-type semiconductor as the support material. Bi₂WO₆ with an orthorhombic structure is constructed by a sandwich substructure of alternating (Bi₂O₂)_n²ⁿ⁺ layers and (WO₄)_n²ⁿ⁻ that are held together by chemical bonds.^15,16 Considering that single-atom alloying strategies involve the incorporation of host elements such as Fe, Co, Ni, Cu, Ag, and Au with a small number of individual guest atoms like Ru, Pd, Ir, Pt, and others,¹⁷ we have opted to designate transition metal Fe single atoms as the active sites. This choice is attributed to their promising characteristics and cost-effectiveness when compared to noble metals.¹⁸ Given the similarity in ionic radius of Fe³⁺ to that of W⁶⁺ and is significantly smaller than that of Bi³⁺,¹⁹ it is highly probable that Fe³⁺ ions can seamlessly substitute for W⁶⁺ ions within the Bi₂WO₆ lattice. In addition, creating controlled oxygen vacancies (OVs) on metal oxide surfaces represents an effective strategy to enhance the photocatalyst's performance by modifying their band structure, increasing active sites, and lowering activation energy barriers, thus improving reactant adsorption and facilitating challenging chemical reactions.²⁰ It has been verified that the introduction of metal heteroatoms could induce the generation of OVs and sustain their presence during the photocatalytic process.²¹ This synergistic effect between metal single atoms and OVs stabilizes adsorbed species on the catalyst surface, ultimately resulting in enhanced catalytic performance.

To the best of our knowledge, numerous studies have explored the utilization of Bi₂WO₆ material in various photocatalytic applications.^21–23 However, there is a notable lack of research dedicated to investigating gas-phase photocatalytic CO₂ reduction and NO removal over Bi₂WO₆-based materials using the SAs approach. In this context, we have designed and employed Fe-Bi₂WO₆ for both applications. The coordination environment and distribution of atomically dispersed Fe were verified by X-ray absorption spectroscopy (XAS) and double spherical aberration-corrected scanning transmission electron microscopy (STEM), respectively. The influence of Fe introduction on the oxygen vacancies' formation was verified by electron paramagnetic resonance (EPR), X-ray photoelectron spectroscopy (XPS), and density functional theory (DFT) calculation results. Eventually, the mechanism of gas-phase photocatalytic CO₂ reduction as well as NO oxidation was investigated. In situ XAS and in situ diffuse reflectance infrared Fourier transform spectroscopy (DRIFTS) were employed to study the CO₂ reduction mechanism, while scavenger trapping tests and 5,5-dimethyl-1-pyrroline N-oxide (DMPO) spin trapping EPR were conducted for NO oxidation. This allocation of research efforts provides a valuable platform for gaining in-depth insights into the structure–property relationship between single atoms and their associated photocatalytic activity.

2. Experimental section

2.1. Chemicals and materials

Bismuth nitrate pentahydrate (Bi(NO₃)₃·5H₂O), sodium tungstate dihydrate (Na₂WO₄·2H₂O), ethylene glycol (EG), iron(III) nitrate nonahydrate (Fe(NO₃)₃·9H₂O), potassium dichromate (K₂Cr₂O₇), potassium iodide (KI), terephthalic acid, benzoquinone, and DI water. All chemicals were of analytical grade and used without further purification.

2.2. Synthesis procedures of Fe-Bi₂WO₆

Initially, 1.9403 g of Bi(NO₃)₃·5H₂O and a desired amount of Fe(NO₃)₃·9H₂O were mixed in 50 mL of EG and sonicated for 30 min. Subsequently, 0.6597 g of Na₂WO₄·2H₂O was added to the above solution and stirred further for 30 min. The resulting mixture is then transferred to a 100 mL Teflon-lined stainless-steel autoclave and heated at 190 °C for 2 h. Finally, the precipitate was collected, washed with DI water several times, and dried at 60 °C for 12 h. The obtained samples with the expected molar ratio of Fe³⁺ to W⁶⁺ were set to be 1%, 1.5%, and 2%, denoted as 1Fe/Bi₂WO₆, 1.5Fe/Bi₂WO_6, and 2Fe/Bi₂WO₆, respectively.

For comparison, the pristine Bi₂WO₆ was prepared without the addition of Fe(NO₃)₃·9H₂O by a similar process.

2.3. Characterizations of materials

The crystal phase of the as-prepared catalysts was examined by X-ray diffraction (XRD) using Cu Kα radiation with a wavelength of 1.5406 Å (Bruker, D2 PHASER with XFalsh). The surface morphologies and microstructures of catalysts were observed by a field-emission scanning electron microscope (FESEM, JEOL, 6700F) and transmission electron microscope (TEM) (JEOL TEM 2100F) equipped with selected area electron diffraction (SAED). High-angle annular dark-field and annular bright-field scanning transmission electron microscopy (HAADF/ABF-STEM) were conducted by JEOL JEM-ARM300F2 equipped with double spherical aberration correctors and energy dispersive spectroscopy (EDS, JED EX-34400MNU). The content of Fe in Fe-Bi₂WO₆ samples was determined by Agilent 710 inductively coupled plasma optical emission spectroscopy (ICP-OES). The UV-visible diffuse reflectance spectra (UV-vis DRS) were obtained by using a Jasco V-670 spectrometer with an integrated method. Nitrogen adsorption isotherms and CO₂ adsorption ability were measured using BELSORB MAX II. XPS was conducted by the Nexsa G2 X-ray Photoelectron Spectrometer System from Thermo Fisher Scientific. UPS measurement was performed using the ULVAC PHI 5000 Versa Probe, and a helium discharge lamp (Specs UVS 10/35) with an energy of 21.2 eV was utilized as a UPS excitation source. X-ray absorption near-edge structure (XANES) measurements of Fe K-edge of as-prepared samples and the samples after stability tests were carried out at the Taiwan Photon Source (TPS)-Beamline of 44A in the National Synchrotron Radiation Research Center (NSRRC, Hsinchu, Taiwan) and collected in fluorescence mode. The in situ XAS of the Fe K-edge was conducted under the conditions of CO₂/H₂O exposure and light irradiation at the Taiwan Light Source (TLS)-Beamline of 17C1 in the NSRRC. The sample chamber provided an airtight environment during the measurements. The Fe K-edge spectra of 1.5Fe-Bi₂WO₆ before and after CO₂/H₂O adsorption and light irradiation were collected by fluorescence mode, and the AM 1.5 solar simulator was used as a light source. The XANES and Fourier transform extended X-ray fine structure (FT-EXAFS) raw data were processed and fitted by Athena and Artemis software. Photoluminescence (PL) and time-resolved photoluminescence (TRPL) were recorded by the Horiba Jobin-Yvon instrument and time-correlated single-photon counting with an excitation wavelength of 405 nm and a frequency of 40 MHz. Electron paramagnetic resonance (EPR) spectroscopy was recorded by a Bruker EMX-Plus EPR spectrometer. The CO temperature-programmed desorption measurements were conducted using AutoChem II 2920 Micrometrics. In situ DRIFTS measurements were performed by a Thermo Nicolet 6700 FTIR spectrometer equipped with a HgCdTe detector and a three-window DRIFTS cell. The isotope labeling experiment was conducted by using ¹³CO₂ to verify the carbon source of the CO product during the photocatalytic CO₂ reduction reaction, and the formed ¹³CO product was monitored by gas chromatography/mass spectrometry (GC-MS, Thermo-Trace 1300 GC + ISQ MS).

2.4. The photocatalytic activity evaluation of catalysts

2.4.1 Photocatalytic CO₂ reduction. The gas-phase photocatalytic CO₂ reduction was performed in a closed quartz reactor without the presence of any sacrificial reagents at room temperature (25 ± 5 °C). The AM 1.5 solar simulator was used as a light source and placed perpendicularly above the quartz window of the reactor. According to our previously published research,²⁴ the photocatalytic CO₂ reduction experiment was conducted as follows: Initially, 20 milligrams of the catalyst were evenly distributed onto a glass holder, which was then placed inside the reaction chamber and sealed with a quartz window. Subsequently, a high-flow rate of 50 sccm of humidified CO₂ was introduced into the chamber and allowed to purge for 15 minutes. Following this, a low-flow rate of 5 sccm of humidified CO₂ was introduced into the reactor for an additional 15 minutes before switching on the light source. After 4 hours of continuous light exposure, the products generated during the photocatalytic reaction were analyzed using SRI 8610C gas chromatography (GG) with a glass column Porapak Q equipped with a flame ionization detector (FID) and a helium ionization detector (HID). For the photocatalytic CO₂ reduction measurement, the humidity was maintained by flowing CO₂ gas through a water bubbler. The stability of photocatalytic CO₂ reduction was tested by performing the same protocol on the same sample every 4 hours for a total of 5 cycles.

2.4.2 The photocatalytic NO removal experiment. The photocatalytic NO removal activity of catalysts was assessed by detecting the NO at the ppb level in a continuous flow reactor at room temperature (25 ± 5 °C), as depicted in Fig. S1.† In detail, 0.2 g of catalyst was ultrasonically dispersed in a certain amount of DI water. The suspension was coated on a Petri dish of 12 cm diameter, dried at 60 °C in an oven, and subsequently placed in a chamber with dimensions of 30 cm × 15 cm × 10 cm (length × width × height). A 300 W Osram light source (230 V) equipped with a 420 nm cut-off filter was positioned vertically, 20 cm above the quartz window. An airstream consisting of 500 ppb NO (humidity level: 70%) flowed through the chamber at a controlled rate of 3 L min⁻¹. The adsorption process was conducted in the dark for 1 h to ensure adsorption/desorption equilibrium before turning on the light source. The concentration of NO was continuously recorded by a chemiluminescence NO_x analyzer (Thermal Environmental Instruments Inc., Franklin, MA, model 42c).

The NO removal efficiency (η), NO₂ conversion efficiency (ε), NO degradation efficiency into NO₃⁻ green product (ρ), and NO₃⁻ selectivity (δ) were calculated using eqn (1)–(4), respectively:

ρ = η − ε (3)

where C^NO_ini and C_ini^NO₂ represent the concentrations of NO and NO₂ at adsorption–desorption equilibrium in the dark, respectively. C^NO_fin and C_fin^NO₂ represent the final concentrations of NO and NO₂ after turning on the light, respectively.

For the durability test of the catalyst, the catalyst was washed with DI water after every run, dried at 60 °C in a vacuum oven, and used for the next run under the same protocol as the above photocatalytic NO removal measurement.

The trapping measurements were also performed to determine the photocatalytic NO removal mechanism. KI, K₂Cr₂O₇, terephthalic acid, and benzoquinone were utilized as scavengers for trapping holes (h⁺), electrons (e⁻), hydroxyl radicals (HO˙), and superoxide radicals (O₂˙⁻), in which 1% of the scavenger was mixed with the catalyst before proceeding with the photocatalytic activity test measurement.

2.5. Computational details

In this study, all DFT calculations were carried out based on the Vienna ab initio simulation package (VASP).^25,26 The exchange and correlation potentials were performed through the generalized gradient approximation (GGA)^27,28 within the Perdew–Burke–Ernzerhof (PBE)^29,30 exchange–correlation functional. The energy cutoff was set to 500 eV for all the calculations. The parameters of the self-consistent field (SCF) force and energy conversion were set to 10⁻⁶ eV and 0.02 eV A⁻¹, respectively. The k-point meshes were set to 2 × 5 × 2 for geometry optimization and 4 × 9 × 4 for electronic investigated calculations. An orthorhombic unit cell of Bi₂WO₆ is built along with the P21 ab space group (No. 29).^31,32 The computed lattice parameters of Bi₂WO₆ as a = 5.437 Å, b = 5.458 Å, and c = 10.872 Å are well agreed with the previous theoretical works^33,34 and other experimental studies.^31,32 The computed average Bi–O and W–O bong lengths are 2.295 Å and 1.864 Å, respectively. These values are well consistent with the experimental^31,32 and the theoretical references,^33–35 which demonstrate our reasonable methodology in this study. A 72-atom 2 × 1 × 1 supercell of bulk Bi₂WO₆ was first relaxed, then the defects were involved in systems with Fe introduction. The DFT + U method³⁶ is applied along with the Hubbard U = 4 for Fe to capture the effects of electronic correlation and Coulomb-driven localization on the structural properties.

The NO oxidation mechanism is determined through the Gibbs free energy (ΔG) valuation computed as follows³⁷

ΔG = ΔE + ΔZPE − TΔS (5)

Where ΔZPE and ΔS are the zero-point energy taken out from the vibrational frequency calculations and the entropy change obtained from the standard thermodynamic tables,³⁸ respectively; T is the temperature at 298.15 K; and ΔE is the computed DFT-reaction energy difference between the reactant and product molecules adsorbed on the surface. Overall, the photocatalytic NO oxidation could be generally described by using the following equations:

NO + * → NO* (7)

where * represents the catalyst's surface, and O*, , and denote the adsorbed species on the catalyst's surface, while O₂, NO₂, and NO₃ signify the gaseous molecules.

3. Results and discussion

3.1. Exploration of material properties

The synthesis of Fe-Bi₂WO₆ photocatalysts with atomic dispersion of Fe is achieved following the procedure depicted in Fig. 1a. The actual amounts of Fe in 1, 1.5, and 2Fe-Bi₂WO₆ samples were determined to be 0.06, 0.11, and 0.18 wt% by ICP-OES (Table S1†). The crystal structure of the as-prepared sample was examined using X-ray diffraction. Fig. 1b indicates the orthorhombic structure of Bi₂WO₆ with diffraction peaks at 28.30, 32.91, 47.14, 55.82, 58.54, 68.78, 76.08, and 78.54° corresponding to (131), (200), (202), (133), (262), (400), (102), and (204) crystal planes (PDF 00-039-0256), respectively. The characteristic peaks of iron and iron oxides are not observable in the XRD patterns of Fe-Bi₂WO₆ samples, which implies the influence of iron introduction into the crystal structure of Bi₂WO₆ is negligible. The similarity in ionic radii between Fe (63 pm) and W (60 pm), along with the significantly smaller radius of W compared to that of Bi (108 pm), suggests the tendency of Fe ions to replace the W positions within the Bi₂WO₆ lattice. This observation aligns well with the results obtained from computational calculations (Fig. S2†), and the computed average bond distances of Bi, W, and Fe with O are summarized in Table S2†. Surface morphology and microstructure analyses were conducted using SEM and TEM techniques. As shown in Fig. S3a,† pristine Bi₂WO₆ has a hierarchical structure with a flower-like shape composed of many thin sheets. For xFe-Bi₂WO₆ (x = 1, 1.5, 2), no obvious differences could be observed, indicating that the original morphology of Bi₂WO₆ is unaffected by the Fe modification (Fig. S3b–d†). The mesoporosity of these materials has been confirmed through nitrogen adsorption–desorption measurements (Fig. S4 and Table S3†). All four samples exhibit type IV isotherms with the H3-type hysteresis loop, which is characteristic of mesoporous structures (Fig. S4a†). The BET surface area of 1.5Fe-Bi₂WO₆ shows a slight increase in comparison with Bi₂WO₆, which indicates that the introduction of Fe enlarges the surface area of samples (Table S3†). The Barret–Joyner–Halenda (BJH) model was used for pore size analysis (Fig. S4b†), which indicates a broad distribution of mesopores ranging from approximately 2 to 50 nm.

Fig. 1 (a) Synthesis process of Fe-Bi₂WO₆, (b) XRD patterns of Bi₂WO₆ and xFe-Bi₂WO₆ (x = 1, 1.5, 2); HRTEM images of (c) Bi₂WO₆, and (d) 1.5Fe-Bi₂WO₆ and inset is respective SAED patterns; (e) HAADF-STEM, (f) ABF-STEM images of 1.5Fe-Bi₂WO₆; (g) intensity profiles of corresponding D1, D2, D3 points in HAADF-STEM image and B1, B2, B3 points in ABF-STEM image, respectively; (h) STEM images and corresponding EDS mappings of Bi, W, O, and Fe elements for 1.5Fe-Bi₂WO₆ sample.

High-resolution TEM (HRTEM) images of Bi₂WO₆ and 1.5Fe-Bi₂WO₆ (Fig. 1c and d) reveal lattice spacings of 0.314 nm and 0.271 nm corresponding to the (131) and (200) crystallographic planes, respectively. Additionally, the SAED patterns, shown in the inset figures within the HRTEM images (Fig. 1c and d), confirm a polycrystalline structure for both Bi₂WO₆ and 1.5Fe-Bi₂WO₆. The SAED patterns exhibit indexed reflections corresponding to the (131), (200), (202), and (133) planes of the orthorhombic structure, which are consistent with the XRD results. To visualize the distribution of atomically dispersed Fe, aberration-corrected HAADF/ABF-STEM was carried out and exhibited in Fig. 1e and f. Based on the Z-contrast, the lighter elements in the Z number will be displayed darker and brighter in the Z-contrast images of HAADF- and ABF-STEM, respectively. As seen in Fig. 1e and f, the dark spots in the HAADF-STEM image and the brighter points in the ABF-STEM image of 1.5Fe-Bi₂WO₆ correspond to isolated Fe atoms. This observation is further confirmed by comparing the relative intensity of atoms at points D1, D2, and D3 in the dark-field image (Fig. 1e) and their corresponding points B1, B2, and B3 in the bright-field image (Fig. 1f), as depicted in Fig. 1g. The STEM-EDS elemental mappings (Fig. 1h) provide additional confirmation of the presence of Bi, W, O, and Fe elements. Furthermore, it clearly illustrates the uniform distribution of Fe throughout the 1.5Fe-Bi₂WO₆ sample, thereby confirming the successful incorporation of Fe into the crystal structure of Bi₂WO₆.

The electronic structures and local coordination environment of Fe in Fe-Bi₂WO₆ samples are corroborated by XANES and EXAFS analysis (Fig. 2a and b). The Fe K-edge rising-edge profiles of Fe-Bi₂WO₆ samples closely resemble the one of Fe₂O₃, suggesting the presence of atomically dispersed Fe in a positive valence state, specifically near Fe³⁺ (Fig. 2a), which aligns well with the findings from EPR results (Fig. 2d). The FT-EXAFS curves (Fig. 2b) of Fe-Bi₂WO₆ samples have a dominant peak at around 1.45 Å correlated with the Fe–O bond, whereas the peaks related to the Fe–Fe bond in both Fe foil and Fe₂O₃ are not detectable. Moreover, wavelet transform (WT) EXAFS analysis was conducted to investigate the Fe species. Fig. 2c clearly displays a single intensity peak at approximately 4.9 Å⁻¹ for the Fe-Bi₂WO₆ samples, corresponding to the presence of the Fe–O from the first nearest-neighbor coordination shell. Notably, the characteristic intensity at 7.7 Å⁻¹ associated with Fe–Fe bonding, as exemplified in the Fe foil, remains unobservable in these Fe-Bi₂WO₆ samples. This result provides evidence that Fe single atoms are firmly anchored to Bi₂WO₆ through Fe–O bonding. Besides, the FT-EXAFS fittings (Fig. S5 and S6†) were conducted, and fitting parameters were listed in Table S4,† showing that the coordination numbers of Fe are 4.06, 4.11, and 4.18 for 1, 1.5, and 2Fe-Bi₂WO₆, respectively. The k-space spectra of Fe-Bi₂WO₆ samples exhibit substantially different oscillations in comparison with those of Fe foil and Fe₂O₃ references, magnifying the unconventional coordination of atomically dispersed Fe in Fe-Bi₂WO₆. The lower coordination numbers of Fe in Fe-Bi₂WO₆ samples in comparison with the coordination number of 6 for Fe₂O₃ imply the existence of Fe on the surface instead of a bulk lattice and the formation of oxygen vacancies. The presence of oxygen vacancies and Fe after the modification with Fe is further confirmed by EPR (Fig. 2e). A strong peak at around g = 4.25 in 1.5Fe-Bi₂WO₆ specifies the presence of Fe³⁺ species in the sample.³⁹ The enhanced EPR signal at g = 2.007 in 1.5Fe-Bi₂WO₆, compared to pristine Bi₂WO₆, confirms a greater presence of oxygen vacancies resulting from the introduction of Fe as a charge compensation mechanism.⁴⁰ This observation confirms that the introduction of single atoms induces oxygen vacancy formation. Notably, in pristine Bi₂WO₆, a signal at g = 2.007 is detected, attributed to oxygen vacancies formed during the synthesis process.

Fig. 2 (a) Normalized Fe K-edge XANES spectra, (b) FT-EXAFS spectra, (c) WT-EXAFS spectra of xFe-Bi₂WO₆ (x = 1, 1.5, 2), Fe₂O₃ and Fe foil references; (d) EPR spectra, and (e) enlarged spectra providing information about oxygen vacancies of Bi₂WO₆ and 1.5Fe-Bi₂WO₆.

The chemical compositions and oxidation states of elements in samples were further studied by X-ray photoelectron spectroscopy (Fig. S7 and S8†). The high-resolution Bi 4f spectrum (Fig. S8a†) of Bi₂WO₆ shows two peaks at 164.42 and 159.09 eV, assigned to Bi 4f_5/2 and Bi 4f_7/2, respectively.⁴¹ The Bi 4f peaks of Fe-Bi₂WO₆ samples exhibit redshifts because the introduction of Fe changes the electronic density around the Bi element. For W 4f (Fig. S8b†), the two peaks at 37.32 and 35.24 eV correspond to W 4f_5/2 and W 4f_7/2.⁴² The O 1s spectrum of Bi₂WO₆ (Fig. S8c†) could be deconvoluted into three peaks at 530.09, 531.30, and 532.45 eV assigned to lattice oxygen, oxygen vacancies, and adsorbed oxygen, respectively.⁴³ The area of the oxygen vacancy peak increases with higher Fe concentration, confirming the oxygen vacancy formation upon Fe modification. Additionally, redshifts in the binding energy of the W 4f and O 1 s spectra are observed in Fe-Bi₂WO₆. Moreover, two distinct peaks are observed in the XPS of Fe 2p (Fig. S8d†), corresponding to Fe 2p_1/2 and Fe 2p_3/2, further verifying the presence of Fe species in Fe-Bi₂WO₆ samples.⁴⁴

The photoabsorption properties and band gaps of all samples were determined using UV-vis diffuse reflectance spectroscopy (Fig. 3a). It's observed that there are redshifts in the absorption edges of xFe-Bi₂WO₆ (x = 1, 1.5, 2). These shifts occur because the modification of Bi₂WO₆ with Fe creates defect levels near the conduction band, resulting in a broadening of the band tails and an extension of the light absorption range. The optical band gap values of the catalysts were calculated using the Kubelka–Munk equation. Bi₂WO₆ is known to have an indirect bandgap, and the determined band gap (E_g) values for Bi₂WO₆, 1Fe-Bi₂WO₆, 1.5Fe-Bi₂WO₆, and 2Fe-Bi₂WO₆ are 2.88 eV, 2.77 eV, 2.74 eV, and 2.69 eV, respectively (Fig. S9†).

Fig. 3 (a) UV-vis DRS, (b) band edge position schematic, (c) PL spectra, (d) TRPL spectra of Bi₂WO₆, and xFe-Bi₂WO₆ (x = 1, 1.5, 2).

To further understand the change in the band structure of catalysts, the UPS measurement was used, as shown in Fig. S10 and Table S5†. The valence band maximum values of Bi₂WO₆ and Fe-Bi₂WO₆ samples are obtained to be 1.58, 1.71, 1.73, and 1.80 vs. NHE, respectively. Combining with the acquired optical band gap values from UV-vis DRS (Fig. S9†), the conduction band minimum values of Bi₂WO₆ and Fe-Bi₂WO₆ samples are calculated to be −1.30, −1.06, −1.01, and −0.89 eV, respectively. The band structures of Bi₂WO₆ and Fe-Bi₂WO₆ samples are illustrated in Fig. 3b, confirming the suitable band positions of as-prepared catalysts for photocatalytic CO₂ reduction reactions. Effective charge separation is crucial for photocatalytic activity. To evaluate this characteristic of the prepared materials, PL and TRPL measurements were conducted. Fig. 3c exhibits the lowest PL emission intensity of 1.5Fe-Bi₂WO₆ than the ones of pristine Bi₂WO₆, 1Fe-Bi₂WO₆, and 2Fe-Bi₂WO₆, indicating better charge separation after modifying with an appropriate concentration of Fe. Furthermore, the TRPL spectra and TRPL fitting results (Fig. 3d and Table S6†) specify the shorter lifetime of 1.5Fe-Bi₂WO₆ (1.07 ns) in comparison to Bi₂WO₆ (1.46 ns), attributed to the suppressed radiative recombination pathway and more efficient charge dissociation and migration in 1.5Fe-Bi₂WO₆.⁴⁵ In addition, shorter TRPL lifetimes often indicate the presence of defects or surface states, which can enhance photocatalytic activity by promoting reactant adsorption and interaction with charge carriers. The enhanced charge separation and transfer increase the probability of interaction with oxygen molecules, thereby facilitating the generation of essential reactive oxygen species for the photocatalytic NO removal process.

3.2. Evaluation of photocatalytic reduction and oxidation activity along with mechanistic insights

3.2.1 Photocatalytic CO₂ reduction. The gas-phase photocatalytic CO₂ reduction was evaluated through a 4 hours batch-type measurement. As depicted in Fig. 4a, the modification with atomically dispersed Fe significantly enhances the photocatalytic reduction of CO₂ to CO. Among the xFe-Bi₂WO₆ samples, 1.5Fe-Bi₂WO₆ demonstrates the highest CO production rate at 36.78 μmol g⁻¹, surpassing the pristine Bi₂WO₆ by 2.7 times. In addition, the CO production rate decreases for 2Fe-Bi₂WO₆, which is due to inefficient charge separation and transfer, as proven in the PL and TRPL results (Fig. 3c, d and Table S6†). To explain the higher CO₂ reduction activity of 1.5Fe-Bi₂WO₆, the CO₂ adsorption ability is one of the important steps. Fig. S11† shows the CO₂ adsorption measurement at an adsorption temperature of 298 K. The presence of Fe significantly boosts the adsorption capacity, and notably, 1.5Fe-Bi₂WO₆ exhibits the highest CO₂ uptake capacity among the samples. This improvement suggests that when the Fe single atom is incorporated into the Bi₂WO₆ surface, it acts as the active site and facilitates electron transfer between the Fe single atom and CO₂ molecule on the catalyst surface, ultimately promoting the activation of CO₂. Meanwhile, the induced surface oxygen vacancies alter the catalyst's electronic structure and expand the light absorption range, as evidenced in the UV-vis spectra (Fig. 3a). Additionally, as reported in previous studies, these oxygen vacancies reduce the energy barrier for dissociating adsorbed COOH intermediates into CO and OH on the catalyst's surface.¹⁶ Efficient product desorption is also a key factor influencing the efficiency and selectivity of the photocatalytic CO₂ reduction process. The CO-TPD experiments, illustrated in Fig. 4b, reveal that CO product desorption over 1.5Fe-Bi₂WO₆ is more effective compared to other catalysts. This improved desorption capability is one of the factors contributing to the superior photocatalytic CO₂ reduction performance of 1.5Fe-Bi₂WO₆. Overall, the exceptional CO₂ adsorption capacity, high surface area, efficient light absorption, effective charge separation, and efficient CO desorption all collectively enhance the activity of 1.5Fe-Bi₂WO₆. Given the significance of CO as a syngas, achieving high selectivity for target products is paramount to reducing post-treatment costs. To gain deeper insights into the catalytically active sites in the photocatalytic CO₂ reduction mechanism, we conducted in situ XAS to examine the valence state and charge transfer between active sites and reactants under the working reaction conditions. Fig. 4c presents the in situ XANES Fe K-edge spectra of the 1.5Fe-Bi₂WO₆ photocatalyst during the reaction test. The non-obvious change in the absorption edge indicates the stable valence state of Fe during the photocatalytic CO₂ reduction process. However, there is a decrease in white line intensity after CO₂/H₂O adsorption, indicating interaction and charge transfer from CO₂/H₂O to Fe on the catalyst surface, leading to the formation of intermediates. Notably, the white light intensity of the Fe K-edge XANES spectra increases after light irradiation, implying an increase in unoccupied d states of Fe or the transfer of photogenerated electrons from Fe to the reactants.^46–48 These in situ XAS results further provide crucial insights into the role of atomically dispersed Fe as the active site in photocatalytic CO₂ reduction. The durability of the catalyst for photocatalytic CO₂ reduction was assessed and illustrated in Fig. 4d, demonstrating the robustness of 1.5Fe-Bi₂WO₆ without structural changes even after undergoing five cycles of CO₂ photoreduction, as confirmed by XAS measurements (Fig. S12 and Table S7†). We also conducted a comparison of gas-phase CO₂ reduction efficiency among Bi₂WO₆-based catalysts (as summarized in Table S8†), revealing that Fe single atoms-modified Bi₂WO₆ stands out as a promising and efficient photocatalyst for this reaction. To validate the obtained hydrocarbon products and eliminate the possibility of carbon contamination, a control photocatalytic experiment was carried out under inert gas and visible-light irradiation. No CO was detected in the blank experiment. Furthermore, to confirm the origin of CO, an isotope-tracer experiment was performed using ¹³CO₂ under the same photocatalytic reaction conditions (Fig. S13†).

Fig. 4 (a) Photocatalytic CO₂ reduction activity, (b) CO-TPD curves of Bi₂WO₆ and xFe-Bi₂WO₆ (x = 1, 1.5, 2) samples, respectively; (c) Fe K-edge XANES spectra of 1.5Fe-Bi₂WO₆ before/after CO₂/H₂O adsorption and under light irradiation; (d) photocatalytic CO₂ reduction stability test over 1.5Fe-Bi₂WO₆.

The adsorption behaviors and formation of intermediates during photocatalytic CO₂ reduction were further investigated by in situ DRIFTS under dark and light conditions. The background spectra were collected before purging CO₂/H₂O and light irradiation. Fig. 5a and b illustrates the adsorption of CO₂ and H₂O vapor on the surface of 1.5Fe-Bi₂WO₆. The peaks around 1654 cm⁻¹ and 1632 cm⁻¹ are apparently observable after purging CO₂ and H₂O vapor without light illumination, assigned to the surface chemisorbed CO₂ and bending mode of adsorbed water (δ-H₂O), respectively.^1,49 The others at about 1384 cm⁻¹ and 1185 cm⁻¹ are attributed to monodentate carbonate (m-CO₃²⁻) and bicarbonate (HCO₃⁻) species.⁵⁰ After light irradiation (Fig. 5c and d), there is a clear reduction in the adsorbed species, suggesting the consumption of absorbed intermediates during the photoreduction reaction. The evolution of a new peak around 1290 cm⁻¹ related to monodentate carbonate (m-CO₃²⁻) species.⁵¹ Furthermore, the peak at 1533 cm⁻¹ gradually evolved, corresponding to COOH species, which is a crucial intermediate for the formation of CO product.^52,53 As reported in previous studies, the formation of COOH intermediate is the rate-determining step during the photocatalytic CO₂ reduction reaction.^54,55 The possible photocatalytic CO₂ reduction mechanism is proposed and illustrated in Fig. 5e. First, the photogenerated carriers will be separated and migrated into the catalyst's active sites upon light irradiation. Subsequently, the CO₂ and H₂O molecules adsorbed on the catalyst's surface become activated, forming m-CO₃²⁻ and HCO₃⁻ intermediates. These intermediates are then further reduced to CO through the formation and dissociation of the COOH intermediate in the presence of H⁺ and e⁻ under light irradiation. Finally, the generated CO is released from the catalyst's surface. The following mechanism is proposed for CO₂ reduction into CO over 1.5Fe-Bi₂WO₆:

1.5Fe-Bi₂WO₆ + hν → 1.5Fe-Bi₂WO₆ (e⁻ + h⁺) (10)

COOH* + H⁺ + e⁻ → CO* + H₂O (14)

CO* → CO (15)

Fig. 5 In situ DRIFTS spectra for photocatalytic CO₂ reduction over 1.5Fe-Bi₂WO₆ recorded (a and b) in the dark for 1 hour and (c and d) under light irradiation conditions for 3 hours; (e) proposed photocatalytic CO₂ to CO reduction pathway.

3.2.2 Photocatalytic NO removal. Photocatalytic NO removal performance was evaluated in a flow reaction system under visible light illumination. Upon reaching the adsorption–desorption equilibrium of NO on the catalyst surface, the lamp was switched on to initiate the reaction. The outcomes of the photocatalytic NO removal process are presented in Fig. 6. It's worth noting that a blank test conducted without catalysts highlights the remarkable photostability of NO, as demonstrated in Fig. 6a. In the case of Bi₂WO₆, only 36.28% of NO is removed after 30 minutes of light irradiation (Fig. 6a), while it exhibits a substantial conversion efficiency of 17.34% for the harmful NO₂ byproduct (Fig. 6b and c). Given the undesirable formation of a toxic NO₂ byproduct, it is crucial to mitigate its formation during photocatalytic NO oxidation. As illustrated in Fig. 6a, the introduction of atomically dispersed Fe led to the enhancement of the NO removal efficiency, achieving 42.61 and 50.37% for 1Fe- and 1.5Fe-Bi₂WO₆, respectively. The NO₂ conversion performances are significantly inhibited in the presence of Fe, resulting in rates of 1.75% and 0.37% for 1Fe- and 1.5Fe-Bi₂WO₆, respectively, while the selectivity towards the NO₃⁻ product increases from 82.66% (Bi₂WO₆) to 99.63% (1.5Fe-Bi₂WO₆) (Fig. 6c). However, when the Fe dopant amount is further increased to 2Fe-Bi₂WO₆, there is a decrease in NO removal efficiency and an increase in NO₂ conversion. This can be attributed to the fact that a higher concentration of Fe dopants can act as recombination reservoirs for photogenerated electrons and holes, as indicated by the PL and TRPL results (Fig. 3c and d).⁵⁶ Hence, the availability of charge carriers needed for the generation of reactive species essential for the oxidation of NO and NO₂ to form NO₃⁻ product is reduced. The accumulation of adsorbed reactants and intermediates onto the catalyst surface without its efficient conversion to final products can result in surface saturation and poisoning, further hindering the exposure of active sites. Similar catalyst behavior linked to surface passivation has been reported in recent studies.^57–59 Moreover, the higher Fe introduction into Bi₂WO₆ causes a gradual downward shift in the conduction band, rendering it thermodynamically unfavorable for reduction reactions, particularly the generation of superoxide radicals. In addition, as shown in UPS results (Fig. S10†), the work function decreases upon the introduction of Fe, indicating improved charge transfer, which could be one of the factors that affect photocatalytic activity. For a more in-depth understanding of the photocatalytic NO removal rate, we have fitted the kinetic curves, which are displayed in Fig. 6d. The calculated rate constants are 11.04 × 10⁻² min⁻¹ for 1.5Fe-Bi₂WO₆, which is 1.5, 1.2, and 1.1 times higher than Bi₂WO₆ (7.53 × 10⁻² min⁻¹), 1Fe-Bi₂WO₆ (9.14 × 10⁻² min⁻¹), and 2Fe-Bi₂WO₆ Bi₂WO₆ (9.97 × 10⁻² min⁻¹), respectively.

Fig. 6 (a) Photocatalytic NO removal efficiency, (b) evolution of NO₂, (c) NO₂, NO₃⁻ conversion efficiency and NO₃⁻ selectivity, (d) the kinetic constants of photocatalytic NO removal reaction for Bi₂WO₆ and xFe-Bi₂WO₆ (x = 1, 1.5, and 2) under visible light; (e) stability test and (f) the selectivity toward NO₃⁻ of 1.5Fe-Bi₂WO₆ for photocatalytic NO removal.

The long-term stability of catalysts is another critical factor to consider in practical applications of the photocatalytic NO removal process. As demonstrated in Fig. 6e, 1.5Fe-Bi₂WO₆ exhibits remarkable stability, with only a 4.34% decrease in NO removal efficiency after 5 runs. Moreover, the selectivity of NO₃ only vanishes from 99.63% to 97.94%, and the NO₂ conversion increases from 0.37% to 2.06%, respectively (Fig. 6f). Furthermore, the structural analysis of Fe single atoms after photocatalytic NO removal was also examined using XAS. The XANES spectrum (Fig. S14a†) of 1.5Fe-Bi₂WO₆ revealed no alternations in the oxidation states of Fe species. The coordination environment around Fe single atoms was verified through EXAFS fitting (Fig. S14b–d and Table S9†), confirming their existence in the single atom form after the photocatalytic NO removal stability test. The coordination number remained lower than that of the Fe₂O₃ reference, indicating the durability of induced oxygen vacancies. These results highlight that 1.5Fe-Bi₂WO₆ exhibits not only excellent stability but also a preference for green product selectivity. In comparison with previously reported studies, as summarized in Table S10†, the atomically dispersed 1.5Fe-Bi₂WO₆ stands out as a potential and promising candidate for effective photocatalytic NO removal, characterized by low NO₂ conversion and good stability.

To gain further insight into the photocatalytic NO removal mechanism, trapping tests with various scavengers and spin-trapping EPR were conducted, as shown in Fig. 7a–d. K₂Cr₂O₇, KI, benzoquinone, and terephthalic acid were utilized to trap correspondingly the photogenerated electrons, holes, superoxide (O₂˙⁻), and hydroxyl radicals (OH˙). As shown in Fig. 7a, the NO removal performance vanishes significantly with the addition of K₂Cr₂O₇, KI, and benzoquinone, which implies the important roles of photogenerated electrons and holes in the photocatalytic NO removal process. The rate constant values of the NO removal reaction in the absence and presence of terephthalic acid, benzoquinone, KI, and K₂Cr₂O₇ were calculated to be 11.04, 10.51, 8.93, 5.94, and 0.34 × 10⁻² min⁻¹, respectively. Furthermore, the presence of hole scavengers significantly enhances NO₂ conversion, highlighting the crucial role of holes in the deep oxidation process of NO₂ into NO₃⁻ (Fig. 7b). EPR measurements were conducted to further confirm the formation of reactive radicals during the photocatalytic NO removal reaction. As illustrated in Fig. 7c and d, no signals of O₂˙⁻ and OH˙ radicals were detected under the dark condition, suggesting that light irradiation is crucial for the formation of reactive free radicals. Under light illumination, the signals of DMPO-O₂˙⁻ (intensity of 1 : 1 : 1 : 1) and DMPO-OH˙ (intensity of 1 : 2 : 2 : 1) adducts were observable for both Bi₂WO₆ and 1.5 Fe-Bi₂WO₆, indicating the generation of these radicals during the photocatalytic process and being consistent with the trapping tests. In addition, 1.5Fe-Bi₂WO₆ shows a higher intensity in the DMPO-O₂˙⁻ and DMPO-OH˙ spectra demonstrating that more reactive radicals are generated after Fe modification.

Fig. 7 (a) Trapping tests and the right panel represent the kinetic rate of photocatalytic NO oxidation over 1.5Fe-Bi₂WO₆ with/without scavengers; (b) NO₂ conversion during photocatalytic NO removal over 1.5Fe-Bi₂WO₆ with/without scavengers; (c and d) DMPO spin-trapping EPR spectra of Bi₂WO₆ and 1.5Fe-Bi₂WO₆ for O₂⁻˙ and HO˙ radicals in the aqueous dispersion, respectively; (e) schematic illustration of photocatalytic NO removal over 1.5Fe-Bi₂WO₆; (f) Gibbs free energy for the formation of intermediates during photocatalytic NO removal over Bi₂WO₆ and Fe-Bi₂WO₆.

Based on the obtained band structure, trapping tests, and EPR results, the mechanism for the photocatalytic NO removal process was proposed, following eqn (16)–(27) and illustrated in Fig. 7e. Firstly, Fe-Bi₂WO₆ absorbs the incident photons and generates electron–hole pairs (eqn (16)). These pairs are then separated, migrate to the surface of the catalyst, and participate in the redox reactions. Notably, the conduction band potential is more negative than the redox potential of O₂/O₂˙⁻, enabling photoinduced electrons to participate in the reduction of adsorbed oxygens, yielding O₂˙⁻ radicals (eqn (17)). These active O₂˙⁻ species will subsequently attend the generation of active OH˙ radicals through the reduction of H₂O₂ (eqn (18) and (19). This explains the detection of OH˙ radicals in EPR spectra, although the valence band potential of 1.5Fe-Bi₂WO₆ is less positive than the redox potential of H₂O/OH˙ (1.99 V). Subsequently, these reactive radicals (O₂˙⁻, OH˙) oxidize the adsorbed NO to generate NO₃⁻ (eqn (20) and (21)). In addition, NO could be oxidized to form NO₂ through eqn (22) and (23) and the desorption of NO₂ from the catalyst's surface induces the rising production of the detrimental by-product. Therefore, it becomes essential to stabilize NO₂ on the catalyst's surface to further transform it into NO₃⁻ (eqn (24)–(27)). In this process, Fe atoms serve as active sites for adsorption and reduction of NO₂ to NO₂⁻. Previous studies have shown that Fe³⁺ species could act as active sites for the adsorption and stabilization of NO₂.^18,60,61 Moreover, more generated oxygen vacancies could boost the surface adsorption and activation of O₂ molecules into active O₂˙⁻ and OH˙ species.⁶² Therefore, the synergistic effect between atomically dispersed Fe and oxygen vacancies is advantageous for the subsequent deep oxidation of stabilized NO₂ by photogenerated holes, O₂˙⁻ and OH˙ active species to produce NO₃⁻ and vanish the evolution of noxious NO₂ byproduct.⁶³ The DFT calculations of NO, NO₂, and NO₃ adsorption energy over pristine Bi₂WO₆ and Fe-Bi₂WO₆ were conducted and shown in Fig. S15.† The adsorption energy of NO and NO₂ on Fe-Bi₂WO₆ is higher than the pristine one, suggesting that NO and NO₂ are strongly stabilized on the surface of Fe-Bi₂WO₆, facilitating subsequent deep oxidation processes. Additionally, Fig. 7f displays Gibbs free energy diagrams for the conversion of NO to NO₃. The formation energies of intermediates over Fe-Bi₂WO₆ are lower than the ones of pristine Bi₂WO₆, indicating the modification of Bi₂WO₆ with Fe decreases the formation energy barrier and is favorable for the formation of NO₃⁻.

1.5Fe-Bi₂WO₆ + hν → e⁻ + h⁺ (16)

O₂ + e⁻ → O₂˙⁻ (17)

O₂˙⁻ + e⁻ + 2H⁺ → H₂O₂ (18)

H₂O₂ + e⁻ → HO˙ + HO⁻ (19)

NO + O₂˙⁻ → NO₃⁻ (20)

NO + 2HO˙ + HO⁻ → NO₃⁻ + H₂O (21)

NO + 2HO˙ → NO₂ + H₂O (23)

NO₂ + 2HO⁻ + h⁺ → NO₃⁻ + H₂O (24)

NO₂ + HO˙ → NO₃⁻ + H⁺ (25)

NO₂ + e⁻ → NO₂⁻ (26)

2NO₂⁻ + O₂˙⁻ + h⁺ → 2NO₃⁻ (27)

4. Conclusions

In summary, we have demonstrated an efficient strategy for engineering single atoms of the low-cost transition metal (Fe)-supported metal oxide Bi₂WO₆. This strategy has yielded remarkable improvements in the photocatalytic performance, particularly the selective reduction of CO₂ and oxidation of NO. The introduction of Fe and induced oxygen vacancies provides a synergistic effect that stabilizes NO/NO₂ and generates more reactive radicals, resulting in a substantial reduction in the production of harmful NO₂ byproducts. Our comprehensive experimental and theoretical investigations have unveiled the pivotal roles of Fe SAs and oxygen vacancies in inhibiting NO₂ formation, ultimately achieving a remarkable 99.63% green product selectivity. Additionally, the atomically dispersed Fe-Bi₂WO₆ photocatalyst exhibits the highest gas-phase photocatalytic CO₂ to CO reduction activity (36.78 μmol g⁻¹) compared to other bismuth tungstate-based photocatalysts. This study not only offers a straightforward strategy for designing functional photocatalysts for efficient solar energy conversion but also opens new avenues for engineering non-precious active centers in metal oxide catalysts for various applications in heterogeneous catalysis.

Data availability

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

Author contributions

Nguyen Quoc Thang: conceptualization, methodology, investigation, formal analysis, writing – original draft. Amr Sabbah: conceptualization, methodology, investigation, validation, writing – review and editing. Chih-Yang Huang: methodology, investigation. Nguyen Hoang Phuong: methodology, investigation, Tsai-Yu Lin: methodology, investigation. Mahmoud Kamal Hussien: methodology, investigation, Heng-Liang Wu: methodology, investigation, resources. Chih-I Wu: methodology, investigation, resources. Nguyet. N. T. Pham: methodology, investigation. Pham Van Viet: investigation, resources. Chih-Hao Lee: investigation, resources. Li-Chyong Chen: conceptualization, validation, writing – review and editing, resources, supervision, project administration, funding acquisition. Kuei-Hsien Chen: conceptualization, validation, writing – review and editing, resources, supervision, project administration, funding acquisition.

Conflicts of interest

The authors declare that they have no competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Acknowledgements

The authors acknowledge the technical support provided by Mr Tzu-Hsien Tseng at the Institutional Center of National Chung Hsing University and Ms. Payal Maharathi at National Taiwan University. This work was financially supported by the National Science and Technology Council (NSTC) of Taiwan, Science Vanguard Project (NSTC 111-2123-M-002-009), Academic Summit Project (NSTC 112-2639-M-002-005-ASP), and NSTC 112-2113-M-001-023. Additional financial supports from the Center of Atomic Initiative for New Materials, National Taiwan University, from the Featured Areas Research Center Program within the framework of the Higher Education Sprout Project by the Ministry of Education in Taiwan (111L9008 and 112L9008), and Academia Sinica (AS-SS-112-01) are also acknowledged.

References

1. A. Sabbah, I. Shown, M. Qorbani, F.-Y. Fu, T.-Y. Lin, H.-L. Wu, P.-W. Chung, C.-I. Wu, S. R. M. Santiago, J.-L. Shen, K.-H. Chen and L.-C. Chen, Nano Energy, 2022, 93, 106809.
2. T. Billo, I. Shown, A. k. Anbalagan, T. A. Effendi, A. Sabbah, F.-Y. Fu, C.-M. Chu, W.-Y. Woon, R.-S. Chen, C.-H. Lee, K.-H. Chen and L.-C. Chen, Nano Energy, 2020, 72, 104717.
3. R. Chen, J. Li, H. Wang, P. Chen, X. a. Dong, Y. Sun, Y. Zhou and F. Dong, J. Mater. Chem. A, 2021, 9, 20184–20210.
4. I. Shown, S. Samireddi, Y. C. Chang, R. Putikam, P. H. Chang, A. Sabbah, F. Y. Fu, W. F. Chen, C. I. Wu, T. Y. Yu, P. W. Chung, M. C. Lin, L. C. Chen and K. H. Chen, Nat. Commun., 2018, 9, 169.
5. M. Qorbani, A. Sabbah, Y. R. Lai, S. Kholimatussadiah, S. Quadir, C. Y. Huang, I. Shown, Y. F. Huang, M. Hayashi, K. H. Chen and L. C. Chen, Nat. Commun., 2022, 13, 1256.
6. C. Zhang, Y. Xu, H. Bai, D. Li, L. Wei, C. Feng, Y. Huang, Z. Wang, X. Li, X. Cui, C. Hu and F. Wang, Nano Energy, 2024, 121, 109197.
7. C. Wu, Q. Tang, S. Zhang, K. Lv, X. Fuku and J. Wang, ACS Appl. Mater. Interfaces, 2023, 15, 30127–30138.
8. B. Qiao, A. Wang, X. Yang, L. F. Allard, Z. Jiang, Y. Cui, J. Liu, J. Li and T. Zhang, Nat. Chem., 2011, 3, 634–641.
9. B. Li, C. Guo, X. Wang, W. Dong, B. Xu, X. Xing, D. Zhou, X. Xue, Q. Luan, W. Tang and C. Hou, Mater. Today Nano, 2023, 21, 100281.
10. G. Ren, M. Shi, S. Liu, Z. Li, Z. Zhang and X. Meng, Chem. Eng. J., 2023, 454, 140158.
11. Y. Shi, C. Zhao, H. Wei, J. Guo, S. Liang, A. Wang, T. Zhang, J. Liu and T. Ma, Adv. Mater., 2014, 26, 8147–8153.
12. Q. Yang, Y. Jiang, H. Zhuo, E. M. Mitchell and Q. Yu, Nano Energy, 2023, 111, 108404.
13. X. Xiong, C. Mao, Z. Yang, Q. Zhang, G. I. N. Waterhouse, L. Gu and T. Zhang, Adv. Energy Mater., 2020, 10, 2002928.
14. C. Gao, J. Low, R. Long, T. Kong, J. Zhu and Y. Xiong, Chem. Rev., 2020, 120, 12175–12216.
15. P. Yang, J. Zhang, C. Chen, X. Guo, J. Zhang, X. Zhang and Z. Wu, ACS Appl. Nano Mater., 2022, 5, 5128–5139.
16. S. Xiong, S. Bao, W. Wang, J. Hao, Y. Mao, P. Liu, Y. Huang, Z. Duan, Y. Lv and D. Ouyang, Appl. Catal., B, 2022, 305, 121026.
17. Y. Wang, Y. Zhang, W. Yu, F. Chen, T. Ma and H. Huang, J. Mater. Chem. A, 2023, 11, 2568–2594.
18. G. Cheng, X. Liu, X. Song, X. Chen, W. Dai, R. Yuan and X. Fu, Appl. Catal., B, 2020, 277, 119196.
19. L. Li, J. Yang, L. Yang, F. Fu, H. Xu and X. Fan, J. Environ. Chem. Eng., 2022, 10, 107708.
20. M. Xiao, L. Zhang, B. Luo, M. Lyu, Z. Wang, H. Huang, S. Wang, A. Du and L. Wang, Angew Chem. Int. Ed. Engl., 2020, 59, 7230–7234.
21. L. Liu, J. Liu, K. Sun, J. Wan, F. Fu and J. Fan, Chem. Eng. J., 2021, 411, 128629.
22. Y. Qiu, J. Lu, Y. Yan and J. Niu, J. Hazard. Mater., 2022, 422, 126920.
23. T. Hu, H. Li, N. Du and W. Hou, ChemCatChem, 2018, 10, 3040–3048.
24. N. Q. Thang, A. Sabbah, L.-C. Chen, K.-H. Chen, L. V. Hai, C. M. Thi and P. V. Viet, Chem. Eng. Sci., 2021, 229, 116049.
25. G. Kresse and J. Furthmüller, Comput. Mater. Sci., 1996, 6, 15–50.
26. G. Kresse and J. Furthmuller, Phys. Rev. B: Condens. Matter Mater. Phys., 1996, 54, 11169–11186.
27. J. P. Perdew, K. Burke and M. Ernzerhof, Phys. Rev. Lett., 1996, 77, 3865–3868.
28. 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.
29. B. Hammer, L. B. Hansen and J. K. Nørskov, Phys. Rev. B: Condens. Matter Mater. Phys., 1999, 59, 7413–7421.
30. M. Ernzerhof and G. E. Scuseria, J. Chem. Phys., 1999, 110, 5029–5036.
31. J. G. Thompson, A. D. Rae, R. L. Withers and D. C. Craig, Acta Crystallogr., Sect. B: Struct. Sci., 1991, 47, 174–180.
32. N. A. McDowell, K. S. Knight and P. Lightfoot, Chemistry, 2006, 12, 1493–1499.
33. H. Djani, P. Hermet and P. Ghosez, J. Phys. Chem. C, 2014, 118, 13514–13524.
34. V. Koteski, J. Belošević-Čavor, V. Ivanovski, A. Umićević and D. Toprek, Appl. Surf. Sci., 2020, 515, 146036.
35. Y. Wei, X. Wei, S. Guo, Y. Huang, G. Zhu and J. Zhang, J. Mater. Sci. Eng. B, 2016, 206, 79–84.
36. B. Himmetoglu, A. Floris, S. de Gironcoli and M. Cococcioni, Int. J. Quantum Chem., 2014, 114, 14–49.
37. J. K. Nørskov, J. Rossmeisl, A. Logadottir, L. Lindqvist, J. R. Kitchin, T. Bligaard and H. Jónsson, J. Phys. Chem. B, 2004, 108, 17886–17892.
38. W. M. Haynes, CRC Handbook of Chemistry and Physics, CRC press, 2016.
39. J. Ma, H. He and F. Liu, Appl. Catal., B, 2015, 179, 21–28.
40. Q. Li, X. Zhu, J. Yang, Q. Yu, X. Zhu, J. Chu, Y. Du, C. Wang, Y. Hua, H. Li and H. Xu, Inorg. Chem. Front., 2020, 7, 597–602.
41. Y. Liu, D. Shen, Q. Zhang, Y. Lin and F. Peng, Appl. Catal., B, 2021, 283, 119630.
42. Y. Jiang, H. Y. Chen, J. Y. Li, J. F. Liao, H. H. Zhang, X. D. Wang and D. B. Kuang, Adv. Funct. Mater., 2020, 30, 2004293.
43. H. Ma, W. Yang, S. Gao, W. Geng, Y. Lu, C. Zhou, J. K. Shang, T. Shi and Q. Li, Chem. Eng. J., 2023, 455, 140471.
44. Y. Xiong, H. Li, C. Liu, L. Zheng, C. Liu, J. O. Wang, S. Liu, Y. Han, L. Gu, J. Qian and D. Wang, Adv. Mater., 2022, 34, e2110653.
45. G. Zhao, W. Li, H. Zhang, W. Wang and Y. Ren, Chem. Eng. J., 2022, 430, 132937.
46. M. Zhou, Z. Wang, A. Mei, Z. Yang, W. Chen, S. Ou, S. Wang, K. Chen, P. Reiss, K. Qi, J. Ma and Y. Liu, Nat. Commun., 2023, 14, 2473.
47. J. Fan, Y. Zhao, H. Du, L. Zheng, M. Gao, D. Li and J. Feng, ACS Appl. Mater. Interfaces, 2022, 14, 26752–26765.
48. S. Ji, Y. Qu, T. Wang, Y. Chen, G. Wang, X. Li, J. Dong, Q. Chen, W. Zhang, Z. Zhang, S. Liang, R. Yu, Y. Wang, D. Wang and Y. Li, Angew Chem. Int. Ed. Engl., 2020, 59, 10651–10657.
49. H. Huang, X. Liu, F. Li, Q. He, H. Ji and C. Yu, Sustainable Energy Fuels, 2022, 6, 4903–4915.
50. M. Wang, M. Shen, X. Jin, J. Tian, Y. Zhou, Y. Shao, L. Zhang, Y. Li and J. Shi, Nanoscale, 2020, 12, 12374–12382.
51. Y. Bo, P. Du, H. Li, R. Liu, C. Wang, H. Liu, D. Liu, T. Kong, Z. Lu, C. Gao and Y. Xiong, Appl. Catal., B, 2023, 330, 122667.
52. Z. Jiang, H. Sun, T. Wang, B. Wang, W. Wei, H. Li, S. Yuan, T. An, H. Zhao, J. Yu and P. K. Wong, Energy Environ. Sci., 2018, 11, 2382–2389.
53. M. K. Hussien, A. Sabbah, M. Qorbani, R. Putikam, S. Kholimatussadiah, D. M. Tzou, M. H. Elsayed, Y. J. Lu, Y. Y. Wang, X. H. Lee, T. Y. Lin, N. Q. Thang, H. L. Wu, S. C. Haw, K. C. Wu, M. C. Lin, K. H. Chen and L. C. Chen, Small, 2024, 20, 2400724.
54. M. Kamal Hussien, A. Sabbah, M. Qorbani, M. Hammad Elsayed, P. Raghunath, T.-Y. Lin, S. Quadir, H.-Y. Wang, H.-L. Wu, D.-L. M. Tzou, M.-C. Lin, P.-W. Chung, H.-H. Chou, L.-C. Chen and K.-H. Chen, Chem. Eng. J., 2022, 430, 132853.
55. X. Jiao, X. Li, X. Jin, Y. Sun, J. Xu, L. Liang, H. Ju, J. Zhu, Y. Pan, W. Yan, Y. Lin and Y. Xie, J. Am. Chem. Soc., 2017, 139, 18044–18051.
56. L. Wen, B. Liu, X. Zhao, K. Nakata, T. Murakami and A. Fujishima, Int. J. Photoenergy, 2012, 2012, 1–10.
57. X. Wu, J. Zhou, Q. Tan, K. Li, Q. Li, S. A. Correia Carabineiro and K. Lv, ACS Appl. Mater. Interfaces, 2024, 16, 11479–11488.
58. R. Hailili, Z. Li, X. Lu, H. Sheng, D. W. Bahnemann and J. Zhao, Environ. Sci.: Nano, 2024, 11, 3301–3316.
59. T. Shen, X. Shi, J. Guo, J. Li and S. Yuan, Chem. Eng. J., 2021, 408, 128014.
60. X. Song, W. Jiang, Z. Cai, X. Yue, X. Chen, W. Dai and X. Fu, Chem. Eng. J., 2022, 444, 136709.
61. Q. Wu and R. van de Krol, J. Am. Chem. Soc., 2012, 134, 9369–9375.
62. Y. Xin, Q. Zhu, T. Gao, X. Li, W. Zhang, H. Wang, D. Ji, Y. Huang, M. Padervand, F. Yu and C. Wang, Appl. Catal., B, 2023, 324, 122238.
63. L. Liu, P. Ouyang, Y. Li, Y. Duan, F. Dong and K. Lv, J. Hazard. Mater., 2022, 439, 129637.

---

Footnote

† Electronic supplementary information (ESI) available. See DOI: https://doi.org/10.1039/d4ta05778c

---