1. Introduction

The year 2024 was the hottest year on record [1], with global average temperatures rising by approximately 1.55°C compared to pre-industrial levels [2]. This increase is closely related to the emission of greenhouse gases: in 2022, the atmospheric CO₂ concentration surged from 278 ppm in pre-industrial times to 420 ppm [3]. The extreme weather events, such as glacier melting and droughts, triggered by global warming pose a severe threat to ecological security. Therefore, CO₂ capture and utilization technologies (CCU) are urgently needed to address environmental and energy crises [4,5].

Among various technologies, photocatalysis has emerged as a promising candidate due to its solar-driven nature and characteristics of ambient temperature and pressure reactions, offering advantages such as low energy consumption, environmental friendliness, and sustainability [6,7]. In recent years, significant progress has been made in photocatalytic research. For example, Liu et al. developed P-g-C₃N₄ nanotubes that achieved a 13.92-fold increase in CH₄ yield compared to traditional g-C₃N₄ through optimized band structure [8]. Abdullah et al. constructed a 2D/2D ZnV₂O₆ -pCN heterojunction system. The modified surface charge enhanced the CH₃OH yield up to 3742 μmol gcat^-1, which is 1.15 and 5 times higher than that of pure ZnV₂O₆ and pCN, respectively [9].

However, photocatalytic reduction of CO₂ technology still faces many challenges, such as low light utilization efficiency [10,11], anatase TiO₂, for example, is only active under ultraviolet (UV) light, and the visible light response of g-C₃N₄ is limited to below 460 nm [12]. Additionally, high carrier recombination rates, competition with hydrogen evolution reactions, poor long-term stability, and low product selectivity severely limit the practical application of photocatalytic technology [13].

Metal compounds, with their tunable band structures, abundant active sites, and good chemical stability, hold great potential [14]. Based on this, this review summarizes the research progress of metal-based photocatalytic materials and elaborates on optimization strategies such as defect engineering, crystal facet control, and heterojunction construction, aiming to provide a theoretical framework for building efficient, stable, and selective photocatalytic systems and to promote the resource utilization of CO₂.

2. TiO₂-Based Photocatalytic Materials

Since Inoue et al. pioneered photocatalytic CO₂ reduction in 1979 [15], a variety of photocatalytic materials have been extensively studied. As the earliest discovered and widely used photocatalytic material, TiO₂ has become one of the most promising photocatalysts due to its excellent photostability, environmental friendliness, and low cost [16]. However, its photocatalytic CO₂ reduction performance is still limited by several factors: (1) insufficient surface active sites for CO₂ adsorption and weak CO₂ activation ability; (2) high carrier recombination rate and low utilization efficiency; (3) a wide bandgap of ~3.2 eV that only responds to UV light, with visible light utilization less than 5% [17,18]. To address these bottlenecks, researchers have developed various modification strategies.

2.1. Crystal Engineering

Crystal engineering primarily involves improving catalytic performance by altering the crystal structure of TiO₂, including elemental doping, morphology control, and crystal facet control. This section mainly discusses elemental doping and morphology control.

Elemental doping can effectively adjust the band structure, enhancing light absorption and carrier utilization efficiency [19]. Asahi et al. first demonstrated that nitrogen doping could shift the valence band upward and reduce the bandgap, extending light response to the visible region [20]. Anna et al. achieved co-doping of Mo/W metal elements, forming functional groups such as hydroxyl radicals on the TiO₂ surface, which significantly improved carrier separation efficiency through synergistic effects [21].

Morphological regulation, by altering parameters such as surface area, crystallinity, and porosity, affects processes like CO₂ activation and adsorption, charge separation, and intermediate desorption [22,23]. Patricia Reñones et al. prepared mesoporous TiO₂ 1-D nanofibers via electrospinning-sol-gel method, which exhibited better nanocrystalline connectivity than traditional TiO₂, promoting rapid charge transport and higher selectivity for CH₄ and CH₃OH [24].

2.2. Heterojunction Construction

Combining TiO₂ with other semiconductors to form heterojunction catalysts (n-n heterojunctions, p-n heterojunctions, Z-scheme heterojunctions, etc.) can significantly enhance carrier separation and optimize photocatalytic performance. Ali's team developed an HCNS@TiO₂ heterojunction catalyst that promoted rapid electron transfer at the interface between HCNS and TiO₂, increasing CH₃OH yield to 11.3 μmol gcat^-1h^-1, which is 10 times and over 5 times higher than that of traditional TiO₂ and g-C₄N₃ [25]. Chen et al. manufactured a ZnIn₂S₄ nanosheet/TiO₂ nanofiber heterojunction via electrospinning-hydrothermal process, with a maximum light absorption edge reaching 550 nm [26].

2.3. Active Site Design

Creating oxygen vacancies (OVs) to improve surface active sites, adjust electronic structure and bandgap, and expand light response and carrier utilization efficiency is a promising approach [27]. Studies have confirmed that surface oxygen vacancies can enhance CO₂ adsorption by binding to CO₂ oxygen atoms and bending the linear CO₂ molecule, reducing the reaction barrier for C-O bond cleavage [28]. Huang's team developed a TiO_2-α/WO_3-δ heterojunction catalyst that induced local electron enrichment by regulating OVs concentration, strengthening the built-in electric field and effectively suppressing carrier recombination, resulting in a CH₄ yield of 28.11μmol g^-1h^-1, which is 18.6 times higher than that of traditional samples [29]. Li et al. combined interstitial carbon doping with oxygen vacancies to prepare TiO2 nanofiber membranes, altering reaction sites to accumulate CO* intermediates at Ti termini rather than desorbing them, enabling continuous hydrogenation to form methane with a yield of 55.17 μmol g^-1h^-1 and a selectivity of 98.3% [30].

Constructing surface frustrated Lewis pairs (FLPs) is an innovative strategy. FLPs, with spatially separated Lewis acid and Lewis base active sites, can induce CO₂ polarization and promote its heterolysis by activating both oxygen and carbon atoms in the CO₂ molecule, thereby enhancing CO₂ activation efficiency. Zhao et al. manufactured a core-shell c-TiO₂@a-TiO_2-x(OH)_y heterostructure via disorder engineering. The mechanism of action of SFLPs is shown in Fig.1a-e,Ti(III) forms SFLPs (Fig.1i-k) with oxygen vacancies (Fig.1g,h) on the TiO₂ surface, creating high-activity sites for CO₂ reaction through protonated hydroxyl groups formed by hydrogen cleavage. The core-shell structure(Fig.1f) enhances light absorption(Fig.1l). Analysis of charge carrier mobility and the localization of energy states via inverse participation ratio (IPR) revealed that a-TiO_2-x(OH)_y has a high concentration of self-trapped polarons and excitons (Fig.1m,n), effectively promoting carrier separation and extending carrier lifetime (Fig.1p). Fig.1o is a schematic diagram of electron migration. This material achieved a CO yield of 5.3 mmol gcat^-1 h^-1, which is 350 times higher than that of c-TiO₂ crystals [31].

Figure 1: (a) SFLPs act on the surface of the c-TiO₂@a-TiO_2-x(OH)_y heterostructure. (b) an unmodified crystalline surface, (c) formation of an SFLP site, (d) a ground-state SFLP site following activation by hydrogen (e) the enhanced activity excited-state SFLP generated by photoexcitation of the ground-state SFLP. (f) HR-TEM micrograph of c-TiO₂@a-TiO_2-x(OH)_y (anatase phase). The amorphous/crystalline interfaces are marked with dotted lines. (g) High-angle annular dark-field scanning transmission electron microscopy (HAADF-STEM) image of c-TiO₂@a-TiO_2-x(OH)_y. (h) Ti-L edge EELS spectra at different positions. (i) Comparison of geometrical and electronic structures between bridging and terminal hydroxyl. j, k HMAS-NMR (j) and ATR-FTIR (k) spectra of c-TiO₂@a-TiO_2-x(OH)_y and c-TiO₂. (l) UV–Vis-NIR diffuse reflectance spectra (DRS). m, n The total density of states (black line, right axis) and the corresponding values of the inverse participation ratio (IPR) (blue dots, left axis) for a-TiO_2-x(OH)_y (m) and crystalline TiO_2-x (n) surfaces. (o) Schematic of charge carrier separation and transfer pathways in c-TiO₂@a-TiO_2-x(OH)_y. The colors of arrows indicate the wavelength of incident light. (p) Time-resolved photoluminescence spectroscopy decay curves of c-TiO₂ and c-TiO₂@a-TiO_2-x(OH)_y.

2.4. Others

In addition to the aforementioned strategies, other approaches include the use of cocatalysts, surface plasmon resonance (SPR), construction of internal electric fields, and multi-dimensional synergistic strategies. Zhang et al. decorated TiO₂ nanotubes (NTs) with carbon quantum dots (CQDs), achieving a methane yield 2.5 times higher than that of traditional TiO₂[32]. Liu et al. synthesized 2.5 wt% Ag/TiO₂ composite materials, with a methanol production rate of 135.1μmol g^-1h^-1 which is 9.4 times higher than that of pure TiO₂[33]. Xing et al. fluorinated single-crystal TiO_2-x to construct an internal electric field, enhancing carrier separation and migration efficiency and increasing methane yield tenfold to 0.98 mmol g^-1h^-1 compared to traditional TiO₂[34].

3. Copper-based Photocatalytic Materials

In photocatalytic reactions, the reduction ability of a semiconductor is primarily determined by the positions of its conduction band (CB) and valence band (VB).Copper-based semiconductors (such as CuO, Cu₂O, and Cu₂S), with their relatively narrow band gaps (approximately 1.7 eV, 2.2 eV, and 1.2 eV respectively), and favorable conduction band (CB) states, exhibit good visible-light response and thus show significant potential in the field of photocatalysis. However, copper-based materials also exhibit several drawbacks. For instance, Cu₂O is prone to high carrier recombination rates, limited charge migration, and photo-corrosion in liquid phases, resulting in poor long-term stability [35]. To address these issues, researchers have developed various strategies.

Zhang et al. modified Cu₂O with Pd to synthesize 100Cu₂O-0.1Pd, which facilitated the transfer of photogenerated holes from Cu₂O to Pd. This modification enhanced charge migration and corrosion resistance, increasing CO yield to 0.13 μmol g^-1h^-1, three times higher than that of pristine Cu₂O [36]. Cui et al. prepared three-dimensional porous Cu₂O with nanoscale dendrites. The dendritic porous structure enhances the light-harvesting ability and electron transfer efficiency, resulting in a CO production rate that is 24 times higher than that of traditional Cu₂O [37]. Khatri et al. fabricated reduced graphene oxide–CuO (rGO–CuO) nanocomposites. The close energy levels of the LUMO (lowest unoccupied molecular orbital) between CuO and graphene facilitates electron transfer, overcoming the high carrier recombination rate in traditional CuO. This leads to a CH3OH production rate of 1228 μmolg^-1h^-1, which is seven times higher than that of pure CuO nanorods [38]. Wang et al. synthesized Cu₂O@ZnTPP heterojunction composites, constructing a Zn-O-Cu electron transfer pathway that enhanced electron migration rates, protected ZnTPP from photo-corrosion, and promoted multi-electron reactions on Cu₂O. This material achieved a CH₄ yield of 120.9 μmol g^-1h^-1, ten times higher than that of traditional Cu₂O, with 98.7% selectivity [39].

4. WO₃-Based Photocatalytic Materials

Tungsten trioxide (WO₃) is characterized by its wide light absorption spectrum (bandgap ~2.5 eV, with absorption extending up to 500 nm), high electron mobility (12 cm²V^-1s^-1), long hole diffusion length, and excellent chemical stability under harsh conditions (e.g., acidic and oxidative environments). These attributes make WO₃ a highly attractive candidate for photocatalytic applications [40]. However, pure WO₃ still falls short of industrial photocatalytic efficiency requirements due to its relatively low CB edge potential (+0.5 eV) and rapid carrier recombination [41]. To enhance the catalytic efficiency of WO₃, various strategies, including defect engineering and heterojunction construction, have been employed.

Regulating the crystal morphology and microstructure of WO₃ is a direct approach to improving its performance. Shi et al. synthesized Cu₂O/WO₃-001 composite nanosheets (Fig.2a-f), whose favorable spatial structure increased the contact area with CO₂. WO₃-001 exhibited a maximum pore size peak at 0.41 nm (Fig.2g), and its CO2 adsorption isotherm showed a higher adsorption capacity (Fig.2h), indicating good porosity and microchannels that facilitate CO₂ adsorption. The possible electron-hole separation process (Fig.2i) involves photogenerated holes remaining in the VB of WO₃ while electrons in the CB of WO₃ transfer to the VB of Cu₂O, achieving effective charge separation.This Z-scheme electron transfer mode significantly enhanced photocatalytic performance. After 24 hours of irradiation, the yields of CO, O₂, and H₂ over this material reached 11.7, 5.7, and 0.7μmol, respectively (Fig.2j), although its stability was moderate (Fig.2k) [42].

Figure 2: SEM and TEM images of WO3 (a), WO3-001 (b, c), Cu2O/WO3-001 (d) and Cu2O/WO3-001 (e, f) sample. (g) Calculated pore size distributions. (h) CO adsorptioisotherm of WO3 and WO3-001 sample. (i) Schematic illustration of the proposed charge transfer mechanisms- the common charge transfer mode and Z-scheme charge transfer mode for Cu2O/WO3-001. (j) Total yields of the products over the Cu2O/WO3-001 catalyst with different irradiation times. (k) Recyclability test of the Cu2O/WO3-001 catalyst under visible-light.

Constructing heterojunctions has shown significant effects on improving WO₃ performance.Gao et al. anchored black phosphorus quantum dots (BPQDs) onto WO₃ nanowires to form a 0D-1D direct Z-scheme heterojunction. The BPQD-WO₃ heterojunction not only achieved high CO conversion efficiency but also innovatively produced C₂H₄ [43]. Zhu et al. synthesized a 2D/1D BiOBr_0.5Cl_0.5/WO₃ S-scheme heterojunction, which, with the assistance of cocatalysts, achieved a CO yield of 16.68 μmol g^-1h^-1, nine times higher than that of traditional WO₃.Notably, the BiOBr_0.5Cl_0.5/WO₃ composite maintained stable photocatalytic performance over time [44].

In addition, there are various pathways such as cocatalysts and oxygen vacancy regulation.Wang et al. modified hexagonal WO₃ with Pt, which promoted local electron delocalization, increased free electron concentration, and significantly enhanced charge separation and transfer efficiency. Moreover, Pt modification improved CO₂ adsorption and activation capabilities. Within 5 hours, 0.5%Pt-WO₃ catalyzed the production of 3.64 μmol CH₄, approximately seven times higher than that of pure hexagonal WO₃[45]. Ben et al. doped C atoms into two-dimensional WO3 nanosheets and introduced oxygen vacancies, achieving a CO yield of 23.2 μmol g^-1h^-1 with 85.8% selectivity [46]. Zeng et al. deposited single-atom Cu and Pt on WO₃, triggering C-C coupling to photoreduce CO₂ into high-value CH₃COOH. The Cu₂Pt₂/WO₃ catalyst achieved an CH₃COOH yield of 19.41 μmol g^-1h^-1 with 88.1% selectivity, significantly higher than that of the original sample [47].

5. Metal-Organic Frameworks (MOFs)

In addition to semiconductor-based photocatalysts, metal-organic frameworks (MOFs) have emerged as novel and promising materials in recent years due to their excellent CO₂ adsorption capabilities and unique structural characteristics. MOFs offer several advantages: (1) high specific surface area and adjustable porous structures that enable efficient CO₂ adsorption and transformation, combined with high crystallinity for superior light absorption and charge transfer capabilities;(2) modular design through diverse metal nodes and organic linkers to tailor photocatalytic performance; (3) high stability and recyclability, facilitating reuse; (4) further performance enhancement through functionalization and construction of composite systems for synergistic catalysis; (5) diverse synthesis methods, many of which are environmentally friendly [48,49]. Optimization strategies for MOFs mainly focus on modifying metal nodes and organic linkers, post-synthetic method (PSM), constructing composites with semiconductors or metal nanoparticles, and optimizing interpenetrating structures. Some recent research progress is briefly introduced below.

Lin et al. demonstrated that synthesis methods can influence surface structure and, consequently, catalytic performance. Their team developed a core-shell structured Fe/Ni-T120 material with uniform mesoporous structures (~2.5 nm) and abundant active sites, achieving a high CO yield of 9.74 mmol g^-1h^-1 with 92.1% selectivity [50]. The photocatalytic activity is closely related to the exposed facets. Cheng et al. investigated the photocatalytic performance of NH₂-MIL-125 (Ti) exposing high-index {112} facets. The {112} facets exhibited CO and CH₄ yields 33 and 31 times higher than those of low-index {001}/{111} facets, respectively. This was attributed to the higher affinity of {112} facets for CO₂ and the narrower HOMO-LUMO gap, which promotes charge generation and transfer [51]. Studies have shown that amino-functionalized linkers are more effective than other functional groups (such as -Br, -OH, -SH, and -NO₂) in enhancing photocatalytic efficiency [52]. Huang et al. synthesized UiO-66-NH₂ via microwave heating, which achieved a CO2 adsorption capacity of 5.8 mmol g^-1 at 273 K and 1 bar, and a CO₂/N₂ selectivity of 66, significantly higher than the 17.8 selectivity of original UiO-66 [53].

6. Other Photocatalytic Materials

In addition to the commonly studied metal oxides, metal sulfides, phosphides, and nitrides also hold great potential.

Cheng et al. prepared a three-dimensional hierarchical Cd_0.8Zn_0.2S (C8Z2S-F) with ultrathin petals. The chemical composition and element states are shown in Fig.3a and b. It features slit-like pores, a narrower bandgap width, and a more negative conduction band potential (Fig.3c-f). It exhibits an outstanding CO selectivity of 89.9% within 3 hours (Fig.3g), with CO and CH₄ yields of approximately 41.4μmol g⁻¹ and 1.4μmol g⁻¹(Fig.3h,i). The mechanism of CO₂ conversion to CO and CH₄ is illustrated in Fig.3j. In CO₂-TPD measurements (Fig.3k), the strong basic adsorption peak of C8Z2S-F at 370°C has a significantly higher desorption temperature and intensity than that of C8Z2S-NP (at 320°C), indicating that C8Z2S-F has a much stronger CO₂ adsorption capacity than C8Z2S-NP. TPR (Fig.3l) shows that C8Z2S-F has a stronger photocurrent density, indicating better carrier separation and migration. The reaction mechanism is shown in Fig.3m [54].

Figure 3: (a) XPS spectra. (b) High resolution XPS spectra. (c) N2 adsorption-desorption isotherms (inset indicates the pore size distribution); (d) UV-vis DRS spectra (inset shows the Kubelka-Munk function vs. photon energy curves). (e) Mott-Schottky plots. (f) Schematic of the electronic band structures of the as-obtained C8Z2S-F sample. (g) Photocatalytic products of CO2 reduction detected by original chromatograms over C8Z2S-F and C8Z2S-NP samples under visible-light irradiation for 3 h; Photocatalytic activity of CO (h) and CH4 evolution (i) over 0.03 g C8Z2S-F and C8Z2S-NP samples under visible light irradiation. Reaction conditions: deionized water (500 μL), Xe lamp(300 W). (j) Schematic of the photocatalytic reduction of CO2 by using C8Z2S-F sample as photocatalyst.CO2-TPD profiles (k) and EIS spectras (l) (inset shows the TPR curves) of C8Z2S-F and C8Z2S-NP samples.(m)Schematic of the possible photocatalytic process during the conversion of CO2 to CO and CH4.

Xu et al. synthesized metal phosphides (M-P/BP) on black phosphorus nanosheets (BP) via in situ growth, such as Co₂P/BP. The ultrafine Co₂P nanocrystals formed on the BP surface effectively adsorbed and activated CO₂ molecules through additional cobalt active sites. The Co-P charge transfer channel significantly enhanced carrier separation efficiency, achieving a CO yield of 255.1 μmol g^-1h^-1, nearly twice that of BP alone (141.8 μmol g^-1h^-1) [55]. Moreover, while semiconductor catalysts like TiO₂ typically require narrower bandgaps to respond to lower-energy infrared light, metal catalysts such as Co₂P/BP exhibit better infrared responsiveness [56].

7. Conclusions

Significant breakthroughs have been made in the field of metal-based photocatalytic CO₂ reduction in recent years. This review summarizes the research progress and performance optimization strategies of TiO₂-based, Copper-based, WO₃-based, and MOF materials, as well as their underlying mechanisms. Advances have been achieved through techniques such as atomic layer deposition and electron beam evaporation to enhance photostability, functional group modification to improve CO₂ adsorption and activation, and heterojunction construction and defect engineering to accelerate carrier migration and utilization.

In the future, metal-based photocatalytic materials for CO₂ reduction can be integrated with digital twin technology to develop dynamic intelligent catalytic systems. They can also be coupled with biological systems to construct biomimetic photosynthetic systems, thereby promoting selective control of products. By incorporating photocatalytic materials into diverse industries such as construction and transportation, innovative solutions can be provided for the global carbon neutrality goal. However, there remain areas that require further exploration, as follows.

1. The carrier recombination mechanism are not yet fully understood, limiting the improvement of carrier efficiency. Advanced in situ characterization techniques, combined with density functional theory (DFT) and other theoretical calculations, are needed to elucidate interfacial charge transfer and surface reaction kinetics mechanisms.

2. The complex multiple-proton and electron coupling processes result in low selectivity for C2+ products. There is a need to reveal the structure-activity relationships of intermediate products and establish predictive models based on descriptors (e.g., d-band center) to explore photocatalysts and pathways with higher selectivity for high-value products.

3. The long-term stability and recyclability of catalysts in large-scale applications urgently need to be verified. It is necessary to develop a catalyst life prediction system in combination with machine algorithms and establish an intelligent ASTM evaluation system. The transition of laboratory achievements to industrialization should be promoted through interdisciplinary integration and technological innovation.