The unchecked global CO₂ emissions arising from wide-spread utilization of fossil fuels have caused serious environmental problems. The development of a renewable energy device with a high energy density and reversible reaction for power generation is still an ongoing challenge. Among numerous efforts, the conversion of CO₂ into value-added chemicals and fuels has been a widespread concern since it realizes the recycling of greenhouse gases [1]. Compared with the currently developed conversion methods (photochemical, biochemical and thermochemical), electrochemical CO₂ reduction is one of the most ideal conversion technologies as the required electricity could be derived from renewable energy sources [2]. Electrochemical CO₂ reduction also possess some intrinsic advantages, such as a stable catalytic efficiency, controllable selectivity, simple reaction unit and potential for industrial application [2].

The electrolyze for electrochemical CO₂ reduction has four components (Fig. 1): an electrolyte that has high conductivity and allows fast mass transport of the reactants and products, a proton membrane to mitigate the oxidation of liquid products, a cathode, and an anode that is coated with highly active and durable catalysts. When an external voltage is applied to the electrodes, the electrochemical CO₂ reduction takes place at the catalyst-electrolyte interface of cathode. Typically, the catalytic process involves four major steps (Fig. 1): (1) chemical adsorption of CO₂ on the catalyst surface; (2) chemical activation of CO₂ to CO₂^•-; (3) multiple electron/proton transfer processes induce the reduction of CO₂; (4) desorption of products from catalyst surface. For step (1), the CO₂ concentration in the electrolyte strongly influences adsorption rate, and a CO₂ saturated electrolyte is adopted to ensure full interaction between CO₂ and the catalyst. Step (2) contains one electron transfer process which forms the key intermediate CO₂^•-. Due to the extremely stable property of CO₂, the activation process needs a significant amount energy to rearrange the linear molecules into a bent radical anion, which is considered to be the rate determining step. Step (3) involves multiple electron/proton coupling processes whose obtained products are always complexes. For step (4), the inefficient desorption of products would reduce the active area of the electrode and block electrolyte diffusion, inhibiting further electrochemical reaction. Overall, many factors influence the efficiency and selectivity of electrochemical CO₂ reduction. Therefore, the acceleration of CO₂ adsorption, the promotion of electron/proton transfer, the reduction of reaction kinetic barriers, as well as the facilitation of obtained desorption products are expected to effectively improve the activity of electrochemical CO₂ reduction.

The electrochemical CO₂ reduction involves several proton-coupled multiple-electron transfer processes. According to different electron/proton transfer amounts, the reaction can be divided into two-, four-, six-, eight-, and twelve-electron pathways in aqueous mediums. The possible reactions for the major products are listed below (Reaction (2)-(7)) alongside the corresponding standard potentials of the reactions, E⁰, with respect to the standard hydrogen electrode (pH 7, 25 ℃, 1 atmosphere gas pressure).

(1)

(2)

(3)

(4)

(5)

(6)

(7)

(8)

The possibility of electrochemical reduction processes is not only decided by the thermodynamic value E⁰, but also relies on kinetic properties. Experimentally, a large overpotential (the difference between the equilibrium and applied potentials) is needed when catalysts are applied and the E⁰ values for different products have similar value. The products obtained at fixed overpotential are commonly a mixture rather than single product. Moreover, the E⁰ values between the electrochemical CO₂ reduction and hydrogen evolution reactions are quite close. As the result, the hydrogen evolution reaction becomes the main competitive reaction. Overall, the reaction system is very complicated, making it quite challenging to reduce CO₂ to the desirable product with high efficiency.

The development of catalysts with high efficiency and selectivity has been the research focus. In the view of reaction kinetics, current catalysts still struggle to overcome the high reaction barrier and often exhibit sluggish reaction rates during electrochemical CO₂ reduction [3]. Moreover, different transformation routes could coexist in one system and the types of products would therefore be complicated, as an electrochemical reaction involves multiple electrons/protons coupling and transfer steps. Contrarily, precious metal-based materials show relatively high activity for electrochemical CO₂ reduction, but their scarcity in nature limits their wide-scale deployment. Inexpensive copper-based materials have been extensively studied for electrochemical CO₂ reduction due to their outstanding conductivity, unique chemical property, and excellent multiple-proton and electron-transfer abilities. Generally, copper-based materials are capable of binding and converting CO₂ to higher value-added products (such as hydrocarbons and alcohols). However, the copper-based materials still suffer from poor product selectivity, activity loss due to catalyst poisoning, and limited efficiency. As a more inexpensive material, carbon-based catalysts have been utilized as catalysts for electrochemical CO₂ reduction due to their high surface area, outstanding conductivity, excellent chemical stability, and remarkable mechanical strength [4-7], with CO being the majority product. However, pure carbon catalysts possess null activity for CO₂ activation or intermediate adsorption. The carbon-based materials show low activity and selectivity toward electrochemical CO₂ reduction. Therefore, the development of new catalysts with high efficiency and selectivity is more appealing for solving the key scientific problems in current-day technology.

Transition metals have valence electrons in d bands which are close to the Fermi level (Fig. 2) [8]. With the d band filled, the relative position of d states center and the Fermi level shifts. As a result, transition metal-based materials show a variety of characteristics signaling a capability for use in catalytic reactions and thus offer many possibilities in a wide range of energy-related applications. Electrochemical CO₂ reduction is a multi-step electron/proton coupling reaction with its intrinsic activity determined by the electronic behaviors on the catalyst surface. In this situation, the variability of the transition metal valence electronic structure makes it an ideal candidate for improving the efficiency and selectivity of electrochemical CO₂ reduction.

MoS₂ possesses unique physical properties and stacked layers similar to graphite and has been considered a promising catalyst for many applications. For CO₂ reduction, the linear scaling relationships between intermediates pose a fundamental limitation to the overall efficiency of transition metal-based catalysts. The overpotential would need to maintain a high value as the relative binding strengths of the different intermediates generally correlate with each other linearly. This would imply that tuning the adsorption energy of a specific intermediate will lead to a change in the adsorption energies of the other intermediates. Nørskov et al. explored the importance of stabilizing ^*COOH, ^*CHO and ^*COH with respect to ^*CO on MoS₂, MoSe₂, and Ni-doped MoS₂ surfaces in breaking the linear scaling relationships using density functional theory (DFT) calculations [9, 10]. The theoretical results indicated the bridging S or Se atoms at the edges of the MoS₂, Ni-doped MoS₂, and MoSe₂ surfaces could bind ^*CHO and ^*COOH, but not ^*CO. In this scenario, the adsorption energy of intermediates can be varied independently, thereby breaking the linear scaling relationships. For all the cases, the S or Se edges show better selectivity towards CO₂ reduction over hydrogen evolution reactions than Mo edges. They predicted the possibility of deviations from the linear scaling relationships if key intermediates bind to different sites. These findings indicate the determination of a kinetic electron/proton transfer barrier and that selectivity is critical for high performance catalyst design.

In 2014, Salehi-Khojin et al. [11] reported the first MoS₂ catalyst for electrochemical CO₂ reduction in an ionic liquid using a two-compartment three-electrode electrochemical cell. The MoS₂ catalyst showed a significantly high current response (65 mA cm^-2 at an overpotential of 654 mV) and high selectivity (Faradaic efficiency of approximately 98%) for CO formation (Fig. 3(a) and (b)). Long-term stability tests indicate the MoS₂ catalysts remain stable for 10 h (Fig. 3(d)). Using DFT calculations, they found that the catalytic activity was primarily related to the edge states of Mo atoms. The Mo-edge atoms display metallic character and have high d-electron density near the Fermi level, which are suitable active sites for CO₂ reduction. Ionic liquid could reduce the reaction barrier for electrons passing into CO₂ by forming ionic liquid-CO₂ complexes. The complexes then physically absorb at the negatively charged cathode, which result in the positioning of CO₂ molecules close to the MoS₂ surface and leads to a high reaction rate (Fig. 3(c)). They also employed a series of transition metal dichalcogenide as high performance electrochemical CO₂ reduction catalysts [12]. The high performance could be ascribed to the active edge sites and efficient electron transfer properties. The calculated free energy diagrams suggest that the formation of CO* from CO₂ is more kinetically favorable, resulting in lower overpotentials. Further study revealed that the edge sites tended to have high CO coverage to maintain a high turnover rate during the catalytic reaction. MoP can also be employed as catalysts for high performance electrochemical CO₂ reduction to HCOOH [13]. The MoP supporting In-doped carbon exhibited the heretofore highest Faradaic efficiency of 96.5% with a current density of 43.8 mA cm^-2. The excellent performance can be ascribed to the low interfacial charge transfer resistance, strong intermediate adsorption ability, and the synergistic effect. Doping is an efficient approach to regulate the electronic behaviors on a catalysts surface. By introducing Nb dopant, the electronic properties show significant deviation from that of pristine MoS₂ because of the different number of valence electrons for Mo and Nb atoms [14]. As the result, the edge Mo atoms exhibit a metallic behavior, which alters the kinetics for intermediate adsorption, enhancing the formation of CO. With 5% Nb doping, the vertically aligned MoS₂ exhibits a CO formation turnover frequency 1 order of magnitude higher than that of pristine MoS₂. Similar to Nb, Bi dopant could also efficiently drive CO generation in the presence of an ionic liquid. When introducing Bi into MoS₂, the obtained catalyst shows high performance for the production of CH₃OH [15]. Furthermore, during electrochemical CO₂ reduction process, CO and H₂ could be produced simultaneously. CO is bound on the catalyst surface and then hydrogenated to CH₃OH. In addition, the Bi dopant can also stabilize CO₂^•- intermediates and thus significantly reduce the overpotential. Yu et al. [16] synthesized Cu-interspersed MoS₂ and utilized it for electrochemical CO₂ reduction. Cu-interspersed MoS₂ led to enhanced electronic conductivity and specific surface area, as well as improved CO₂ adsorption capacity. This resulted in a significant increase in overall efficiency and selectivity.

The carbon monoxide dehydrogenase enzyme, with the Ni-[Fe₄S₄] functional group, has been shown to efficiently and reversibly catalyze the reduction of CO₂ to CO. DFT calculations indicate the Fe sites in enzyme are the primary active sites during CO₂ reduction [17]. Costentin et al. [18] found the introduction of phenolic groups could speed up catalytic process by using electrogenerated iron(0) complexes. Further substitution of the four parahydrogens by trimethylammonium groups would produce a water-soluble molecular catalyst for electrochemical CO₂ reduction [19]. When molecular Fe on carbon nanotubes via noncovalent interactions are employed, the composite shows high selectivity and rapid catalytic process towards CO₂ to CO conversion [20]. Wallace et al. [21] found that the addition of ionic liquid could increase catalytic activity of molecular Fe. The ionic liquid affects a positive shift in the potential required for the reduction of Fe¹ to Fe⁰ rather than a proton source to facilitate the CO₂ transportation. This leads to a lower overpotential and substantially higher current density [21]. The other group further reported that immobilization of a monolayer of molecular Fe via a phosphonic acid anchor could allow for CO₂ to CO conversion with very low catalyst loading [22]. For inorganic Fe-based catalysts, DFT calculations found that the surface structure of FeS played an important role in the activation of CO₂ [23]. The CO₂ molecules tend to adsorb on the (011) and (111) FeS surfaces, preferentially at Fe sites. The charge transfer accrued from the Fe species to the CO₂ molecules and thus facilitated the activation of CO₂ molecules. On the (111) surfaces, Fe sites exhibit the best performance for CO₂ reduction. Kolpak et al. [24] employed Fe as individual d state components on WC. After introducing Fe, the adsorption energies of CO and O decreased compared to pure WC, changing the preferred reaction pathway from oxophilic to carbophilic. The two different interfaces could then provide different active sites, which enable an alternative approach for overcoming scaling relations and thus optimize the reaction steps. This new model can explain both site preference and binding-energy trends. This approach highlights a new avenue to tailor catalysts with desirable product selectivity and optimized activation overpotential. Experimentally, Li et al. [25] found that the primary role of the Fe dopant was identified to reduce the reaction barriers, exhibiting the lowest onset overpotential of 0.19 V when obtaining CO.

Inorganic Co-based catalysts have emerged as a promising candidate for HER applications owing to their abundant reserves, thermal stability, and low cost. Inorganic Co-based catalysts are considered nearly non-catalytic for electrochemical CO₂ reduction. In 2016, Xie et al. [26] developed the first inorganic Co-based catalyst for electrochemical CO₂ reduction by modifying morphology and oxidation states. They constructed an ideal model of an intact metal atomic layer and then created oxide on its surface. The hybrid 4-atom-thick layer structure ensures that most of the metal atoms serve as either surface atoms or surface ions, which allow for the two typical active sites for surface catalytic reactions (Fig. 4). The partially oxidized Co catalyst shows higher intrinsic activity and selectivity towards HCOOH production at lower overpotentials than pure Co. The electrochemical CO₂ reduction activity gradually increases as the Co₃O₄ content increases, indicating that the presence of intrinsically more active site is associated with the oxidation states. Based on the Co/Co₃O₄ system describe above, Xie et al. [27] further reported atomic layers for Co₃O₄ as promising catalysts for electrochemical CO₂ reduction. The obtained atomic-layer catalysts possess abundant active sites and high electrical conductivity, which greatly promote the reaction kinetics. The atomic Co₃O₄ layers donate one fast electron during pre-equilibrium transfer to form the CO₂^•- intermediate and thus exhibit higher intrinsic activity than bulk Co₃O₄. The thinner structure also provides the Co₃O₄ layers with many dangling bonds, which become the active site during catalytic reaction and ensure higher intrinsic catalytic activity. Using DFT calculations, they revealed that the ultrathin Co₃O₄ layers had a more dispersed charge density near the Fermi level, which enhance electronic conductivity and promote faster carrier transport which participate in the reduction reactions (Fig. 5). Along with this work, Xi et al. [28] recently reported the fabrication of vacancy-rich Co₃O₄ single-unit-cell layers, the subjection of vacancy-rich Co₃O₄ as efficient catalysts for electrochemical CO₂ reduction, and the identification of intrinsic activity by exploring the composition configurations and their correlations with electrochemical performance. There is theoretical and experimental evidence that suggests that the proton transfer process is the rate-limiting step, while the presence of oxygen (Ⅱ) vacancies can stabilize the intermediate compounds. As the result, the vacancy-rich Co₃O₄ single layers showed high current response and HCOOH selectivity. Lee et al. [29] prepared a Co₃S₄-Co₃O₄ core-shell octahedron structure for electrochemical CO₂ reduction applications. The synergistic interaction between the composites can reduce the reaction barriers and enable fast electron transfer, facilitating the kinetics for intermediate adsorption. Molecular Co is also proven to be efficient catalysts for electrochemical CO₂ reduction. Sun et al. [30] constructed a molecular Co nanotube model and systematically studied the influence of geometric and electronic structures and stability on electrocatalytic activity. Large diameter nanotubes were found to be beneficial for CO₂ to CO conversion, while the high curve nanotubes with smaller radii could further catalyze CO reduction to produce CH₄. The pH of electrolysis also plays an important role for the formation of CO and methane. An appropriate pH would affect the protonation process from a direct adduction into an indirect pathway and thus promote the electrochemical CO₂ reduction process [31]. When employing molecular Co on carbon-based materials, the complexes significantly improve the catalytic activity and the products selectivity [32]. The electrochemical CO₂ reduction performance could be further benefitted with molecular level structure optimization. Moreover, carbon nanotubes can also facilitate the adsorption and reaction of CO₂ in a given aqueous medium [33].

In the carbon monoxide dehydrogenase enzyme, Ni atoms could lower the energy pathway and improve the selectivity for CO₂ to CO conversion [17]. Ni-based materials are considered as potential catalysts for electrochemical CO₂ reduction. Nanomaterials exhibit unexpected properties when they are downsized to atomic scale because of their unique electronic and geometric structures. Hashimoto et al. [34] reported the first inorganic Ni-based catalyst (Ni-N-graphene oxide) for high performance CO₂ conversion and provide atomic-level insight into the influence of Ni-N bonds. N and Ni potentially influence the selectivity of electrochemical CO₂ reduction. Wu et al. [35] constructed a Ni-N₄ structure through a topochemical transformation strategy. The outer carbon layer can ensure the suitable dispersion of Ni atoms and thus provide a more active reaction site. This work supports the approach to construct high performance catalysts with abundant active sites and advances the catalysts' design for electrochemical CO₂ reduction applications. With the assistant of metal-organic frameworks, Li et al. [36] synthesized single Ni atoms for electrochemical CO₂ reduction. The synthesis is based on ionic exchange between Zn nodes and adsorbed Ni ions (Fig. 6(a)). Ni and N were homogeneously dispersed on the whole structure (Fig. 6(b)-(g)). This structure enables abundant low-coordinated sites on the surface and thus provide excellent performance. The composite shows a different mechanism for the reduction process with a Tafel value of 249 mV dec^-1. The electrolysis in this system could act as a proton donor and a transporter of CO₂, thus promoting the efficiency of the electrochemical CO₂ reduction. Li et al. [37] prepared atomically dispersed Ni on Ni-doped graphene and identified the monovalent Ni atomic center as the catalytically active site. The obtained single-Ni-atom catalyst exhibits high intrinsic CO₂ reduction activity, achieving high current density of 110 mA cm^-2 at an overpotential of 1 V (Fig. 7(a)) and turnover frequency of 14, 800 h^-1 at an overpotential of 0.61 V (Fig. 7(c)) with 97% Faradaic efficiency (Fig. 7(b)). The catalyst also showed long-term stability as it maintained a at high current density for 100 h (Fig. 7(d)).

The employment of Mo, Fe, Co, and Ni-based materials as catalysts has numerous possibilities in electrochemical CO₂ reduction applications. Herein, we focus on the recent advances of Mo, Fe, Co, and Ni-based materials in the CO₂ reduction process. Through different approaches, highly efficient and selective catalysts were obtained. However, despite these advances, there are still some challenges in this field:

(1) High energy barriers for the formation of the key intermediate CO₂^•- always causes a large overpotential. In addition, the insufficient mass transfer ability of CO₂ to the catalyst surface also limits the efficiency of the reduction process. Catalysts that simultaneously exhibit small overpotential (e.g. < 0.2 V) and high current densities (e.g. > 1 A cm^-2) are still required to be developed for commercial applications.

(2) The electrochemical CO₂ reduction involves multiple electron/proton coupling processes and different reaction pathways could exist on a fixed overpotential. It remains a challenge to efficiently reduce CO₂ to desirable products.

(3) The electrochemical CO₂ reduction process always leads to the formation of both gaseous and liquid products. The rapid desorption of products would destroy the microstructure and block electrolyte diffusion. In addition, the active sites of catalysts can also be blocked or poisoned by reaction intermediates, by-products, and impurities from the electrolyte. The stability of catalysts reported so far is yet to reach 1000 h.

(4) The electrochemical CO₂ reduction process is a complex reaction. The precise role of the Mo, Fe, Co, and Ni-based materials, as well as the electrolysis, in improving the activity is still not clear, and is sometimes conflicting. The development of Mo, Fe, Co, and Ni-based materials as electrochemical CO₂ reduction catalysts is still in its infancy, both with respect to the synthesis of advanced catalysts and understanding its detailed mechanisms. Research and development of this field of study is therefore urgently needed.

(1) Exploring new catalysts for electrochemical CO₂ reduction will be one of core research goals in the next few years. With regard to the development of catalysts, the potential candidates, including transition metal nitrides, phosphides, carbides, and borides can be explored for use in CO₂ reduction.

(2) The surface morphology structure plays the central role in determining the activity and selectivity of the catalyst during electrochemical CO₂ reduction. If placed in the correct morphology and electronic structure, a nearly non-catalytic compound may become very active catalyst. The electrochemical CO₂ reduction involves the interaction between solids, liquids, and gases. The local concentration of CO₂ surrounding typical catalysts and the contact between different phases, together with the surface area of electrocatalysts, limit the efficiency of electron/mass transfer to the active sites. The innovative design and synthesis of different unique nanostructures would significantly increase the catalytic active sites, shorten the paths for charge/mass transport, promote the interaction with CO₂, and facilitate desorption of products, thus improving electrochemical reaction performance.

(3) In principle, the intrinsic activity and selectivity of catalysts are determined wholly by its surface electronic structure. The surface electronic structure strongly influences the multiple-step electron and proton coupling processes, thus determining the intrinsic activity and selectivity towards electrochemical CO₂ reduction. For Mo, Fe, Co, and Ni-based materials, the electrocatalytic properties are related to its d state electrons. For example, the p-state (e.g. N, S) doping would perturb or potentially hybridize with the continuous d-states. As the result, the scaling relations between the carbonaceous adsorbates would not be applicable and the overpotential would therefore decrease. As such, the optimization of d-state structures related to various intermediate adsorption reactions [12] hold great promise for the improvement of intrinsic catalytic activity and selectivity.

(4) The fundamental understanding of electrochemical properties, electron transfer kinetics, and multiple-electron coupled mechanisms are limited. Currently, there are still barriers for precisely detailing the geometric parameters of catalysts in a large-scale setting. The measured performance only reflects the average properties of catalysts. This may lead to the conclusion of a statistically false-positive result if the geometric parameter is random. To better understand the mechanisms of electrochemical CO₂ reduction at an atomic level, the utilization of more precise models is highly desirable. In addition, the integration of theoretical simulation and in situ characterization techniques would beneficial for clarifying the catalytic mechanisms.