Recent Advances in Heterogeneous Electroreduction of CO₂ on Copper-Based Catalysts

Abstract

Facing greenhouse effects and the rapid exhaustion of fossil fuel, CO₂ electrochemical reduction presents a promising method of environmental protection and energy transformation. Low onset potential, large current density, high faradaic efficiency (FE), and long-time stability are required for industrial production, due to economic costs and energy consumption. This minireview showcases the recent progress in catalyst design and engineering technology in CO₂ reduction reaction (CO₂RR) on copper based-catalysts. We focus on strategies optimizing the performance of copper-based catalysts, such as single-atom catalysts, doping, surface modification, crystal facet engineering, etc., and reactor design including gas diffusion layer, membrane electrode assembly, etc., in enhancing target electroreduction products including methane, methanol, ethylene, and C₂₊ oxygenates. The determination of the correlation and the developed technology might be helpful for future applications in the industry.

1. Introduction

In recent years, the increasing usage of fossil fuels has not only accelerated the exhaustion of limited natural energy, but has also led to enormous CO₂ emissions, resulting in energy shortages and greenhouse effects. A great deal of research has been undertaken to solve these problems. Among many suggestions, electrochemical CO₂ reduction reaction (CO₂RR) appears a promising way to produce value-added chemicals and simultaneously decrease CO₂ concentration in the atmosphere [1,2]. Powered by intermittent renewable electricity, and using water as the proton donor, CO₂RR converts CO₂ molecules into a wide range of reduced carbon compounds, including carbon monoxide (CO), methane (CH₄), formic acid (HCOOH), methanol (CH₃OH), ethylene (C₂H₄), ethanol (C₂H₅OH), and acetate (CH₃COOH) [3]. The collected products are either directly used as fuels or further converted into value-added chemicals [4,5]. With these advantages, CO₂RR has become an important research area and many scientists have tried to put the associated ideas into practice.

The low current density, low faradaic efficiency (FE), and high overpotential occurring with CO₂RR remain a challenge for product generation. The disadvantages of the stable C=O bond (about 750 kJ mol⁻¹) [6], and the complex pathway of proton-coupled electron transfer, contribute to the chemical sluggishness of CO₂RR [7]. The similar range of redox potentials for different CO₂RR products and the hydrogen evolution reaction (HER) contribute to the poor Faradaic efficiency (FE) of specific products [8]. Although many electrochemical catalysts have been designed to enhance the activity and selectivity, and good catalytic performance was obtained, another issue of catalytic stability was revealed in further application. Stability should be maintained for at least 4000 h for industrial application, and more than 20,000 h are required to make production appealing economically [9,10]. However, various factors in electroreduction, including surface reconstruction, flooding behavior, etc., caused unsatisfactory stability only lasting for tens of hours, far from industrial requirements [11,12]. Thus, research was carried out targeting large current density, high FE, low overpotential, and long stability.

Since Hori’s first work in 1985 [13], many efforts in fields of mechanism exploration and catalyst synthesis have been made to disclose the reaction process and to overcome the obstacles mentioned above [14,15,16,17,18,19,20,21,22]. Methods for catalyst synthesis including morphology [23], facet engineering [24,25], metal doping [26,27], and surface modification [28,29] could modulate the binding strength between the intermediates and active centers, thus favoring reaction rates and rearranging product distribution. Copper is one of the few metals that can reduce CO₂ to hydrocarbons and alcohols with decent efficiency, and copper-based catalysts have received much attention [30,31]. The uniqueness of Cu as a CO₂RR electrocatalyst is explained by the fact that it is the only metal that has negative adsorption energy for *CO and positive adsorption energy for *H [2]. Furthermore, the moderate binding of *CO to Cu provides a balance between activation of CO₂ and hydrogenation of *CO, which is the key step towards hydrocarbons. In view of the unique advantage of Cu, we will focus on Cu-based catalysts and strategies to improve their catalytic performance.

In addition to catalyst design, other improvements to the catalytic system have been presented in recent years. Gas diffusion electrodes (GDE) were widely adopted to feed reactant gas directly to the interface between the catalyst and electrolyte, enhancing the CO₂ concentration on the catalysts’ surface and promoting the current density. Other methods including changing the CO₂ concentration and electrolyte engineering were also reported to enhance catalytic performance [32,33].

In this review, the advances in efficient copper-based catalysts and CO₂RR systems in recent years are summarized. First, the reaction mechanisms for generating C₁ and C₂₊ products are briefly introduced. Since low overpotential is required to minimize the energy cost, we next introduce catalyst synthesis and engineering to reduce overpotential. Single-atom catalysts (SACs) with maximum atom utilization efficiency present unique structures for the structure-function relationship. The electronic structures can be finely tuned through changing their near-range coordination environment and long-range interactions, lowering the overpotential of CO₂RR products [34]. Then, strategies involving catalyst and system design to achieve high FEs and current densities are summarized. Finally, the degradation factors affecting CO₂RR are discussed and methods for stability improvement are presented. Based on the interaction between the catalyst surface and the intermediates, valuable catalyst design and engineering technologies are proposed to achieve low overpotential, high FE, large current density, and long stability of copper-containing catalysts.

2. Electroreduction Pathways

CO₂RR activity is carried out through multiple proton and electron transfer steps involving many possible intermediates. The catalytic performance can be tuned and determined largely by key reaction intermediates. Advances in operando spectroscopy and computational techniques provide significant scope for exploring the evolution of surface-bound species and rationalizing a pathway to the desired product. The reaction pathways of CO₂ reduction with key intermediates on copper-based catalysts are summarized (Figure 1).

The reaction pathways for the formation of C₁ products are relatively simple. C₁ products from CO₂RR include HCOOH, CO, CH₃OH, and CH₄. Generally, the formate pathway, which involves the *OCHO intermediate, is considered a dead-end, and the *OCHO is bound to the catalyst surface via one or two oxygen atoms [19,35]. For CO, *COOH is first generated with carbon atom binding to the catalyst surface, followed by *CO formation through dehydrogenation of *COOH, which is finally desorbed from the catalyst to release CO [36]. *CO is the initial key intermediate for further reduction, and the selectivity is determined by the binding energy of *CO. If the binding energy is too weak, the *CO will be desorbed from the catalyst surface to generate CO. If the catalyst binds *CO very strongly, the catalytically active sites are poisoned and the HER becomes dominant. Only the catalysts with the moderate binding energy of *CO can produce muti-electron reduction products [37]. *CO is further reduced to CH₄ through the *CHO pathway on these catalysts, forming *CHOH and then *CH₂OH and finally branching into two routes (CH₃OH and CH₄) [38]. It has also been observed that *C is formed by dehydration of *COH and constitutes another pathway to generate CH₄ [39]. Furthermore, *CH₃O formed by protonation of *CHO is also an important intermediate for CH₄ generation [40,41].

The C₂₊ pathways are much more complex and many controversies still exist. The electroreduction pathway towards the main products includes C₂H₄, C₂H₅OH, CH₃COOH, and n-C₃H₇OH. In the formation of the multi-carbon compounds, C–C coupling is the most important step and starts between different C-containing intermediates. *CO dimerization is the most widely accepted step, having been evidenced by theoretical calculation and in situ spectroscopic observation [42,43]. Then, the *CO dimer is further reduced to *OCHCH₂, which serves as the key intermediate in determining the selectivity of ethylene and ethanol [44,45]. It is proposed that the *CO dimer follows another pathway to produce CH₃COOH via *COCHOH intermediate [3]. Additionally, some studies suggest the possibility of coupling between *CHO and *CO, especially at higher overpotentials. After coupling, the intermediate *COCHO is further reduced to ethanol, ethylene, and acetate [18]. The coupling occurs between *CO and *CH₂ via *CO insertion, which produces acetate through multiple proton and electron transfer steps [34,46]. Although it was suggested that n-propanol forms by C–C coupling between CO and C₂H₄ precursors [47], the elementary steps for C₃ production have not yet been well proposed.

3. Advances in CO₂ Electroreduction

The recent research into CO₂ electroreduction focuses on reducing overpotential, enhancing the FE of the target product, enlarging current density, and maintaining stability. Different synthesis methods of the catalysts and catalysis engineering have been introduced to promote catalytic performance.

3.1. Low Overpotential

Overpotential is defined as the additional potential (above the thermodynamic requirement) to drive a reaction at a specific current density [48]. It can be simulated by theoretical calculations such as density functional theory (DFT). The adsorption energies of the evolving intermediates were calculated to build up the thermodynamic energy diagram for CO₂RR, and the overpotential was attributed to the most unfavorable step in the electroreduction pathway [27,40], which was called the rate-determining step (RDS). Although some methods of calculation correction have been developed to assess the kinetic barrier [49,50], the difference in adsorption energy of intermediates was found to influence the overpotential, and other effects such as local CO concentration and charge transfer resistance also played a vital role. Recent research on reducing the onset potential for CO₂RR products is shown in

Table 1. Recent research on onset potential for CO₂RR products (* The onset potential is not clearly determined and this value is the lowest applied bias with decent FE).

Major Product	Catalyst	Onset Potential (V vs. RHE)	Reference
CO	Cu-N2/CN	−0.33	[51]
	FeN5	−0.2 *	[52]
	Ni–N3S	−0.17	[53]
	Fe3+-N-C	−0.2	[54]
	Co–N–Ni/NPCNSs	−0.2	[55]
	CoPc©Fe-N-C	−0.13	[56]
Formate	single-atom Snδ+ on N-doped graphene	−0.18	[57]
	BiN4/C	−0.51	[58]
Methane	AuAgPtPdCu	−0.3 *	[59]
Ethylene	Organosuperbases modified Cu-NC	−0.43	[60]
	F-Cu	~−0.2	[61]
Ethanol	Cun (n = 3 and 4) cluster	−0.3~−0.4	[62]
	Au/Cu	−0.7 *	[63]
	FeTPP[Cl]/Cu	−0.42	[64]
Acetate	Cu–Cu2O/Cu	~−0.2 *	[65]

. The potential for the target product is quite different. Low overpotential is required for CO and formate, while high overpotential is needed for deep electroreduction products of methane, ethanol, and ethylene.

3.1.1. Two-Electron Electroreduction Products

Generally, two-electron electroreduction products with a relatively simple process require lower overpotential, compared with other reduction products such as methane and ethylene. Since the properties of active metal centers in SACS can be finely tuned through changing the near-range coordination environment and long-range interactions, SACs could effectively lower the overpotential of CO₂RR products. Considering the promotion effect of unsaturated coordination on catalytic activity, Zheng et al. fabricated coordinatively unsaturated single-atom with nitrogen sites anchored on graphene (Cu-N₂/CN) [51]. Aberration corrected high-angle annular dark field scanning transmission electron microscopy (AC HAADF-STEM) and X-ray absorption spectroscopy (XAS) demonstrated that the single-atom Cu species were uniformly distributed and coordinated with two N atoms (Figure 2a,b), and inductively coupled plasma mass spectrometry (ICP-MS) determined the Cu content of 1.45 wt% in Cu–N₂/GN nanosheets. The coordinatively unsaturated Cu not only promoted the adsorption of CO₂ on the catalyst surface, but also accelerated the electron transfer from Cu–N₂ sites to *CO₂. The Cu–N₂/CN catalyst produced CO at a low overpotential, with a maximum FE of 81% at −0.50 V vs. RHE, and onset potential −0.33 V vs. RHE (Table 1). They found that the electronegative N atoms near the coordinatively unsaturated metal center further reduced the energy barrier, lowering the onset potential to −0.30 V vs. RHE [66]. Though we focus on Cu-based catalysts in this review, their effect on reducing the overpotential of CO₂RR, especially for two-electron products, is inferior to catalysts with other metals. Zhang et al. prepared singly dispersed FeN₅ sites supported on N-doped graphene with an additional axial ligand coordinated to FeN₄ through thermal pyrolysis [52]. The AC HAADF-STEM images showed the atomically dispersed Fe and an absence of larger clusters in as-prepared catalysts. In electrochemical tests, the over-coordinated catalyst exhibited a high FE of 97% for CO production at low overpotential. DFT calculation disclosed that the additional coordination number weakened the *CO binding strength of FeN₅, and facilitated the desorption of *CO, which changed the RDS and lowered the overpotential. The coordinated adjustment was also an effective way to modify the electronic structures of the metal center and enhance the catalytic performance. Yang et al. synthesized an S-doped Ni-SAC (Ni–N₃S) by pyrolysis treatment [53]. The single-Ni-atom catalyst was prepared by pyrolysing a mixture of the amino acid (l-alanine or l-cysteine), melamine, and nickel acetate in argon, with the addition of a sulfur precursor (l-cysteine)The catalytic results showed that the S doping reduced the onset overpotential at only 70 mV, 100 mV lower compared to the Ni–N₄ catalyst without S doping. Ni K-edge X-ray absorption near-edge structure (XANES) spectra indicated that the non-centrosymmetric ligand strength of Ni–N₃S highly distorted the geometry, which was considered to promote the adsorption of reactants and intermediates, and reduce the overpotential. Cl and N dual-coordinated Mn-SAC ((Cl, N)-Mn/G) synthesized by Zhang et al. improved CO₂RR, and the catalytic activity was obtained at low overpotential [67]. The d-band center of (Cl, N)-Mn/G was lower than that of MnN₄, thus weakening the strong adsorption for *CO. DFT calculation proved that the energy barrier of the RDS (desorption of *CO) decreased from 1.64 eV for MnN₄ to 0.65 eV for (Cl, N)-Mn/G (Figure 2c). As there were four types of N atoms existing in SACs, different types of N atoms coordinated with the metal center and resulted in the variation of catalytic performance. Gu et al. prepared Fe³⁺-N-C coordinated with pyrrolic N, which displayed a CO partial current density of 94 mA cm⁻² at an overpotential of 340 mV [54]. In this catalyst, the pyrrolic N coordination rendered the Fe^3+/2+ reduction potential more negative than the Fermi level of the carbon support, leading to the stabilization of the Fe³⁺ ions in CO₂RR conditions. Then, the stabilized Fe³⁺ induced faster CO₂ adsorption and weaker CO absorption, which contributed to the high activity with low overpotential.

Other strategies to boost the catalytic performance of metal centers were also proposed. Pei et al. reported a catalyst with Co-Ni bimetallic sites connected by an N bridge, namely Co–N–Ni/NPCNSs, achieving CO Faradaic efficiency of 96.4% at −0.48 V vs. RHE, with lower overpotential than the catalysts with single metal sites (Figure 2d) [55]. Experimental characterization and DFT calculation suggested that the charge transfer between Co and Ni via the N bridge promoted the formation of *COOH intermediates, thus accelerating the electroreduction. Similarly, anchoring Fe-N sites with the molecular catalyst cobalt phthalocyanine (CoPc) (denoted as CoPc©Fe-N-C) exhibited a low onset potential of −0.13 V vs. RHE (Table 1), with enhanced CO current density and a broadened potential window of high CO FE [56]. Furthermore, the overpotential of CoPc©Fe-N-C was shown to be lower than other samples at the density above 300 mA cm⁻², suggesting that the synergistic effect between different metal centers may act as an effective strategy in future applications at large current density. By introducing the electron-withdrawing cyano group, the CoPc anchored on carbon nanotubes exhibited exceptional selectivity and activity for CO at a low overpotential of 0.52 V, ascribing to the promotion of the active sites Co (I) for the reduction of CO₂ [68]. For the product of HCOOH, the overpotential was reduced by modifying the metal center. The positively charged single-atom Sn was anchored on N-doped graphene, lowering the energy barriers of CO₂ activation and protonation. The reduction potential of −0.74 V vs. SCE corresponding to an overpotential of only 60 mV was observed [57]. Exclusive Bi-N₄ sites were fabricated on porous carbon networks by Zhang et al., exhibiting high intrinsic activity for CO₂ conversion at a low overpotential of 0.39 V [58].

3.1.2. Multi-Electron Products

In general, Cu-based catalysts have been used in deep CO₂ electroreduction. Although Cu SACs produced CH₄ and other deep reduction products, the overpotential was relatively high [69,70]. Xu et al., found that Cu metallic clusters were synthesized through reversible transformation by an amalgamated Cu–Li method, converting the CO₂ to ethanol with an extraordinary 91% FE at −0.7 V vs. RHE (Figure 3a) [62]. Operando XAS identified the transformation from atomically dispersed Cu atoms to Cu_n clusters (n = 3 and 4) under CO₂RR conditions. According to the operando measurements and DFT calculation, the proposed reaction mechanism is illustrated in Figure 3b. Under open-circuit potential, the uniformly dispersed Cu ion was anchored by oxygen atoms from the hydroxyl group and water. After imposing a negative bias potential, the Cu²⁺ was reduced to Cu₃ or Cu₄ ligated by the surface hydroxyl group. Zhu et al., reported that a 3D dendritic cooper-cuprous oxide composite catalyst (Cu–Cu₂O/Cu) was obtained via in situ growth and decomposition of the corresponding complex film [65]. The catalyst formed in situ exhibited a high C₂ FE of 80% under a low applied potential of −0.4 V vs. RHE, attributing to the low charge transfer resistance and favorable C–C coupling.

Among the evolving intermediates of CO₂RR, *CO was an important intermediate and its binding strength and surface coverage influenced the overpotential for deep reduction products. The electric field was considered to stabilize adsorption species such as *CO, and to promote C–C coupling [71]. Fan et al. modified commercial Cu nanoparticles (NPs) with water-insoluble organosuperbases to enhance C₂₊ production [60]. The modifier with high pK_a would be protonated to a “proton sponge”, presenting a positive charge under CO₂RR. The positive species near the double-layer surface induced a large local electric field to stabilize *CO and promote C–C coupling, decreasing the onset potential of C₂H₄ from −0.57 V to −0.43 V. Furthermore, overpotential is reduced at the large current density, revealing the future potential for realizing the large-scale electrosynthesis of C₂₊ compounds from CO₂. Morales-Guio et al. fabricated the copper catalyst with tandem gold to increase the local CO concentration on the surface, and the reduction products (>2e⁻) were generated more than 100 times faster than on Cu catalyst. The detection overpotential for alcohol was reduced by 260 mV (Figure 3c,d) [63]. Theoretical calculation suggested that the estimated local CO concentration on the tandem catalysts’ surface was above solubility and much larger than that on Cu, due to the CO spillover from gold, contributing to the enhanced performance at low overpotential. Similar work was reported by Li et al, who functionalized the Cu surface with porphyrin-based metallic complexes for CO₂RR, and a low onset potential of −0.42 V vs. RHE for ethanol was observed [64].

Reducing the energy barrier to the following steps of *CO was taken into account to lower the overpotential for multi-electron electroreduction products. Nellaiappan et al. fabricated a nanocrystalline high-entropy alloy (HEA: AuAgPtPdCu) to convert CO₂ into gaseous products [59]. The main products of CH₄ and C₂H₄ were obtained at a low applied potential of −0.3 V vs. RHE, with FEs of 38.0% and 29.5%, respectively. DFT calculation suggested that the step *OCH₃ to *O was the RDS, and the HEA catalyst showed excellent performance at low overpotential. Ma et al. modified Cu with halogen and found that the onset potential of C₂H₄ decreased obviously as the electronegativity of the halogen increased (Figure 3e) [61]. Although the enhancement of CO adsorption may play a role due to the presence of Cu⁺, isotopic experiments and DFT calculation coupled with previous research indicated that electronegative fluorine enhanced the dissociation of H₂O into *H, thus lowering the barrier of hydrogenation of *CO, the RDS of CO₂RR to C₂H₄ on Cu given by DFT. In general, the overpotentials of multi-electron electroreduction products are difficult to reduce, due to the sluggish kinetics of the evolving reaction following *CO. Many researchers have recently focused on FEs and current densities and have made great progress, but the overpotential of CO₂RR should be paid more attention to reduce the cost of electricity to drive the reaction, to make CO₂RR competitive economically. Though researchers achieved limited activities at low overpotential, these provided insights into the mechanisms of CO₂RR and gave valuable information to support electrocatalyst design with high effciency at low overpotential.

3.2. High Faradaic Efficiency

Faradaic efficiency (FE) is the ratio between the amount of product detected and the amount of product theoretically formed, based on the charge passed through the circuit, and it expresses the selectivity of an electrocatalytic reaction. A high FE for CO₂RR is desirable to minimize the total current required for a target production rate and save the downstream separation costs. In the following section, we summarize recent strategies to raise Fes, according to the products shown in

Table 2. Recent progress on promoting current density and FE for CO₂RR products.

Major Product	Catalysts	Current Density (mA·cm−2)	Faradaic Efficiency (%)	Electrode Configuration	Reference
CO	5-fold twinned Ag NWs	~2.2	99.3	Non-GDE electrode	[72]
	ZrO2@Ni-NC	200	98.6	GDE	[73]
Formate	nBuLi-Bi	500	92		[74]
	Bi-NRs@NCNTs	6	90.9	Non-GDE electrode	[75]
Methane	SA-Zn/MNC	31.8	85		[76]
	Cu-DBC	203	80	GDE	[77]
	CuGaO2	717	71.7		[41]
Methanol	Cu1.63Se(1/3)	41.5	77.6	Non-GDE electrode	[78]
	Ag,S-Cu2O/Cu	122.7	67.4		[79]
Ethylene	Cu(OH)2-D/Cu	250	58	GDE	[25]
	F-Cu	1600	65		[61]
	Cu-Al	400	80		[26]
Ethanol	Cu3Ag1	25	63	Non-GDE electrode	[27]
	Cun (n = 3 and 4) cluster	~2	91		[62]
	Ag0.14/Cu0.86	250	41	GDE	[80]
Acetate	Cu2Ag3	~0.9	21.2	Non-GDE electrode	[81]
n-Propanol	CuSX-DSV	9.9	15.4		[82]

. High FE was achieved for CO and formate, even at a large current density scale, but it remains a challenge for deep electroreduction products. GDE plays a positive role in achieving current density as well as FE for the target products.

3.2.1. CO and Formate

High FEs for CO and formate in CO₂RR were comparatively easy to reach, even in early studies. Several recent studies have addressed catalyst design for these two-electron electroreduction products. It is noteworthy that non-copper catalysts such as Ag, Au, Sn, etc., are inclined to obtain two-electron electroreduction products.

Metal-based electrocatalysts usually possess high conductivity and easily facilitate electron transfer to the active surface sites, thus achieving good performance. Liu et al. synthesized five-fold twinned Ag nanowires through a facile bromide-mediated polyol method, which efficiently produced CO and achieved a maximum FE of 99.3% at −0.956 V vs. RHE [72]. The DFT calculation revealed that the improved performance over the catalysts originated from the diameter and length effects. as well as the special Ag (100) enclosed nanostructure and the increased edge-to-corner ratio. Gao et al. prepared a series of cadmium sulfide (CdS) nanostructures using a simple microwave heating strategy [83]. The enhancement of the electric field at the tips of CdS nanoneedles increased the neighbouring K⁺ concentration, which interacted with CO₂ via a non-covalent feature and facilitated the reaction. As a result, the CO formation FE for CdS nanoneedles (91.1%) greatly surpassed that for CdS nanorods (42.4%) and nanoparticles (25.1%). Fan et al. reported a chemical lithium tuning method to create abundant active grain boundaries (GB) on Bi nanoparticles [74]. Due to the small particle size and enriched GBs, the catalysts achieved a maximal formate FE of 97%, and maintained above 90% over a wide electrochemical window from −0.72 V to −1.05 V.

As a promising alternative to metal catalysts, carbon-based materials also showed high FEs for CO and formate production. Ren et al. reported that the isolated diatomic Ni–Fe sites supported on N-doped carbon achieved a CO FE above 90% over a wide potential range [84]. Theoretical DFT studies revealed that the diatomic site underwent a distinct structural change after adsorption of CO₂, which reduced the energy barrier for the formation of *COOH and desorption of CO. Zhang et al. observed a 90.9% FE for formate production in Bi nanorods with nitrogen-doped carbon nanotubes (Bi-NRs@NCNTs) [75]. The favorable nanocapillary and nanoconfinement effects of hollow Bi-NRs@NCNTs synergistically facilitated the mass transfer, adsorption, and concentration of reactant CO₂ molecules onto the active sites, promoting the formation of HCOO⁻. Chen et al. reported a 2D/0D composite catalyst composed of bismuth oxide nanosheets and nitrogen-doped graphene quantum dots (Bi₂O₃-NGQDs) for the efficient electrochemical reduction of CO₂ to formate [85]. The electrochemical tests demonstrated that the Bi₂O₃-NGQDs exhibited almost 100% FE for formate at -0.9 V vs. RHE. The DFT calculation disclosed that the high FE of the catalyst was attributed to the increased adsorption energy of *CO₂ and *OCHO intermediate after combination with NGQDs. Therefore, non-copper catalysts play a significant role in the generation of two-electron electroreduction products of CO and formate.

3.2.2. Methane

To raise the FE of CH₄, C–C coupling should be prohibited to produce C₂ products. SAC was shown to be an effective strategy for producing CH₄ exclusively, because the large distance between isolated Cu sites resulted in poor C–C coupling. Guan et al. demonstrated this effect in experiments by tuning the Cu concentration in SAC [69]. Through changing the calcination temperature, Cu-SACs with different Cu concentrations were fabricated. A high Cu concentration of 4.9%_mol was obtained, and the distance of neighboring Cu sites enabled C–C coupling to produce C₂H₄, with FE of CH₄ of only 13.9%. After raising the calcination temperature and lowering the Cu concentration to 2.4%_mol, the distance became so large that C–C coupling was prohibited, which resulted in a high FE of 38.6% for CH₄ production. A regular pattern was seen in Figure 4a. With the temperature elevated, the Cu concentration decreased, decreasing the current density of C₂H₄ and continuously increasing the ratio of CH₄ to C₂H₄. To further enhance the FE of methane on SAC, Cai et al. fabricated carbon-dots-based SAC margined with CuN₂O₂ sites, using the partial-carbonization method [70]. The introduction of oxygen ligands modified the electronic structures of the center atoms, leading to a high FE of 78% for methane production. SACs based on other metals also effectively enhanced the FE of methane. Han et al. chose Zn as the metal center and anchored it onto microporous N-doped carbon [76]. When Zn was the metal center, the formation route of CH₄ involved the formation of *OCHO instead of *COOH, which was crucial for CO generation, and the CO₂-to-CO pathway was largely suppressed. As a result, the catalyst exhibited a high FE of 85%, and a partial current density of 31.8 mA cm⁻² for methane at a potential of −1.8 V vs. SCE.

Catalysts with single Cu sites promoted the FE of methane. Wang et al. designed a single atomic Cu substituted CeO₂ catalyst with multiple oxygen vacancies to optimize the CO₂RR to CH₄ [87]. The single-atomic Cu on CeO₂ (110) stabilized three oxygen vacancies around each Cu site, which was evidenced by X-ray photoelectron spectroscopy (XPS) and XAS. These features were beneficial for CO₂ adsorption and activation, enabling an FE of 58% towards methane. Zhang et al. fabricated a conductive metal-organic framework with Cu-O₄ sites, which showed a high FE of 80% and a current density of 203 mA·cm⁻² when tested in the flow cell (Figure 4b) [77]. Due to the smaller charge transfer resistance and favorable adsorption of *H on the O atom, the Cu-O₄ site was feasible to accept an (H⁺/e⁻) pair reduced to +1 valence, which was beneficial to the CO₂ adsorption and reduction. Recently, our group fabricated ultrathin rhombohedral-phase CuGaO₂ nanosheets by an induced anisotropic growth strategy, to convert CO₂ to CH₄ efficiently [41]. On the surface of CuGaO₂ nanosheets, the (001) surface was highly exposed, which possessed abundant single-interlayered Cu(Ⅰ) with a large Cu–Cu distance of 3.013Å, and prohibited C₂H₄ formation. The DFT calculation indicated that the CO pathway was energetically unfavorable due to the high CO desorption barrier. The DFT suggested that water dissociation to generate the first adsorbed *H on CuGaO₂ was comparatively easier than CuAlO₂. Considering that the pathway towards CH₄ involved multiple hydrogenation steps, the simpler process of water dissociation enhanced CH₄ production. With these features, the CuGaO₂ nanosheets exhibited efficient CO₂-to-CH₄ electroreduction with FE of 71.7% at a high CH₄ partial current density of 717 mA cm⁻², substantially outperforming the previous catalysts.

While the mechanism of single-site Cu catalysts prohibited C–C coupling, lowering the intermediate concentration also decreased C–C coupling and enhanced the FE towards methane. Wang et al. directly changed the partial pressure of the CO₂ gas stream towards Cu and observed that methane FEs in dilute CO₂ increased at high reaction rates [32]. According to Henty’s law, local CO₂ concentration was proportional to the partial pressure, and the local *CO₂ coverage was positively related to CO₂ concentration.Therefore, tuning the CO₂ partial pressure effectively changed the *CO₂ coverage, which further decreased *CO coverage according to DFT studies. Finally, the value of (ΔE_CHO-ΔE_OCCOH), considered as the descriptor for the *CHO vs. C–C coupling, decreased when *CO coverage was reduced, thus comparatively enhancing the methane FE. In experiments, the CO₂RR performance at various CO₂ concentrations (25%, 50%, 75%, 100%) was evaluated by tuning the volume ratios of CO₂ to N₂ in the gas streams. At high current densities, the ratio between CH₄ and C₂H₄ increased with decreasing CO₂ concentration (Figure 4c), in accord with the theoretic prediction mentioned above. Although low CO₂ concentration was unfavorable for CO₂RR, the CO₂ concentration of 75% exhibited the highest methane FE of 48%, with a partial current density of 108 mA·cm⁻² at −1.416 V vs. RHE, obviously higher than the pure CO₂ gas stream which mainly produced ethylene. Xiong et al. fabricated yolk-shell nanocell structures which involved an Ag core and a Cu₂O shell resembling a tandem nanoreactor [88]. By fixing the Ag core and changing the Cu₂O envelope size, the CO flux arriving at the shell was regulated, which further modulated the *CO coverage and *H adsorption and facilitated CH₄ production. As a result, the best catalyst among structures with different envelope sizes exhibited a high CH₄ Faraday efficiency of 74% and a partial current density of 178 mA·cm⁻² at −1.2 V vs. RHE.

The electrolyte component changes the selectivity of CO₂RR, and the organic additives in the aqueous electrolyte have been reported to enhance the FE of methane, partly attributed to the increase of CO₂ solubility. Qui et al. added methyl carbamate (MC) containing an –NH₂ group into 0.5 M aqueous solution of NaHCO₃, with Cu foil as the catalyst [33]. The –NH₂ group in MC is prone to generate an –NH₃ group in an aqueous solution, and the affinity effect of the Lewis acid (–NH₃⁺) with the Lewis base (CO₂) increased the solubility of CO₂. Furthermore, the generated –NH₃⁺ group promoted the adsorption of *CO and *CHO and stabilized the *CHO intermediate via the facilitation of partial electron transfer from Cu to C, and the formation of H-bonds. As a result, the FE of CH₄ increased by 30% with the addition of MC. Organic additives influenced the morphology of catalysts. Han et al. used ethylenediamine tetramethylenephosphonic acid (EDTMPA) to promote CO₂ electroreduction to methane [89]. Experimental and theoretical studies have shown that EDTMPA molecules were preferentially adsorbed on Cu (110) during the CO₂RR, which not only induced the selective generation of Cu (110) to shave an inherently high *CO binding strength, but also formed a local environment that promoted proton transfer from water to the catalyst surface. After adding EDTMPA, the FE of methane increased to 61% and 64% in H cell and flow cell, respectively, demonstrating the effect of EDTMPA in CH₄ promotion at current density with a different order of magnitude.

3.2.3. Methanol

As the most widely used materials in CO₂RRM, Cu based materials were also effective for enhancing the FE of methanol. Yang et al. reported that Cu selenide NPs exhibited outstanding performance for CO₂RR to methanol [78]. At a low overpotential of 285 mV, the current density was up to 41.5 mA cm⁻², and FE reached 77.6%. The control experiment suggested that Se in the catalysts was crucial for efficient methanol production, compared with Cu oxide and sulfide, and the Se atom in the catalysts was unsaturated, further enhancing the performance. Zhao et al. fabricated single atom immobilized MXene for methanol production by selective etching of hybrid A layers (Al and Cu) in quaternary MAX phase (Ti₃(Al_1−xCu_x)C₂), illustrated in Figure 4d [86]. According to XAS and DFT calculation, this etching process resulted in Cu atoms with unsaturated electronic structures, which delivered a low energy barrier for the RDS (formation of *CHO), thus achieving a high FE of 59.1% towards methanol, with good stability. Li et al. studied the influence of dual doping in a Cu₂O/Cu host on the CO₂-to-methanol process, using cation (Ag, Au, Zn, Cd) and anion (S, Se, I) [79]. The results indicated the Ag and S dual doping exhibited good performance, achieving a methanol FE of 67.4% at potential of -1.18 V vs. RHE, with a high current density of 122.7 mA·cm⁻² in an ionic liquid/H₂O electrolyte (Figure 4e). The DFT calculation indicated that the anion S modified the electronic structures of neighboring Cu atoms, which reduced the energy barrier to *CHO formation, and the cation Ag mainly increased the reaction barrier to HER.

Other non-Cu catalysts have been reported to obtain high FE for methanol. Zhang et al. designed ultrathin Pd nanosheets partially capped by SnO₂ NPs to selectively produce methanol [90]. The combination of Pd and SnO₂ not only improved the adsorption capacity of CO₂ on the catalyst surface, but also weakened CO poisoning on Pd, facilitating the formation pathway of CH₃OH. By tuning the Pd/Sn ratio, the 2D hierarchical structure achieved a maximum FE of 54.8% for methanol at −0.24 V vs. RHE. Ji et al. reported the Fe₂P₂S₆ nanosheets efficiently catalyzed the CO₂RR to alcohols in aqueous electrolyte, with a total FE_{methanol+ethanol} of 88.3%, and 65.2% for methanol FE [91]. An Fe atom on the Fe₂P₂S₆ surface was regarded as the active site for alcohol formation, according to the DFT calculation. In addition to metal catalysts, other catalysts without metal have also been reported. For example, boron phosphine NPs were demonstrated to efficiently convert CO₂ to methanol, achieving a high FE of 92.0% for CH₃OH at −0.5 V vs. RHE [92]. Based on DFT calculations, it was demonstrated that the outermost B sites strongly adsorbed CO₂, after which P donated electrons to B, which synergistically activated CO₂. In addition, the desorption of CO and OCH₂ were energetically unfavorable, which contributed to the high selectivity of the CO₂-to-CH₃OH conversion process.

3.2.4. Ethylene

The RDS of C₂H₄ formation was C–C coupling between *CO or *CHO. After C–C coupling, C₂₊ products were generated. Therefore, facilitating C–C coupling was crucial for raising the FE of ethylene in electrochemical CO₂RR [18,93]. Hori et al. studied CO₂RR on Cu single crystal and showed that the product distribution was dependent on the facet of the catalyst surface [94]. Specifically, CH₄ and C₂H₄ were the main products on Cu (111) and Cu (100), respectively, while Cu (110) suppressed FEs of CH₄ and C₂H₄ but favored the formation of C₂₊ oxygenates. Facet tuning became an effective strategy for promoting electrochemical CO₂RR. Recently, Gao et al. prepared Cu₂O NPs with different crystal facets and found that the Cu₂O NPs enclosed with both (111) and (100) facets exhibited excellent performance, with a maximum FE of ethylene of 59% [24]. The XPS indicated that these Cu₂O NPs were well preserved after the stability test, with only tiny NPs on the surfaces ascribed to the Cu NPs formed during the electroreduction process. The DFT calculation suggested that the surface of t-Cu₂O combined the strong CO adsorption capacity of Cu₂O (100) and the easy C₂H₄ desorption of Cu₂O (111), thus promoting ethylene production. Besides, the difference from the Fermi level of Cu₂O facilitated the charge transfer between the two facets, and further promoted the multi-electron kinetics involved in ethylene production. Previous studies on facet dependence were mostly performed in H-cells with slow reaction rate, while Gregorio et al. demonstrated that facet-dependent selectivity was retained in gas-fed flow cells at large current densities [95], which suggests that the facet dependence found in the lab could provide useful information for future applications at large current densities. Furthermore, hydrogen production was suppressed due to the high alkaline conditions in the flow cell, which further increased the conversion of CO₂RR. In addition to these low-index surfaces, high-index surfaces with abundant active sites changed the product distribution. Zhong et al. prepared the Cu catalyst derived from Cu(OH)₂, CuO, and Cu₂O, respectively, and the Cu(OH)₂-derived catalyst (Cu(OH)₂-D/Cu) exhibited the best performance [25]. Transmission electron microscopy (TEM) images showed the Cu(OH)₂-D/Cu with a high-index stepped surface, such as Cu(310) = 3(100) × (110) and Cu(210) = 2(110) × (100), composed of Cu(100) and (110) (Figure 5a). In situ spectroscopy revealed the Cu(OH)₂-D/Cu with a high-index facet promoted the adsorption of CO intermediate, while the DFT calculation demonstrated that the activation energy of CO dimerization on Cu(210) and Cu(310) was lower than that on the low-index facet. Performance testing in H-cell showed that the Cu(OH)₂-D/Cu exhibited higher selectivity for C₂₊ products than CuO-D/Cu and Cu₂O-D/Cu, achieving ≈59% at −0.98 V, compared with ≈45% for CuO-D/Cu and about 21% for Cu₂O-D/Cu at ca. −1.03 V and −1.23 V, respectively. To overcome the solubility limitation of CO₂ in the aqueous solution and meet the requirements for industry application, the performance of Cu(OH)₂-D/Cu was assessed with a flow-cell electrolyzer at 1 M KOH. Resulting from the alkaline environment, the FE of H₂ was suppressed, and the Cu(OH)₂-D/Cu catalyst achieved a high FE of 87% for C₂₊ product, with 58% for ethylene at −0.54 V vs. RHE at total current density of 250 mA·cm⁻²_. The strong performance of Cu(OH)₂-D/Cu at current densities with different orders of magnitude suggests that facet engineering is an effective strategy to enhance the FEs of C₂₊ products in future applications.

Surface modification by organic addiction was proved effective to enhance the FE of ethylene. Wakerley et al. modified the Cu dendrites with 1-octadecanethiol to tune the selectivity of CO₂RR [97]. Contact angle measurements illustrated a drastic improvement in hydrophobicity after modification, with contact angles increasing from 17° to 153°.This hydrophobic feature coupled with hierarchical structures resulted in gas trapping at the interface between the catalyst surface and electrolyte, which facilitated the mass transport of CO₂ and decreased that of H⁺. Therefore, the FEs for electrochemical CO₂RR were obviously enhanced after modification, with methane going from 0% to 7%, ethylene from 9% to 56%, and ethanol from 4% to 17%, but with slightly lower current density. Wei et al. coated the Cu surface with a 50 nm thick film of polyaniline (PANI), enhancing the selectivity of CO₂RR for both Cu foil and Cu NPs [28]. The hydrophobic nature of the polymer suppressed the HER, and the in situ spectroscopy revealed that the Cu/PANI interface improved the coverage of *CO and strengthened interactions with *CO, leading to a higher probability of C–C coupling. Hence, when PANI coating was applied to Cu nanoparticles, the FE of C₂₊ hydrocarbon reached 80% and the FE of ethylene over 40%. The modification of the function group of the organic additive further enhanced the selectivity of C₂H₄. Li et al. investigated the influence of eleven kinds of molecules derived by the electro-dimerization of arylpyridiniums on Cu catalysts [29]. The in situ Raman spectroscopic interrogation indicated that different functional groups caused differences in the electron-donating abilities of molecules, and further caused the change of the relative populations of atop-bound CO (CO_atop) and bridge-bound CO (CO_bridge). Combined with CO₂RR performance, a volcano-shaped trend between the FE of ethylene and the ratio of Cu_atop and Cu_bridge was found, with the ratio near 1 obtaining the highest C₂H₄ selectivity (Figure 5b), suggesting that C–C coupling between Cu_atop and Cu_bridge was favorable. Accordingly, the DFT calculation indicated that in the presence of the organic additive, the adsorption of CO on atop sites increased compared with that on bridge sites. Furthermore, the CO dimerization between Cu_atop and Cu_bridge possessed the lowest energy barrier, according to with the experimental results. To further raise the FE of ethylene, the researchers synthesized the molecule with optimal electro-donating properties and many electro-donating N atoms. Hence, the ethylene FE on this catalyst (named Cu-12) reached 72% at −0.83 V, with a current density of 232 mA·cm⁻² (Figure 5c), surpassing Cu and the eleven catalysts with other additives.

While previous research demonstrated that Cu-based alloy enhanced C₂H₄ production, various guest elements and active sites made the discovery of the efficient alloy catalysts time-consuming. Recently, Zhong et al. used DFT calculation in combination with active machine learning and found that Cu-Al electrocatalysts efficiently reduced CO₂ to C₂H₄ [26]. 12,229 surfaces and 228,969 adsorption sites from 244 different copper-containing intermetallic crystals were obtained from The Materials Project, and then a subset of these sites was assessed using DFT simulation to calculate their CO adsorption energies (ΔE_co). These data were used to train a machine learning model which was employed to predict ΔE_co on the adsorption sites. Combined with volcano scaling relations which have optimal ΔE_co near −0.67 eV, adsorption sites with optimal ΔE_co were predicted and then simulated using DFT to provide additional training data. In total, about 4000 DFT simulations were carried out to yield a set of candidates for experimental testing. Among these, Cu-Al exhibited the highest abundance of sites and site types with near-optimal ΔE_co, illustrated in Figure 5d. Therefore, Cu-Al catalysts were prepared and tested experimentally, and the de-alloyed Cu-Al catalysts on polytetrafluoroethylene (PTFE) substrates showed excellent performance. Under electrochemical CO₂RR conditions, the FE of ethylene achieved 80% at a current density of 400 mA·cm⁻². This research illustrated that computation and machine learning were useful tools to guide the experimental exploration of efficient CO₂RR catalysts, which may be widely used in the future.

3.2.5. C₂₊ Oxygenates

As mentioned above, ethylene was usually the major product after C–C coupling, while C₂₊ oxygenates, including C₂H₅OH, CH₃COO⁻, and n-C₃H₇OH, generally accounted for a small part of the C₂₊ product. Therefore, if C₂₊ oxygenates were the target products in the electrocatalytic CO₂RR, elaborate catalyst design was required to improve the FE of C₂₊ oxygenates.

It is usually believed that bimetallic catalysts composed of Cu and a metal capable of catalyzing CO₂ to CO, such as Ag, Au, or Zn, would enhance the selectivity of C₂₊ oxygenates in the electrocatalytic CO₂RR [80,98]. Among these, the Cu-Ag catalysts have shown a great potential for obtaining high FEs of C₂₊ oxygenates. Li et al. fabricated bimetallic Cu-Ag catalysts with different ratios via co-sputtering [80]. The X-ray diffraction (XRD) and XAS characterization revealed that Ag and Cu were in an alloy phase. For the good catalysts (Ag_0.14/Cu_0.86), the FE of ethanol reached 41% with a current density of 250 mA cm⁻² at −0.67 V vs. RHE, higher than the pure copper control (29%). According to the DFT calculation, the introduction of Ag disrupted the uniform Cu sites, resulting in a broader peak of CO adsorption during in situ Raman spectrum analysis, and destabilized the ethylene intermediates, thus promoting ethanol formation. Lv et al. developed a Cu₃Ag₁ electrocatalyst by galvanic replacement of an electrodeposited Cu matrix [27]. From XPS and XAS results, it was concluded that the introduction of Ag in this catalyst facilitated the electron transfer from Cu atoms to Ag atoms, leading to electron deficiency in the Cu matrix. DFT calculation revealed that an alcohol-pathway intermediate like *OCH₂CH₃ was more stabilized on electron-deficient Cu atoms than on pure Cu. As a result, the Cu₃Ag₁ catalyst enabled a 63% FE for CO₂-to-alcohol production at −0.95 V vs. RHE. For acetate production, Wang et al. synthesized ultrasmall Cu_nAg_m bimetallic NPs to enhance the acetate FE [81]. They fabricated a series of these bimetallic NPs through electroreduction on electrochemically polymerized poly-Fe(vbpy)₃(PF₆)₂ film electrodes on glassy carbon. It was found that the FE of CH₃COO⁻ increased with the addition of Ag to the nanoparticles, reaching a maximum of 5.5% for Cu₂Ag₃ catalysts (Figure 5e). Through further optimization of the reaction conditions, the FE of acetate reached 21.2% at −1.33 V vs. RHE in a −0.5 M KHCO₃ solution with 8 ppm benzotriazole at 0 ℃.

Combining Cu catalysts with carbon-based materials raises the FE of C₂₊ oxygenate. Chen et al. used N-doped graphene quantum dots (NGQ) as the second component to modify CuO-derived nanorod (Cu-nr) [99]. Characterizations such as TEM and XAS indicated the uniform dispersion of the NGQ on the Cu-nr. During the electrochemical reduction, the structures were retained, and the FEs of C₂₊ alcohols reached 52.4% with a current density of 282 mA cm⁻² at −0.9 V vs. RHE, obviously higher than bare NGQ and Cu-nr. Control experiments and DFT studies showed that the combination of materials exhibited the synergistic effect and stabilized *C₂H₃O, the intermediate to C₂₊ alcohols, thus enhancing activity and selectivity. Wang et al. reported that confinement using a nitrogen-doped carbon (N-C) layer on Cu catalysts promoted ethanol production [100]. They fabricated N-C/Cu by sputter deposition of a layer of N-C on the surface of sputtered Cu NPs, achieving an ethanol FE of 52% at −0.68 V vs. RHE. The different plots of electron density generated by DFT showed that the N-C layer was beneficial for electron transfer to adsorbed *CO on Cu, compared to the carbon layer, facilitating the generation of the C–C coupled intermediate. Furthermore, the capping layer stabilized the C-O bond of *CHCOH, suppressing the deoxygenation process and favoring the ethanol pathway instead of ethylene. The carbon-based materials not only stabilized the reaction intermediates, but also influenced the dispersion of active sites. Using reversible transformation, the Cu SAC synthesized by an amalgamated Cu–Li method converted the CO₂ to ethanol with an extraordinary 91% FE at −0.7 V vs. RHE (Figure 3a) [62]. During the process of catalyst synthesis, the carbon surface formed hydroxyl groups due to the highly basic environment caused by the dissolved LiOH. These hydroxyl groups were found to play a crucial role in stabilizing the transiently formed Cu clusters, which were responsible for the high selectivity towards ethanol.

For C₃ products, it remains difficult to achieve high FEs in electrochemical CO₂RR. The reason was that the formation of C₃ product requires the stabilization of C₂ intermediate, and a number of complex active sites not only for C₁–C₁ coupling but also for C₁–C₂ coupling. Furthermore, the mechanism towards C₃ products has been far less well understood compared to C₂ products. Nonetheless, some strategies have been proposed to raise the FEs of C₃ oxygenates in electrochemical CO₂RR, especially for n-C₃H₇OH. Rahaman et al. reported the fabrication of a dendritic Cu by a stepwise method, which enhanced the FE of n-C₃H₇OH by a process involving the electrodeposition of dendritic Cu catalysts under mass transfer control using Cu(Ⅱ) ions followed by 3 h thermal annealing at 300 ℃ [96]. The initial electrodeposition catalysts mainly produced HCOO⁻ and C₂H₄, but the annealed samples directed the major product towards C₂₊ alcohols (Figure 5f), with the FE of n-C₃H₇OH reaching 13.1%. It was proposed that the presence of smaller nm-sized Cu NPs/Cu crystals and nanocavities on the surface of the dendritic Cu played a key role in promoting C–C coupling and C₂₊ alcohol production. Using an electrochemical lithium method, Peng et al. modulated the density of sulfur vacancies on CuS by tuning the cycle number, and synthesized CuS_X catalysts with a double-sulfur vacancy (DSV) [82]. The DFT calculation revealed that the DSV formed on hexagonal copper sulfide allowed enriched negative charges near two adjacent co-planar vacancies and stabilized the *OCCO dimer. In addition, the coupling energy between a third *CO and the dimer was decreased in the presence of the DSV. As a result, after 10 cycles the CuS_x catalysts involving the rich DSVs, exhibited an FE towards n-propanol of 15.4% at −1.05 V vs. RHE, higher than other samples (Figure 5g). The decrease in FE_n-propanol with numerous cycles was ascribed to the damage of the formed DSV sites.

3.3. Current Density

The current density is another key parameter in electrochemical CO₂RR, describing the activity of the reaction. However, as mentioned in the introduction, the stable C=O bonds and sluggish kinetics result in poor current density in CO₂ reduction. Therefore, to enable electrochemical CO₂RR applicability, strategies based on reactors and catalysts were needed to promote current density to an industrial-relevant level (>200 mA·cm⁻²) [101].

Early studies operated in H-cells showed low current density (≤100 mA·cm⁻²) due to the mass-transport limitation of CO₂. Flow cells with GDE configuration were widely used to promote the transport of the reactant and significantly raise the current density of CO₂RR.

Carbon-based gas diffusion layers (GDLs) were often used in previous studies on gas diffusion electrodes (GDEs) for CO₂RR [102]. As shown in Figure 6a, recently developed GDEs typically contain GDL consisting of a microporous substrate (MPS), a microporous layer (MPL), and a catalyst layer (CL) [103]. The MPS is usually made of conductive carbon fibers or titanium foam coated with PTFE, providing mechanical stability and electrical contact, as well as the distribution of CO₂ gas through its macro-scale pores. MPLs are comprised of carbon nanoparticles and polytetrafluoroethylene (PTFE), which enhance the interfacial electrical connection and prevent flooding in the GDE. CLs are prepared by coating MPLs with a mixture of catalyst particles and ionomers, as binders. Ionomers hold catalyst particles together and concurrently offer ionic conductivity to CLs.

The flow cell consists of three flow channels for the CO₂ gas, catholyte, and anolyte, as shown in Figure 6b [104]. The catholyte and anolyte liquid streams are separated by an ion-exchange membrane, which prevents CO₂RR products from crossing over to the anode where they can be oxidized back into CO₂, and restricts the evolved O₂ at the anode from crossing over to the cathode and stealing electrons for ORR. The GDE separates the catholyte and gas channel. The catalyst layer, located on the liquid facing the front of the GDE, is in contact with the electrolyte, while gas-phase CO₂ is continuously delivered to the catalyst through the back of the GDE, thus lowering the CO₂ diffusion to the surface of the catalyst by three orders of magnitude and achieving a high current density above 200 mA·cm⁻² [29,41,77,105]. SACs were found to achieve a large current density with the use of GDE [106,107,108]. Recently, the self-supported single-atom nickel-decorated porous carbon membrane catalyst was directly used as a gas diffusion electrode, combining gas diffusion and catalyst layers into a single architecture [109].

The improvement in current density not only satisfies the basic demand of the industrial process, but also provides ways to compare the performance of catalysts at current densities of different orders of magnitude within the laboratory, thus giving more guidance for future applications. Additionally, the high feeding rate of CO₂ combined with a short diffusion distance, allowed highly alkaline electrolytes, increased conductivity, and suppressed HER, further promoting electrocatalytic performance.

Figure 6. (a) Schematic illustration of the structure of gas diffusion electrode [103]; copyright Royal Society of Chemistry, 2020. (b) Schematic illustration of the structure of a flow cell [104]; copyright John Wiley & Son, 2019. (c) Illustration of the promotion of gas and ion transport by the hydrophobic -CF₂ layer and hydrophilic −SO₃⁻ layer, respectively [105]; copyright AAAS, 2020. (d) Current density on F-Cu catalysts compared by other typical electrocatalysts (squares and circles represent the electrocatalytic reactions conducted in an H-shape cell and flow cell, respectively) [61]; copyright Springer Nature, 2020. (e) The stability of Cu-12 with N-aryl-dihydropyridine-based oligomer [29]; copyright Springer Nature, 2019. (f) Long-term catalytic performance test of CO₂ reduction to ethylene in 7 M KOH catalyzed by the graphite/carbon NPs/Cu/PTFE electrode, compared with a traditional carbon-based gas diffusion electrode (GDE) (insets show the cross-section SEM and energy-dispersive X-ray spectroscopy mapping of the sample after 150 h of continuous CO₂ reduction operation) [110]; copyright AAAS, 2018. (g) Schematic illustration of the structure of a membrane electrode assembly (MEA) cell [111]; copyright MDPI, 2020.

Figure 6. (a) Schematic illustration of the structure of gas diffusion electrode [103]; copyright Royal Society of Chemistry, 2020. (b) Schematic illustration of the structure of a flow cell [104]; copyright John Wiley & Son, 2019. (c) Illustration of the promotion of gas and ion transport by the hydrophobic -CF₂ layer and hydrophilic −SO₃⁻ layer, respectively [105]; copyright AAAS, 2020. (d) Current density on F-Cu catalysts compared by other typical electrocatalysts (squares and circles represent the electrocatalytic reactions conducted in an H-shape cell and flow cell, respectively) [61]; copyright Springer Nature, 2020. (e) The stability of Cu-12 with N-aryl-dihydropyridine-based oligomer [29]; copyright Springer Nature, 2019. (f) Long-term catalytic performance test of CO₂ reduction to ethylene in 7 M KOH catalyzed by the graphite/carbon NPs/Cu/PTFE electrode, compared with a traditional carbon-based gas diffusion electrode (GDE) (insets show the cross-section SEM and energy-dispersive X-ray spectroscopy mapping of the sample after 150 h of continuous CO₂ reduction operation) [110]; copyright AAAS, 2018. (g) Schematic illustration of the structure of a membrane electrode assembly (MEA) cell [111]; copyright MDPI, 2020.

Based on the flow cell with GDE, modifying the catalyst’s surface with ionomer further enhanced the CO₂ mass transfer, thus increasing the current density. Arquer et al. spray-coated a solution of perfluorinated sulfonic acid (PFSA) ionomers onto hydrophilic metal catalysts to establish a catalyst: ionomer planar heterojunction [105]. The PFSA ionomers containing -SO₃⁻ (hydrophilic) and -CF₂ (hydrophobic) groups, formed colloids with hydrophilic -SO₃⁻ groups exposed to solvent in a polar electrolyte, which was proved by ex situ and in situ surface-enhanced Raman spectroscopy (SERS). Therefore, this controlled assembly into distinct hydrophobic and hydrophilic layers offered differentiated pathways whereby gas transport was promoted through the hydrophobic layers, and ion transport was enhanced by the hydrophilic layers (Figure 6c), thus promoting CO₂RR due to the benefit from a three-phase reaction interface. Enhanced gas transport with decent water availability was evidenced by the oxygen reduction reaction (ORR) and HER. To maximize the three-phase reaction interface, they fabricated the 3D catalysts by ionomer bulk heterojunction consisting of Cu NPs and PFSA, and spray-casted on a PTFE/Cu/ionomer gas diffusion support. Hence, the catalyst achieved CO₂ electroreduction in 7 M KOH with a peak ethylene partial current density of 1.3 A·cm⁻².

Other strategies for overpotential decrease or selectivity enhancement also increased the current density. For example, the fluorine-modified Cu (F-Cu) catalysts mentioned earlier achieved high current density in an alkaline flow cell [61]. The modification of F enhanced CO adsorption, water activation, and the hydrogenation of *CO to *CHO, and the kinetics of C₂₊ formation was promoted. Hence, the current density reached 1.6 A·cm⁻² with an FE of 80% for C₂₊ products, surpassing typical electrocatalysts (Figure 6d). The electrochemical lithium tuning strategy to form vacancies was used to raise the FE of n-propanol, as already mentioned, and also enhanced the current density [82,112,113]. During the process, the Li⁺ intercalated into the lattice of synthesized Cu₃N to form Li₃N, based on the partial conversion reaction of Cu₃N + Li⁺ + e⁻ → Cu₃N_x + Li₃N (0 < x < 1). The nitrogen vacancy density was adjusted by controlling the charge–discharge current. With an optimal current of 50 μA, the Cu₃N_x catalysts exhibited a partial current density towards C₂ products of 307 mA·cm⁻², with a high FE of 81.7%. The DFT calculation revealed that the nitrogen-vacancy-rich Cu₃N promoted *CO adsorption and C–C coupling, thus enhancing the electrocatalytic performance. Since the Li₂CuO₂ involved Li species, the Li vacancies were directly formed in the discharging process. It was suggested by DFT that the lithium vacancies (V_Li) resulted in the short distance to the adjacent [CuO₄] layer and reduced the coordination number of Li⁺ around Cu, thus lowering the energy barrier of C–C coupling and enhancing the activity and selectivity. Therefore, with the V_Li percentage around 1.6%, the Li_2-xCuO₂ catalysts showed a high current density of 706 mA·cm⁻² with an FE of 90.6% towards C₂₊ products. A similar strategy was adopted to obtain a high current density for formate production. Yan et al. developed a surface-lithium-doped tin (s-SnLi) catalyst by controlled electrochemical lithiation [114]. DFT calculations indicated that the Li dopants introduced electron localization and lattice strains on the Sn surface, thus enhancing the activity and selectivity of the CO₂ electroreduction to formate. As a result, the s-SnLi electrocatalyst exhibited strong performance, with a partial current density of 1.0 A·cm⁻² for producing formate, and a corresponding FE of 92 %. For CO production, Wang et al. developed a porous Ni-N-C catalyst containing atomically dispersed NiN₄ sites and nanostructured zirconium oxide (ZrO₂@Ni-NC) via a post-synthetic coordination coupling carbonization strategy [73]. In a flow cell, the ZrO₂@Ni-NC delivered a current density of 200 mA·cm⁻² with CO selectivity of 98.6% at −1.58 V vs. RHE. The experimental results demonstrated that the introduced nanostructured ZrO₂ accelerated the transfer rate of the first electron and altered the RDS from conventional CO₂ adsorption to the protonation of *CO₂⁻ intermediate to form *COOH intermediate, thus achieving a high reaction rate.

3.4. Stability

Considering the demand for large-scale application of electrochemical CO₂RR, stability is also an important parameter, as it determines whether cost efficiency in mass production can be achieved. It has been suggested that the short life of the system causes an exponential increase in maintenance costs and the infeasibility of the practical economy [12]. However, the stability issue affecting the reaction has often been overlooked compared to activity and selectivity. The stability issue involves the maintenance of catalysts and electrochemical systems. In this section, the factors causing poor stability are introduced, followed by strategies to inhibit effectively these deactivation processes, which include surface capping, cutting-edge design, and improved cell configuration.

The catalysts’ surface morphology has been acknowledged as one of the most important factors determining catalytic activity and selectivity. However, metal-based catalysts are susceptible to surface reconstruction, leading to unsatisfying stability. For example, the high-index facets of Cu showing high FE for ethylene usually underwent reconstruction due to their high surface energy during electrocatalysis. An increase in the average size of crystalline NPs occurred due to the tendency for the smaller nanosized particles to adjust their surface energy via various growth mechanisms such as Ostwald ripening and coalescence [115,116]. An effective measure to preserve the morphological change was the usage of surface capping ligands as surface passivating reagents. Ma et al. prepared Au NPs on carbon support modified by polyvinyl alcohol (PVA), exhibiting a consistent FE of CO above 95% and a current density of 27 mA·cm⁻² for 24 h reaction time [117]. It was suggested the longer durability compared to that of bare Au/C arose from the Au NPs’ immobilization on carbon support by the PVA surface capping ligands. Zhang et al. showed that N-heterocyclic carbene ligands enhanced the stability of CO₂RR [118]. In addition to these surface capping ligands, other organic additives also extended the operational time. Li et al. fabricated an N-aryl-substituted tetrahydro-bipyridine as an additive on sputtered Cu catalysts, by the electrochemical dimerization of N-arylpyridinium additives [29]. The catalysts exhibited an ethylene FE of 65% at −0.83 V vs. RHE, and after further optimization by increasing the number of electro-donating N atoms, the sputtered Cu catalysts modified by electro-oligomerized N,N’-(1,4-phenylne) bispyridinium showed remarkable stability over 190 h, with high current density and C₂H₄ selectivity (Figure 6e).

Although alloyed metals were observed to enhance the electrocatalytic performance in CO₂RR, they were restricted by the disadvantage that each metal tended to aggregate into a single metal state during the reaction, and the metal with low surface energy preferred to compose an outer layer by segregating from the alloy [119,120]. To solve these issues, a cutting-edge design was more recently proposed. Sun et al. fabricated AuFe@Fe core-shell NPs by leaching out Fe from prepared AuFe alloyed NPs, which showed stable CO production for 90 h [121]. The DFT calculation suggested that the high stability arose from dimpled Au surfaces, which mitigated Au diffusion and aggregation. Zhu et al. synthesized Au₃Cu alloy with abundant vacancies by acid etching from as-prepared Au₃Cu alloy NPs [122]. The prepared alloy NPs were stable under CO₂RR condition for 30 h, with a high CO FE of 94.3% at −0.43 V vs. RHE.

In addition to the deactivation of catalysts, the overall stability of CO₂RR was also dependent on the reactors. As mentioned earlier, the flow cell with GDE configuration enhanced the mass transfer and increased the current density, but faced severe problems such as flooding and salt accumulation. It was a challenging task to achieve the long-term stability of CO₂RR.

Previous studies of GDEs for CO₂RR mainly used carbon-based GDLs [102], and the issue of flooding caused by the penetration of H₂O into GDE resulted from the surface of carbon in the GDE turning hydrophilic during long-term operation. The diffusion of gaseous reactant towards the catalysts was prohibited and the performance was impaired. To overcome this issue, the modification of GDE was adopted. For example, Ma et al. modified the GDE by coating 20 μL 1H,1H,2H,2H-perfluorooctyltrichlorosilane (5 wt% in toluene) onto the gas diffusion layer to enhance hydrophobicity [61]. The modified catalyst was stable for 40 h at a fixed current density of 400 mA·cm⁻², towards ethylene and ethanol. Dinh et al. fabricated GDL with separate PTFE and carbon NP layers to decouple the hydrophobic and current collection requirements of traditional carbon-based GDLs [110]. The pure PTFE layer acted as a stable hydrophobic GDL that prevented flooding, and the Cu catalysts were sandwiched by the two layers stabilized by the carbon NPs and graphite. After the fabrication, the graphite–carbon NPs/Cu/PTDE electrode operated for 150 h without loss of ethylene selectivity, demonstrating a 300-fold increase in the reaction time compared with that of the Cu–carbon GDL (Figure 6f). In fact, the GDL has often been overlooked because of the availability of a variety of commercial GDLs for fuel-cell application. It is worth noting that most of the GDLs tested so far are not stable for CO₂ reduction, mainly because of salt accumulation [123,124,125]. The creation of next-generation GDLs with mechanical and chemical properties suited for CO₂ reduction is necessary to enhance the current density and stability of CO₂RR.

The salt accumulation issue on the GDE was mainly attributed to the formation of carbonates by the reaction of CO₂ with OH⁻, blocking the pores of GDE and hindering the permeability and conductivity. The side reaction resulted in carbon loss and the decrease of pH, which was unbeneficial to C–C coupling, and increased energy costs. Regularly refreshing the electrolyte suppressed this process, but with elevated operating costs and energy consumption. Operating in acid conditions seemed to be an effective way to alleviate the salt accumulation, as carbonate formation was largely suppressed. It was reported that the acidic fuel cell was effective for at least an order of magnitude longer than the alkaline fuel cell [126]. However, the dominant HER in an acidic medium could lead to low selectivity towards the target product. Although some research has suggested utilizing the cation effect to suppress HER and made some progress on the issue of selectivity during catalysis, the FE towards multi-electron electroreduction product remained unsatisfying and did not meet the expectations for stability [127,128]. Therefore, a number of efforts are still needed to enhance the selectivity and stability of CO₂RR in the acidic flow cell.

The recent development of membrane electrode assembly (MEA) cells may be a potential strategy to enhance stability. The configuration of MEA is shown in Figure 6g [111]. In an MEA cell, the cathodic GDE and anodic catalyst were directly pressed on both sides of the ion exchange membrane, thus the flooding of the GDE and the carbonate formation were circumvented because there was no catholyte. Li et al. reported that the MEA achieved a stable operation for 190 h during electrochemical CO₂RR to C₂H₄, with an FE of ethylene of around 60% and a current density of 120 mA·cm⁻² in neutral medium, demonstrating the long durability of the MEA cell [29]. As there is no electrolyte in the cathode, the humidity of the CO₂ input stream was found to be important as the proton in the solution is needed for electroreduction [111]. Increasing reactant humidification resulted not only in high current density operation but also low power requirements for reducing CO₂ to CO [129]. However, despite excellent stability, the MEA system suffered from low current density and high cell voltage due to low local pH and conductivity. Recently, the direct electrolysis of carbonate led to stable operation under CO₂RR conditions. The catalysts were loaded on hydrophilic carbon paper without carbonate formation, thus avoiding flooding and salt accumulation. Inspired by this concept, Li et al. constructed a system and achieved continuous operation for 145 h with an Ag catalyst, outperforming a great deal of previous research [130]. However, other problems existed in this system. Only two electron reduction products were obtained, and the FEs towards them decreased rapidly as the current density increased. The cell voltage was relatively large due to the bipolar membrane inside the device, far from a mature technique [131,132,133].

4. Conclusions and Prospects

In this minireview, we have summarized recent progress in CO₂RR on copper based-catalysts, including catalyst synthesis and reactor design. The drawbacks of high overpotential, low faradaic efficiency, small current density, and short-time stability have prevented further industrial application due to economic costs. Catalyst synthesis and the catalysis engineering have been improved to obtain target products, with high current density at low overpotential.

To produce electroreduction products at low overpotential, SACs have been employed in CO₂RR due to their good catalytic performance. Single-atom fabrication by the stabilized ligands enhanced the catalytic reduction reaction at decreased overpotential. For multi-electron electroreduction products, Cu-based catalysts were used in metal clusters, special morphology, and tandem catalysts, or modified by base chemicals. Thus, the binding and coverage of the *CO intermediates were enhanced, resulting in multi-electron electroreduction at low overpotential.

To obtain the target product economically, high FE in CO₂RR is necessary. For CH₄, traditional SACs are favorable for the prohibition of C–C coupling. Cu sites were stabilized by CeO₂, MOF, and CuGaO₂, exhibiting high CH₄ partial current density. The nanocell catalysts combining an Ag core and Cu₂O shell resembled a tandem nanoreactor, achieving high CH₄ FE by changing the Cu₂O envelope size. For methanol, Cu-based materials were also discussed; strategies that boosted the FE of methanol included fabrication of other NPs, MXene immobilization, and doping. For ethylene, the strategies for the catalyst design were concentrated on C–C coupling. Crystal facet tuning and surface modification were found to enhance the generation of ethylene. For C₂₊ oxygenates, bimetallic catalysts containing Cu and another metal were suggested for stabilizing oxygenate intermediates, contributing to the product of C₂₊ oxygenates. It may be possible to obtain C₃ products of n-propanol, using Cu materials with further fabrication and the introduction of co-planar vacancies.

To enlarge the current density, a flow cell with GDE technology was introduced to overcome the limitation of CO₂ dissolution. A highly alkaline electrolyte was reported to increase the conductivity and promote electroreduction. Modifying the catalyst’s surface with ionomer was thought to enhance the CO₂ mass transfer, thus increasing the current density. It was suggested that the strategies used for low overpotential and high FE are also effective to increase the current density.

To maintain the stability of the reaction, surface reconstruction should be prohibited. Stabilized ligands and alloyed metals were used to protect the active metal species of the catalysts during electroreduction. Fabrication of the GDE and GDL was adopted to minimize the flooding effect and an acidic electrolyte was used to limit the effect of salt accumulation on the catalytic performance in the flow cell. A developed MEA was achieved to avoid flooding and salt accumulation, but improvements remain necessary to address overpotential, FE, and current density.

Recent strategies including single-atom catalysts, crystal facet engineering, bimetallics, doping, surface modification, metal–carbon composition, and the developed technologies of flow cells, GDL, and MEA have brought improvements in the overpotential, FE, current density, and stability of CO₂RR. Thus, the combined route of catalyst synthesis and operational technology presents a promising pathway for obtaining the target products of methane, methanol, ethylene, and C₂₊ oxygenates at low economic cost. It can be expected that the practical application of CO₂RR will be achieved in industry in the future.