Recent Advances in Niobium-Based Materials for Photocatalytic Solar Fuel Production

Abstract

The search for renewable and clean energy sources is a key aspect for sustainable development as energy consumption has continuously increased over the years concomitantly with environmental concerns caused by the use of fossil fuels. Semiconductor materials have great potential for acting as photocatalysts for solar fuel production, a potential energy source able to solve both energy and environmental concerns. Among the studied semiconductor materials, those based on niobium pentacation are still shallowly explored, although the number of publications and patents on Nb(V)-based photocatalysts has increased in the last years. A large variety of Nb(V)-based materials exhibit suitable electronic/morphological properties for light-driving reactions. Not only the extensive group of Nb₂O₅ polymorphs is explored, but also many types of layered niobates, mixed oxides, and Nb(V)-doped semiconductors. Therefore, the aim of this manuscript is to provide a review of the latest developments of niobium based photocatalysts for energy conversion into fuels, more specifically, CO₂ reduction to hydrocarbons or H₂ evolution from water. Additionally, the main strategies for improving the photocatalytic performance of niobium-based materials are discussed.

1. Introduction

Increasing energy demand concomitantly with global warming is a major world concern. Currently, fuels and electricity supplies are still strongly dependent on nonrenewable sources. For instance, in 2018, the worldwide consumption of fossil energy was almost 36.5 billion barrels of oil, 3850 billion cubic meters of natural gas, and 3800 million tons oil equivalent of coal per year. As a consequence, more than 33,000 million tons of carbon dioxide was emitted, which represents an annual increasing of 2.0%, the fastest growth for seven years [1]. This scenario can trigger serious problems for the current and next generations such as the insufficiency of fossil fuel resources, air pollution, and also climate changes as a result of the greenhouse effect [2].

In the face of this issue, the development of methods to obtain energy in a clean and sustainable way becomes imperative. Therefore, solar light as a primary energy resource stands out for its advantages in terms of spread availability and accessibility in an inexhaustible and inexpensive way around the world. In addition, obtaining energy from the sun may cause less damage to the environment in relation to waste production, polluting gases emission, or impacts to different ecosystems. The possibility to convert light into valuable solar fuels from water or even CO₂ is a promising approach to provide clean energy and, at the same time, contribute to diminishing the global warming effect [3]. Additionally, the total sum of renewable energy from all the reserves is around 1% of the solar energy supplied to the earth surface by the sun, that is, 173,000 TW. This amount of energy is more than 9000 times larger than the current total energy consumption of the world (17.91 TW in 2017). The challenge lies on the development of economically viable technologies for solar energy harvesting, storage, and utilization [4].

Thus, since Fujishima and Honda [5] reported the possibility of promoting the photo-oxidation of water on the surface of TiO₂ photoanodes, semiconductor materials have been extensively investigated as photocatalysts to store solar energy in chemical bonds. Among different photocatalysts, it is worth pointing out Nb(V)-based materials that have been investigated owing to their suitable electronic and morphological properties for light-driving reactions. Nb(V) is characterized by its high affinity to oxygen and its oxides exhibit different properties depending on the preparation and desired application [6,7]. Niobium pentoxide is considered the most thermodynamically stable Nb-based compound with more than 15 different structural configurations. Depending on the crystalline phase, its band-gap energy can vary from 3.2 eV to larger values such as 5 eV [8,9,10].

Besides Nb₂O₅ polymorphs, Nb(V) can also combine with oxygen to form extended lamellar structures resulting in polyoxy anions, the niobates. They are basically constituted of stacked negative charged layers with intercalated cations [11]. This configuration provides a considerable surface area and allows easy modification either by intercalation, superficial modification, or formation of nanosheets and nanoscrolls [12,13,14]. Furthermore, Nb(V) is also frequently applied as a doping agent. It is commonly combined to TiO₂ and considered one of the most promising dopants for modifying the matrix crystalline structure and minimize the electron–hole recombination [15,16,17,18].

In this paper, the recent developments of niobium-based materials for photocatalytic solar fuel production, especially those in the last three years, are systematically reviewed. The main strategies for improving the photocatalytic performance for H₂ production from water splitting and CO₂ reduction to hydrocarbons, focused on niobium species, will be presented. In Section 2, application to H₂ evolution is discussed, followed by Section 3 related to CO₂ reduction.

2. H₂ Evolution and Water Splitting

2.1. Niobium Pentoxide (Nb₂O₅)

Nb₂O₅ is a typical n-type semiconductor that, along with other oxides, has been widely studied as a photocatalyst thanks to its electronic properties such as excellent chemical and thermal stability, large abundance, rich morphologies, and several polymorphs [19,20,21,22]. It is a white solid powder, insoluble in water, and can be dissolved just by fusion with strong basic or acid fluxes. In general, its structure is formed by NbO₆ octahedral, which can result in an amorphous state or may crystallize in a wide range of polymorphs with different physical properties [6,8]. Some of those polymorphs are pseudohexagonal (TT-Nb₂O₅), orthorhombic (T-Nb₂O₅), and monoclinic (B, H, R, and N-Nb₂O₅) [19]. The most known nomenclature is based on Brauer’s system, in which the crystalline phases are identified by letters corresponding to the temperature at which they are obtained [8].

Despite these properties, bulk Nb₂O₅ is not able to achieve a considerable photoactivity owing to its low specific surface area and fast recombination rate of the photogenerated charge carriers [21]. In this sense, several approaches have been applied to increase its photocatalytic performance. One of them is to control the shape and morphology of particles in order to tune the Nb₂O₅ properties. This can be possible, for instance, by selecting a suitable precursor to obtain the desired configuration of Nb₂O₅. Wen and coauthors firstly synthesized orthorhombic Nb₂O₅ (O-Nb₂O₅) by a hydrothermal reaction using K₄Nb₆O₁₇ 4.5H₂O as a precursor. This sample exhibited rectangle nanosheets with a dominantly exposed (010) plane (Figure 1a–d) [22]. After acid treatment of K₄Nb₆O₁₇ 4.5H₂O to get H₄Nb₆O₁₇ 3H₂O, the ion-exchanged material was treated hydrothermally for 24 h at different conditions of temperature (195–215 °C) and pH (1.5–11.5), as shown in Figure 1e. At first, by changing the temperature at pH 3.5, X ray diffraction (XRD) results showed that the O-Nb₂O₅ phase began to be formed at 200 °C, and at 210 °C, the peaks of H₄Nb₆O₁₇ 3H₂O disappeared completely and all observed peaks could be indexed as the O-Nb₂O₅ phase. From the Figure 1e, it can be also observed that the O-Nb₂O₅ was obtained in all selected pHs; however, the authors reported that, at pH 3.5, the material showed higher crystallinity and more regular morphology. A commercial O-Nb₂O₅ sample with spherical morphology was used for comparison and exhibited a band gap value of 3.2 eV, in contrast to 3.0 eV for the as synthesized O-Nb₂O₅ rectangle nanosheets. Then, the photoelectrochemical activity of both samples was evaluated by measuring the photocurrent resulted from H₂ evolution under 370 nm UV light irradiation (30 W·m⁻²). Both photoelectrodes show very low dark current, but under illumination, the O-Nb₂O₅ nanosheet-based photoelectrode exhibited photocurrents 4.3 times higher than the commercial sample. The authors could justify the improved performance to a larger number of charge carriers coming from the exposed (010) facet with lower band gap energy, fewer recombination centers, and also faster charge transfer by the regular rectangle nanosheets with high crystallinity.

Later, the same strategy was employed by the same group to obtain crystalline hexagonal Nb₂O₅ (H-Nb₂O₅) nanobelt clusters with an exposed (−110) plane [23]. In this case, a fiber-like K₂Nb₂O₆·H₂O material was used as precursor. In order to improve the photocatalytic performance of H-Nb₂O₅ under visible light, metallic Ag was deposited by in situ photoreduction. UV/vis diffuse reflectance spectra showed that the absorption edges were 375 nm for the H-Nb₂O₅ nanobelt clusters and 423 nm for the Ag-modified sample. Then, both samples were evaluated photoelectrochemically under simulated sunlight illumination (100 W m⁻²) at zero applied voltage. The Ag-modified sample achieved a photocurrent of 10.73 μA, 535 times higher than that observed for bare H-Nb₂O₅, which was attributed to the improved light-harvesting efficiency as well as the better charge transfer performance.

Zhou and coauthors reported monoclinic Nb₂O₅ nanorod superstructures by a similar route using Sn₂Nb₂O₇ nanoparticles as the precursor [24]. SEM images showed that Nb₂O₅ nanorods were obtained with lengths around 2 µm and diameters of 200 nm. The light absorption properties of the Nb₂O₅ nanorods were compared to those of commercial Nb₂O₅ powders and a considerable blue shift was observed in the absorption edge, which could be related to the nanosize of the former. Photocatalytic H₂ production was also evaluated with 0.5 wt% Pt as co-catalyst in 25% methanol aqueous solution, employing a 500 W high-pressure mercury lamp as light source. The Nb₂O₅ nanorod superstructures sample showed a H₂ evolution rate of 91 µmol h⁻¹, whereas the commercial one exhibited a rate of 0.6 µmol h⁻¹, ca. 150 times lower than that for the Nb₂O₅ nanorod. Moreover, Nb₂O₅ nanorod superstructures maintained an almost constant photocatalytic H₂ production rate after 24 h. The higher performances were explained by the authors based on the superstructures of the material (Figure 2), in which photogenerated electrons could be transferred along the 1D nanorods direction and also transported to the neighboring nanorods, leading to a more efficient separation of photogenerated charge carriers.

Clearly, such morphological and structural transformations lead to changes in the optical and electronic properties of niobium oxide. In this sense, Zhang et al. reported an enclosed porous structure of pseudo-hexagonal tief-tief (TT) Nb₂O₅ nanowires, which showed a significant decrease in the band gap from 3.22 to 2.95 eV in relation to the nonporous Nb₂O₅ nanowires, and then an enhancement of H₂ evolution under visible light irradiation (λ > 420 nm) [25]. The material was prepared by a solvothermal synthesis using ammonium niobate(V) oxalate hydrate and oleic acid mixed with trioctylamine. The sample was calcined at different temperatures. Thermal treatment at 600 °C and 700 °C causes the formation of orthorhombic phase, while at 580 °C, the pristine TT-phase was formed, so this last condition was selected for further analysis. In TEM images, it was observed that the porous nanowires were shorter than 1 µm and exhibited well separated pores inside the structure, whereas nonporous ones were longer than 5 µm and showed ripple-like contrast owing to bending-induced strain. The authors could explain the formation of such porosity to the replacement of oxalate anion by oleic acid and trioctylamine in the Nb center, resulting in a structure with hydrophobic core and surrounding Nb species. Then, the subsequent heating treatment induced the formation of pores from the cores and the Nb₂O₅ crystalline nanowires from the coordination structure. However, the samples were not characterized by isotherms of N₂ adsorption to be properly classified as porous material. The authors suggested that under-coordinated ions in the twisted lattice around the pores could generate mid-gap states, and hence lower band gap energies. As a result, the H₂ production rate of the porous sample was 243.8 μmol g⁻¹ h⁻¹, considerable higher than the nonporous one, which exhibited H₂ evolution rate of 113.1 μmol g⁻¹ h⁻¹ under similar conditions, and even higher than that for P25 TiO₂ (129.6 μmol g⁻¹ h⁻¹).

Besides the morphology, other strategies can be applied to increase the photocatalytic activity of Nb₂O₅ species. The introduction of defects into the crystal lattice could be used to engineer the electronic transport properties along the oxide structure or even to create new trap states inside the forbidden band gap region [26]. Zhao and coauthors reported improved solar-light absorption via partial reduction of Nb₂O₅ nanorods with active exposed (001) surfaces, which result in the black niobium oxide [27]. In summary, the previous prepared Nb₂O₅ was thermally treated in an aluminum reduction device, where the sample and aluminum powder were positioned independently in a two-zone tube oven, and then the Nb₂O₅ and aluminum powders were heated to 500 °C and 800 °C, respectively. This procedure led to the formation of Nb⁴⁺ centers and hence substantial oxygen vacancies that resulted in a black powder, and thus an enhancement in visible and infrared light absorption compared with the pristine Nb₂O₅. The reduction of Nb₂O₅ could be confirmed by Raman and X-ray photoelectron spectroscopies. The Nb⁴⁺ centers were also characterized by electron paramagnetic resonance. High resolution transmission electron microscopy indicated predominant (001) crystal plane exposed on the nanorod surfaces, suggesting good crystallinity of samples before and after reduction. The samples were evaluated as photoanode using dip-coated FTO glasses in a 1 M NaOH electrolyte under 100 mW cm⁻² illumination with a 150 W Xe lamp. At 1.23 V, the reduced Nb₂O₅ sample showed a photocurrent density 138 times higher than that for Nb₂O₅. The carrier density was calculated by Mott–Schottky plots and was 3 × 10⁴ times higher for the reduced sample, evidencing an enhanced electrical conductivity, charge transport, and separation. Finally, the samples were evaluated to produce H₂ with 0.5 wt% Pt and 20% methanol solution under full-sunlight irradiation from a 300 W Xe lamp, which, for the reduced Nb₂O₅, resulted in 13 times higher production rate of 274.8 µmol h⁻¹ g⁻¹ in comparison with Nb₂O₅ nanorods (21.1 µmol h⁻¹ g⁻¹).

An extended visible light activity of orthorhombic Nb₂O₅ was achieved by Kulkarni et al. as result of nitrogen doping using a solid state reaction of urea and niobium salt [20]. At 500 °C, ammonia is released and works as nitrogen source to be introduced into the Nb₂O₅ lattice, filling oxygen vacancies. Field emission scanning electron microscopy (FE-SEM) analysis showed that the urea concentration was crucial for the observed morphologic changes. Concerning the optical properties, the modified samples were analyzed by UV/vis diffuse reflectance and compared to the pristine Nb₂O₅, which had a band gap energy of 3.4 eV. In general, all the N-doped Nb₂O₅ samples showed significant absorption in the range of 400–600 nm and a higher red shift in the band gap absorption when the urea concentration was increased (2.6–2.4 eV). Density functional theory (DFT) calculations described that N 2p bands falls above the O 2p bands, and so the top of the valence band gives rise to the reduction in the band gap. At the highest Nb/urea ratio investigated (1:15), this trend was not observed and a higher band gap edge than the expected trend (2.5 eV) was obtained. This would have resulted from nitrogen saturation of the niobia matrix. Photoluminescence measurements indicate that the optimal Nb/urea ratio was 1:10. This sample exhibited the lowest luminescence intensity among the other samples, indicating the efficient electron/hole separation. These results were in agreement with the photocatalytic H₂ evolution experiments.

Another strategy to achieve higher photocatalytic performances is coupling to different Nb₂O₅ compounds with special attributes to create an efficient heterostructured photocatalyst. Owing to its flexibility in morphological, electronic, and structural properties, Nb₂O₅ is a promising candidate to design an effective heterojunction. Huang and coauthors reported a synergic combination of highly ordered mesoporous Nb₂O₅ (MNb) nanostructure to two-dimensional N-doped graphene (NGR) through a good interfacial connection (Figure 3) [28]. UV/visible diffuse reflectance spectroscopy evidenced gradual shifts to longer wavelengths as the NGR quantity was increased, so the band gap energies were shift from 3.12 eV up to 2.67 eV. Photoluminescence analysis suggested that NGR supported an efficient separation of charge carriers owing to the fast electron transfer from the conduction band (CB) of MNb to NGR sheets.

Huang and coauthors reported an in situ growth of mesoporous Nb₂O₅ microspheres on the surface of g-C₃N₄ [21]. In this case, assuming the photocatalytic activity of g-C₃N₄ under visible light, the enhanced photoactivity was mainly achieved thanks to the effective interfacial charge transfer between g-C₃N₄ and Nb₂O₅. Therefore, it is mandatory to ensure sufficient interfacial contact, and this condition was succeeded by the authors via a so-called in situ self-assembly process of synthesis. The introduction of Nb₂O₅ microspheres on g-C₃N₄ caused an expressive enhancement on the specific surface area. UV/visible diffuse reflectance spectroscopy showed that g-C₃N₄ and the composites presented an absorption edge at ca. 450 nm, while pure Nb₂O₅ had slight a band edge of ca. 420 nm. The slightly visible light absorption of Nb₂O₅ was justified by traces of carbonate species from the thermal decomposition of the triblock copolymer used in the synthesis. Photoluminescence spectroscopy evidenced that pure g-C₃N₄ exhibits stronger emission than the composites. Photocatalytic activities for H₂ evolution over the samples were evaluated and the composite with 38.1 wt% of Nb₂O₅ exhibited the highest H₂ evolution rate. With higher amounts of the Nb₂O₅ (69.6 wt%), the photocatalytic performance was still higher than pure g-C₃N₄, but slightly decreased in relation to the composite with 38.1 wt% of Nb₂O₅. This fact could be related to a less effective interfacial contact between the two semiconductors and to a considerable reduction of light absorption by the composite. Other successful combinations between Nb₂O₅ and g-C₃N₄ are reported elsewhere [29,30].

Therefore, it can be observed that Nb₂O₅ pristine or modified has been employed successfully as a photocatalyst for H₂ evolution. The role of the crystalline phase, morphology/shape, defects/oxygen vacancies, and heterojunction formation were investigated and showed great contributions for H₂ evolution activity of Nb₂O₅.

2.2. Niobium Layered Compounds

Layered niobates are composed by a repetition of [NbO₆] octahedral units connected through adjacent or opposite sharing edges or corners, resulting in an extended 2D layered arrangement. This configuration is built by the stacking of negative charged layers separated by cations in the interlayer space [11,31]. The first niobate phases were reported in 1955 by Reisman and Frederic [32]. In this work, they could investigate the products K₃NbO₄, KNbO₃, K₄Nb₆O₁₇, KNb₃O₈, and “K₆Nb₄₄O₁₁₃” from the reaction of K₂CO₃ and Nb₂O₅ by differential thermal analysis [32,33]. Then, in 1969, the K₄Nb₆O₁₇ and KNb₃O₈ crystals were further investigated by Nassau and coauthors [33]. In 1982, a new series of phases MCa₂Nb₃O₁₀ (M = Li, Na, K, Rb, Cs, NH₄, Tl) was reported by Dion et al. [34]. Similarly, some years later, Jacobson and coauthors were able to prepare and investigate the interlayer reactivity of the series of layered compounds K[Ca₂Na_n−3Nb_nO_3n+1] [35]. Since then, a class of layered perovskite with the general formula A_x[B_m−1Nb_nO_3n+1] is known as Dion–Jacobson compounds (where A represents an alkaline monocation; B an alkaline earth ion; m = 1, 2, and 2 ≤ n ≤ 7; n indicates the number of [NbO₆] chains that form each perovskite-like slab) [11]. Moreover, plenty of different niobates are currently known such as alkali (M⁺NbO₃), columbite (M²⁺Nb₂O₆), and rare-earths orthoniobates (YNbO₄) [6].

These materials are known as promising photocatalysts not just because of their different structures and band gap energies, but also their extended 2D layered arrangement, which provides a considerable surface area and allows easy modification either by intercalation, superficial modification, or formation of nanosheets and nanoscrolls through exfoliation with bulky n-alkylammonium salts [11,31,36]. Recently, our group investigated the influence of the medium, co-catalyst precursor, and heat treatment in the photoactivity of Pt-modified hexaniobate (K_4−xH_xNb₆O₁₇) for H₂ evolution [31]. The effect of the Pt(0) precursor and the deposition method (absorption or impregnation) was also investigated. These different conditions of the samples’ preparation mainly affected their configuration and morphology. At the end, the H₂ evolution performance was dependent on those conditions, where a ‘soft’ photoreduction of platinum precursors was identified as the best method for the preparation of those photocatalysts.

Suspended hexaniobate layers (K_4−xH_xNb₆O₁₇) in tetrabutylammonium hydroxide (TBAOH) showed a good performance in designing the photocatalytic film through the layer-by-layer (LbL) deposition technique [36]. Thin films were assembled by alternative immersions of FTO substrate into exfoliated hexaniobate with pre-adsorbed [Pt(NH₃)₄]²⁺ suspension (pH = 8) and poly(allylamine hydrochloride) solution (pH = 4) up to 25 bilayers. Further thermal treatment at 500 °C removed the organic species, leading to a fuzzy assembly of hexaniobate nanoscrolls with Pt evenly distributed in the surface, as shown in the FE-SEM image (Figure 4). In this case, it could be concluded that the scrolled morphology was favored in relation to the opened sheets. The Nb-based films were photoactive to produce H₂ from 20% (v/v) methanol/water solutions under UV-irradiation. The bare hexaniobate LbL films were able to photocatalyze the H₂ evolution, with the apparent quantum yield being proportional to the number of deposited bilayers. This behavior evidence that no photocatalytic surface is lost owing to material deposition. Moreover, when the Pt nanoclusters were added to the film composition, the observed H₂ evolution rates were approximately two times higher, reaching 4.0% ± 0.5%.

Oshima and coauthors reported the photocatalytic activity of Pt-deposited KCa₂Nb₃O₁₀ for water splitting [37]. With an adsorption method, cationic precursor of Pt ([Pt(NH₃)₄]Cl₂) was attached on niobate surface (1.3 wt%) by the TBA⁺/Ca₂Nb₃O₁₀⁻ suspension, followed by restacking with potassium hydroxide and reduction with H₂ gas in various temperatures. The authors noted that the deposition of Pt occurred in the interlayer spaces of the restacked nanosheets. Additionally, the increase in the annealing temperature up to 973 K decreased the Pt content in the KCa₂Nb₃O₁₀ interlayer space. Diffuse reflectance and X-ray photoelectron spectroscopies suggested that the adsorbed Pt was reduced from an oxidized to the metallic state according to the applied temperature, and that interlayer Pt species were less susceptible to reduction. The samples were evaluated for water splitting in NaI 10 mmol L⁻¹ solution and using a 300 W xenon lamp (λ ≥ 300 nm). The maximum rate of both H₂ and O₂ production was achieved with the sample treated at 473 K, and higher reduction temperatures led to a decline in the photoactivity. This fact could be explained by an improved contact between Pt and the niobate and the formation of suitable active sites. At higher temperatures, the Pt species were located mostly at the external surface. The authors also investigated the impact of the Pt valence state on water splitting activity, comparing the performance of samples heated at the 573 K under H₂ and air atmosphere. By TEM, both methodologies led to the formation of Pt with the same size. However, more electron-deficient Pt species were identified on the sample treated in air, which performed a better overall water splitting reaction than the sample treated at H₂.

Distinct species to form the layered Dion–Jacobson materials AB₂Nb₃O₁₀ also play an essential role in the physical properties of the material and especially in their photocatalytic performance. From this point of view, Kulischow et al. investigated a family of layered Dion–Jacobson perovskite-type materials AB₂Nb₃O₁₀ (A = K, Rb, Cs and B = Ca, Sr, Ba) and their photocatalytic activity for hydrogen production [38]. Seven different niobate were prepared by the molten salt method mixing BCO₃ or B(NO₃)₂, (B = Ca, Sr, Ba), Nb₂O₅, A₂CO₃ with ACl (A = K, Rb, Cs), and at specific reaction temperatures (750–1200 °C). UV/vis diffuse reflectance revealed that the band gap energies varied only with different B-cations, seeming to be independent of A-cations. In the case of Ca²⁺ as B-cation, the band gap value was 3.6 eV, whereas it was 3.3 eV for Sr²⁺ and 3.0 eV for Ba²⁺. This trend suggested a dependence on the ionic radii of the B-cations that filled the spaces among NbO₆ octahedral units. With bigger B-cation, the overlap of the Nb 4d orbitals was greater, stabilizing the conduction band energy and, as a result, the band was shifted negatively. The prepared niobates were evaluated for H₂ evolution in 10% methanol aqueous solution without a co-catalyst, and then Rh was in situ photodeposited (0.05 wt% Rh loading) on the catalyst. Without co-catalyst, the KCa₂Nb₃O₁₀ sample exhibited the highest H₂ evolution rate; however, after the Rh addition, the highest overall hydrogen production was achieved by CsCa₂Nb₃O₁₀. In this case, photocatalytic behavior was influenced by the type of the A-cation at the interlayer, suggesting a dependence on the ionic radii in the order Cs > Rb > K, with the exception of RbSr₂Nb₃O₁₀. The B-cation influenced the hydrogen generation rates owing to more cathodic conduction band edges in the order Ca > Sr > Ba.

Zhou and coauthors showed that the performance of KCa₂Nb₃O₁₀ for H₂ evolution could be enhanced through a nitrogen and tetravalent niobium doping to generate yellow and black [Ca₂Nb₃O₁₀]⁻ nanosheets [39]. After the preparation of KCa₂Nb₃O₁₀ by solid state reaction, the material was treated at 800 °C for 5 h under NH₃ flow, resulting in the black niobate (N⁻/Nb⁴⁺-codoped). The previous addition of a certain stoichiometric excess of K₂CO₃ prevented the formation of oxygen defect, and thus the reduction of Nb⁵⁺ ions, which resulted in the yellow niobate (N-doped). Then, the samples were acid-exchanged, followed by the exfoliation reaction with tetrabutylammonium hydroxide (TBAOH). X-ray absorption fine structure (XAFS) data on the Nb K-edge (18.986 keV) of all the samples indicate that doping did not change the perovskite-type crystal structure, as all three samples showed very similar overall profiles. However, an increase of the pre-edge peak (18.985 keV) of black niobate could be attributed to the partial reduction from Nb⁵⁺ to Nb⁴⁺ and the increased Nb–N bonding for yellow niobate, whereas the black one, showing a simultaneous decrease of both Nb–O and Nb–N, could point out the dominating generation of oxygen vacancies during the ammoniation process without K₂CO₃ additive. The samples were evaluated for photocatalytic H₂ production using 20 vol% methanol aqueous solution under full range irradiation with an Xe lamp. Undoped nanosheets show very low photocatalytic activity (11.4 mmol h⁻¹) compared with the activity of yellow and black niobate nanosheets (42.9 and 154.7 mmol h⁻¹, respectively). In this sense, the self-doping with Nb⁴⁺ could significantly promote an enhanced separation efficiency of photogenerated carriers as well as the light absorption beyond single N³⁻ doping. When Pt was loaded at 0.5 wt%, the H₂ evolution rates were three times higher, 190.1 and 429.5 mmol h⁻¹ for yellow and black niobate nanosheets, respectively. The last one could achieve energy conversion efficiency for solar hydrogen production of 2.7%.

Later, the same group could also investigate the combination of elemental doping, liquid exfoliation, and composition control for a series of typical Dion–Jacobson phases KCa₂Na_n−3Nb_nO_3n+1 photocatalysts [40]. As seen in the scheme presented in Figure 5, KCa₂Na_n−3Nb_nO_3n+1 were prepared with various values of n in order to tune the thickness of the perovskite layer. Theoretically, using n = 3–6, the desired thickness of the layer should increase from 2.4 to 4.0 nm. The materials were obtained from the molar ratio of K/Na/Ca/Nb = 1.05/2/1.05(n−3)/n through calcination at 1200 °C, followed by the N/Nb⁴⁺-doping via heat treatment under NH₃ flow. This method resulted in samples with different colors, varying from white to black. Samples before and after NH₃ treatment were analyzed by UV/vis diffuse reflectance, where red-shifts of the absorption edge were observed in all N/Nb⁴⁺ doped samples. Even after liquid exfoliation of bulk doped samples to obtain nanosheets, their suspension in water kept the colors from the bulk materials. TEM and atomic force microscopy (AFM) images revealed ultrathin sheets with several micrometers in size and thicknesses of 2.4, 2.8, 3.3, and 4.0 nm for the n = 3, 4, 5, and 6 homologues, respectively. The photocatalytic performance of the doped nanosheets was evaluated towards hydrogen evolution from 20 vol% methanol solution under full arc irradiation (Xe lamp, 300 W). Considering all samples with and without 0.5 wt% Pt, the n = 4 product exhibited the highest H₂ production rate. This fact can be related to the migration length, which affects the electron–hole separation step. The higher photocurrent measurement was also found for n = 4 with a sequence of n = 4 > 5 > 3 > 6, which is consistent of H₂ evolution experiments.

The photocatalytic activity of HCa₂Nb₃O₁₀ for hydrogen production was also extended for visible irradiation region by combination with CdS [41]. In this composite, after light excitation of CdS, the photo-generated electrons were transferred to the calcium niobate nanosheets thanks to their matched conduction band position of −1.32 eV for CdS and −1.18 eV for niobate (V vs. Ag/AgCl, pH = 7.0). So, the observed increase of the electron paramagnetic resonance (EPR) signal after illumination is the result of the efficient electron transfer that further induced the growth of oxygen vacancies. The samples were evaluated for photocatalytic H₂ evolution in (0.1 M) SO₃²⁻/(0.15 M) S²⁻ solution under visible light, in which the exfoliated niobate was not able to be excited. The best performance was achieved by samples containing 53.6 wt% of CdS with a H₂ production rate of 16.5 μmol h⁻¹, about four times higher than pure CdS. Further increase of CdS resulted in lower activity, which was related to the aggregation of CdS nanoparticles. Later, a similar approach was applied by Hu and coauthors to design nanohybrid of CdS and HCa₂Nb₃O₁₀ and get photocatalytic H₂ production under visible light [42]. It has been shown that CdS has more negative conduction band energy than the employed niobate, indicating the spontaneous transfer of photoinduced electrons from CdS to niobate nanosheets.

Xiong and coauthors could improve the photocatalytic hydrogen production activity of HNb₃O₈ nanosheets by loading Cu with a facile photodeposition method [43]. Exfoliated HNb₃O₈ niobate was dispersed in Cu²⁺ solutions with corresponding loadings of 2%, 1%, 0.5%, and 0.25% which was photodeposited in situ in triethanolamine solution. Firstly, the authors evaluated the samples for H₂ evolution activity with 10 vol% of triethanolamine aqueous solution under the simulated solar light using a 300 W Xenon lamp. The photocatalytic performance increased after addition of Cu²⁺ up to 0.5%, reaching the maximum rate of 59.1 µmol h⁻¹. With higher amounts of Cu, the shielding effect was detrimental for the photoactivity. In addition, when methanol was used as the sacrificial agent, the photocatalytic hydrogen production activity achieved the value of 98.2 µmol h⁻¹. The authors argued that in situ photodeposition induced selective deposition on the photocatalytic reduction sites of HNb₃O₈ nanosheets rather than arbitrary distribution.

The same type of niobate was recently applied to engineer a stable nanocomposite photocatalyst. Xia and coauthors reported a method to growth NiS in a highly dispersive way on the HNb₃O₈ surface [44]. In general, the random deposition and easy agglomeration in large particles of NiS are described as being problematic for the photocatalytic H₂ evolution. Therefore, an electrostatic adsorption/self-assembly methodology was developed to obtain NiS on the HNb₃O₈ nanosheet as an efficient photocatalyst for the first time. The photocatalytic H₂ evolution was performed with a 300 W Xenon lamp in a 10% triethanolamine (TEOA) aqueous solution, where the bare NiS and layered KNb₃O₈ showed no or very low H₂ production. In this condition, for the composite with 1 wt% NiS, the H₂ evolution rate was 1519.4 µmol g⁻¹ h⁻¹, higher than the physical mixture of the components (582.5 μmol g⁻¹ h⁻¹). Time-resolved fluorescence decay analysis was applied to calculate the lifetime of the photogenerated electrons and holes. The kinetic data showed a slower decay for the composite, which indicates that the applied methodology resulted in more efficient utilization of photogenerated electrons and holes.

The potential of niobates structures for water splitting has also been demonstrated by some theoretical works. The perovskite-type niobate NaNbO₃ was selected by Wang and coauthors to investigate the effect of anionic monodoping with N, C, P, and S dopants, as well as with (N + N), (C + S), and (N + P) codoping pairs by hybrid density functional theory calculations [45]. At first, the direct band gap of pure cubic NaNbO₃ was predicted to be 3.30 eV, with the conduction band mainly formed by Nb-4d orbitals and valence band by the O-2p orbitals, and almost no Na-related states around the band edge were found, indicating that the Na atoms have negligible effects on the electronic structures near the Fermi level. In this sense, one of the O atoms in the cell was firstly substituted by one of four different species (N, C, P, and S), corresponding to a doping concentration of 2.5%. In the case of N-doped NaNbO₃, a hole would be generated as the N dopant has less valence electrons than the O atom. The higher energy of N-2p states (around 2.0 eV) induced a localized unoccupied impurity state above the Fermi level as well as several occupied impurity states around the valence band of NaNbO₃. For C doping, the substitution would result in two holes in the system, however, no unoccupied localized impurity states were found above the Fermi level. As a result, the total magnetic moment was higher and there were several impurity states above the valence band, which shifted up by about 1.58 eV. The doping with P resulted in a spin-polarized ground state as well, and, owing to the P-3p higher energy than the O-2p, several empty impurity states were found near the conductive band. When S was applied as a dopant, no empty states were found within the band gap, but several localized impurity states were localized below the Fermi level, right above the valence band of the niobate, which resulted in a narrowed band gap. This filled impurity states originated from the hybridization of S-3p and O-2p orbitals. Thus, the band gap values for N-, C-, P-, and S-doped NaNbO₃ were calculated to be 1.88, 1.61, 1.04, and 2.34 eV, respectively. Despite the narrower band gap energies, some disadvantages were observed as the undesirable empty gap states from N and P monodoping, which may act as trapping and recombination centers, and the elevated valence band of NaNbO₃ after C doping, precluding the oxygen evolution reaction. In this sense, the codoping possibility using pairs of (N + N), (C + S), and (N + P) was also investigated. It resulted in the band values as shown in Figure 6. In comparison with the pristine niobate, for (N + N) codoping, two fully filled states were generated within the band gap and its effective energy was 1.24 eV lower. The (C + S) dopping led to mid-gap states above the valence band of host system, and then a narrow band gap of 1.42 eV. For both systems, the band edges matched properly with the redox potentials of water, indicating that the photocatalyst would be suitable for overall photocatalytic water splitting. However, in the case of (C + S) codoping, the conductive band was shown to no be longer good for a spontaneous water splitting process even with a band gap value of about 1.33 eV. This study suggested that codoping can be a successful approach to improve the visible light photocatalytic performance of perovskite NaNbO₃.

Kaneko and coauthors deeply investigated the origin of the visible light absorption and the optical band gap of 1.9 eV in the d1 metallic strontium niobate (SrNbO₃) [46]. By computational methods, firstly, they could verify at which electronic states the photoexcitation takes place, from CB → B₁ transition or B₋₁ → CB (B₋₁ denotes the band below the conductive band (CB), and B₁ denotes the band above the conductive band). By the band structures and projected density of states calculations, it was found that B₋₁ state lies at around −4 eV and consists of O(p). Thus, from the atomic orbitals constituting each band, it was inferred that the optical gap corresponds to the Nb(d) → Sr(d)/Nb(d) transition, and so the optical gap was attributed to the CB → B₁ transition. The shifts of Fermi level as well as the changes in the optical gap concerning the Sr defects and O vacancies were investigated. The authors could conclude that the optical gap hardly depends on the amount of Sr defects. Sr and O are divalent cation and divalent anion, respectively, and the valence of the Nb ion constituting CB did not change even if the Sr defect and the O vacancy were produced in the same amount. Thus, if the valence of the Nb ions constituting CB does not change, the Fermi level will not shift either. Consequently, the optical gap, which is mainly caused by the excitation from the Fermi level to B₁, does not change by Sr_1−xNbO_3−x. In addition, in the case of Sr_0.875NbO₃ and Sr_0.875NbO_2.875, it was found that the latter evidenced the dependence of the optical gap on the amount of Sr defects and its light absorption intensity was larger than that of the former.

Such doped niobates have been prepared and evaluated as photocatalysts. Yu et al. reported at first time the performance of carbon-doped KNbO₃ as well as the effect of MoS₂ addition in the material for photocatalytic H₂ evolution [47]. Both pristine and C-doped KNbO₃ presented the orthorhombic phase, however, the latter resulted in peaks of 30–32° shifted to a lower angle, indicating that the carbon atom was doped into the lattice of KNbO₃. In addition, the C species were identified as C⁴⁺ by X ray photoelectron spectroscopy (XPS) analysis, which meant that it did not replace the position of lattice oxygen. In addition, it was also confirmed that the substitution of Nb⁵⁺ by C⁴⁺ occurred and also generated some oxygen vacancies to compensate the charge balance. The pristine niobate showed only UV light absorption with band gap of 3.07 eV. C-doped niobate presented very similar behavior, however, it showed a difference in the peak tailing in the range of 400–600 nm. The samples after the addition of MoS₂ in 0.2 wt% also presented the peak tailing in the visible-light region. H₂ evolution was performed in methanol solution (20% v/v) under 300 W Xe lamp irradiation and in situ photodeposition of metallic Pt (0.37 wt%). The C doping of KNbO₃ could enhance the H₂ evolution from 5 µmol g⁻¹ h⁻¹ to 142 µmol g⁻¹ h⁻¹ in comparison with the pristine sample. The decoration of MoS₂ on C-KNbO₃ could further increase the rate of H₂ evolution to 1300 µmol g⁻¹ h⁻¹. The samples were also evaluated under visible light, where pure niobate did not show activity, while C-KNbO₃ and MoS₂/C-KNbO₃ resulted in rates of 4.2 µmol g⁻¹ h⁻¹ and 9.3 µmol g⁻¹ h⁻¹, respectively. These activities were much lower compared with values under UV/vis light, indicating that the composite is more suitable for working in the UV region of irradiation.

Columbites are also founded in the literature with further applications in photocatalytic research. The effect of structural distortion of SnNb_xO_y in comparison with the crystalline SnNb₂O₆ regarding the photocatalytic activity was investigated by Huang et al. [48]. The crystalline sample exhibits a band gap energy of 2.10 eV, while the band gap energies for disordered and amorphous samples were 2.33 eV and 2.43 eV, respectively. Furthermore, the electronic band gap could be tuned by the structural disorder engineering, as an increase of structural lattice disorder led to an increase of band gap energies. By XPS and UV/vis absorption analysis, the valence band edge of crystalline SnNb2O6 was about 1.42 V with the conduction band of −0.68 V, whereas for the disordered sample, the values were 1.98 V and −0.35 V, respectively. As a consequence, the latter exhibited H₂ evolution rates 11 times and 1.7 times higher than that of crystalline and amorphous samples, respectively, under visible light irradiation (λ ≥ 420 nm). The photoinduced charge carrier behaviors under pulse laser at 532 nm were investigated by transient absorption spectroscopy (TAS). The three samples presented broad and continuous absorption in the visible range at a delay time of 100 ns, indicating the separation of charge carries at different trap states. This behavior was especially pronounced in disordered SbNb_xO_y, which indicated a better charge separation. The transient absorption decay kinetics also expressed the same, with effective lifetimes for the amorphous, disordered, and crystalline samples of 0.52 ms, 0.56 ms, and 0.26 ms, respectively.

The influence of type of polymorph on the photocatalytic activity of CuNb₂O₆ columbite was further explored by Kamimura et al. [49]. Two different polymorphs, monoclinic and orthorhombic, revealed that the local crystal structure of the CuO₆ octahedral was strongly correlated to the optical absorption property. Both samples exhibited the same band gap, 2.7 eV. However, two additional broad bands in the near-infrared (NIR) region were observed for the orthorhombic sample, being ascribed to Cu²⁺ d–d transitions in a distorted octahedral ligand field. In principle, this transition is forbidden by the Laporte selection rule owing to the symmetric CuO₆, but in the case of the orthorhombic phase, it was partially allowed because of the Jahn–Teller effects of orthorhombic CuNb₂O₆. The photocatalytic H₂ evolution was performed using a 10 vol% methanol solution and an Air Mass (AM) 1.5G solar light irradiation. The two polymorphs of CuNb₂O₆ produced H₂ under these conditions, but the monoclinic sample was more active than the orthorhombic one. Thus, to elucidate if the d–d absorption affects the photocatalytic activity, the previous results were compared to those using a near-infrared (NIR) cutoff filter. When the monoclinic sample was tested, the rate of H₂ evolution rate did not change expressively when the NIR filter was applied. In this case, the loss of 20% of activity could be explained by the low transmittance of the employed cutoff filter in the visible region. In contrast, the active for orthorhombic CuNb₂O₆ decreased 90% in relation to that observed without an NIR filter. On contrary to the monoclinic phase, which was insensitive to the excitation of the d–d transition in Cu²⁺ ions, for the orthorhombic one, the electron 3d⁹ configuration of Cu²⁺ readily induces structural distortion by Jahn–Teller effects in an octahedral crystal field. The empty 3d orbitals of Cu²⁺ could work as deep traps for the photo-generated electrons, and thus increase the recombination rate, resulting in a significant decrease in photocatalytic activity in the presence of an NIR cut-off filter. With NIR photons, the photo-generated electrons filled 3d orbitals of Cu²⁺ via d–d absorption, a state that inhibited the recombination of photogenerated electrons in the conduction band. These distinct mechanisms were schematically represented on Figure 7. So, in the case of orthorhombic structure, the photocatalytic performance was dependent on fully filled Cu 3d-orbitals.

Chun and coauthors employed a two-step hydrothermal process to form nanosized ZnNb₂O₆ columbite [50]. The samples were applied to photocatalytic H₂ evolution in 20 vol% methanol under irradiation of 500 W high-pressure mercury lamp. The sample prepared by the new method produced an optimal activity of 23.6 µmol H₂ h⁻¹ g⁻¹, while a bulk ZnNb₂O₆ prepared by solid state exhibited a H₂ evolution rate of 9 µmol h⁻¹ g⁻¹. This fact was attributed to its appropriate crystallinity and high specific surface area (61 m² g⁻¹). Further, the authors evaluated the effect co-catalyst deposition on nano-ZnNb₂O₆. In this case, both pristine water and methanol aqueous solution media were applied. In pure water, the bare photocatalyst showed an undetectable amount of H₂. The optimal co-catalyst was found to be Pt, as the H₂ generation rate achieved up to 680 and 3200 µmol h⁻¹ g⁻¹ in pure water and methanol aqueous solution, respectively. Moreover, with the modified sample and at optimal conditions, a maximal apparent quantum yield (AQY) value of 4.54% was obtained in pure water. The value increases to 9.25% with the assistance of methanol as the sacrificial agent.

The manipulation of composition, morphological, and surface properties of three-dimensional hierarchical Nb₃O₇(OH) superstructures was achieved through Ti(IV) incorporation, as reported by Betzler and coauthors [51]. Firstly, the undoped Nb₃O₇(OH) consisted of blocks of corner-sharing NbO₆ octahedra, forming hollow cubic superstructures of nanowire networks (Figure 8a). When an average amount of 5.5% titanium was incorporated into the crystal lattice, it formed a spherical morphology, hollow and built up from nanowires as well (Figure 8b). With a higher amount of Ti(IV) (10.8%–31.2%), a cubic morphology with smaller nanowires arranged to form the walls of hollow cubes was observed (Figure 8c). All the samples presented the same nanowire arrangement; however, it became smaller, shorter, and flatter as the quantity of Ti was increased. Both XRD and energy-dispersive X ray (EDX) measurements indicated a homogeneous distribution of titanium, however, with more than 12% Ti, the crystal lattice of Nb₃O₇(OH) was not able to accommodate it, and then the formation of niobium-doped anatase TiO₂ plates was favored for further titanium excess. The specific surface area was also affected and it changed from 79 m² g⁻¹ for Nb₃O₇(OH) (morphology a) to 132 m² g⁻¹ and 173 m² g⁻¹ for morphologies b and c, respectively. The three samples showed the same band gap energy of 3.2 eV, but an additional absorbance in the 550−850 nm region was observed, which came from oxygen vacancies in the TiO₂ plates. Further, the samples were evaluated to H₂ evolution under an Xe lamp irradiation (600 mW cm⁻²), with 10 vol% methanol and in situ photodeposited Pt at 8 wt%. The Nb₃O₇(OH) resulted a rate of 870 μmol g⁻¹ h⁻¹, and then, as the Ti concentration increases, the H₂ rates zre raised to 1773 μmol g⁻¹ h⁻¹ for morphology b and 1988 μmol g⁻¹ h⁻¹ for morphology c. The same trend was also expressed by transient absorption measurements, where morphology a showed an averaged lifetime of 54 ± 2 ps, while morphologies B and C presented lifetimes of 64 ± 2 ps and 73 ± 4 ps, respectively. The authors explained this behavior by the fewer amounts of hydroxyl groups required for charge neutrality with the replacement of Nb⁵⁺ by Ti⁴⁺, as they are known for acting as non-radiative recombination sites. The doubled H₂ production rate of morphology b, compared with morphology a, was most likely the result of a combination of its fewer recombination sites and larger surface area. The further, but less pronounced increase of the H₂ production rate for morphology c came from the presence of TiO₂ plates, which did not have such a strong effect on the photocatalytic reaction.

It is clearly noted that niobates offer the possibility to be combined with innumerous co-catalysts. Previous cited works and other examples from the literature are summarized in

Table 1. Summary of the photocatalytic activity of niobates combined with distinct materials towards H₂ evolution from aqueous solution.

Niobate	Co-Catalyst	Light Source	Sacrificial Reagent	H2 Formation Rate	Ref.
HCa2Nb3O10	CdS, Pt	300 W Xe, >400 nm	Lactic acid	52.6 μmol h−1	[41]
HCa2Nb3O10	CdS	300 W Xe, >420 nm	methanol	450 μmol g−1 h−1	[42]
HNb3O8	Cu2+	300 W Xe	methanol	98.2 µmol h−1	[43]
HNb3O8	NiS	300 W Xe	triethanolamine	1519.4 μmol g−1 h−1	[44]
C-doped KNbO3	MoS2, Pt	300 W Xe	methanol	1300 µmol g−1 h−1	[47]
H1.78Sr0.78Bi0.22Nb2O7	Ni-CH3CH2NH2	300 W Xe	methanol	372.67 μmol h−1	[52]
KNb3O8	g-C3N4	1000 W Hg λ > 400 nm	dimethylhydrazine	25.0 µmol h−1 g−1	[53]
K3H3Nb10.8O30	g-C3N4	1000 W Hg λ > 400 nm	dimethylhydrazine	27.0 µmol h−1 g−1	[53]
HNb3O8	g-C3N4	1000 W Hg λ > 400 nm	dimethylhydrazine	37.0 µmol h−1 g−1	[53]
Ca2Nb2TaO10	g-C3N4, Pt	300 W Xe, >400 nm	triethanolamine	43.54 µmol h−1	[54]
Ba5Nb4O15	g-C3N4	3 W LEDs 420 nm	oxalic acid	2673 μmol h−1 g−1	[55]
AgNbO3	g-C3N4, Pt	300 W Xe, >420 nm	methanol	88.0 µmol g−1 h−1	[56]
KTa0.75Nb0.25O3	g-C3N4, Pt	300 W Xe, >420 nm	methanol	86.2 μmol·g−1·h−1	[57]

.

Besides the previous cited materials, niobium is also flexible in terms of designing different layered compounds with distinct elements. Fujito and coauthors reported a visible light response of layered perovskite oxychloride Bi₄NbO₈Cl for water splitting [58]. This material consists of single-layer NbO₄ perovskite units separated by (Bi₂O₂)₂Cl blocks. It presented a band gap of 2.39 eV with valence and conductive bands of 2.11 eV and −0.28 eV, respectively. Experimental and theoretical results suggested that the valence band was formed by highly dispersive O-2p orbitals coming from the interactions within and between Bi−O and Nb−O slabs. As expected from the conductive band value, Bi₄NbO₈Cl showed activity for H₂ evolution from an aqueous methanol solution under UV irradiation (300 W Xe lamp) and Pt (0.5 wt%) loading. However, the obtained rate of ~0.1 μmol h⁻¹ was considered low and a higher performance for O₂ was found. Apparent quantum efficiency for O₂ evolution was determined as ~0.4% with FeCl₃ as an electron acceptor, under monochromatic light at 420 nm (~25 mW/cm²). So, through a Z-scheme mechanism, the system was coupled to a H₂-evolving photocatalyst of Rh-doped SrTiO₃ and Fe³⁺/Fe²⁺. In this way, as in the schematic shown in Figure 9, a simultaneous evolution of H₂ and O₂ under visible light was successfully observed.

Wakayama and coauthors reported a novel compound obtained from a mixture of layered RbNdNb₂O₇ and Rb₂CO₃ [59]. The new layered niobium oxynitride, Rb₂NdNb₂O₆N, resulted in a structure as Rb₂NdNb₂O₆N H₂O composed by double-layer [NdNb₂O₆N]²⁻ perovskite slabs separated by two Rb cations and one H₂O molecule. The material showed a band gap of 2.5 eV with visible light absorption, whereas the precursor RbNdNb₂O₇ essentially absorbs UV light with a band gap energy of 3.7 eV. In the former case, additional N 2p orbitals in the valence-band formation caused a valence-band maximum that was much more negative. In contrast, as the conduction band consists mainly of Nb 4d orbitals, its value remained basically unaffected. Thus, the H₂ evolution was evaluated in the presence of TEOA as an electron donor in dimethyl sulfoxide (DMSO) containing 1 mL of water. The sample was irradiated with visible light (λ > 400 nm) by 300 W xenon lamp, and Pt was previously deposited at 0.5 wt%. As shown in Figure 10, a stable H₂ evolution was observed for 20 h from the Rb₂NdNb₂O₆N sample, in contrast to the precursor, which did not show a response under visible light irradiation.

Therefore, it can be observed that niobium layered compounds form a versatile Nb-based materials class that exhibits great activity for H₂ evolution reaction. Additionally, several procedures to tuning the photocatalytic performance niobium layered compounds were discussed, such as changes in the structural and morphology, doping and co-doping, and combination with other semiconductors.

2.3. Nb-Doped Materials

Two of the main drawbacks in the photocatalysis heterogeneous are the rapid recombination of electron/hole pair and semiconductor activation only under UV irradiation. A strategy to increase the photocatalytic performance of semiconductors by overcoming these challenges is the introduction of impurities able to change the electronic behavior of the material, in one word, the doping. Niobium is well known in the literature as a dopant agent, especially for titanium-based semiconductors, as its higher charged ion can substitute the Ti⁴⁺ site in the oxide structure, and simultaneously play the role of donor, improving carrier concentration and conductivity [60]. For TiO₂, the defect state provided by the doping is located upon the conduction band minimum and contributes electrons to the unoccupied Ti 3d orbital without introducing additional states in the bandgap. However, in order to satisfy the charge compensation, the substitution can be accompanied by the incorporation of other defects as oxygen vacancies or even Ti³⁺ defects below the conductive band, which lead to a visible range absorption [61]. In

Table 2. Summary of Nb-doped photocatalysts employed for H₂ evolution from aqueous solution.

Semiconductor	Co-Catalyst	Light Source	Sacrificial Reagent	H2 Rate	Ref.
TiO2	-	300 W Xe > 420 nm	ethanol	1146 µmol g−1	[15]
TiO2	Pd	500 W Hg-Xe 420–680 nm	methanol	~0.6 mmol g−1 h−1	[62]
TiO2	Pt	500 W Hg-Xe 420–680 nm	methanol	~0.1 mmol g−1 h−1	[62]
TiS3	-	200 W halogen	Na2SO3 (electrolyte)	2.2 μmol min−1·cm2	[63]
Cu5Ta11O30	Pt	1000 W Xe arc > 420 nm	methanol	6 μmol h−1	[64]
Ta3N5	-	400–700 nm	water	41.4 µmol g−1 h−1	[65]
KTaO3	Pt	300 W Xe	methanol	728 µmol g−1 h−1	[66]
BiVO4	-	100 mW cm−2Xe	NaHCO3 (electrolyte)	0.18 mmol h−1	[67]

, examples of Nb-doped semiconductors for H₂ evolution are listed.

The performance of the Nb-TiO₂ photocatalyst was evaluated using various sacrificial agents such as methanol, ethanol, and 2-propanol by Fontelles-Carceller [68]. The photocatalyst was prepared with an Nb/Ti ratio of 0.025:0.975, which was confirmed by XPS measurements. In principle, the presence of Nb does not significantly change the band gap (3.15 eV) or the specific surface area (80.5 m² g⁻¹) of the photocatalyst in comparison with pure TiO₂ (3.18 eV of band gap and 82.1 m² g⁻¹ of surface area). The H₂ evolution was investigated by the photocatalysts loaded with 0.5 wt% of Pt in presence of a 3:7 (v/v) alcohol/water mixture under irradiation of an Hg-Xe lamp (500 W). For both TiO₂ and Nb–TiO₂, the maximum rates were observed for methanol, followed by ethanol and 2-propanol. In all tested conditions, as shown in Figure 11, the doped sample showed a significantly better result than the non-doped sample. Both samples presented maximum rates for methanol, followed by ethanol and 2-propanol. In addition, the products formed during the photocatalytic reaction were investigated and, for methanol and ethanol, basically the same compounds were identified. Both samples generated formic acid and metyl formate by methanol oxidation with the addition of formaldehyde for the Nb-doped sample under UV illumination. When ethanol was applied, in all conditions of irradiation and applied samples, acetaldehyde and ethyl acetate were found. Nevertheless, when 2-propanol was applied as a sacrificial reagent, the Nb-doped sample generated acetone and propanone–diisopropyl–acetal, whereas bare TiO₂ produced only acetone. So, it was possible to affirm that the presence of Niobium at the surface even in a small proportion was able to generate a new compound not observed for the pristine sample. In this sense, in situ infrared studies were also carried out under dark and illumination conditions and in the presence of the alcohol in water vapor. In principle, under the dark, it was possible to verify the adsorption of the alcohols on the surface of both samples. Under UV irradiation, the infrared spectra showed that alcohol and alcoxy species were consumed and/or desorbed from the surface of the samples. For methanol and ethanol, carboxylate/carbonate species were found as the main surface product. This could indicate the attack from hole to the alcohol to form oxidized species related to the corresponding alcohol/alcoxy adsorbed species. In the case of 2-propanol, carboxylate species were less present at the surface owing to the low coverage of products from the relatively easy desorption of acetone. Regarding the H₂ evolution, the authors suggested that the differences in activity are correlated quantitatively with the average oxidation potential of the alcohol. When comparing methanol with ethanol using infrared spectroscopy, they showed essentially the same behavior of the titania surfaces against the alcohol and evolving molecules coming from the hole-attack.

Different materials that were doped by niobium ions to make them able to split water have been reported in the literature. The improvement of WO₃ for water splitting by photoelectrochemical process was achieved by Nb-doping [69]. The morphology of the material turned from nanorod to nanotriangle, which also increased the porosity of the WO₃ films. Also, the band gap increased to 2.74 eV by Nb-doping in comparison with WO₃ (2.63 eV) and additional oxygen vacancies were induced. Besides that, the carrier densities significantly increased from ~3.44 × 10¹⁹ cm⁻³ to 1.27 × 10²² cm⁻³, indicating that Nb⁵⁺ ions act as donor dopants in the WO₃ lattice. It could improve the charge separation and electron transport, and then the doped sample exhibited higher photocurrent, incident photon to charge carrier efficiency (IPCE) (39% to 52% at 300 nm and 1.23 V vs. RHE), and photoconversion efficiencies.

Also, with the objective to promote photoelectrochemical water splitting, niobium could support the activity of BiVO₄ films [67]. In this case, the authors concluded that Nb⁵⁺ ions were not placed in the crystalline structure of BiVO₄, rather as niobium oxide or oxochloride. Firstly, the addition of 10% of Nb(V) resulted a higher IPCE, which reached 75% in comparison with pristine BiVO₄ of 55% at 2.08 V versus RHE (380–550 nm). The H₂ evolution experiments were run by an H-shaped cell with pristine or modified BiVO₄ film as a photoanode coupled with a cathode of Pt and carbon paste. After 16 h of irradiation at 2.08 V, the pristine sample reached a maximum H₂ evolution rate of 0.09 mmol h⁻¹, whereas Nb-modified BiVO₄ increased about two times to 0.18 mmol h⁻¹. The authors argued that the higher performance of the modified sample would be related to the modification of the nanostructure of the film, but also owing to other effects not obvious by the obtained data. A high efficiency for the photoelectrochemical water splitting reaction was achieved by doping of SnO₂ nanotubes (NTs) with Nb and N co-doping [70]. Niobium could be ascribe as an effective dopant owing to its abundant electronic states and lower ionic radii of Nb⁴⁺ (69 pm) compared with Sn⁴⁺ (71 pm), resulting in minimum lattice strain. As shown in Figure 12a, the addition of Nb could improve the absorption of light at lower wavelengths, suggesting a decrease in the band gap of the doped samples, with an increase in Nb content. Then, a further doping with N could result in a significant decrease of band gap to 1.99 eV from 3.1 eV of (Sn_0.95Nb_0.05)O₂. The lowering of the optical band gap for this sample could provide an IPCE of 10% at 500 nm and potential of ~0 V (vs. RHE), which was described by the author as the highest IPCE value obtained for semiconductor materials explored so far for photoelectrochemical (PEC) water splitting. The generated H₂ and O₂ gases at the cathode (Pt wire) and the photoanode are shown in Figure 12b. The theoretical concentration of H₂ from Faraday’s Law was also calculated, which was very similar to the experimental one, suggesting a Faradaic efficiency close to 100%. Besides that, the ratio of produced gases was around 2, showing a stoichiometric decomposition of water into H₂ and O₂. From these as well as the theoretical results, it was possible to conclude that the Nb and N co-doping in SnO₂ afford better light absorption properties, increased carrier density, and facile electrochemical charge transfer.

3. Photoreduction of CO₂

Artificial Photosynthesis

Carbon dioxide (CO₂) emission from fossil fuels is the main concern when discussing global warming and related climate changes. Because the global energy demand will continue to increase, the consumption of fossil fuels will continue to be high for the next few decades, which undoubtedly brings the necessity of large-scale CO₂ mitigation from the atmosphere [71]. The observation of natural photosynthesis in green plants motivated the development of photocatalytic CO₂ reduction (artificial photosynthesis), which is one promising solution to transform the current fossil fuel-based economy into CO₂-neutral energy systems using the sustainable “photon” economy and mitigate the greenhouse gas effect [3,72,73,74]. Another advantage of the photocatalytic CO₂ reduction is the indirect solar energy storage on carbon-based molecules. Therefore, the artificial photosynthesis acts as killing two birds with one stone in view of protecting the environment and simultaneously supplying renewable energy. However, CO₂ is very stable/inert regarding chemical reactions, meaning that CO₂ conversion demands a considerable amount of energy; therefore, it is quite difficult to convert it to reusable chemicals [75].

In the beginning of the 1900s, Giacomo Ciamician published a seminal paper with the first idea about the artificial photosynthesis, with the suggestion that society should shift from consuming fossil fuels to generating sustainable energy from the sun [76]. Later, Inoue et al. reported the photoelectrocatalytic reduction of CO₂ to form organic compounds such as formic acid, formaldehyde, methyl alcohol, and methane. In the presence of various semiconductors, the dependence between the formed products and the semiconductor conduction band energy was demonstrated [77].

The CO₂ photoreduction process is a multielectron process (2, 4, 6, or 8 electrons) that can yield diverse products and occurs through different steps that involve adsorption, activation, and dissociation of the C=O bond. The adsorption and activation step is especially challenging because CO₂ is highly stable and inert [78]. In

Table 3. Common products from CO₂ reduction, their simplified half reactions, and redox potentials at pH = 7.

Reactions	E (V vs. RHE)
CO2 + 2e− + 2H+ → CO + H2O	−0.53
CO2 + 2e− + 2H+ → HCOOH	−0.61
CO2 + 4e− + 4H+ → HCOH + H2O	−0.48
CO2 + 6e− + 6H+ → CH3OH + H2O	−0.38
CO2 + 8e− + 8H+ → CH4 + 2H2O	−0.24

, some of the possible CO₂ reduction reactions are summarized along with their respective redox potentials. Selected potentials are compared with Nb-based photocatalysts at Figure 13. The formation of first intermediate (CO₂^•−) state imposes a significant energy to the reaction and is frequently identified as the rate determining step. Once CO₂^•− is formed, it may be subsequently reduced via the protonation of its oxygen atom, resulting in the formation of ^•COOH. This intermediate can be reduced to CO and released from the semiconductor surface. Alternatively, CO₂^•− may also be reduced via the protonation of its carbon atom to form HCOO^•, which is further reduced to formate (HCOO⁻) [79,80,81]. Therefore, the CO₂ reduction mechanism is very complex and imposes kinetic and thermodynamic limitations.

The semiconductors that have been investigated for CO₂ conversion present low activity, selectivity, stability, and high band gap values; therefore, challenges remain on how to make this process feasible. The CO₂ reduction reaction competes with H₂ evolution in protic media, thus efficient catalysts must exhibit high activity for CO₂ conversion with low or negligible activity for H₂ evolution. This requires a very high activation energy barrier for H₂ evolution or a change in the reaction pathway to slow down the kinetics of H₂ formation [82].

Few semiconductors have been identified with high surface affinity to CO₂ and, simultaneously, with suitable electronic properties for CO₂ reduction via a photo-activated mechanism. In this sense, Nb-based materials are promising for application in CO₂ photoreduction owing to their properties, such as high acidity, excellent textural properties, and interesting electronic properties. Additionally, it is well-known that the product selectivity is significantly affected by the nanostructure design and the architecture of the photoactive components [83].

Efficient CO₂ reduction involves the optimization of synthetic procedures and the development of different architectures in order to achieve the structural, electronic, morphological, and surface properties required for this application. Despite that the application of Nb-based semiconductors for CO₂ photocatalytic reduction has only recently been explored, some reports have revealed the potential of these structures for artificial photosynthesis. It can be highlighted that some materials, such as alkali niobates (NaNbO₃ and KNbO₃), columbite niobates (CuNb₂O₆ and ZnNb₂O₆), niobium oxide (Nb₂O₅), and niobic acid (HNb₃O₈). Additionally, Nb-based materials have been involved in some strategies to improve the CO₂ photocatalytic reduction activity and selectivity, such as doping and formation of composites

Shi et al. [84] evaluated the CO₂ photocatalytic performance of NaNbO₃ with different morphologies for the first time. NaNbO₃ in orthorhombic phase were prepared by a conventional solid-state reaction and hydrothermal method, with micron-sized particles and homogeneous nanowires, respectively. It was observed that NaNbO₃ decorated with Pt was efficient to convert CO₂ in CH₄, while there was almost no CH₄ to be detected over pristine NaNbO₃. Additionally, the CO₂ reduction performance of Pt–NaNbO₃ nanowires was higher than the micron-sized particles, owing to the higher specific surface area of Pt–NaNbO₃ nanowires.

It is well-known that the photocatalytic reactions (oxidative or reductive processes) are particularly affected by the semiconductor crystal structure [85]. Therefore, the CO₂ photocatalytic reduction performance of the cubic and orthorhombic NaNbO₃ was evaluated. The samples were obtained using inorganic and organic Nb-precursors by a furfural alcohol-derived polymerization–oxidation [85]. Both samples exhibited the same morphology features, that is, cuboids particles with a size between 30 and 50 nm, likely owing to the use of poloxamers with surfactant behavior (Figure 14). The cubic Pt–NaNbO₃ presented lower band gap energy than the orthorhombic Pt–NaNbO₃. Cubic NaNbO₃ exhibited a better photocatalytic performance for H₂ evolution and CO₂ reduction for CH₄ than orthorhombic NaNbO₃; the authors attributed this behavior to its unique electronic structure, which benefits the generation and migration of photo-generated electrons and holes. This behavior was confirmed for the same group in another publication [86]. The authors also investigated the effect of annealing temperature in the CO₂ reduction performance, and a deleterious effect with the increase in the temperature was observed.

On the basis of previous reports, Fresno et al. evaluated the CO₂ photoreduction performance of NaNbO₃ obtained by solid state reaction at 900 °C and compared it with NaTaO₃ photoactivity [87]. Owing to the high annealing temperature, orthorhombic NaNbO₃ with low surface area was obtained (~1 m²·g⁻¹). The main CO₂ reduction products were CO, CH₄, and CH₃OH, together with H₂ coming from the concomitant reduction of water competing with that of CO₂. It was observed that NaNbO₃ and NaTaO₃ give rise to similar conversions in the CO₂ reduction reaction. Additionally, they demonstrated that NaNbO₃ was photoactive even without a co-catalyst, although the main product was CO; the authors attributed this difference in the selectivity to the absence of a co-catalyst (Pt) [87].

Inspired in the Degussa P25, which consists of anatase and rutile TiO₂ (i.e., heterostructure) and shows higher activities than either pure anatase or rutile in water splitting and gaseous pollutant photodegradation [88], Ye’s group proposed the investigation of heterojunction formation between NaNbO₃ in cubic and orthorhombic phases for the CO₂ photocatalytic reduction [89]. The NaNbO₃ samples were synthesized at 400 °C to 600 °C based on the polymerized complex method. Pure cubic and orthorhombic NaNbO₃ were obtained at 400 and 600 °C, respectively, while the NaNbO₃ samples with mixed phases were formed at the temperature ranging from 400 to 600 °C (Figure 15b,c). Along with the phase transition from the cubic phase to the orthorhombic phase, the absorption edge of NaNbO₃ shows a slight and gradual blue shift, which is consistent with the previous report that cubic NaNbO₃ has a narrower band gap than orthorhombic NaNbO₃ (Figure 14c). Additionally, the presence of heterojunctions between the two phases was verified; the smaller ones less than 10 nm (cubic phase) and the bigger ones about 30 nm (orthorhombic). For the CO₂ photocatalytic reduction experiment, Pt was used as a co-catalyst and CH₄ was the main product. All the mixed-phase samples were more efficient in CH₄ formation than the pure phases. The authors state that the photo-excited electrons migrate from the conduction band of the orthorhombic phase to the trapping sites on the cubic surface, thus avoiding the electron–hole recombination in orthorhombic NaNbO₃ and improving the charge separation efficiency in the mixed-phase NaNbO₃ samples.

Niobates materials are also efficient photocatalysts for CO₂ photoreduction, because a layered materials configuration is favorable for the separation and transportation of photo-excited carriers (i.e., electrons and holes) [90,91]. In this sense, niobic acid (HNb₃O₈) is a promising material, as it exhibited the valence band and the conduction band edges at desirable potential levels. The layered structure facilitates electron transfer, and the protonic acidity is favorable for water adsorption through hydrogen bonding [91]. It is expected that these properties should exert a certain positive impact on CO₂ photoreduction by water [91].

Therefore, Li et al. [92] evaluated the photocatalytic CO₂ reduction performance of HNb₃O₈ obtained through two different methods, hydrothermal (HT) and solid state reaction (SSR), and compared it with that of KNb₃O₈. HNb₃O₈ is isostructural with KNb₃O₈, and both samples were crystallized in an orthorhombic symmetry. The samples obtained through the HT method exhibited thin nanobelts with several micrometers in length, while the samples obtained by the SSR method showed containing irregular particles in micron size. The surface area value increased from 2.7 m² g⁻¹ of KNb₃O₈ (SSR) to 28.8 m² g⁻¹ of KNb₃O₈ (HT), and from 6.5 m² g⁻¹ of HNb₃O₈ (SSR) to 39.4 m² g⁻¹ of HNb₃O₈ (HT). CH₄ was the only one hydrocarbon product detected in the gas phase during CO₂ photoreduction for all samples. The rate of CH₄ formation over KNb₃O₈ (HT) was seventeen times higher than that over KNb₃O₈ (SSR), and the rate of CH₄ formation over HNb₃O₈ (HT) was about twenty times higher than that over HNb₃O₈ (SSR). Additionally, the solid acid sample (HNb₃O₈) exhibited higher activity than the corresponding potassium salt (KNb₃O₈). Also, both samples exhibited higher activity than degussa TiO₂–P25. Therefore, the sample morphology, specific surface area, and protonic acidity of HNb₃O₈ play important roles in the photocatalytic reduction of CO₂ to CH₄.

In order to improve the CO₂ photocatalytic reduction performance of HNb₃O₈, the intercalation of guest components into the interlayer space was evaluated, and lamellar HNb₃O₈ was purposely pillared with silica [93]. The pillarization of HNb₃O₈ with SiO₂ was verified by interlayer distance expansion, from 1.13 nm to 2.85 nm. Owing to the notably expanded interlayer distance, the surface area value of SiO₂-HNb₃O₈ was as large as 197.3 m² g⁻¹, in contrast to 6.5 m² g⁻¹ of the nonpillared HNb₃O₈. All of the samples were loaded with 0.4 wt% Pt by photodeposition to evaluate the CO₂ photoreduction. The yield of CH₄ over the HNb₃O₈ sample was almost three times higher than that achieved over Nb₂O₅. The improved activity for the solid acid material might be ascribed to its layered structure and the stronger adsorption ability to H₂O through hydrogen bonding. The rate of CH₄ formation was further enhanced with the SiO₂-HNb₃O₈ sample. Additionally, the activity of the current SiO₂-HNb₃O₈ photocatalyst was higher than that of previously reported HNb₃O₈ nanobelts [92]. Compared with nonpillared HNb₃O₈, the SiO₂ pillared HNb₃O₈ sample has a notably expanded interlayer distance and much greater surface area. Thus, the reactive active sites at the interlayer space of SiO₂-HNb₃O₈ should be more accessible to the substrate. To investigate the role of protonic acidity for CO₂ reduction, the activity of SiO₂-HNb₃O₈ was compared with that of SiO₂-KNb₃O₈ and anatase TiO₂ in different molar ratios of H₂O/CO₂ (0.06 or 0.25). The SiO₂-HNb₃O₈ sample showed higher activity than SiO₂-KNb₃O₈ and anatase TiO₂; it is noteworthy that the activity of the SiO₂-HNb₃O₈ sample was more significantly enhanced than the other two samples at an elevated water content. This confirms that water molecules could react easily with the protons at the interlayer space of lamellar solid acids through hydrogen bonding.

Despite columbite niobates exhibiting interesting electronic properties and suitable conduction and valence band energies for CO₂ photoreduction [6], this class of material has been explored only in a limited manner. Kormányos et al. evaluated the CO₂ photoelectrochemical reduction activity of monoclinic type-p CuNb₂O₆ and orthorhombic type-n ZnNb₂O₆ semiconductors [94]. The CuNb₂O₆ and ZnNb₂O₆ samples showed an indirect bandgap of 1.77 eV and 3.55 eV, respectively. The photoelectrochemical (PEC) performance of the p-CuNb₂O₆ sample was tested in CO₂- and N₂-saturated 0.1 M NaHCO₃ solution. The photocurrents are two times higher (at E = −0.4 V) when the solution was saturated with CO₂. According to the authors, the dramatic increase in the presence of CO₂ is clearly attributable to the ability of CuNb₂O₆ to photoelectrochemically reduce CO₂. However, the CuNb₂O₆ sample’s efficiency and selectivity was not evaluated, as the products formed from CO₂ photoelectrochemical reduction were not monitored.

Recently, da Silva et al. investigated the role of Nb₂O₅ surface acidity for CO₂ reduction performance [95]. Nb₂O₅ catalysts were prepared through a modified peroxide sol–gel method using different annealing temperatures, and it was verified that the increase in the annealing temperature decreases the surface acidity. The authors related the activity and selectivity of the Nb₂O₅ samples to their surface acidity; high surface acidity prompted the conversion of CO₂ to CO, HCOOH, and CH₃COOH, whereas low surface acidity led to the conversion of CO₂ to CH₄. It was verified that CO is the main intermediary during the CO₂ photoreduction in all conditions. This work unveiled the importance of surface acidity for CO₂ photoreduction.

Despite the remarkable properties of NaNbO₃ and interesting performance for CO₂ photocatalytic reduction, the relative wide band gap (3.4 eV) limits its photocatalytic activity only to be active in the ultraviolet light region (4% of the solar spectrum). Therefore, there is a growing necessity to develop photocatalysts with an efficient response under visible-light irradiation. In this sense, Shi et al. design a heterostructure between NaNbO₃ and C₃N₄ in order to extended the absorption spectral range for visible region [96]. The NaNbO₃ sample was obtained through hydrothermal synthesis, and the heterostructure was tailored by annealing NaNbO₃ with melamine (C₃N₄ precursor) at 520 °C for 4 h. The XRD patterns of g-C₃N₄/NaNbO₃ revealed coexistence crystalline phase of NaNbO₃ and g-C₃N₄ in the composite. NaNbO₃ nanowires are found to randomly deposit and distribute on the surface of g-C₃N₄ sheets, which results in forming a heterostructured g-C₃N₄/NaNbO₃ material. The CO₂ photoreduction under visible irradiation showed that g-C₃N₄/NaNbO₃ improved over eight-fold the activity than that of bare C₃N₄ in the CH₄ formation. Such a remarkable enhancement of photocatalytic activity was mainly attributed to the improved separation and transfer of photogenerated electron-hole pairs at the intimate interface of g-C₃N₄/NaNbO₃ heterojunctions. The same group evaluated the CO₂ photoreduction performance of g-C₃N₄/KNbO₃, and the photocatalytic activity of g-C₃N₄/KNbO₃ for CO₂ reduction was almost four times higher than that of individual g-C₃N₄ under visible light irradiation [97].

Another Nb-based material that is worth mentioning is Sr₂Bi₂Nb₂TiO₁₂, an Aurivillius-type perovskite. Yu et al. investigated, for the first time, the Sr₂Bi₂Nb₂TiO₁₂ as photocatalyst for CO₂ photoreduction [98]. The introduction of oxygen vacancies on the surface of Sr₂Bi₂Nb₂TiO₁₂ extends photoresponse to cover the whole visible region and also tremendously promotes the separation of photoinduced charge carriers. Compared with the bulk material prepared by traditional solid-state reaction, the Sr₂Bi₂Nb₂TiO₁₂ nanosheets with the optimal oxygen vacancies concentration yields a substantially high CO evolution rate of 17.11 μmol g⁻¹ h⁻¹.

Niobium has been used also as dopant, as it can improve the photocatalytic activities of anatase TiO₂ [61,99,100,101], by changing the crystal structure, electrical properties, and absorption characteristics of TiO₂. It was proposed that Nb would dope into TiO₂ lattice and decrease the band gap energy of the photocatalyst, and thereby Nb substitution on the Ti site creates a Nb⁵⁺ defect state located upon the conduction band minimum. Therefore, Nb doped TiO₂ would work as a visible light-driven photocatalyst [61,100,101].

Nogueira et al. investigated the role of Nb-doped TiO₂ for CO₂ photocatalytic reduction to methanol [102]. The samples were obtained through the Pechini method, which promoted the substitution of Ti⁴⁺ for Nb⁴⁺ at different concentrations. It was observed that the surface area tends to increase with the concentration of Nb. The concentration of formed methanol was directly proportional to the Nb concentration in the TiO₂, the authors attributed the formation of methanol to the presence of Nb in the lattice of TiO₂.

The evaluation of CO₂ photoreduction performance of Z-scheme formed between Nb–TiO₂ and g-C₃N₄ was also evaluated [103]. It was confirmed that the band gap energy of the Nb–TiO₂ (2.91 eV) was lower than that of the TiO₂ (3.2 eV). The formation of Z-scheme Nb–TiO₂/g-C₃N₄ was evidenced through XRD and TEM analysis. CO, CH₄, and O₂ were formed from CO₂ photoreduction when the Nb–TiO₂ sample was used. Nb–TiO₂/g-C₃N₄ not only converted CO₂ to CO, CH₄, and O₂, but also generated HCOOH. Additionally, 50Nb–TiO₂/50g-C₃N₄ was the best material for the CO₂ reduction.

It is well known that, while metals and semiconductors usually exhibit great stability, but poor selectivity for CO₂ reduction, enzymes and metal complexes can selectively convert CO₂ into CO, but they lack long-term stability. Therefore, Faustino et al. evaluated the role of hexaniobate nanoscrolls (K_xH_(4−x)Nb₆O₁₇) as suitable substrates for the immobilization of two kinds of Re(I)-based molecular catalysts (namely I and II) for CO₂ photoreduction [104]. Hexaniobate nanoscrolls were obtained through solid state reaction with subsequent proton exchange and exfoliation with tert-butylammonium hydroxide. The Re(I) complexes were immobilized by an adsorption process that was verified for UV/vis absorption and Fourier-transform infrared (FTIR) spectroscopies. As a result of the hexaniobate exfoliation, the specific surface area was increased from 2 m² g⁻¹ to 110 m² g⁻¹. The sensitized oxides were employed as visible-light photocatalysts for CO₂ reduction in 5:1 dimethylformamide (DMF)/TEOA exposed to a 300 W Xe lamp (λ = 420 nm). The possible gaseous products were analyzed by mass spectrometry with CO being the only detected photoproduct. The complexes immobilized on the surfaces of the oxides exhibited much higher turnover number (TON_CO) than those in solution. For comparative purposes, the complexes were immobilized on TiO₂ as well. While for complex I, the immobilization on the niobate nanoscrolls led to 20% higher TON_CO in comparison with that observed on TiO₂, for complex II, the opposite behavior was observed; that is, the immobilization on TiO₂ led to better photoactivities. Such behavior was rationalized based on the recombination kinetics of each composite and on the reductive potential of the injected electrons on the niobate or TiO₂ conduction bands. For comparison, a summary of the photoactivity of Nb-based photocatalysts towards CO₂ reduction is presented in

Table 4. Summary of the photocatalytic activity of Nb-based compounds combined with distinct materials towards CO₂ reduction.

Nb-Based Material	Co-Catalyst	Light Source	Main Products	Observed Rate	Ref.
NaNbO3	Pt	300 W Xe	CH4	653 ppm h−1 g−1	[84]
NaNbO3	Pt	300 W Xe > 300 nm	CH4	10.4 μmol h−1 m−2	[85]
NaNbO3	None	300 W Xe > 420 nm	CO, CH4, CH3OH, H2	NA	[87]
NaNbO3 heterojunction	Pt	300 W Xe > 300 nm	CH4	~10 μmol h−1 m−2	[89]
HNb3O8	None	350 W Xe	CH4	3.5 μmol g−1 h−1	[92]
SiO2-Pillared HNb3O8	Pt	350 W Xe	CH4	2.90 µmol g−1 h−1	[93]
Nb2O5	None	UVC lamp, 0.167 mW cm−2	CO, CH4, CH3COOH, HCOOH	~10 μmol L−1 g−1	[95]
C3N4/NaNbO3	None	300 W Xe λ > 420 nm	CH4	~6 µmol h−1 g−1	[96]
C3N4/KNbO3	Pt	300 W Xe λ > 420 nm	CH4	0.25 µmol h−1	[97]
Re(I) polypyridyl complexes/KxH4−xNb6O17	None	300 W Xe λ > 420 nm	CO	2.9 *	[104]

* Turnover frequency (TOF), h^–1.

.

4. Concluding Remarks and Perspectives

Over the past few years, great efforts have been spent by the scientific community to develop clean renewable energy sources. Photocatalysis can be described as an efficient method to produce solar fuels from available sources as H₂O and CO₂. Nevertheless, it is always a challenge to develop efficient materials able to promote solar-to-fuel conversion. In this sense, niobium-based materials offer diverse possibilities to engineer compounds showing important characteristics for photocatalytic processes such as efficient light absorption, charge separation, and charge transfer reaction. Herein, we could provide a glimpse from the last years on the extensive flexibility offered by niobium-based compounds for CO₂ reduction and notably for H₂ evolution. Niobium oxides as Nb₂O₅ and the wide range of niobates offer untold possibilities of structure, morphologies, and electronic configuration, which can supply desired properties for photocatalytic reactions. Moreover, niobium cations as a dopant have provided the opportunity to tune band energy positions and increase the efficient of other materials. Thus, niobium-based compounds are promising materials that still offer opportunities of new investigations and insights regarding photocatalytic applications.