Review on Heteroatom Doping Carbonaceous Materials Toward Electrocatalytic Carbon Dioxide Reduction

Carbon dioxide (CO2) reduction into chemicals or fuels by electrocatalysis can effectively reduce greenhouse gas emissions and alleviate the energy crisis. Currently, CO2 electrocatalytic reduction (CO2RR) has been considered as an ideal way to achieve “carbon neutrality.” In CO2RR, the characteristics and properties of catalysts directly determine the reaction activity and selectivity of the catalytic process. Much attention has been paid to carbon-based catalysts because of their diversity, low cost, high availability, and high throughput. However, electrically neutral carbon atoms have no catalytic activity. Incorporating heteroatoms has become an effective strategy to control the catalytic activity of carbon-based materials. The doped carbon-based catalysts reported at present show excellent catalytic performance and application potential in CO2RR. Based on the type and quantity of heteroatoms doped into carbon-based catalysts, this review summarizes the performances and catalytic mechanisms of carbon-based materials doped with a single atom (including metal and without metal) and multi atoms (including metal and without metal) in CO2RR and reveals prospects for developing CO2 electroreduction in the future.


Introduction
The global demand for energy is rapidly increasing due to economic development. Currently, ∼75% of human energy consumption still depends on fossil fuels represented by coal [1]. The carbon dioxide (CO 2 ) released from fossil fuel combustion cannot be transformed by plants in time, and the atmospheric CO 2 concentration is increasing year by year, causing serious global warming [1,2]. It was revealed that the atmospheric CO 2 concentration exceeded the safety limit valve (415.2 ppm vs. 350 ppm [3]) in 2019, and Earth's surface temperature has risen 1.2 °C in the past 100 years [4], which will undoubtedly greatly affect the harmonious development of human society. How to reduce the atmospheric CO 2 concentration and realize the sustainable development of society has become an important challenge.
In 2015, the United Nations proposed goals to achieve "carbon neutrality" and "emission peak" by a certain period of time. On the one hand, it is necessary to find renewable clean energy (such as wind energy, solar energy, etc.) to replace fossil fuels and reduce CO 2 emissions from the source. On the other hand, technical approaches should be adopted to reduce the atmospheric CO 2 concentration, such as CO 2 storage [5] and CO 2 conversion [6]. CO 2 can be converted into high-value-added carbon chemicals, such as methane, methanol, formic acid, and syngas, through technical means [7][8][9][10][11], which seems to be more efficient than CO 2 storage. CO 2 is a rich and cheap C1 resource. In nature, CO 2 can be reduced by plants using photosynthesis to realize the carbon cycle. Inspired by this, how to use chemical means to convert CO 2 into high-value-added chemicals has attracted great attention from scientists. CO 2 conversion technologies mainly include chemical conversion [12], photochemical conversion [13][14][15], and electrochemical conversion [16][17][18][19]. Photochemical and electrochemical conversions use energy conversion devices to realize the carbon cycle by inputting clean energy light or electricity. This has been of wide interest because of the characteristics of simple operation, small floor area, and environmentally friendliness. This process can not only greatly reduce the CO 2 content in the air but also store clean energy, such as solar energy and wind energy, in the form of chemical energy, which can effectively and perfectly achieve the carbon cycle, as shown in Fig. 1. Compared with CO 2 photocatalytic reduction, the electrocatalytic reduction of CO 2 (CO 2 RR) can control the target product synthesized by adjusting the applied potential, which is flexible and easy to adjust [20][21][22] while avoiding catalyst deactivation caused by electron-hole recombination [18,23]. Thus, developing CO 2 RR technologies has high research significance and commercial value for the carbon cycle and energy storage.
Working Principle of CO 2 RR CO 2 RR requires an electrocatalytic device (Fig. 2) that is composed of a cathode, an anode, and an electrolyte, including a working electrode (WE), reference electrode (RE), and counter electrode (CE). The electroreduction of CO 2 takes place at the WE. The oxygen precipitation reaction occurs at the CE at the positive stage, that is, 2H 2 O → O 2 + 4H + + 4e − . The WE, which is made of a catalyst, is the object of the CO 2 RR study. Its intrinsic properties will directly determine the catalytic activity and conversion efficiency of electrocatalytic CO 2 .
CO 2 RR occurs in four processes [24]: (1) CO 2 adsorption, (2) CO 2 activation, (3) product formation, and (4) product separation and diffusion. CO 2 is very thermodynamically stable because of its strong C = O bond energy (bond energy 750 kJ/mol), which is much larger than those of C − H, C − O, and C − C bonds [25,26]. CO 2 conversion must first overcome the huge energy barrier of the C = O bond, which can be realized by a single electron attacking CO 2 molecules to form * CO ⋅− 2 free radical anions CO 2 (g) + e − → * CO ⋅− 2 (standard reduction potential − 1.9 V [3]). The products of CO 2 RR vary according to the number of transferred electrons and can be classified into three types based on the number of carbon atoms, namely, C1 chemicals, such as CO [27,28], CH 4 [29,30], HCOOH [31,32], and CH 3 OH [33]; C2 chemicals such as CH 2 CH 2 [34], CH 3 CH 2 OH [35], and CH 3 COOH [32]; and C3 chemicals, such as propanol [36], and acetone [37]. Table 1 lists the electroreduction reactions on the WE and the corresponding standard reduction potentials. The hydrogen evolution reaction has a very similar standard reduction potential to electrocatalytic CO 2 to radical anions in aqueous electrolytes, so they are competitive reactions. The final product depends on the inherent properties of the catalyst [3]. Since the formation of C2 and C3 chemicals requires more electron transfer and higher overpotential, reducing CO 2 to C1 chemicals is easier than reducing to C2 and C3 chemicals. At low applied voltage, the C≡O bonds in the intermediate *CO cannot be struck, and CO is the main reduction product. At higher applied voltage, the output electrons gain enough energy to attack *CO, and the reduction products of multielectron transfer such as CH 3 OH, C 2 H 4 , and C 2 H 5 OH can be obtained. Catalysts commonly used for CO 2 RR include metal [38], nonmetal [39,40], and molecular catalysts. Metal catalysts (including pure metals, metal oxides, etc.) have high catalytic

Carbon-Based Catalysts
Carbon-based materials have attracted much attention due to their diversity, low cost, high conductivity, and high throughput [44]. Nowadays, carbon-based materials have been widely used in energy storage, catalysis, and other fields [45,46] and are also used as effective catalysts for CO 2 RR [47][48][49][50][51]. Carbon-based materials in aqueous solution have chemical inertness under negative potential, which can inhibit H 2 release and improve CO 2 reduction activity. Hence, carbon-based materials are promising CO 2 electroreduction catalysts.
Electrically neutral carbon atoms have no catalytic active sites, but their catalytic performance can be improved by heteroatom doping, through which reaction sites for adsorption and activation of CO 2 can be created, thus effectively regulating the catalytic activity. There are many types of heteroatoms, including N [52][53][54][55], B [56], F [57], and multi atoms [58][59][60], that can be doped into carbon-based materials. More and more carbon-based catalysts have recently been developed and used for CO 2 RR.

Synthesis Methods of Doped Carbon Materials
Methods for preparing doped carbon materials are post-treatment synthesis and in situ synthesis. Post-treatment synthesis methods are post-treatment modifications of traditional carbon materials using heteroatom-containing precursors that aim to introduce heteroatoms into carbon materials. For example, the commonly used N-containing precursors are ammonia, ethylenediamine, hexamethylene tetramine, and urea. Most of the heteroatoms introduced by post-treatment synthesis exist on the surface of the carbon materials, and a few of them can enter the carbon skeleton, hence a not high heteroatom content in carbon materials. Their disadvantages include cumbersome procedures and high costs.
In situ synthesis methods introduce heteroatoms directly into carbon materials using high-heteroatom-content precursors. Taking N-doping as an example, common precursors are mainly melamine, benzylamine, acetonitrile, aniline, and N-containing heterocyclic compounds. The commonly used in situ synthesis methods mainly include templating, chemical vapor deposition, hydrothermal synthesis, sol-gel, and high-temperature pyrolysis. These methods can introduce the heteroatom-containing functional groups uniformly into the carbon structure. However, at present, most in situ synthesis methods are still in the experimental stage, and some problems abound. (1) The in situ synthesis system is limited, and sometimes, the required tissue cannot be formed by reaction. (2) The density of the obtained product is not high. (3) The reaction process is not easily controllable, and the formation of the mesophase has an adverse impact on the properties of materials.

Parameters for Evaluating Catalyst Performance
(1) Onset potential/overpotential The resistance of CO 2 RR makes the initial potential smaller than its thermodynamic equilibrium potential. Overpotential η is the difference between these two potentials. In CO 2 RR, the role of the catalyst is to reduce the reaction activation energy and overpotential. The overpotential directly reflects the performance of the catalyst. (2) Current density and partial current density Current density reflects the speed of electron transfer and thus the speed of electrocatalytic reaction. Current density can be divided into total current density and partial current density. Total current density is calculated as the ratio of the current value measured by the electrochemical workstation to the geometric area of the working electrode, J, mA/cm 2 . Partial current density is the current density contributed by each product, which is calculated by multiplying the total current density and Faradaic efficiency (FE) of each product. The partial current density reflects the selectivity of some products to some extent. The Tafel slope is the slope of the approximately straight-line segment on the semilogarithmic curve obtained when plotting the overpotential and the logarithm of partial current density. It indicates the degree to which the electrode potential needs to be changed to achieve a certain current.

Research Progress on CO 2 Electrocatalysts
Carbon atoms in carbon-based materials can be assembled into various sizes and structures with different advantages, such as zero-dimensional (carbon quantum dots), onedimensional (carbon nanotubes (CNTs), carbon nanofibers), two-dimensional (graphene, carbon nanosheets), and threedimensional (porous carbon spheres) carbon nanomaterials. Electrically neutral carbon atoms have limited adsorption capacity for CO 2 . Doping can make carbon atoms electronegative, thus inducing charge redistribution and creating reaction sites for CO 2 adsorption and activation. The selectivity and activity of carbon-based materials for CO 2 RR are different due to different hybrid states or nanostructures of the carbon atom. In addition, the selectivity and activity of heteroatom-doped carbon materials for CO 2 are different. In Fig. 3, based on the type and number of doped atoms, different from the previous classification method, this review discusses the single-atom-doped (including metal and without metal) and multiatom-doped (including metal and without metal) carbon-based catalysts. Through this classified discussion, on the one hand, we can easily grasp the effect of heteroatom doping on the improvement of the catalytic activity of carbon-based materials. On the other hand, it may be helpful for researchers to explore new ways to improve the catalytic activities of carbon-based materials. Table 2 lists some representative carbon-based catalysts and their performances in CO 2 RR.

Monatomic Doping
Studies have been published on heteroatom doping in carbon materials, including N, B, S, F, and metal atoms. Atom doping improves the electronegativity of carbon materials and creates active sites for CO 2 RR.

Nonmetallic Monatomic Doping
N doping N and C atoms have similar radii but different electronegativities. The incorporation of N into carbon-based materials has shown excellent conductivity and more active sites. Hence, N atoms are very important doping atoms in carbon-based materials.
Carbon quantum dots (CQDs) are a class of zero-dimensional carbon nanomaterials with significant fluorescence properties and many advantages such as low cost, environmental friendliness, and good electrochemical performance. They have a good application prospect in the field of catalyst preparation. N-doped CQDs have been prepared and used for CO 2 RR. Wu et al. [35] synthesized nanosized N-doped graphene quantum dots (NGQDs) with a sawtooth profile and used them to electrocatalytically reduce CO 2 to hydrocarbons. The products included CH 4 , C 2 H 4 , and oxygenates, with a maximum FE of 90% and selectivity for C2 + C3 products exceeding 55% (Fig. 4a). The results showed similar catalytic efficiency and product distribution of NGQDs to copper-based nanoparticles. NGQDs showed high activity and selectivity in promoting the reduction of CO 2 to CO.
Heteroatom doping can induce charge redistribution and activate reaction sites. The different properties of functional groups will also have different results for the charge redistribution of carbon atoms, which will affect the electrocatalytic activity. In 2021, Zhang et al. [61] functionalized GQDs through functional groups (-OH or -NH 2 ), which were used for CO 2 to CH 4 conversion, and found that electron-donating groups can promote the electroreduction of CO 2 to CH 4 while electron-withdrawing groups can inhibit CO 2 electroreduction. For the electron-donating group, the yield of CH 4 was positively correlated with the electrondonating ability and quantity. At room temperature, the FE of CH 4 can reach 70.0%, and the partial current density can reach − 200 mA/cm 2 . In Fig. 4b, GQDs showed high selectivity and high yield.
Carbon nanofibers are one-dimensional carbon nanomaterials with high strength, good thermal conductivity, and  [82] high conductivity. N-doped carbon nanomaterials have been prepared. Kumar et al. [52] carbonized polyaniline into carbon nanofibers as a catalyst for the electrocatalytic reduction of CO 2 to CO (Fig. 5a). The current density of the catalyst is more than 13 times that of silver. It is found that the high electronegativity of N and the high positive charge on the carbon atom result in the strong catalytic activity of the catalyst. The structure of the nanofibers and a single fiber can form fractal wavy active centers, which improves the catalytic activity.
CNTs are also classic one-dimensional carbon nanomaterials with high conductivity and are widely used in electrochemistry and hydrogen storage. N-doped CNTs (NCNTs) have also been prepared. Sharma et al. [62] used acetonitrile as the N source to synthesize NCNTs with a great N content at 850 °C for the electrocatalytic reduction of CO 2 to CO, with FE as high as 80%. They also found that the change in pyridine N content in NCNTs leads to a change in FE (Fig. 5b). This also confirms that pyridine N is the main active site in the electrocatalytic reduction of CO 2 to CO.
Dicyandiamide can also be doped into carbon materials as an N source. Wu et al. [63] prepared NCNTs with dicyandiamide as the N source. The catalyst can efficiently and stably reduce CO 2 to CO at an overpotential of − 0.18 V (vs. RHE) and selectivity of 80% (Fig. 5c). N doping in CNTs improves the conductivity of the catalyst, increases the catalytic active sites, promotes CO 2 activation, and reduces hydrogen evolution. Density functional theory (DFT) calculations show that pyridine nitrogen is the highest selective site for CO generation. The catalyst electrode showed strong stability after continuous operation for 10 h.
Recently, Shi et al. [64] have prepared NCNTs with polyaniline as the main N source and used them for the electrocatalytic reduction of CO 2 to CO. At an overpotential of − 0.83 V (vs. RHE), the partial current density of CO can reach 13.8 mA/cm 2 . At an overpotential of − 0.63 V (vs. RHE), the FE reaches a maximum value of 93%. In Fig. 5d, the catalyst shows high selectivity and high efficiency. For CO 2 in simulated flue gas (15% CO 2 , 77% N 2 , and 8% O 2 ), the catalyst can directly reduce CO 2 to CO, with FE as high as 70%. The catalyst shows an extremely high activity, which may be mainly due to the rich imine N in the catalyst [65].
Carbon nanosheets are two-dimensional carbon nanomaterials with good electrochemical properties. Li et al. [66] used the post-etching strategy to prepare wrinkled N-doped porous carbon nanosheets (WNCS) for CO 2 RR. The initial overpotential of the catalyst is as low as − 0.19 V (vs. RHE). When the overpotential is − 0.49 V (vs. RHE), the FE is 84%. The two-dimensional nanoporous structure of the WNCS promotes the rapid charge transfer and is conducive to the rapid adsorption of CO 2 to the pyridine N active center. DFT results show that pyridine N can provide a favorable binding site for intermediate * COOH and promote catalytic reduction.
Graphene is a two-dimensional carbon material with excellent electrical properties, which is known as the "superstar" of nanomaterials. It is widely used in materials science, energy, biomedicine, and so on. N-doped Fig. 4 a Performance of NGQDs in electro catalytic reduction of CO 2 . Reproduced with permission from Ref. [35]. Copyright © 2016 Nature Publishing Group. b Performance of functionalized GQD for electrochemical reduction of CO 2 . Reproduced with permission from Ref. [61]. Copyright © 2021 Nature Publishing Group graphene has been successfully prepared and used for CO 2 RR. Sun et al. [67] directly synthesized N-doped graphene by free radical self-polymerization (Fig. 6a). The porosity and N concentration are adjusted by controlling the pyrolysis temperature. Compared with the templatefree precursor, the catalyst has a higher surface area and larger pore size. When CO 2 is reduced to CO, the FE of the catalyst is 83.0%, and the partial current density of CO is 1.79 mA/cm 2 at overpotential of − 0.9 V (vs. RHE). The catalyst has abundant graphite-N species and a hierarchical pore structure, which enhances the activation performance of CO 2 RR.
Porous carbon is a typical three-dimensional carbon material with a large specific surface area and pore volume. Recently, to broaden the application of porous carbon, heteroatom-doped porous carbon has been gradually developed and applied to CO 2 RR. Li et al. [68] developed and adopted an ammonia corrosion strategy to synthesize N-doped porous carbon (NPC) materials from coal as effective metal-free catalysts for CO 2 RR. The catalyst exhibits well-developed porous nanostructures and highly exposed N defects. They found that appropriate pyridine-N and pyrrole-N have synergistic effects on catalysis. At room temperature, the FE reaches 92% at overpotential of − 0.6 V (vs. RHE) (Fig. 6b). The catalyst can efficiently catalyze CO 2 to produce CO, which provides a new idea for the preparation of porous carbon electrocatalysts from coal.
Metal-organic frameworks play an important role in the preparation of N-doped carbon materials. Ye et al. [69] prepared NPC with an oxygen-rich metal-organic framework (Zn-MOF-74) as the precursor, which has a highly porous structure and a large amount of active nitrogen (pyridineand graphite-N). The NPC has a high concentration of CO 2 RR active sites, and the active sites can be adjusted. Because of the large pore size, the active center of the catalyst is highly exposed, providing a large number of active  (Fig. 6c).
B Doping Incorporating B atoms can significantly improve the properties of carbon-based materials. For example, doping diamond with B greatly improves its conductivity, corrosion resistance, oxidation resistance, and other properties. Based on this, B-doped carbon-based materials have been gradually developed. Sreekanth et al. [56] prepared a B-doped graphene catalyst with boric acid as the B source in an argon atmosphere at 900 °C for CO 2 RR with saturated bicarbonate as the electrolyte. The FE of HCOOH reached 66% at 1.4 V (vs. SCE) overpotential. B-doped carbon-based materials exhibit excellent electrocatalytic reduction properties for CO 2 . Nakata et al. [70] synthesized B-doped diamond (BDD) using trimethyl borate and acetone solution as the B and C sources, respectively. Under environmental conditions, the catalyst electrode can electrochemically reduce CO 2 to HCHO in seawater. BDD shows a high FE of 74%.
F Doping F has stronger electronegativity than N, B, and S. Doping F can induce the polarization of adjacent carbon atoms, resulting in greater charge density, thus having stronger adsorption capacity for intermediates such as *COOH. Recently, it has been gradually proven that doping F atoms in carbon-based materials can significantly enhance electrocatalytic performance. In 2018, Xie et al. [57] synthesized a fluorocarbon (FC) catalyst with BP 2000 (a kind of oil, used as pristine carbon) and polytetrafluoroethylene. Because of F doping, FC has more structural defects. In addition, FC has shown excellent electrocatalytic activity for CO 2 RR. At a low overpotential of 510 mV (vs. RHE) and a small Tafel efficiency of 81 mV/dec, the maximum FE reaches 90%, surpassing the metal-free catalyst at that time. F doping activates adjacent carbon atoms and promotes the stability of key intermediate *COOH on F-doped carbon materials. The results also show that CO on FC shows high current density and H 2 shows low current density, which indicates that F doping effectively inhibits hydrogen evolution.
S Doping There are currently few reports on single S-doped carbon-based materials used for CO 2 RR. Single S-doped carbon materials have many shortcomings when they are used in CO 2 RR. For example, surface vulcanization and sulfonation often lead to a significant decline in specific surface areas. In addition, S is easy to leach from the adsorbent in the S-doped carbon materials, resulting in the reduction of adsorption capacity, which affects their performance in CO 2 RR. Carbon-based materials co-doped with S and other atoms show good performance for CO 2 RR. The relevant introduction about S and other atoms co-doped carbon-based materials will be discussed in the "Polyatomic Doping" subsection.

Metal Monatomic Doping
Noble-metal catalysts show excellent performance in electrocatalysis. However, because of the scarcity of noble metals, the costs of preparing them are high. Transition metals have a d orbital electronic structure similar to noble metals, which is expected to reduce the preparation cost of catalysts. At present, researchers have prepared carbon-based material catalysts doped with metal atoms. Esmaeilira et al. [71] synthesized four transition-metal carbide nanoflakes (W 2 C, Mo 2 C, Nb 2 C, and V 2 C) by carburization and liquid stripping technology and used them for CO 2 RR to CH 4 . The results show that these materials have high selectivity for CH 4 . For W 2 C, the maximum current density of CH 4 is − 421.63 mA/ cm 2 , and the FE of CH 4 is 82.7% ± 2%. The EE of CO 2 RR is about 62.3% under the minimum energy path of CO 2 to CH 4 by W 2 C. DFT calculations show that water dissociation, CO 2 chemisorption, and C-O bond fracture are almost Fig. 6 a Preparation of N-doped graphene catalyst and its application in CO 2 RR. Reproduced with permission from Ref. [67]. Copyright © 2021 Chemical Industry Press of China. b Nitrogen-doped porous carbon for CO 2 electrocatalysis. Reproduced with permission from Ref. [68]. Copyright © 2019 Elsevier. c Nitrogen-doped porous carbon for CO 2 electrocatalysis. Reproduced with permission from Ref. [69]. Copyright © 2020 Wiley 1 3 spontaneous on the W 2 C surface, which is the main reason for the high activity, selectivity, and low potential observed.

Polyatomic Doping
Single-atom doping into carbon-based materials can significantly improve the CO 2 electrocatalytic activity of the catalyst. Compared with single-atom doping, multipleatom doping into carbon-based materials can provide more active sites. Because of the synergistic effect between the doped atoms, the electrocatalytic activity for CO 2 is more prominent.

Nonmetallic Polyatomic Doping
CO 2 RR over a metal-free N-doped carbon catalyst provides a solution for CO 2 reuse. However, because of the low intrinsic reactivity of the N dopant, the overpotential is large and the selectivity is poor. In 2019, Pan et al. [59] synthesized N, F co-doped porous carbon layers by introducing the electron-rich F to activate pyridine-N sites (Fig. 7a). For CO, the catalyst achieves 90% FE within 40 h at a 490 mV (vs. RHE) low overpotential, which is significantly higher than that of the N-C catalyst without F. DFT calculations show that electron feeding from nearby F atoms increases the charge density and delocalizes the electron density of pyridine N states. The synergistic effect induced by doping creates a new way to control the performance of CO 2 RR.
Although F can significantly improve the catalytic activity, its toxicity limits its application. Pan's group explored replacing F with S doping into carbon materials, prepared a N, S co-doped carbon layer catalyst, and used it for CO 2 RR [72]. They found that S addition can greatly improve the catalytic activity and selectivity of CO 2 reduction. At a low overpotential of 490 mV (vs. RHE), the FE of CO is as high as 92%, and the partial current density of CO is 2.63 mA/ cm 2 , which is much better than those of sulfur-free N-C (Fig. 7b). DFT calculations show that S addition to N-C leads to the formation of higher-density pyridine N as the active center for CO 2 reduction. In addition, by doping S, the free energy barrier of the key intermediate *COOH can be reduced to improve the catalytic activity and selectivity. In the same year, Han et al. [73] prepared N, S dual-doped carbon nanoweb catalysts for CO 2 RR. They found that the catalyst can quickly convert CO 2 to CO. At a low overpotential of 490 mV (vs. RHE), the FE is as high as 93.4%, and the partial current density is − 5.93 mA/cm 2 (Fig. 7c). The minimum value of the Tafel slope is 93 mV/dec. In addition, within 20 h of electrolysis, the current density distribution is very stable, and the FE of CO exceeds 90%.

Metal Containing Polyatomic Doping
NGQDs show excellent catalytic performance in CO 2 RR. By co-doping with metal atoms, they can play a synergistic role and show excellent catalytic performance. Fu et al. [74] transferred surfactants on single-crystal gold nanoparticles to NGQDs through a wet chemical reduction method and synthesized NGQDs wound SCAu (NGQDs SCAu) (see Fig. 7a) for CO 2 RR. CO 2 can be efficiently converted into CO at an ultra-low potential of − 0.25 V (vs. RHE). The maximum FE can reach 93%. The onset potential of CO is as low as − 0.15 V (vs. RHE) (Fig. 8a), which is much lower than that of Au nanocatalysts. DFT calculations show that this is due to the synergistic effect between pyridine N and SCAu in NGQDs, which improves the adsorption energy of *COOH on the catalyst, thus promoting the reduction of CO 2 to CO.
Because of the expensiveness of Au, alternative cheaper transition metals are added to carbon-based materials to prepare metal-doped carbon catalysts. The common pure metal Fe has almost no CO 2 electrocatalytic properties. However, the introduction of Fe into the carbon skeleton and the formation of metal-N co-doped carbon materials can activate their CO 2 electrocatalytic properties. Cheng et al. [75] designed a Fe-N co-doped carbon nanofiber catalyst, which can effectively catalyze the CO 2 to CO conversion (Fig. 8b). The process has nearly 100% selectivity and about 95% FE. The catalyst shows remarkable durability and stability. Pan et al. [76] designed a new nanostructured electrocatalyst by doping N and Fe through the calcination strategy protected by SiO 2 . The electrocatalyst consists of a CNT/carbon nanosheet (CNT/CNS) structure, which provides abundant N-induced defects and high active sites of Fe − N species for CO 2 RR. CNTs also provide a smooth path for charge transfer in CO 2 RR. At a low overpotential of 0.48 V (vs. RHE), the maximum FE of CO reaches 69%. The synergistic effect of structure design and doping optimization can effectively improve electrocatalytic activity and catalytic efficiency.
Ni and Fe belong to group VIIIB elements with similar properties. In recent years, Ni-based carbon materials have also been successfully synthesized. In 2021, Niu et al. [77] prepared a double-chain nickel-based catalyst (Ni@NC@ NCNT) by solvothermal evaporation calcination. The performance of Ni@NC@NCNT for CO 2 RR is much better than that of single-chain Ni@NC. The catalyst has a large specific surface area and rich pyridine-N and Ni δ+ sites to ensure its high CO 2 adsorption capacity, as well as good stability. The FE of CO is as high as 94.1%, and the partial current density is as high as 48.0 mA/cm 2 (Fig. 8c). After 43 h of continuous electrolysis, its performance remains unchanged.
In the same year, Gang et al. [78] prepared a mixed CNS/ CNT catalyst with Ni, N co-doped by one-step chemical vapor deposition (CVD) (Fig. 8d). In the CVD process, Ni nanoparticles were coated with a graphite carbon layer, which inhibited the competitive hydrogen evolution reaction and achieved high CO selectivity. The partial current density of CO is 28.9 mA/cm 2 , and the FE is 91% at − 0.74 V (vs. RHE) overpotential at room temperature. When the applied voltage is − 2.0 V (vs. RHE), the selectivity for CO is more than 85%. These carbon-based catalysts show excellent performance in the current density and selectivity of CO.
Zn is a common transition metal, and Zn-doped carbon materials also show excellent CO 2 RR characteristics. Chen et al. [79] prepared single center zinc coordinated nitrogen co-doped graphene (Zn-N-G) and used it to catalyze CO 2 to produce CO (Fig. 8e). The Zn-N-G catalyst showed excellent intrinsic activity for CO 2 reduction. At an overpotential of 0.39 V (vs. RHE), the FE is 91%. The isolated Zn-N x exists in the catalyst and is the key active site for CO formation. This work provides a basis for further understanding of the active sites on metal-N-C catalysts.
Sun et al. [80] achieved high efficiency of CO 2 RR to CO by adjusting the interface between CdO and carbon black (CB). Through the performance test of CO 2 RR to CO by CdO/CB composites, they found that CdO/CB composites with 20 wt% CdO can obtain a total FE of 92.7% at − 1.0 V (vs. RHE) potential, which is much higher than that of pure CdO (69.5%) under the same conditions. The FE of CO is up to 87.4%. They found that the high catalytic performance of CdO/CB composites is attributed to the interface and high contact area between CdO and CB. In addition, the catalyst showed super stable catalytic activity.
Monatomic catalysts have become a research hotspot because of their high atomic efficiency and good activity. However, in the catalytic process, the influence of the support structure in monatomic catalysts is often not noticed.
Xiong et al. [81] synthesized a series of hollow mesoporous carbon spheres supported Ni-N4 monatomic catalysts (Ni/ HMCS-3-800) with different support structures and with excellent catalytic activities for CO 2 RR (Fig. 9a). In the potential range of -0.7 to -1.1 V (vs. RHE), the FE of the catalyst for CO is as high as 95%. They found that the shell thickness and compactness of carbon spheres effectively regulate the chemical environment of N-doped species in the carbon skeleton and promote the activation of CO 2 molecules. This provides a new idea for improving the catalytic performance of monatomic catalysts.
Recently, Du and Chen [82] reported that CNT/Ni@ mC were successfully prepared by modifying CNTs with a layer of mesoporous carbon shell anchored by nickel (Ni) nanoparticles. Dispersed Ni nanoparticles and N doping in mesoporous carbon provide abundant active sites, so the catalyst shows very excellent catalytic performance. Selective reduction of CO 2 to CO obtains a maximum FE of CO of 98% at the crossing point of − 0.81 V (vs. RHE) (Fig. 9b) and partial current density of CO of 60 mA/cm 2 .
In addition, Cu shows excellent performance in CO 2 RR. In 2021, Li et al. [83] dispersed atomic copper on N-doped carbon nanosheets and prepared Cu − N − C composites by the metal-organic skeleton derivation method. Because of the unique structural characteristics of monatomic Cu, the catalyst can more effectively promote CO 2 RR. The Cu − N − C catalyst has better catalytic performance than the traditional copper electrode, and the yield is as high as 71%. Also, Cometto et al. [84] modified exfoliated graphite carbon nitride (GCN) with a single copper atom and used it for CO 2 RR. Experimental and theoretical studies determined the adsorption sites and coordination of copper atoms. They found GCN to be a promising substrate that can effectively stabilize metal atoms because of the strong Cu-N interaction in its structure.
Metal oxides also show excellent performance in CO 2 RR. Shi et al. [65] synthesized metal oxide/N-doped carbon catalysts (ZrO 2 /NC, CeO 2 /NC) with high activity by surface pyrolysis. In CO 2 RR, the ZrO 2 /NC and CeO 2 /NC catalysts have higher catalytic activity than NC catalysts without metal oxides. At room temperature, the maximum FE of CO reaches 90%, and the partial current density of CO is 5.5 mA/cm 2 (Fig. 9c). They found that the metal oxide carbon interface plays an important role in improving the catalytic activity and selectivity of CO. Meng et al. [85] designed an integrated three-dimensional bismuth oxide ultra-thin nanosheet/carbon foam electrode through the sponge effect and electrically reduced CO 2 to formic acid with high efficiency and selectivity for the first time. The catalyst has a unique three-dimensional array foam structure, which realizes high-efficiency mass transfer by optimizing the exposed active sites. The electrode has achieved excellent partial current density, high FE (up to Fig. 8

Conclusions and Prospects
Herein, research progress on carbon-based materials doped with different heteroatoms in CO 2 RR is reviewed in detail, and breakthroughs in identifying active sites, improving catalytic activities, and clarifying the catalytic mechanism are comprehensively analyzed. The catalytic activity of carbon-based materials for CO 2 RR is highly dependent on the carbon structure and heteroatoms. The synthesis of efficient carbon-based catalysts needs to be adjusted, and the carbon/ heteroatom precursors should be optimized. The goal is to enhance the catalytic activity, clarify the catalytic mechanism, and improve the product yield. However, we still face many major challenges. Therefore, the prospect of efficient CO 2 RR is discussed in this paper.
(1) Reasonably designed and prepared bimetallic-doped carbon-based catalysts. When the two metals are made into an alloy, the product distribution of CO 2 RR will not show superposition but will produce a highly selective product distribution different from that of a single metal. Reasonable design and preparation of bimetallic carbon-based catalysts may regulate the electronic structure of the metal interface, effectively improve the electrochemical performance, and achieve high selectivities of products. Fig. 9 a Hollow mesoporous carbon spheres supported Ni-N 4 catalyst for CO 2 RR. Reproduced with permission from Ref. [81]. Copyright © 2020 Wiley. b CNT/ Ni@mC catalyst for CO 2 RR. Reproduced with permission from Ref. [82]. Copyright © 2022 Elsevier. c ZrO 2 /NC and CeO 2 /NC catalysts for CO 2 RR. Reproduced with permission from Ref. [65]. Copyright © 2021 Springer. d Ultra thin bismuth oxide nanosheet/carbon foam catalyst for CO 2 RR. Reproduced with permission from Ref. [85]. Copyright © 2020 Wiley-VCH (2) Constructing a carbon-based catalyst with a core-shell structure. The catalyst with this structure may change the binding strength between the catalyst and CO 2 RR intermediates by adjusting the surface electron distribution while possibly changing the local atomic arrangement of the active site, thereby affecting the selective adsorption of different intermediates. Hence, carbonbased materials with a core-shell structure may greatly improve the performance of CO 2 RR. (3) Introducing a cocatalyst into a carbon-based catalyst may greatly improve its catalytic activity. The cocatalyst itself has no catalytic activity, or the catalytic activity is very low, which can be ignored. Because of the synergistic effect, the introduction of a cocatalyst may change the adsorption and desorption strength of intermediates, so the catalyst as a whole shows unexpected catalytic performance and achieves the effect of 1 + 1 > 2. The catalytic activity of the catalyst may be greatly improved, as well as the selectivity or stability of the product. (4) Optimization of the electrocatalysis system. At present, electrocatalytic CO 2 reduction is mostly carried out in an H-type electrolyzer. As the reaction proceeds, the products gradually increase, and the local microenvironment in the electrolytic cell changes, which will restrict the reaction efficiency. In addition, the electrolyte in the electrocatalysis system is mostly an aqueous solution, which limits the dissolution of CO 2 . Based on this, organic, seawater, and solid electrolytes have been gradually applied to the CO 2 RR system, which has great potential. Hence, it is particularly important to optimize the electrocatalysis system and find a suitable electrolyte.