Recent Advances and Perspectives of Electrochemical CO₂ Reduction Toward C₂₊ Products on Cu-Based Catalysts

Abstract

Renewable-electricity-powered electrochemical CO₂ reduction reactions (CO₂RR) to highly value-added multi-carbon (C₂₊) fuels or chemicals have been widely recognized as a promising approach for achieving carbon recycling and thus bringing about sustainable environmental and economic benefits. Cu-based catalysts have been demonstrated as the only candidate metal CO₂RR electrocatalysts that catalyze the C–C coupling. Unfortunately, huge challenges still exist in the highly selective CO₂RR to C₂₊ products due to the higher activation barrier of C–C coupling and complex multi-electron reaction. Key fundamental issues regarding both active species and product formation pathways have not been elucidated by now, but recent developments of advanced strategies and characterization tools allow one to comprehensively understand the Cu-based CO₂RR mechanism. Herein, we review recent advance and perspective of Cu-based CO₂RR catalysts, especially in terms of active phases and product formation pathways. Then, strategies in catalysts design for CO₂RR toward C₂₊ products are also presented. Importantly, we systematically summarized the advanced tools for investigating the CO₂RR mechanism, including in situ/operando spectroscopy techniques, isotope labeling, and theoretical calculations, aiming at unifying the knowledge of active species and product formation pathways. Finally, future challenges and constructive perspectives are discussed, facilitating the accelerated advancement of CO₂RR mechanism research.

Graphical Abstract

1 Introduction

In the past hundreds of years, especially after the Second Industrial Revolution, the development of human society has been heavily dependent on fossil fuels as the main energy sources, such as coal and oil. Due to the excessive consumption of fossil fuels, the concentration of CO₂ in the atmosphere increases sharply from 270 ppm (1 ppm = 1 mL m⁻³) in the early 1800s to 410.45 ppm in June 2019 [1,2,3,4,5,6,7], bringing about a series of problems including global warming and energy crisis. In terms of reducing the concentration of atmospheric CO₂, replacing fossil fuels with various clean and renewable energy such as solar and wind is one potential approach to reduce the emission of CO₂. On the other hand, converting CO₂ to value-added fuels through CO₂ reduction reactions (CO₂RR) has recently received extensive attention and been devoted to reducing the present CO₂. It is noted that the renewable energy sources were difficult to be used because of their intermittent nature and stored by limited storage devices. Therefore, CO₂RR using renewable electricity can not only be converted into fuels and chemical industry to achieve carbon recycling, but also solve the problems of the intermittent and unstable renewable energy utilization [8,9,10,11,12].

Since Hori et al. systematically investigated the CO₂RR performance of various metal electrodes in the 1980s and 1990s [13], extensive studies used to improve the activity and selectivity of CO₂RR catalysts have been reported. Among various CO₂RR electrocatalysts investigated so far, Cu-based materials have been considered to be the only metallic electrocatalyst capable of reducing CO₂ to multi-carbon oxygenates and hydrocarbons (denoted as C₂₊ products) [14,15,16,17,18]. Therefore, the development of various Cu-based electrocatalysts has greatly improved the overall performance for decades. Nevertheless, high overpotential is required due to the stable structure of CO₂ molecules and the relatively low Faradaic efficiencies (FEs) are observed due to the competing hydrogen evolution reaction (HER). Furthermore, the extremely strong reducing power leads to the wide product distribution of Cu-based CO₂RR electrocatalysts, reducing the selectivity of a specific product. The complexity of the CO₂RR mechanism not only is reflected by the high probability that C₂₊ products share the same intermediates and have cross-reaction pathways, but also arises from the fact that the product distribution is highly surface-sensitive. In particular, four possible intermediates for the formation of C₂H₄ and EtOH (i.e., *COCHO, *CHCOH, *CCH, and CH₂CHO*) have been proposed and the adjusted ratio of their FEs has been extensively studied [19,20,21,22,23]. Although multiple factors (i.e., physical and chemical properties, the change of the surface state, defects, etc.) have been reported to play a key role in the catalytic activity of Cu-based catalysts, surface speciation on the surface of various types of Cu during CO₂RR remains still under debate [24, 25]. Therefore, we believe that tracking the evolution of surface speciation and gaining a deeper insight into reaction pathways will guide the future design of efficient electrocatalysts for producing CH₄ and C₂₊ products.

With the rapid development of density functional theoretical (DFT) calculations and advanced characterization technologies, great efforts have been devoted to exploring the mechanism of Cu-based CO₂RR catalysts, especially the determination of the chemical state of Cu sites and the exploration of CO₂RR pathways. Therefore, we aim to comprehensively review recent advances and perspectives in understanding the mechanism of Cu-based CO₂RR catalysts, which have brought about considerable improvements in identifying the active sites and possible reaction pathways. This review would begin by introducing the CO₂RR mechanism on Cu-based catalysts, including the discussion of the active sites, the summary of main C₂₊ product pathways, and the main factors affecting CO₂RR performance. As shown in Scheme 1, we then discuss strategies in catalyst design for CO₂RR toward C₂₊ products, as well as advanced tools for investigating CO₂RR mechanisms. An overview of strategies and advanced tools improves understanding of the active phase of catalysts, the selectivity of C₂₊ products, key intermediate states, and the rate-determining step (RDS). Furthermore, the future development and challenge, and more deeper insights into the mechanism of Cu-based CO₂RR catalysts are also discussed.

Scheme 1

Outline of this review. Recent strategies and advanced tools are further summarized to improve understanding of the CO₂RR mechanism, especially the discussion of active sites and possible reaction pathways

2 Mechanism on CO₂RR on Cu Catalysts

2.1 An Overview of CO₂RR Mechanisms on Cu-Based Catalysts

Designing a complete and coherent schematic diagram to understand the overall mechanism of CO₂RR is a challenging task, owing to the highly complicated and multiple processes with multi-electron/proton transfer via various reaction pathways in an aqueous solution [24,25,26,27,28,29,30]. For Cu-based catalysts, the distribution of products is broad, such as main C₁ (CO, HCOOH, and CH₄) and C₂ (H₂C₂O₄, C₂H₄, C₂H₆, and EtOH) and C₃ (PrOH) products. Table 1 summarizes the half-reactions of C₂₊ products that can be produced from the CO₂RR with the corresponding reversible hydrogen electrode (RHE) potential. Despite the variety of the products, the first step of CO₂RR is the initial activation of CO₂ molecules. However, the formation of the CO₂·⁻ radical intermediate via one-electron injection process requires a very negative redox potential [−1.9 V vs. standard hydrogen electrode (SHE)] because CO₂ is a very stable molecule. The adsorption of CO₂ generates CO by breaking the C–O bond or formate through the formation of the C–H bond, in which the obtained *CO is a key intermediate for CH₄, alcohols, and C₂₊ products.

Table 1 Standard equilibrium potentials for the electrochemical CO₂ reduction to CO and further reduced C₂₊ products

In general, the whole formation of C₂₊ products mainly included four steps: (1) the formation *CO; (2) the formation of the C–C bond; (3) post-C–C-coupling; (4) desorption of products. However, in the case of the formation of C₂₊ products on Cu-based catalysts, clarifying the formation path of the product is a huge challenge. Most possible C₂₊ pathways during CO₂RR are summarized in the previous reviews [2, 9, 26, 31, 32], which clearly demonstrate that the mechanism is complex due to the presence of many intermediates and pathways. In addition, the selectivity and pathways of C₂₊ products are highly dependent on the active phase of Cu-based electrocatalysts and the surrounding electrolyte, such as the chemical state, exposed crystal face, pH value, and composition of the electrolyte.

2.2 True Active Sites: Cu⁰ or Cu⁺

The complexity of CO₂RR on Cu-based catalysts arises from the abundance of C₂₊ products, involved multiple proton/electron transfer steps, and the activity and selectivity of C₂₊ products are highly surface-sensitive, especially for oxide-derived Cu (OD-Cu) catalysts [16, 33]. Though various OD-Cu catalysts have been designed and significantly improved the activity and selectivity of C₂₊ products during CO₂RR, the mechanism of the significant enhancement toward C₂₊ products remains controversial. Fundamentally, since the standard reduction potential of Cu⁺/Cu⁰ redox is − 0.36 V, Cu-based catalysts are thermodynamically metallic following the Pourbaix diagram under the potentials of CO₂RR [24]. Therefore, many researchers believe that the active phase is metallic Cu, and the related mechanism of enhanced activity and selectivity on OD-Cu catalysts during CO₂RR is attributed to unique structures, grain boundaries, low-coordinated atoms, defective sites, and roughened morphology [34]. However, on one hand, the state of Cu can easily vary due to the reversible change between the oxidation state and the reduction state of Cu under electrochemical reaction condition. On the other hand, the kinetics of Cu reduction is slow, which is confirmed by many characterizations [35]. Furthermore, many groups confirm that the Cu⁺ sites remain on the surface of catalysts and the key active sites during CO₂RR.

By combining experimental and theoretical research, it has been proposed that under the condition of CO₂ reduction, the presence of metastable states (such as residual oxygen, underground oxygen, and the Cu oxide state) significantly enhances the activation of CO₂ and reduces the energy barrier of CO–CO dimerization to promote C₂₊ pathways [36, 37]. DTF calculations suggested that during the initial step of CO₂ activation, the strong H₂O adsorption site on top of the Cu⁺ region can provide hydrogen bonding to stabilize the CO₂ on the adjacent Cu⁰ region. For the followed *CO dimerization step, due to the opposite Mulliken charges of *CO on Cu⁺ and Cu⁰, this electrostatic force significantly improved the kinetics of CO–CO coupling, thus improving the selective of C₂₊ products. Mistry et al. reported that the presence of Cu⁺ played a key role in lowering the onset potential and improving C₂H₄ selectivity by using operando X-ray absorption spectroscopy (XAS) and cross-sectional scanning transmission electron microscopy (STEM) [35]. Recently, the influence of the Cu oxidation state on the catalytic performance has been widely discussed and summarized. Interesting, by summarizing various Cu catalysts related on Cu⁺ species [24], Chen and coworkers found a link between the chemical state and selectivity. They proposed that the mixed Cu⁺ and Cu⁰ states favored the selectivity of C₂ products, while the Cu⁺- and Cu⁰-dominated states enhanced the selectivity of C₁ products.

Although experimental and theoretical studies indicate that the Cu⁺ states play a key role in improving the selectivity of C₂₊ products, no direct experimental studies showed that this high selectivity was attributed to the Cu⁺ states. Remarkably, Xu and coworkers firstly reported that CuO_x and CuO_x/(OH)_y species are unlikely to be the active sites for facilitating the formation of C₂₊ products by combining variation trends of product selectivity and spectroscopic results [25, 38]. Furthermore, Xu’s group also found that the Cu–O species were pH dependent by in situ surface-enhanced Raman spectroscopy (SERS), which could be completely reduced in the nearly neutral electrolyte [39]. This phenomenon is generally confirmed by recent reports related to Raman spectroscopic results. Importantly, in addition to Raman spectroscopic characterizations, fully reduced Cu–O species were observed in CO₂-saturated 0.1 M (1 M = 1 mol L⁻¹) KHCO₃ during CO₂RR by electron energy loss spectroscopy (EELS), in situ electrochemical TEM and the quasi-in situ X-ray photoelectron spectroscopy (XPS) analysis, but associated catalysts still maintained high selectivity of C₂₊ products [40, 41]. The above results were discussed in detail in Sect. 4.1, and those investigations further confirmed that surface oxygen-containing species were indeed completely reduced, indicating that Cu–O species are not the active sites for the formation of C₂₊ products. Furthermore, the differences of CO adsorption bands on mechanically polished polycrystalline Cu (referred to as Cu-MP) and anodized Cu-MP were monitored by using time-resolved SERS in CO₂-saturated 0.1 M KHCO₃ [42], which was consistent with that of polycrystalline Cu and OD-Cu reported by Xu and coworkers in CO-saturated 0.05 M KOH, indicating that changes of nanostructure and nanomorphology may play key roles in improving selectivity of C₂₊ products. In addition, the coordination number of OD-Cu catalysts was obtained by fitting extended X-ray absorption fine structures (EXAFS) curve, which was lower than that of Cu foil, indicating that undercoordinated Cu sites might play a key role in improving the selectivity of C₂ products.

2.3 Possible Reaction Pathways to C₂₊ Products

Due to the competitive undesired HER and higher activation barrier of C–C coupling as well as competition of C–C, C–O and C–H bond formations, the formation of C₂₊ products from CO₂RR on Cu-based electrocatalysts is a difficult and complex multi-electron reaction. There are two proposed mechanisms for C–C bond formation: the Eley–Rideal mechanism (E–R), in which gas CO is the reactant, and the Langmuir–Hinshelwood mechanism (L–H), in which catalyst surface adsorbed CO is the reactant. Calle-Vallejo et al. suggested that CO–CO coupling mainly proceeded through the E–R mechanism on Cu(1 0 0) by using a purely thermodynamical model, in which gaseous CO may insert itself into the bond between the Cu surface and adsorbed CO [21]. Major pathways for C₂₊ products are usually associated with the L–H mechanism to form a *C₂O₂ intermediate, which can be hydrated to generate C₂₊ products [43, 44]. Importantly, the energy barriers of the C–C coupling via the E–R mechanism were higher than that of the L–H mechanism [45]. The C–C coupling mechanism was controversial and affected by many factors, including applied potentials, cation, tensile strain, and CO partial pressure, which could be beneficial to increase *CO coverage and reduce activation energies for *CO dimerization [46,47,48,49,50]. Therefore, understanding the C–C coupling mechanism and determining the reaction path of products will be more conducive to the design of catalysts for highly selective C₂₊ compounds rather than C₁ chemicals. Possible reaction pathways for C₂₊ products (C₂H₄, C₂H₆, EtOH, and PrOH) are summarized in this section and that of other products can be found in other reviews [27, 32, 51].

2.3.1 C₂H₄ Pathway

To date, although two different pathways of the C₂H₄ formation based on the overpotential have been proposed during CO₂RR and CO reduction reactions (CORR) on the Cu single-crystal surface, the mechanism of C₂H₄ formation is still under debate on Cu-based catalysts. In earlier studies, the direct CH₂–CH₂ coupling has been proposed to form C₂H₄ [52], while recent studies suggested that the formation of C₂H₄ was more likely through CO dimerization [49, 53]. For the first time, Hori et al. reported that the C₂H₄ formation was pH independent, while the CH₄ formation depended on the pH [13]. This indicated that the RDS of CH₄ pathway involved a coupled proton–electron transfer, whereas that of C₂H₄ did not. However, the process of C–C coupling was still controversial. Koper and coworkers used differential electrochemical mass spectrometry (DEMS) technique to explore the intermediate of C–C coupling [54]. They concluded that the first step of the C₂ pathway was CO–CO coupling, followed by the protonation (CO–CHO) to form a surface-bonded enediol or enediolate, or an oxametallacycle, which would explain the selectivity toward C₂H₄. They also proposed two different pathways of C₂H₄ formation with different overpotentials by using DEMS [55]. The high-overpotential pathway shares the same RDS intermediate as CH₄, occurred both on Cu (100) and (111) facets, while the low-overpotential pathway involves the RDS of CO–CO coupling on Cu (100) facet.

Apart from the DEMS technique, employing in situ Fourier transform infrared spectroscopy (FTIR) [56], OC–COH took place on the Cu (100) facet, while not found on the Cu (111) facet. For Cu-based catalysts, the intermediate of CHO–CHO [57] and CH₂–CO [58,59,60,61] were proposed as the pathway of C–C coupling to elucidate the mechanism of C₂H₄ formation. Wang and coworkers proposed a H-assisted CHO–CHO coupling mechanism for C₂H₄ formation combined DFT calculations and experiments on Cu (111) [62]. In addition, the intermediate of CH₂–CO in the C₂H₄ pathway could be detected on Cu-based composite catalysts. For instance, on Cu-Ag composites [61], CO₂ could be first reduced to CO and then bind to either Cu or Ag sites, while CO on the Cu sites could be reduced further to CHO or CH_x intermediates. The CO on Ag sites may then insert itself into *CH₂ to form *COCH₂ intermediates of C₂H₄ pathway.

2.3.2 EtOH Pathway

It is generally believed that EtOH is proposed to share the similar intermediates at the early stages of CO/CO₂ reduction such as C₂H₄. Therefore, the C–C coupling is also considered to be the key step for EtOH formation, including two reaction pathways involved CO–CO couplings and *CHO or *CH_x intermediates inserted CO. Zhang et al. demonstrated that ionic liquids, as a chemical trapping agent, could selectively suppress the formation of products during CO₂RR [63]. They proposed that EtOH can be produced through the pathway of *CH₂ intermediates inserted CO, while C₂H₄ can be produced through two independent pathways. This mechanism was widely present in Cu-based alloy catalysts, which are discussed in detail in Sect. 3. For the CO–CO coupling pathways, *CHCOH and *OCHCH₂ have been reported as sharing intermediates that determine the selectivity of C₂H₄ and EtOH [22, 23]. Calle-Vallejo and coworkers reported that the reaction barriers of CO/CO₂ reduction to EtOH were larger than that of C₂H₄ by combining experimental and computational studies, and hence it was easier to form the latter [64], which was consistent with the selective tendency of Cu-based catalysts. In addition, H. Sargent and coworkers reported that reaction intermediates starting from *HCCOH become more favorable for EtOH by introducing Ag on the Cu surface [23]. They also found an alternative catalytic approach to promote EtOH by increasing the energy barrier of intermediates starting from *OCHCH₂ to C₂H₄ [22].

2.3.3 C₂H₆ Pathway

Ethane, as a minor product on the Cu-based catalysts during CO₂RR, has only been reported occasionally [65,66,67,68,69,70,71]. In analogy with EtOH, C₂H₆ can be formed through the similar intermediates for the electrocatalytic reduction of CO/CO₂, including two important pathways: (1) *CO dimerization pathways, and (2) coupling of *CH_x accompanied by further hydrogenation [65, 66, 72]. The intermediate of *OC₂H₅ derived from *C₂O₂ was reported as the key in determining selectivity between C₂H₆ and EtOH. Therefore, C₂H₆ and EtOH should be detected together, which is consistent with previous reports [65, 73]. However, on one hand, Zhang et al. demonstrated only EtOH was detected by using ionic liquids via the intermediate of *OC₂H₅ on the Cu foam catalyst [63]. On the other hand, Bertheussen et al. reported that C₂H₆, AcH, and EtOH were observed during CORR on OD-Cu electrodes [74], while only C₂H₆ was not detected through the CH₂CHO reduction, indicating that the formation of C₂H₆ from *OCH₂CH₃ can be ruled out. In addition, D. Handoko et al. reported that the formation of C₂H₆ was likely to originate from the *CH₃ dimerization by adding diacetyl as the reactant, rather than further hydrogenation of *C₂H₄ or *OCH₂CH₃ [66].

2.3.4 PrOH Pathway

In order to improve the FE of PrOH, much research effort has been dedicated. More recently, the FE for PrOH product during CO₂RR was enhanced through the modification of Cu-based catalysts, which include reconstructed nanocrystals [75], dense pack [76], two-step activation [77], Cu²⁺ ion cycling [78], and core–shell structures [22]. However, the mechanism of C–C coupling, especially in PrOH generation of CO₂RR and CORR, remains a challenge. So far, it has been proven that C₂H₄ is an intermediate in the formation of PrOH in CO₂RR and CORR. Ren et al. reported that significant amounts of PrOH could be observed when using a 1:1 mixture of C₂H₄ and CO, while little or no detectable amounts of PrOH are produced when only CO or C₂H₄ were reduced [75]. During CO₂RR, PrOH generation was optimized when both C₂H₄ and CO are formed at a high rate. They concluded that PrOH could be formed through the coupling of C₂H₄ with CO. H. Sargent and coworkers designed a strategy via binding of C₂ intermediates to boost selectivity from C₂ to C₃ products [79]. The combined product distributions showed that the increase production of PrOH corresponded to the decrease in C₂H₄, indicating that the C₂ intermediate coupling with CO for PrOH generation may be related to C₂H₄. In addition, Han and coworkers observed that PrOH and EtOH followed a similar trend of increasing selectivity with the corresponding decrease in C₂H₄, indicating the presence of a common intermediate in the reaction process [80]. Further DFT calculations suggested the stabilization of *OCH₂CH₂ intermediate was favorable for the formation of PrOH and EtOH, which was in agreement with previously reported mechanisms for PrOH production [22].

However, it was also generally accepted that the formation of PrOH involved the dimerization between CO and hydrogenated carbon (e.g., *CH₂), followed by proton/electron transfer to form propionaldehyde, and then further reduced to PrOH. For instance, Yang and coworkers proposed that PrOH formation required coupling between CO and *CH₂ by analyzing Tafel slopes and potentials of CH₄ and PrOH [76]. Zhang et al. demonstrated that the formation of PrOH involved *CH₂ by using the chemical trapping agent of ionic liquids, which was also the intermediate of EtOH and C₂H₄ [63]. In addition, Xu and coworkers reported that a small proportion of PrOH come from cross-coupling between CO and AcH when using CO and AcH by combining isotope labeling and spectroscopic techniques, while the main PrOH was produced through self-coupling of CO [81]. Therefore, they demonstrated that the C–C coupling between CO and AcH was unlikely to be the main pathway during CORR.

2.4 Factors Influencing the Formation of C–C Bond in C₂₊ Products

In general, pathways of CO₂RR to various C₂₊ products are highly influenced by applied overpotential, liquid electrolyte, and properties of the electrocatalyst. In the CO₂RR process, liquid electrolytes, including ionic liquids [82, 83], organic electrolytes [84, 85], and aqueous electrolytes, can realize ionic transport and provide a reaction environment. Most of the research on CO₂RR has been carried out in aqueous electrolytes and recent advances in other electrolyte types are presented in other reviews [34, 86]. The intrinsic properties of electrocatalysts (facets, defects, and morphology) significantly affect the adsorption energy of intermediates and the energy barrier of CO₂RR, which changes the pathway of products. In this section, we will discuss in detail the three main factors that influence the product pathways: pH, cations, and facet effects.

2.4.1 pH Effect

The pH is an essential factor in determining the activity and selectivity due to the complex protonation process during CO₂RR, especially inhibiting the HER and turning the ratio of C₁/C₂ products for Cu-based catalysts [27, 87]. In general, as the pH increases, the selectivity of the product shifts from H₂ and CH₄ to multi-carbon products such as C₂H₄ [88]. However, the effect of pH on CO₂RR products is very complex, which may be related to the multiple proton-coupled-electron transfer and the RDS in different reaction pathways [78, 89, 90]. The difference of the formation, the starting potential, and the Tafel slope of CH₄ and C₂H₄ indicates that they follow different reaction paths [91]. For the first time, Hori and coworkers proposed that the reaction rate for CH₄ is pH dependent on the normal hydrogen electrode (NHE) scale due to the coupled proton–electron transfer of *CO during the RDS [58], while that of C₂H₄ was shown to be pH independent. By contrast, Schouten’s group reported that the onset potentials of CH₄ and C₂H₄ on Cu (111) electrodes were both pH dependent through online electrochemical mass spectrometry (OLEMS) [92]. This can be explained by the coexistence of two different C₂H₄ formation pathways. The high-overpotential pathway on Cu (100) and (111) facets shares the same RDS with CH₄ (*CO protonation) and the low-overpotential pathway on Cu(100) involves the CO–CO coupling [55]. Contrary to the second pathway, the first pathway, which includes CH₄ and C₂H₄, shows a pH dependence on the SHE scales as there is a coupled proton/electron transfer of CO involved. It was concluded that both the CO–CO coupling or proton–electron transfer from water during the RDS are conceivable elucidations for the pH independence of the C₂₊ pathway on the SHE scales.

The bulk pH, however, is different from that at gas–liquid–solid three-phase interfaces due to the unbalance between the supply and the consumption of protons [27]. The shift of local pH during CO₂RR depends on the bulk pH in the electrolyte, buffer capacity of the electrolyte, the current density, and the morphology of catalysts [93, 94]. The selectivity toward C₂ products on Cu-based catalysts can also be tuned by inhibiting the production of CH₄ and H₂ in the buffer electrolyte, which depends on the nature and concentration of electrolyte [35, 47, 95, 96]. Furthermore, by considering solvent effects using implicit solvation model, DFT calculations presented that the C₁ and C₂ paths on Cu(111) are competing and products are pH-dependent underlying CORR [97]. At low pH, the C₁ pathway forms CH₄ through *COH to *CHOH, while C₂₊ pathways are suppressed kinetically. At neutral pH, the C₁ and C₂₊ pathways share a common *COH, in which the branching C–C coupling is achieved through the pathway of CO–COH. At high pH, selectivity for C₂₊ products through the early C–C coupling arises by kinetically blocking C₁ pathways [98]. Based on the above analysis, especially for Cu-based catalysts, higher pH electrolytes are more favorable for the generation of C₂₊ products [88]. For instance, Sargent and coworkers reported that the Cu catalyst reduced CO₂ into C₂H₄ with 70% FE at − 0.55 V_RHE in a strong alkaline electrolyte (7 M KOH). Combined experimental and theoretical insights, strong alkaline media can accelerate the kinetics of CO₂RR by lowering the CO₂ reduction and the C–C coupling energy barrier and stabilizing the adsorbed OCCO through a stronger dipole attraction. Therefore, more researchers are investigating higher concentration of KOH as electrolytes to further enhance the FE of C₂₊ products [99].

However, as shown by experimental investigations and DFT calculations, the current mechanistic understanding of the formation of C₂₊ products suggests that the C₂₊ pathway of CO–CO coupling is pH independent, in which there is no proton transfer in this L–H process. Recently, Lu and Xu pointed out that the reactions performed at a higher electrolyte pH are actually subjected to a larger overpotential by ΔpH × 59 mV, and the catalytic effect of the highly alkaline electrolyte could be exaggerated when comparing C₂₊ products formation rates [100]. Systematically varying the concentration of Na⁺ and OH⁻, they first found that the rate of C₂₊ product increased significantly with the increase in OH⁻ concentration under the same RHE scale, which was similar to the phenomenon in the previous study. However, the RDS of C₂₊ products is pH independent, so it is reasonable to compare the C₂₊ products formation rate at the same SHE scales. Based on the same SHE scales, they found that the formation rate of C₂₊ products did not change with the increase in OH⁻ concentration, and at the same OH⁻ concentration, that increased significantly with the increase in Na⁺ concentration. Therefore, they concluded that the increase in Na⁺ concentration, rather than OH⁻, improved the formation rate of C₂₊ products during CORR at the SHE scales. The introduction of crown ether led to a sharp decrease rate of C₂₊ products formation, further confirming the effect of the nature and concentration of cations for CORR.

2.4.2 Cation Effect

Much research effort reported that the cations in the electrolyte play a key role in influencing the activity and selectivity of catalysts at the reaction interface during CO₂RR and CORR [47, 50, 98, 101]. The involved mechanisms have been discussed widely since it was first reported decades ago. Akira and Hori observed that the selectivity of C₂H₄ and alcohols was progressively higher than that of CH₄ and H₂ with ionic size increasing from Li⁺ < Na⁺ < K⁺ < Cs⁺ [102]. They proposed that this cationic effect stems from the adsorption tendency of cation on electrode surfaces, which is mainly determined by the reaction energetics and the degree of cation hydration. The hydration capacity could be stronger with ionic size increasing, and the larger size cation is not only more likely to be adsorbed, but also can be rejected by H⁺ to reduce the selectivity toward CH₄ and H₂. In addition, Thorson et al. proposed that adsorbed cations at the electrode surface could stabilize the intermediate *CO²⁻ to promote the CO₂RR [103]. Kim et al. observed that the generation of CO was more efficient in K⁺-based electrolytes than that of Na⁺ [104]. Further DFT calculations clarified that the local electric field induced by alkali metal cations is beneficial to enhance the stability of intermediates such as *COOH and *CO, and thus lower the thermodynamic energy barrier [105, 106]. However, by taking into account the solvent effect, this effect was not observed by the Ab Initio Molecular Dynamics (AIMD) simulations [107].

In addition to the effect of cation adsorption and stabilizing intermediates, the mechanism of electronic field effects induced by cations was also mentioned, which was strongly dependent on the size of hydrated cation [108,109,110]. Recently, Chan et al. reported that the multi-scale modeling approach was able to have an extraordinary quantitative consistency with experimental trends in cation effects, supporting the effect of cations by changing the interfacial electric field [108]. Due to the less surrounded H₂O for larger cations, the size of hydrated cations was arranged in order: Cs⁺·xH₂O < Rb⁺·xH₂O < K⁺·xH₂O < Na⁺·xH₂O < Li⁺·xH₂O. Therefore, the smallest size of Cs⁺·xH₂O showed the strongest electric field, which could facilitate the adsorption of intermediates and the generation of multi-carbon products [110, 111]. For example, on O₂-plasma-activated Cu, the current density and selectivity of C₂₊ products increased with the increase in alkali metal cation size [110]. Further DFT calculations showed that larger cations are more conducive to adsorption, resulting in the stabilization of the CO₂RR intermediates to enhance the generation of C₂₊ products. However, based on the Poisson-Nan-Planck model, the interfacial electric field of Cs⁺ is weaker than that of K⁺ due to the reduced dielectric constant [112]. It was noted that the promotion effect of alkali metal cations may be related to not only the electric field but also the pK_a for the hydrolysis. Bell and coworkers proposed that the FE of C₂H₄ and EtOH increased with the increase in alkali metal cation size [113], which was ascribed to the hydrolysis capacity of cations. With the increase in cation size, the pK_a decreased, which reduced the pH value to increase the CO₂ concentration near the cathode.

Although a variety of possible ways in which cations influence surface-mediated electrocatalytic reactions have been proposed, no consensus has been reached due to the lack of direct experimental evidence. The Stark tuning rate is usually used to represent the electric field strength, which is obtained by fitting the slope of the peak frequency of the *CO stretch band and the corresponding applied potential [100, 114, 115]. Using surface-enhanced infrared absorption spectroscopy (SEIRAS) [115], Waegele and coworkers found that the electric field intensity was too small to have a significant effect on the observed changes in C₂H₄ selectivity, but sharp peaks generated by the interaction of *CO and H₂O were detected. They suggested that the C₂H₄ collection efficiency was due to the changing interaction between *CO and H₂O in the presence of different cations. In addition, recent studies by Xu and coworkers demonstrated that the reaction rate can be significantly increased by increasing the cation concentration without the change of interfacial electric field intensity, which is determined by the Stark tuning rate [100]. This result indicated that the interfacial electric field intensity is not a key parameter of CORR. Furthermore, Xu et al. also systematically studied on the impact of cationic effects, including electricity and non-electric field (NEF) strength components [116]. In this work, the rates of CORR and HER increased with the increase in cation size on polycrystalline Cu catalysts, while the FE of CORR products also increased from Li⁺ to K⁺, but leveling off for the larger cations. The Stark tuning rate measurements by using in situ SEIRAS investigations in five different alkaline metal cation electrolytes implied that the size of electrochemically related cation followed the sequence: Cs⁺·xH₂O ~ Rb⁺·xH₂O ~ K⁺·xH₂O < Na⁺·xH₂O < Li⁺·xH₂O. The variation trend of the reactivity rate and Stark tuning rate is similar in these cation-containing electrolytes. They drew a conclusion that the cationic effect has both the favorable interfacial electric field strength and the unfavorable NEF strength component.

2.4.3 Facet Dependency

Experimentally, it is well established that the activity and selectivity of catalysts during CO₂RR and CORR are strongly dependent on the surface crystal facets of Cu-based catalysts [27, 117, 118]. It has been confirmed that the Cu (100) facet particularly increases the selectivity for C₂ products, especially C₂H₄, while Cu (111) favors preferentially the generation of CH₄ [109]. The generation of C₂H₄ on the Cu (100) facet follows two different pathways depending on the applied overpotentials. At low overpotentials, C₂H₄ is pH dependent and exclusively produced through CO–CO coupling. At high overpotentials, the formation of C₂H₄ and CH₄ via a *CHO pathway are strong pH dependent on the SHE scales, similar to that on the Cu (111) facet. In addition, Cu (110) facet preferentially increases the selectivity for CH₄ containing trace amounts of C₂H₄. Apart from CH₄ and C₂H₄, liquid, oxygenated compounds also could be detected on the low-index Cu facets [e.g., Cu (111), Cu (110), Cu (100)] [96, 101]. Hori et al. reported that the Cu (111) facet preferentially formed HCOOH compared to alcohols, while the opposite could be detected on the Cu (100) facet. The Cu (110) facet yielded similar selectivity toward liquid and oxygenated compounds [91].

Similar to the low-index of the Cu (100) facet, the stepped Cu (211) facet is favorable for CO–CO coupling compared with Cu (111), and thus is beneficial for the formation of C₂₊ products [78, 119,120,121]. A. Nilsson et al. reported the C₂H₄ and CH₄ products of three Cu single-crystal surfaces (100), (111), and (211). They offered a qualitative interpretation that the activity of Cu (100) and Cu (211) is quite similar, whereas that of Cu (111) is the lowest [119], which has been previously demonstrated by H. T. Wang and coworkers [2]. Furthermore, high-index Cu crystal facets [i.e., Cu (311), Cu (511), Cu (711), and Cu (911)] tended to be highly selective for C₂H₄ and other liquid and oxygenated compounds [69, 91, 122,123,124,125,126], For example, Y. Huang and coworkers reported that activated Cu NWs catalysts with the Cu (511) plane ([3(100) × (111)]) stepped surface exhibit a high FE of C₂H₄ (77.40% ± 3.16%) over ~ 200 h [123]. DFT calculations revealed that the stepped surface was thermodynamically favorable for C₂ products compared with the surface of Cu (100), but unfavorable to the formation of C₁ products and H₂. Different from the plane of Cu-[n(100) × (111)] [40, 91, 101], Cu-[n(100) × (110)] (n > 2) facets are highly EtOH selective, while the stepped Cu(210) is more selective toward CH₄ [40, 127]. In addition, F. Jaramillo et al. demonstrated that Cu (751) over the Si (111) substrate was higher selective for > 2e⁻ oxygenate generation at low overpotentials than that of Cu (111) and Cu (100) [126]. In addition to single-crystal Cu, grain boundary (GB) Cu also presented high activity and selectivity of C₂₊ products during CO₂RR [128, 129]. For instance, Gong and coworkers reported that the GB-rich Cu could be obtained to drive a remarkable FE of 73% for C₂₊ products [130]. DFT calculations further revealed that the Σ3 twin boundary can increase the CO binding energy to improve the kinetics of CO dimerization.

It is important to determine the crystal face structure of Cu catalysts for understanding the catalytic mechanism to design new catalysts. However, Cu catalysts would not maintain a constant structure and morphology under CO₂RR operating conditions, resulting in the change of Cu lattice structure and exposure of unsaturated active sites, which has been widely studied [12, 131, 132]. For example, Kim et al. investigated the lattice structure evolution of the polycrystalline Cu electrode via operando electrochemical scanning tunneling microscopy (ECSTM) [133]. Under CO₂RR conditions with a fixed negative potential (− 0.9 V vs. SHE), the surface of polycrystalline Cu gradually reconstructed, first to Cu (111) within 30 min, then to Cu (100) after another 30 min, and remained stable after that, which was consistent with subsequent reports [134, 135]. In addition, the morphology and crystal face changes of Cu catalysts under CO₂RR operating conditions were observed by using in situ TEM, operando grazing incidence X-ray diffraction, and electrochemical atomic force microscopy [136,137,138,139,140]. Therefore, an in-depth understanding of the dynamic evolution of Cu catalysts can determine the true active site and further understand the relationship between structure and catalytic performance.

3 Strategies in Cu-Based Catalysts Design for CO₂RR toward C₂₊ Products

3.1 Dimension Control

The evaluation of electrocatalytic performance depends heavily on the inherent physical and chemical properties of materials, which can be fine-tuned by modifying the morphology of the material, such as its size and shape [2, 27, 34, 141, 142]. Therefore, the exploration of nanostructure is of great significance in improving the selectivity of C₂₊ products on Cu-based CO₂RR catalysts. So far, diverse morphologies have been developed for rapid CO₂RR processes. According to the spatial configurations, Cu-based nanomaterials can be classified into five categories as follows: single-atomic catalysts (SACs); zero-dimensional (0D) nanomaterials, such as metal nanoparticles (NPs) and nanoclusters; one-dimensional (1D) nanomaterials, such as nanowires (NWs), nanotubes (NT) and nanorods (NRS); two-dimensional (2D) nanomaterials, such as nanosheets (NSs); and three-dimensional (3D) nanomaterials, having nanometer sizes at different directions.

3.1.1 SACs

Due to the unique electronic structure and unsaturated coordination environment, SACs have been showed enhanced activity and unique selectivity toward CO₂ reduction into CO, CH₄, HCOOH, CH₃OH, C₂H₄, C₂H₆, EtOH, and CH₃COCH₃, respectively [143,144,145,146,147,148,149,150,151,152,153]. For instance, Zhao et al. [150] designed single-atom Cu on N-doped porous carbon catalysts for reducing CO₂ to CH₃COCH₃ as the major product (FE 36.7%) with a production rate of 336.1 μg h⁻¹ in Fig. 1a. DFT calculations revealed that the active site of single Cu with four N atoms reduces the activation energy of CO₂ and the reaction free energy of C–C coupling (Fig. 1b). However, Karapinar et al. [145] suggested that single Cu sites were converted into very small Cu NPs by using the operando XAS experiment during CO₂RR, which is likely to be the catalytically active species, displaying a unique EtOH selective with the FE of 55%. In addition, Qiao’s group [147] developed a Cu-C₃N₄ complex with dual active centers for reducing CO₂ into C₂ (C₂H₅OH, C₂H₆, and C₂H₄) species. DFT computations revealed that g-C₃N₄ framework could not only provide additional active centers, but also effectively uplift the d-band center of Cu toward the Fermi level to enhance the adsorption strength of the intermediates during CO₂RR.

Fig. 1

a FEs of CO₂RR products on Cu-SA/NPC. b Free energy diagrams CO₂ reduction to CH₃COCH₃ on two types of Cu-SA/NPC. Reproduced with permission from Ref. [150]. Copyright © 2020, Springer Nature. c FEs of reaction products during CO₂RR on Cu NPs. Reproduced with permission from Ref. [157]. Copyright © 2014, American Chemical Society. d Schematic illustration of Cu NWs with stepped surface and its FE of reaction products during CO₂RR. Reproduced with permission from Ref. [123]. Copyright © 2020, Springer Nature. e TEM images of n-Cu NSs, size histograms of nano-defects and f the FE of C₂H₄ for n-Cu NS, Cu NS, and Cu NP. (e, f) Reproduced with permission from Ref. [175]. Copyright © 2020, American Chemical Society. g Illustration of the microfluidic CO₂ flow cell (left) and the hierarchical electrode design (right). h C₂₊ and C₁ FEs on Cu-D and Cu-P electrodes. (g, h) Reproduced with permission from Ref. [185]. Copyright © 2021, American Chemical Society

3.1.2 0D structure

Differing from bulk Cu-based materials, 0D nanomaterials can provide a higher electrochemical surface area to enhance the activity and selectivity of Cu-based catalysts during CO₂RR. Combined with theory and experiment, ~ 2 nm metallic Cu clusters have been demonstrated to be active sites for the high selectvity of CH₄ due to the increased adsorption strength of CO intermediate and the reduced energy barrier of the CO $\to$ CHO* step with decreasing particle size during CO₂RR [154,155,156]. Differing from Cu clusters, Cu NPs produced CH₄ and C₂H₄ as the predominant hydrocarbon product from CO₂RR [4, 117]. For example, the catalytic activity and selectivity of H₂ and CO were significantly increased for Cu NPs with the decrease in Cu NP size, particularly for NP size below 5 nm, while the selectivity of CH₄ and C₂H₄ was inhibited (Fig. 1c) [157]. Huang and coworkers [158] reported a high-yield synthesis of unique star decahedron Cu NPs, which included a large number of surface defects, twin boundaries, tension strains, and multiple stacking faults. This unique structure led to lower overpotentials by 0.149 V for CH₄ than commercial Cu NPs and high FE of (52.43% ± 2.72%) for C₂H₄ production at (− 0.993 ± 0.012 9) V_RHE. DFT calculation revealed that the existence of twin boundaries significantly reduced the energy barrier of *CHO at low overpotential to promote the generation of CH₄, while at high overpotential the lower formation energy of *OC–CHO was more conducive to the formation of C₂H₄. Anna et al. [159] synthesized three different sizes of Cu NC cubes (24, 44, and 63 nm), in which the cubes with side length of 44 nm showed the highest activity and selectivity toward CO₂RR (${\text{FE}}_{\text{CO}_{2}\text{RR}}$ = 80%) with a 41% FE for C₂H₄. However, the ratio of selectivity for C₂H₄ and CH₄ increased with the increase in cube size. This result indicated that an optimal ratio of edge sites over (100) plane-sites played a key role in maximizing C₂H₄ selectivity and the activity of CO₂RR. As discussed above, the size of NPs has an effect on CO₂RR selectivity. With the size of NPs increases, the relative atom number on the corners and edges decreases, which is not conducive to the formation of CH₄, while the increase in atomic numbers in the plane promotes the formation of C₂H₄. In general, the selectivity of C₂H₄/CH₄ increases with the increasing size of NPs.

3.1.3 1D structure

Compared with 0D electrocatalysts, 1D nanostructures are beneficial for exposing the specific crystal plane, improving the electron transport performance, and maintaining long-term durability due to the preferential crystal growth and no defects on the surface. Various NWs electrocatalysts including powder and self-supporting NWs have been designed and the mechanisms have been extensively studied [123, 160,161,162,163,164,165,166,167]. Yang and coworkers [163] reported that ultrathin fivefold twinned Cu NWs exhibited high CH₄ selectivity with trace amounts of C₂H₄ due to the existence of the high-density twin boundary edge during CO₂RR. In addition, Sun’s group [160] reported 50 nm Cu NWs with especially selective for CO reduction into C₂H₄ and C₂H₆. Furthermore, Huang’s group [123] reported that Cu NWs with highly active stepped surfaces by in situ electrochemical activation demonstrate a remarkable FE of C₂H₄ (> 70%) and exceptionally high stability over 200 h (Fig. 1d). DFT studies reveal that the stepped surface was in favor of C₂ products compared with the Cu(100) surface and revealed a higher barrier for the C₁ pathways and HER. Similarly, Xia’s group [162] demonstrate that partially oxidized Cu NWs achieve an FE as high as 57.7% for C₂H₄ at − 1.0 V_RHE, which can be mainly due to the rough surface and the presence of defective sites and cavities.

Differing from powder NWs, the main products self-supporting NWs were complex, such as CO [168,169,170,171,172], CH₃OH [173], C₂H₄ [164], C₂H₅OH [164], C₂H₆ [165], and PrOH [166], which may result from differences in the length and density of Cu NWs. For example, Ma et al. synthesized Cu NWs array electrodes with different lengths and densities [65]. HCOOH formation can always be detected, while PrOH was observed on Cu catalysts with NWs length not less than (2.4 ± 0.56) mm. As the length of Cu NWs increased, the formation of C₂H₆ (FE = 2%) with a small amount of EtOH was observed. They proposed a formation route of C₂H₆ through the intermediate (CH₃CH₂O) in the CO–CO pathway. More importantly, with the increase in Cu NWs length and density, the local pH value increased, which can enhance CO–CO coupling to promote the formation of C₂H₄.

3.1.4 2D structure

Compared with bulk structures, 2D nanomaterials have larger specific surface area to expose specific crystal faces and facile electron and/or ion transfer properties, which are conducive to improving activity and selectivity of CO₂RR [40, 174,175,176]. For instance, Zhang and coworkers designed Cu NSs with defects at the nanoscale (2–14 nm) for CO₂RR into C₂H₄ in Fig. 1e [175]. Experimental and DFT calculation results reveal that highest FE of C₂H₄ FE (83.2% in Fig. 1f) with a current density of ~ 60 mA cm⁻² at − 1.18 V_RHE was attributed to enhanced adsorption of reaction intermediates and hydroxyl ions on the surface of nano-defective structure to synergistically promote C–C coupling for C₂H₄ formation. In addition, Kang’ group [176] reported that the Cu NSs exhibit a higher acetate FE (48%) with a partial current density of 131 mA cm⁻² during CORR. Further analysis suggested that the reduction of exposed (100) and (110) surfaces inhibited the formation of C₂H₄ and EtOH to enhance acetate selectivity.

3.1.5 3D structure

Similar to self-supported NWs, the major products of 3D nanostructure were complex, such as CO [177], HCOOH [67], C₂H₄ [178, 179], EtOH [180], C₂H₆ [181, 182], and PrOH, which may result from the influence of local pH, retention time of the intermediate, gas permeability or liquid diffusion. For instance, Broekmann and coworkers [181] prepared mesoporous Cu foam with C₂ (C₂H₄ and C₂H₆) FE reaching 55% at − 0.8 V_RHE. The systematic CO₂ electrolysis study shows that the surface aperture with moderate pore size was beneficial to improve the FE of C₂ by providing more available C–C coupling sites and increasing the retention time of the intermediate (in particular CO and C₂H₄). In another study, Broekmann’s group [179] also found that μm-sized pores on a 3D skeleton structure could capture the reaction intermediates (e.g., C₂H₄) more effectively and promote the complete reduction of C₂ products. In addition, the main product of Cu nanofoam with pore sizes prepared by Sargent was C₂H₄ [178].

Apart from micron-sized pores, nano-porous structures could also increase the local pH value and prolong the retention time of intermediates to enhance C₂ selectivity [182]. Cu mesopore electrodes with the precise adjustment of the pore widths and depths were synthesized by sputtering Cu on anodized aluminum oxide. In general, the FE of C₂ products increased with the decrease in hole width and depth, whereas that of the C₁ products was decreased. As the pore width was reduced from 300 to 40 nm with the hole depth 40 nm, the FE of C₂H₄ was increased from 8% to 38%. Interestingly, when the pore depth is increased from 40 to 70 nm with the hole width 40 nm, the main C₂ product was converted to C₂H₆ with an FE of 46%. Though the effect and mechanism of different morphologies on the activity and selectivity of CO₂RR are different, the morphologies with more edges, edges, or sharp tips seem to be more conducive to promoting C₂₊ products, such as nano-dendrite [183, 184]. Recently, Gao et al. reported a 3D bionic Cu catalyst on GDL, which mimicked the unique hydrophobic structure of Setaria’s leave (Fig. 1g) [185]. This unique structure was conducive to the establishment of a gas–liquid–solid triple-phase boundary. The experimental results showed that hierarchical Cu structure achieved high FE (64% ± 1.4%) of C₂₊ products with a current density of (255 ± 5.7) mA cm⁻² (Fig. 1h), as well as excellent stability over 45 h in the flow reactor at 300 mA cm⁻², which greatly exceeded the performance of the wettable Cu electrode.

3.2 Oxide-Derived Cu

Recently, oxide‐derived catalysts have attracted widespread attention for CO₂RR. Various oxide-derived metal electrocatalysts, including Au, Ag, Co, Sn, In, and Cu, have been prepared and their mechanisms have been studied in detail. The high selectivity and stability of oxide‐derived Cu-based catalysts mainly possess beneficial properties such as the low-coordinated [169, 171, 186], metallic Cu⁰ atoms [66, 69, 187,188,189], Cu⁺/Cu⁰ site [190], Cu⁺ site [35, 184, 191, 192], increased grain boundaries [68, 193,194,195,196,197] and subsurface oxygen [198, 199]. Although the actual active site is still debated, OD-Cu has shown excellent performance in reducing the required overpotential and improving selectivity. Oxide-derived Cu-based catalysts were prepared by annealed, electrochemical, and plasma. In this section, the selectivity of C₂₊ products on oxide-derived Cu-based electrocatalysts will be described in detail as followed.

3.2.1 Annealed/OD-Cu

Annealing is a convenient and useful way to improve the activity and selectivity of Cu-based catalysts, which is affected by the temperature and the corresponding gas atmosphere during CO₂RR or CORR. For example, W. Li et al. reported OD-Cu electrodes prepared by annealing in air at 500 °C [193], which produces EtOH and HAc with the FE of 57% at modest potentials. In addition, by using the TEM nano-diffraction analysis and temperature-programmed desorption (TPD) experiments [194], they found that the grain boundary density of OD-Cu and the CO binding sites decreased with increasing annealing temperature in N₂ (Fig. 2a, b). They concluded that the active sites on the surface of OD-Cu, especially strong binding sites of *CO, were supported by grain boundaries. To quantify the effect of GBs during CORR [195], they prepared five electrodes by depositing Cu on carbon nanotubes with different annealing temperatures in N₂. With the increase in annealing temperature, the amount of GBS in Cu NPs decreases gradually, confirmed by TEM. The activity was linearly related to the density of GBs (Fig. 2c), indicating that GBs were responsible for creating the vast majority of the active surfaces.

Fig. 2

a FEs for CORR in 0.1 M KOH. b Surface-area corrected j_CO at − 0.4 V_RHE versus the percentage of binding sites. (a, b) Reproduced with permission from Ref. [194]. Copyright © 2015, American Chemical Society. c Correlation between j_CO and GB surface density. Reproduced with permission from Ref. [201]. Copyright © 2015, American Chemical Society. d SEM images with a scale of 5 μm at different applied potentials after at least 1 h of reaction. e FEs of CO₂RR products on ERD Cu. (d, e) Reproduced with permission from Ref. [184]. Copyright © 2020, Springer Nature. f Q_defects/Q₁₀₀ and product selectivity as a function of t_a/t_c applied. Reproduced with permission from Ref. [127]. Copyright © 2020, Springer Nature. g EDS maps of Cu foils treated with O₂ plasma for 20 W 2 min, 100 W 2 min, and 100 W 2 min + H₂ plasma. h Summary of hydrocarbon selectivity of plasma-treated Cu foils at − 0.9 V_RHE. (g, h) Reproduced with permission from Ref. [35]. Copyright © 2016, Springer Nature

Interestingly enough, the major product was EtOH rather than C₂H₄ on OD-Cu prepared by annealing in air [74, 186, 193, 194, 200]. Apart from annealed Cu foil, Wang and coworkers reported CuO nanowires obtained by annealing in air for 8 h [186], in which the optimized Cu nanowires achieved 50% FE toward production of EtOH for CORR. By combining structural analysis and DFT calculations, the high FE of EtOH be related to the unsaturated (110) surface sites on the Cu NWs. Furthermore, the FE of EtOH was higher than C₂H₄ on the Cu foil annealed in air at 500 °C during CO₂RR and CORR [200], while that of C₂H₄ was higher on the surface of Cu (100), Cu (111), and polycrystalline Cu. In addition, the main products on annealed Cu electrocatalysts for CO₂RR were CO or HCOOH [68, 169, 171, 201]. The total FE of CO and HCOOH increased with the increase in annealing temperature in air [68], while this was opposite in 5% H₂/Ar [169, 171]. Wang and coworkers systematically studied CO₂ reduction on high-density Cu NWs, in particular the surface structure effects on the formation of *CO. They proposed a structure–property relationship between highly dense nanowires and identified open facets [e.g., (110) and reconstructed (110)], which may be the active sites for CO₂ reduction into CO at the low overpotentials.

3.2.2 Electrochemical/OD-Cu

Different from the annealed/OD-Cu, the electrochemical/OD-Cu showed higher FE of hydrocarbons than that of multi-carbon oxygenates [66, 70, 77, 184, 192, 202], including electrodeposition and anodizing, which could also promote activity and selectivity of C₂₊ products during CO₂RR. For instance, Sargent and coworkers [184] synthesized Cu nano-dendrites by electro-redeposition with a C₂₊ FE of ~ 73% (~ 45% C₂H₄, 22% EtOH, 9% PrOH) in Fig. 2d, e. They attributed the enhanced selectivity to the increased local pH and the presence of Cu⁺ at negative potentials. In addition, the electrodeposition of dendritic Cu resulted in the CO₂RR product selectivity toward C₂H₄ (FE = 34.3%) without multi-carbon oxygenates [77], while the thermal annealing treatment of dendritic Cu directed the CO₂RR product selectivity toward EtOH and PrOH (FE_total = 24.8%), which was higher that of C₂H₄. They proposed that the generation of hydrocarbon on the surface of electropolished Cu relied on a coupled C₁/C₂ reaction pathway that shares common intermediates such as *COH and *CH₂, whereas that on the annealed Cu catalyst depended on the CO–CO coupling pathway. However, they also reported that both annealed and electrodeposited Cu skeleton catalysts showed profound selectivity toward C₂H₄ and C₂H₆ [179], which suggested that this preference was related to the presence of (100) textured Cu. More importantly, Ren et al. found that the FE of C₂H₄ and EtOH can be systematically regulated by altering the thickness of the deposited overlayers [202].

Interestingly, on the electrodeposited/OD-Cu, the highest FE of C₂ products corresponds to potential in the range of − 0.9 to − 1.2 V_RHE, indicating that the formation of hydrocarbon products on the electropolished Cu is more likely to depend on the intermediates of *CHO or *COH rather than CO–CO coupling. Therefore, both CH₄ and C₂H₆ were observed [66, 70, 77, 179, 192, 202]. In particular, the FEs of CH₄ on the electrodeposited Cu electrodes were 55% and 59% [192, 202], respectively. Similar to the electrodeposited Cu, the peak potential of the highest FE of C₂ products on the anodized OD-Cu also located on the higher potential [42, 127, 140, 203, 204]. For example, by tuning the applied pulse potential [127], the surface structure and composition of Cu catalyst can be adjusted simultaneously during CO₂RR. The evolution of morphology was detected by using cyclic voltammetry and in situ atomic force microscopy, and the chemical states on the surface of catalysts were detected by quasi-in situ XPS. The results established a correlation between increased C₂₊ products (76% at − 1.0 V_RHE in CO₂-saturated 0.1 M KHCO₃ solution) in Fig. 2f and the presence of Cu (100) steps, Cu₂O and Cu (100) defects, which synergistically promoted C–C coupling.

3.2.3 Plasma/OD-Cu

Oxygen plasma treatment is a simple and scalable technique for rapidly changing the chemical state, tunning morphology, creating defects, and embedding atoms on catalyst surfaces at room temperature, which has been used to enhance activity and selectivity of CO₂RR [35, 191, 205]. For example, Mistry et al. showed that oxygen plasma-activated OD-Cu catalysts enabled a higher C₂H₄ selectivity (> 60%) than other plasma-treated Cu foils at − 0.9 V_RHE by suppressing CH₄ formation (Fig. 2 h) [35]. Combined STEM–EDS and operando XAFS, Cu²⁺ are found to be reduced quickly, while Cu⁺ species were remarkably resistant to reduction and remained on the surface during CO₂RR in Fig. 2g, which played a key role for CO₂RR into C₂H₄. Later, Gao et al. conducted plasma-activated Cu NCs catalyst with a tunable (100) facet and ion (O²⁻ and Cl⁻) content [191], which exhibited drastically FE of ~ 45% for C₂H₄ and ~ 22% for EtOH with current density of ~ 35 mA cm⁻² at − 1.0 V_RHE. By combining STEM–EDS results with in situ XPS, they proposed that the presence of surface and subsurface oxygen species was the key of affecting the binding of CO and Cu to achieve high activity and C₂₊ products selectivity. In addition, they also reported that O₂-plasma-treated Cu foil exhibited drastically FE of ~ 58.9% for C₂₊ products with current density of ~ 38 mA cm⁻² at − 1.0 V_RHE [205].

Recently, Scholten et al. prepared a series of well-oriented Cu (100) and Cu (111) single-crystal catalysts to determine the real catalytic active site [140]. All samples were first initially prepared via sputtering in an ultra-high vacuum (UHV), and then treated by Ar sputtering, O₂ Plasma, and electropolishing treatment, respectively. DFT calculations predicted that the single-crystal Cu electrode will generate hydrocarbon products, but their experimental results showed that well-oriented Cu single-crystal catalysts treated by UHV with atomically ordered arrangement generated H₂ during CO₂RR. Only electropolished and O₂-plasma-treated Cu catalysts produced significant amounts of hydrocarbons. The quasi-in situ XPS results ruled out the effect of the oxidation state or the presence of subsurface oxygen in the treated Cu catalysts on the different selectivity trends. ECSTM images confirmed the existence of step edges/bunches on the surface of treated Cu catalysts.

3.3 Organics Decoration

Organic molecule modification of Cu-based catalysts can not only improve the activity and selectivity of C₁ products (e.g., CO and HCOOH) [206,207,208], but also improve that of C₂ products (e.g., C₂H₄ and EtOH) [28, 141, 209]. Due to the interaction between organic groups and intermediates, the surface modification by organic molecules played an important role in regulating the electronic structure of Cu, changing the morphology, or affecting binding strengths of intermediates. For example, amino acid modifier was an effective way to dramatically enhance the FE of hydrocarbons on Cu electrodes during CO₂RR. Wang and coworkers for the first time reported that various amino acids (i.e., glycine, dl-alanine, dl-leucine, dl-tyrosine, dl-arginine, and dl-tryptophan) were used as surface additives on Cu electrodes [210], and found that amino acids containing –COOH and –NH₂ functional groups enhanced the hydrocarbon selectivity, while that could be weakened by using additives in the absence of –COOH or –NH₂. In particular, glycine-modified Cu NWs films, only containing –COOH and –NH₂ functional groups, provided higher hydrocarbon FE of 34.1%, nearly twofold than that of the bare Cu NWs films. Combining theoretical calculations with experimental observations, they concluded that both –COOH and –NH₂, especially –NH₂, could stabilize *COOH and *CHO intermediates to promote the formation of hydrocarbon. Moreover, a volcano-type dependency of hydrocarbon selectivity upon the amount of glycine modifier suggested that too high Glycine coverage may block active catalytic sites and inhibit CO₂RR.

Similar to amino acids, polyacrylamide (PAM)- or polyaniline (PANI)-modified Cu also stabilizes intermediates to improve selectivity of C₂ products due to the presence of –NH₂ functional group [206, 211, 212]. Sunyhik et al. reported that the Cu foam modified by PAM can double the FE of C₂H₄ as compared to unmodified foam (Fig. 3a, b) [211]. According to DFT calculations, PAM not only donates charge to the Cu surface through the Cu–O bond between Cu surface and its carbonyl group to activate CO for dimerization, but also stabilized the CO dimer intermediate via H bond due to the presence of –NH₂ functional group. Gewirth, A. and coworkers reported a Cu-polyamine hybrid catalyst through the co-electroplating scheme with the FE (87% ± 3%) for C₂H₄ at − 0.47 V_RHE [213]. The SERS results indicated that the higher CO₂RR activity toward C₂H₄ resulted from higher surface pH, higher CO content, and higher stabilization of intermediates on the Cu-polyamine electrode. Zhuang and coworkers reported a Cu/PANI interface by coating a PANI solution on Cu foil and the FE of C₂₊ products was increased from 15% to 60% (Fig. 3c, d) [214]. This excellent performance was attributed to an improvement in the coverage of *CO by using IR spectroscopy.

Fig. 3

a A schematic of CO–CO coupling for Cu with or without poly(acrylamide) modification. b Bar chart of FEs of Cu foams for CO₂RR without treatment and with modified with poly(acrylamide). (a, b) Reproduced with permission from Ref. [211]. Copyright © 2018, American Chemical Society. c Schematic illustration of Cu-PANI electrode and d its FE of CO₂RR products and current density. (c, d) Reproduced with permission from Ref. [214]. Copyright © 2020, American Chemical Society. e Molecular structures of additives 1–11. f Trend for C₂H₄ FE and calculated Bader charge. g FE of C₂H₄ on Cu and Cu-12. (e–g) Reproduced with permission from Ref. [217]. Copyright © 2020, Springer Nature. h Schematic illustration of free-base porphyrins possessing different linker lengths. Reproduced with permission from Ref. [219]. Copyright © 2017, American Chemical Society

Apart from the amines-containing molecules, pyridinium and benzimidazole (BIMH) were also employed as the competent modifiers to stabilize key reaction intermediate on the catalyst surface for selectivity escalation [215,216,217]. For instance, Han et al. reported that the selectivity of C₂₊ products can be tuned from 26.0% to 76.1% via electrochemical coupling of N-aryl pyridinium on the polycrystalline Cu [216]. Strikingly, H. Sargent and coworkers proposed a strategy to stabilize the intermediate during CO₂RR through the introduction of a series of N-substituted pyridinium-based molecules on the surface of Cu (Fig. 3e) [217]. By quantifying the CO configurations through in situ SERS, they found that the ratio of atop-bonded CO (CO_L) to bridge-bonded CO (CO_B) was positively correlated with the Bader charge of nitrogen atoms. A volcano-shaped relationship was observed between the FE of C₂H₄ and nitrogen Bader charge (Fig. 3f). Therefore, the pyridinium additives-modified Cu has moderate electron donation capacity to reach the optimal FE of C₂H₄, achieving a C₂H₄ FE of 72% (Fig. 3g). In addition, Zhong et al. reported BIMH-modified Cu foil, which can directly convert CO₂ to C₂₊ products with the FE of ~ 77% through enhancing formation of *COOH, verified by DFT simulations [215]. In particular, in addition to C₂H_4, molecular modifications have also been favorable for the generation of EtOH [218, 219]. Sargent et al. reported a CO₂-to-EtOH conversion with an EtOH FE of 41% at 124 mA cm⁻² by functionalizing the surface of Cu with porphyrin-based metallic complexes (FeTPP[Cl]) [218]. Compared to bare Cu, the local concentration CO on FeTPP[Cl]/Cu obviously improved based on Raman spectroscopy analysis. DFT calculations further indicated that higher local CO coverage contributed to C–C coupling and led the reaction toward EtOH. In addition, Gong et al. presented a modular synthetic approach via self-assembly of supramolecular iron porphyrin with a terminated thiol group bound to the Cu surface (Fig. 3h), achieving up to 57% EtOH FE during CORR [219].

In addition to stabilize intermediates and reduce the energy barrier of C–C coupling, N,N′-ethylene-phenanthrolinium dibromide (1-Br₂), and ethylenediaminetetraacetic acid disodium salt (EDTA-2Na) were beneficial for tuning the morphology of electrodes [220, 221]. Herein, Thevenon et al. reported a convenient method for fabricating of nanostructured Cu cube using 1-Br₂ as a molecular additive [220]. The FE of C₂₊ products was as high as 70%, and the surface morphology remained unchanged over 40 h. Mechanistic studies revealed that the protective organic layer formed by its dimerization not only kept the stable long‐time electrocatalysis, but also promoted the FE of C₂₊ products. Liu et al. constructed porous hollow Cu microspheres (H-Cu MPs) by using the method of EDTA-2Na-assisted electrodeposition [221]. The synthesized H-Cu MPS was spherical and uniformly dispersed, while Cu-Poly without EDTA-2Na presented a block-like morphology, which led to a double increased C₂H₄ FE from 23.3% to 50.1%. DFT calculations revealed that the adsorbed EDTA anions regulated the morphology, but also reduced the formation energy of the dimer to increase yield of C₂H₄.

3.4 Halogen Incorporation

Compared to oxygen, the effect of halogen ions (right next to the oxygen family), particularly F, Cl, Br, and I, has also been investigated and halide-derived Cu-based electrocatalysts showed extraordinary activity and selectivity of C₂₊ products for CO₂RR. Although the mechanism of the effect of halide-derived Cu was complex and contradictory, it mainly focused on the construction of unique Cu nanostructures or specifically adsorbed halogen ions on the surface of Cu. In earlier studies, the halide‐assisted reconstruction resulted in nanostructure Cu to explore more (100) planes at surfaces, resulting in a significant increase in C₂ products selectivity [110, 119, 196, 205, 222, 223]. Recently, Hu and coworkers [71] present a wet chemistry strategy to prepare a series of well-defined cuprous halide (e.g., CuCl, CuBr, and CuI) microcrystals in Fig. 4a. Interestingly, I‐derived Cu nanofibers showed a particular selectivity toward C₂H₆ at a low-overpotential and dendritic Cu from Br reduction dominantly presented a favorable formation of C₂H₄ at large overpotential, which all favor the higher selectivity of C₂₊ products, whereas Cl‐derived Cu NCs manifest being more favorable for the selectivity of CO and HCOOH in a wide range of applied potential. Their findings revealed that the morphology effect of halide-derived Cu was regarded as the main factor affecting the selectivity of CO₂RR products. Similarly, Kim et al. [224] prepared Cu-halide-derived catalysts with different nanostructures by using anodic halogenation. For all Cu-halide-derived catalysts, CO and HCOOH are subject to form at a low overpotential, while C₂ products gradually become predominant with C₂H₄ as the major product and a small amount of C₃ products gradually form when the potential turns more negative. The higher C₂₊ products were attributed to the high density of defect sites that promote C–C coupling and low roughness that inhibits HER.

Fig. 4

a Schematic illustration of the synthesis of CuCl, CuBr, and CuI and the Cu nanostructure via electrochemical reduction. Reproduced with permission from Ref. [71]. Copyright © 2019, American Chemical Society. b Total FEs and partial current density of C₂₊ products. Reproduced with permission from Ref. [225]. Copyright © 2019, Wiley. c Correlations between current density and halide adsorption potential. Reproduced with permission from Ref. [205]. Copyright © 2017, American Chemical Society. d The FEs of C₂₊ products over X-Cu catalysts. e Bulk halogen contents determined by EDX. (d, e) Reproduced with permission from Ref. [62]. Copyright © 2020, Springer Nature. f EtOH and C₂H₆ reaction pathway outlining the selectivity determining *OCH₂CH₃. Reproduced with permission from Ref. [73]. Copyright © 2020, Wiley

Apart from the construction of unique Cu nanostructures, the presence of residual halides can enhance activity and selectivity of C₂₊ products during CO₂RR. For instance, Gao et al. [225] reported that the presence of residual halides reduced the overpotential and increased the rate of C₂₊ products in Fig. 4b. This enhancement effect increased with the increase in halide ions adsorption capacity (I⁻  > Br⁻  > Cl⁻) by combining the trend of CO₂ electroreduction performance with XPS characterizations, which was consistent with his recent research (Fig. 4c) [205, 226]. Furthermore, Wang and coworkers confirmed that F⁻ existed on the surfaces of these catalysts by XPS characterizations, whereas Cl⁻, Br⁻ or I⁻species were detected [62]. The FE and normalized formation rates of C₂₊ products increased in the sequence of no halide < I < Br < Cl < F at all corresponding potentials (Fig. 4d), which was attributed to the difference of CO adsorption energy caused by halogen electronegativity and H₂O activation energy. The surface contents of halogens in the X-Cu (where X = F, Cl, Br, or I) catalysts measured by XPS were ~ 6 mol% (mol% means the molar percentage) (Fig. 4e) higher than their bulk contents (~ 2 mol%). Different from the research of Gao and coworkers, Wang proposed that it was difficult to disentangle the intrinsic effort of halide species due to the different catalyst morphologies.

So far, the mechanisms and products of halide-derived Cu-based catalysts are chaotic for CO₂RR. Qiao and coworkers [73] reported I-derived Cu using calcination and subsequent electrochemical reduction, which exhibited significantly greater C₂H₆ selectivity than C₂H₄ and EtOH (Fig. 4f). By combining with in situ XAS and Raman spectrum, they found experimentally for the first time that the generation of C₂H₆ and EtOH shared the same intermediate of oxygen-bound ethoxy and the stability of the intermediate was beneficial to controlling the formation of C₂H₆. Apart from the products of C₂H₆ and EtOH, the high CH₄ selectivity of 83% and a high C₂H₄ selectivity of 93% were achieved on Cu catalysts by using the epoxide-assisted hydrolysis method with Cl⁻ or I⁻, respectively [227]. Experiments and DFT studies show that the CO_L and CO_B on the local Cu⁰/Cu⁺ sites are favorable to couple into C₂H₄, whereas the CO_B on the Cu⁰ sites is suitable for the hydrogenation of CH₄. It was noted that the surface contents of halogens (Cl or I) measured by XPS were not detected after electrochemical CO₂ reduction, which was consistent with previous research [71, 119, 196, 222, 223].

3.5 Cu–M Bimetallic Alloys

Cu, as a catalyst to produce C₂₊ products, has attracted the most attention due to the neither too strong nor too weak bond between *CO and Cu. Compared with pure Cu, Cu-based bimetallic catalysts have demonstrated promising results in selectively generating C₂₊ products in the past few years [228, 229]. It is generally assumed that In and Sn metals are known to mainly generate HCOOH during CO₂RR, while Au, Ag and Zn metals are more likely to generate CO [230]. Thus, Cu–M bimetallic electrodes (where M = In, Sn) maintained a higher selectivity to CO or HCOO⁻. However, Cu–M (where M = Au, Ag, Zn) not only showed increased CO selectivity in some cases, but also have demonstrated synergistic effects, increasing selectivity of C₂₊ products [101]. In general, the CO₂RR performance of bimetallic catalysts varies with the change of composition and morphology, both of which might be modified. The products of various Cu-based alloy catalysts during CO₂RR were complex and summarized in previous reports [17, 101, 228,229,230,231], and we address recent progress on Cu-based alloy catalysts with C₂₊ products in the following sections.

3.5.1 Cu–Ag Alloys

Recently, Cu–Ag alloys system has been intensively proposed and showed enhanced CO₂RR selectivity and activity toward C₂₊ products than pure Cu. In general, the surface Cu-rich samples are preferred for C₂₊ products during CO₂RR, while that of Ag-rich is mainly CO, highlighting the importance of the metal atomic ratio in alloyed catalysts. The facilitated FE of C₂₊ products in the Cu–Ag alloys system has been usually associated with the enhanced coverage of *CO to suppress of HER, the diffusion of CO from Ag to Cu that promotes C–C coupling (CO spillover) [61, 232,233,234,235,236,237,238] and the optimized binding energy of reaction intermediates [23, 239,240,241]. For example, as shown in Fig. 5a, Gao et al. developed Cu NWs decorated with Ag islands by reducing Ag-covered Cu₂O NWs [234]. The production of CO on Ag sites led to the increase in local CO concentration to promote the formation of the C–C bond on Cu sites (Fig. 5b), which was confirmed by calculating the efficiency of CO spillover (Fig. 5c). Chen and coworker developed Ag NWs coated with various thicknesses of the OD-Cu layer via a two-step method for CO₂RR [233]. According to in situ Raman spectroscopy, the peaks greatly declined and then remained steady under − 0.55 V for 5 min in 0.1 M KHCO₃ due to the possible dynamic equilibrium between Cu⁰ and Cu⁺. Using in situ XAS, they found a strong correlation between the valence state of Cu and the distribution of products. They concluded that the CO generation can overcome the limitation of mass transport between Ag sites and nearby Cu sites to reach a high FE of EtOH. However, Cuenya et al. developed well-defined Cu₂O NCs (35 nm) uniformly covered with Ag NPs (5 nm) by reducing Ag-covered Cu₂O NCs. Compared to pure Cu₂O NCs, the FE of the Ag-covered Cu NCs electrocatalyst for C₂₊ liquid products (30%) displayed a twofold increase at − 1.0 V_RHE [238], while the formation of HCOOH and H₂ was inhibited. Operando XAS showed that Cu₂O NCs were partially reduced under CO₂RR conditions accompanied by the re-dispersion of Ag NPs, while the surface Cu₂O species was completely reduced through the Raman spectroscopy data. Raman spectroscopy data further revealed that the introduced Ag site caused a CO binding configuration and the significant variations of the bond between *CO and Cu sites, which were mainly caused by the Ag–Cu site, and were essential for C–C coupling. These proposed that CO spillover can effectively take place on the Ag sites near Cu, and the alloying of Cu with Ag could change the binding energy of CO to promote the generation of C₂₊ liquid products.

Fig. 5

a Presentation for the synthesis of Cu₂O and Cu₂O-Ag NWs. b Proposed mechanism of C₂H₄ formation on CuAg catalyst. c Quantitative analysis of CO generated on Ag sites. (a–c) Reproduced with permission from Ref. [234]. Copyright © 2019, American Chemical Society. d Binding illustration for Ag/Cu catalyst to produce EtOH and C₂H₄. e FEs for Ag_0.14/Cu_0.86 toward CO₂RR products. (d, e) Reproduced with permission from Ref. [23]. Copyright © 2019, American Chemical Society. f EDS mappings of Ag/Cu. g C₂H₄ FEs of Ag/Cu, AgCu, and Cu. (f, g) Reproduced with permission from Ref. [242]. Copyright © 2019, American Chemical Society. h Proposed mechanism for CO₂ reduction to EtOH on Cu_xZn catalysts. i FE of EtOH and the average FE_EtOH/FE$_{\text{C}_{2}}{_{\text{H}_{4}}}$ ratio on Cu_xZn. (h, i) Reproduced with permission from Ref. [59]. Copyright © 2019, American Chemical Society

Apart from the effect of CO spillover, Cu–Ag alloys were beneficial to increase the binding energy of the intermediates during CO₂RR (e.g., *CO, CH₃CHO*, and *CHCHOH), which improved the FE of alcohol production. For instance, as shown in Fig. 5d, Li et al. [23] developed the bimetallic Ag/Cu catalyst prepared by co-sputtering deposition to stabilize the intermediates. Experimentally, the synthesizing Ag/Cu catalysts achieve an EtOH FE of 41%, which was higher than that of Cu (29%) in Fig. 5e. DFT calculations revealed that multiple bind sites on Ag/Cu surface made the C₂H₄ reaction pathway unstable, but promoted the formation of EtOH intermediates, thus improving the selectivity of EtOH. Zheng et al. [241] also revealed that electron transfer at the interface of Cu and Ag was favorable for the intermediates CH₃CHO* and CH₃CH₂O* during CO₂RR based on DFT calculations, which resulted in 126-fold enhancement in the alcohol/C₂H₄ ratio. In addition, Zhang and coworkers found that the intermediate of *CO₂⁻ bonded to the surface of Ag-decorated Cu strongly through C and weakly through O by using in situ Raman spectroscopy. This unique adsorption was conducive to the formation of C₂H₄. Furthermore, Wang et al. developed Ag-doped Cu to promote C₁–C₂ coupling, which was more conducive to improving the selectivity of PrOH during CORR. Experimentally, Ag-doped Cu catalyst achieved a PrOH FE of 33% at 4.5 mA cm⁻² with a total C₂₊ FE of about 80%. DFT calculations showed that an asymmetric active site was provided due to strain and ligand effects, which contained two adjacent Cu atoms with distinct electronic structures to support both C₁–C₁ and C₁–C₂ coupling. Different from Ag–Cu alloy catalysts, Ag/Cu composites were investigated and showed excellent catalytic activity during CO₂RR. Du and coworkers constructed homogeneous Ag–Cu alloy and Ag/Cu composites by uniformly distributing tiny Ag NPs on the surface of large Cu NPs in Fig. 5f [242]. Compared with Ag–Cu alloy, Ag/Cu catalyst was favorable to the formation of C₂ products due to the presence of Cu–Ag interface, especially C₂H₄ (Fig. 5g).

3.5.2 Cu–Zn Alloys

Cu–Zn bimetallic catalysts have been considered to be an effective way to improve the catalytic performance of CO₂RR, and attracted much attention due to their low cost and non-toxicity. Because of the high selectivity of Zn nanomaterials for CO₂RR to CO, it was expected that a bimetallic Cu–Zn bimetallic nanomaterial would generate hydrocarbons via the method of “relay catalysis”. Several previous works reported that the main product of CuZn alloy catalysts was CO or HCOOH as the main product, which could probably be due to the reduced adsorption energy of both surface *CO₂ and *CO species due to the lower Cu contents in Zn-rich alloys [243, 244]. Recent studies have demonstrated enhanced C₂₊ products FE on Cu–Zn catalysts for CO₂RR. For instance, Ren et al. [59] reported that the selectivity of EtOH and C₂H₄ can be changed by tuning the content of Zn in the Cu_xZn film catalyst, in which the highest FE of CO₂ reduction to alcohols was 29.1% in Fig. 5h. The authors demonstrated that the amount of CO on Cu site could be increased due to the spillover of CO from Zn to Cu site, which was further reduced to *CHO or *CH (Fig. 5i). Furthermore, the weak adsorption energy of *CO on the Zn sites also can further reduce to EtOH by inserting the bond between Cu sites and *CH₂ to form *COCH₂. In addition, B–Cu–Zn catalysts were favorable toward the formation of high selectivity for EtOH versus C₂H₄, tunning from 0.57 to 1.04 with the increase in Zn content [245]. CuZn bimetallic catalysts prepared by in situ electrochemical reduction bi-metal-oxide, achieved 41% FE for C₂₊ liquids at only − 0.68 V_RHE [246]. Operando Raman spectroscopy showed that the stretching modes of CO binding on Cu were changed by Zn compared to pristine Cu and the adsorbed *CH₂ or *CH₃ were believed to form *COCH₃ intermediate and further reduce to EtOH. It is worth noting that CuZn bimetallic was completely reduced [246], while B–Cu–Zn catalysts were partially reduced [245]. This difference may depend on the reduction time, zinc oxide thickness, and synthesis method.

Different from the higher FE of EtOH for Cu–Zn bimetallic catalysts, hydrocarbons containing no alcohols can also be formed. For instance, Du and coworkers proposed an engineering strategy to stabilize the Cu⁺ species by constructing Cu/ZnO_x interfaces [247]. In situ Raman spectroscopy measurement demonstrated that OD-Cu is easier to be reduced than Cu/ZnO_x in a CO₂-saturated 0.1 M KHCO₃. They proposed that the higher FE of CH₄ and C₂H₄ on Cu/ZnO_x catalysts is attributed to the presence of the surface Cu⁺ species. By tuning the concentration of Zn [248], size-controlled (~ 5 nm) Cu_100−xZn_x NPs showed a drastic increase in CH₄ selectivity (~ 70% FE) with trace amount of C₂H₄ for Zn contents from 10 to 50, while the H₂ production was suppressed. Hou reported that Zn@Cu NPs exhibited the highest C₂H₄ FE of 43.1% without alcohols [249]. In addition, The C₂H₄ FE was as high as 91.1% on Cu oxides/ZnO electrocatalysts. Furthermore, Du and coworkers also prepared Cu–Zn Alloy and Cu–Zn mixture catalysts [243]. Compared with Cu–Zn mixture, Cu–Zn material showed significant selectivity to C₂H₄ accompanied with the simultaneous decrease in CO, indicating that both stabilization of *CO and its transfer from Zn to neighboring Cu promoted the formation of C₂H₄.

3.5.3 Cu–Au/Pd/Ni/Al Alloys

Compared with Cu–Ag and Zn systems, the Cu–Au system has obvious similarities. As the Au content changes, the different components of the Cu–Au system show enhanced selective generation of CO [250, 251], C₂H₄ [252], and even alcohols [253,254,255]. For example, Luo reported AuCu alloy NPs embedded Cu submicrocone arrays [253]. Enhanced EtOH selectivity with an FE of ~ 31% was gained in KHCO₃ aqueous solution, while C₂H₄ formation was relatively inhibited. DFT calculation results indicated that the introduction of Au enhanced the binding energy of *OCHCH₃ intermediate to improve the selectivity of EtOH. In addition, Zhang and coworkers successfully synthesized heterophane 4H/fcc Au@Cu nanostructures, which achieved the FE$_{\text{C}_{2}}{_{\text{H}_{4}}}$ of 46.7% at − 1.17 V_RHE. In addition, Cu–M (where M = Ni, Pt, Pd) generally leads to an increase in HER activity compared to pure Cu, but hydrocarbons and alcohols are also produced in some cases. For instance, Han and coworkers reported that Cu–Pd bimetallic catalysts with the C₂H₄ FE of 45.2% at − 1.2 V_RHE [256]. CO and CH₄ were also detected on Cu–Pd and Cu–Pt bimetallic catalysts [257,258,259]. For Cu–Ni bimetallic catalyst, Zhang et al. synthesized Ni–Cu NW by uniformly decorating atomic nickel clusters on the surface of defect-rich Cu, which reached the highest C₂₊ products FE of 62% for at − 0.88 V_RHE [260]. DFT calculations found that the binding sites changed from the Ni–Ni bridge of *CO₂ and *COOH to the Ni–Cu bridge of *CO, which favored the C–C coupling step. Importantly, it should be noted that Sargent and coworkers used machine learning (ML) and DFT calculations to construct a mutually feedback screening framework to study the CO₂RR performance of Cu-based alloy catalysts under different atomic configurations. The surface adsorption energy of key intermediate *CO on the surface of over 16 000 different metals or alloys with different atomic arrangements was studied. They found that the low Miller exponent surfaces of Cu–Al alloy containing 4%–20% Al had the optimal adsorption energy for *CO and *H, which significantly reduced the energy barrier of *COOH, C–C coupling and C–H, effectively promoting the formation of C₂H₄. Using the method of physical vapor deposition, they prepared a series of large-area nano-porous Cu–Al catalysts on the gas diffusion electrodes by controlling the surface composition, achieving an 80% FE of C₂H₄ with an ultra-high current density of 600 mA cm⁻². This work provided deep understanding of the core physicochemical mechanisms of CO₂RR, improving the performance of electrocatalytic reduction of CO₂ to key multi-carbon products, and promoting the practical application of CO₂RR technology.

4 Advanced Tools for Investigating the Mechanism of CO₂RR

At present, research on the mechanism of CO₂RR in the literature is mostly based on spectroscopy characterization techniques, isotope labeling, and DFT calculations. The characterization of electrocatalysts during CO₂RR is particularly important because of the dramatic transformation. As described in the above sections, Cu oxide is reduced to form a metallic Cu interface during CO₂RR, resulting in the generation of more active sites. In order to investigate the active phase of the catalyst and the relationship between structure and performance during CO₂RR, in situ or operational spectroscopy characterization techniques under controlled conditions were used. In addition, isotope labeling and DFT calculations are used to determine active sites of the catalyst and pathways of the product. These techniques can contribute to a deeper understanding of the mechanism of CO₂RR by (1) identifying catalytic active sites, (2) detecting key intermediate configurations, and (3) exploring reaction pathways.

4.1 Raman Spectrum

Different from other vibrational spectroscopy techniques, Raman spectroscopy is not only widely used to characterize the adsorbed species due to no obvious interference signal at the interface of electrodes and electrolytes, but also used to detect the evolution of catalysts because of its high spatial resolution (< 1 μm) [29, 261]. However, Raman spectroscopy, based on molecular bond vibrations, is not suitable for the detection of pure metal catalytic systems. In addition, due to the low detection sensitivity of Raman spectroscopy, Electrochemical SERS (EC-SERS) was developed with high detection sensitivity by increasing the Raman cross section to obtain appropriate data, which was used to investigate the intermediate adsorption and the change of catalyst chemical states in electrocatalysis processes [29, 262].

Thanks to its high sensitivity and rapid response ability, in situ EC-SERS is more suitable for detecting reaction intermediates during CO₂RR. Based on the Raman spectroscopy of *CO on Cu surfaces during CO₂RR, *CO is known to be a key intermediate for CO₂ reduction into C₂₊ products. The related signals of *CO at 280–288 and 360–370 cm⁻¹ could be assigned to the restricted rotation of adsorbed CO and Cu–CO stretching (Fig. 6a), respectively [59, 80, 234, 246, 263,264,265]. The wave numbers of CO stretching modes were complex and widely distributed, which mainly included CO_L and CO_B [39]. Remarkably, Xu’s group systematically studied the correlation between the Raman band of *CO and pH (Fig. 6b–d). The CO_L would be converted into CO_B as the pH of the electrolyte increased [39]. The presence of microenvironment on two samples also caused the subtle differences of CO adsorption bands. This observation presented by Xu and coworkers was further corroborated by papers found as far as possible [16, 25, 234, 237, 238, 246, 265,266,267], while CO_L detected in 1 M KOH has also been reported at − 0.6 V [80, 185, 268, 269]. It is worth noting that the CO_L was detected in 2 M KOH at − 0.61 V due to the retention of Cu–O species in the catalyst and the peak strength of CO_L decreased with the decrease in Cu–O residues during CO₂RR [268].

Fig. 6

Operando Raman spectra of a *CO and b *C–H during CO₂RR. (a, b) Reproduced with permission from Ref. [234]. Copyright © 2019, American Chemical Society. In situ SER spectra on polycrystalline Cu MPs in c, f 0.1 M KHCO₃, d, g 0.05 M K₂CO₃, e, h 0.1 M KOH. (c–h) Reproduced with permission from Ref. [39]. Copyright © 2020, American Chemical Society. Raman spectra of i Cu MPs, j Chem-Cu, and k OD-Cu in KOH electrolyte. l Comparison of product FE on four types of Cu. (i–l) Reproduced with permission from Ref. [25]. Copyright © 2020, American Chemical Society

In addition to *CO, CO₃²⁻, *COOH, and *C–H_x were observed on various catalysts to demonstrate preferred reaction pathways by using in situ EC-SERS. For example, two peaks at 1 010–1 032 and 1 065–1 080 cm⁻¹ corresponded to the C–OH stretching of HCO₃⁻ and the total symmetric C–O stretch of carbonate, respectively [240, 282, 283]. The bands observed at 1 640 cm⁻¹ belonged to the COOH* band and shifted to lower wavenumbers of 1 510 cm⁻¹ through ¹³C isotopic substitution due to the isotopic effect. The adsorbed HCO₃⁻ species firstly saturated and could be further reduced to COOH*. These results suggest that COOH* played a key role during CO₂RR [284]. Based on operando EC-SERS and the systematic analysis of electrocatalytic behavior, Ren et al. reported that free CO and *CH₃ were believed to be capable of forming *COCH₃, further exclusively reducing to EtOH [246]. In addition, CH-containing intermediates in the region from 2 700 to 3 000 cm⁻¹ (Fig. 6b) also were identified as the key intermediate of C₂H₄ and C₂H₆ [73, 234, 269, 271]. Importantly, it was a huge challenge to assign these mountains more precisely due to a variety of species and adsorption modes of reaction products containing complexity C–H bonds [265].

Recently, in situ EC-SERS has been widely used to detect the evolution of oxygen-containing species [i.e., CuO, Cu₂O, CuO_x/(OH)_y] on Cu-based catalysts during CO₂RR. Multiple Raman peaks around 218, 427, 528, and 616 cm⁻¹ were attributable to Cu₂O [59, 202, 279, 280]. In clear contrast, CuO was known to have a high strength peak at 288 cm⁻¹ and a low strength peak at 337 cm⁻¹, which was not detected. In addition, the Raman bands of Cu(OH)₂ located at about 292 and 483 cm⁻¹ were also not detectable [265, 280]. Importantly, CuO_x/(OH)_y on Cu-based catalysts could be detected during CO₂RR [25, 39, 42, 237]. Remarkably, Xu’s group systematically studied the pH dependence of Cu surface speciation for the first time by in situ EC-SERS (Fig. 6f–i) [39]. Interestingly, the SERS band of Cu–O species is absent between − 0.4 and − 0.8 V_RHE in CO-saturated 0.1 M KHCO₃ (pH = 8.9), and their onset potential becomes more positive with the increase in pH values of electrolytes. Furthermore, the SERS band of the Cu₂O peak disappeared in CO₂-saturated 0.1 M KHCO₃ for all samples at − 0.4 V_RHE [238], which was consistent with the quasi-in situ XPS analysis. In addition, the peak strength of vibrational patterns located at 518 and 624 cm⁻¹ from Cu_x(OH)_y and Cu₂O specie on the catalyst surface has dropped to zero at potential <  + 0.2 V_RHE in CO₂-saturated 0.5 M KHCO₃ (pH = 7.2) [237], which was even more corrected than the potential measured by the Operando X-ray diffraction (XRD) due to possible technical differences. The above studies showed that Cu–O species at low pH were completely reduced, which was consistent with recent reports in Table 2. However, the peaks of the Raman fingerprints of Cu–O species still existed and were not completely reduced in CO₂-saturated 0.5 M KHCO₃ [233, 245].

Table 2 Summary of recently reported CORR/CO₂RR catalysts characterized by in situ/operando Raman spectroscopy

Importantly, recent studies have clarified that the presence of Cu–O species on the surface of Cu-based catalysts may not play a key role in the formation of C₂₊ products during CO₂RR [25, 202, 280, 285]. For instance, Mandal et al. investigated the Cu oxidation state of Cu₂O catalysts with different morphologies (NWs, nanocrystals, and NPs) during CO₂RR by in situ Raman spectroscopy [280]. They found that CO₂RR products were observed only after the Cu₂O on the surface layer of catalysts was reduced to metallic Cu, which may be attributed to the fact that the reduction of Cu₂O was kinetically and energetically favorable than CO₂RR. Ren et al. and Dutta et al. have reported this similar phenomenon [202, 285]. Dutta et al. reported that the CO₂-related species during CO₂RR occurred preferentially on the metal Cu rather than the Cu–O species by using the operando Raman spectroscopy [285]. The disappearance XRD peaks of crystalline Cu₂O were observed by operando XRD, which happened before the transformation of Cu(I) to Cu(0) was completed by XAS detection. Through the in-depth analysis of XAS and XRD, it was further found that Cu₂O eventually formed metallic Cu due to lacking of long-range transitional order at most applied cathode potentials. Furthermore, Xu and coworkers also investigated the speciation of commonly used Cu surfaces (including Cu foil, Cu MPs, Cu NPs, EP-Cu film, and OD-Cu) with the potentials of CO reduction reaction at CO-saturated 0.01 M KOH and 0.045 M K₂SO₄ (pH = 11.7) through in situ SERS (Fig. 6j–l) [25]. At negative potential, CuO_x and CuO_x/(OH)_y species existed on all five catalysts, but the relative abundance of CuO_x on Cu foil was higher, whereas that of Cu(OH)_y on Cu MPs or Cu NPs was higher. This suggested that the oxidized species were related to the initial degree of oxidation of Cu. The electrochemical CO reduction showed that the variation trend of product selectivity for five catalysts was different (Fig. 6m). Combining the variation trend of product selectivity and in situ SERS results on these catalysts, they concluded that Cu–O species were unlikely to be active in promoting C–C coupling pathways during CORR.

Apart from in situ EC-SERS, Cu–O species at low pH were completely reduced by using EELS and in situ TEM. For example, Lei et al. [41] prepared HQ-Cu (containing Cu, Cu₂O, CuO) and AN-Cu [containing Cu, Cu(OH)₂]. The distribution and evolution of Cu–O species were studied by electron microscopy and EELS and ultrathin specimens on an electrode were obtained through a focused ion beam. They found that Cu–O species on all electrodes have all been reduced to metallic Cu in CO₂-saturated 0.1 M KHCO₃ during the steady stage of the CO₂RR, regardless of the initial states, indicating that Cu–O species were not active sites for C₂₊ products. The pretreated electrode of HQ-Cu and AN-Cu, fully reducing to Metallic Cu by electroreduction at − 5.0 V vs. Ag/AgCl, showed even higher selectivity of C₂₊ products than the unpretreated. The grain boundaries and high exponent surfaces, resulting from the fragmentation process during the electrochemical oxidation–reduction cycle, promoted the C–C coupling. Furthermore, Peter and coworkers reported novel 2D CuO NS catalysts with high C₂₊ products and industrially relevant currents in gas diffusion electrodes at 0.1 M CO₂-saturated KHCO₃ solution [40]. The (001)-oriented CuO NS slowly evolved into highly branched Cu under applied bias by operando XAS and in situ TEM. The electrochemical transformation of CuO into disordered and undersaturated coordination metallic Cu took more than 2 h by operando XAS during CO₂RR, which were held responsible for high C₂₊ products.

4.2 IR Spectrum

Considering that the frequency of adsorbed functional groups is suitable to the range of IR spectroscopy, it has become an important tool to monitor adsorbed species on the surface of metal electrodes in the electrochemical process [286]. Four IR spectroscopy techniques can be accessible to investigate CO₂RR such as (1) transmission; (2) diffuse reflectance; (3) reflection–absorption, and (4) attenuated total reflection (ATR). The ATR-IR spectroscopy is commonly used for CO₂RR due to the minimization of electrolyte interference. The interactions between the attenuation waves and the adsorbed molecules enhance the absorption signal on the surface of thin metal electrodes, and this technique is often called SEIRAS.

Due to the higher IR absorption cross section of CO, ATR-SEIRAS is particularly sensitive to adsorbates containing C–O bonds. Waegele et al. reported that CO_B was an irreversible and inert surface adsorbed species under alkaline conditions during CO₂RR by using ATR-SEIRAS and DFT calculations [287]. Therefore, CO_B was rarely reported [227], while CO_L was generally reported and detected on the surface of higher C₂₊ products selectivity catalysts among the research papers [62, 227, 269, 288,289,290]. It was noted that only strong CO_B binding sites were detected on MP-Cu in 0.05 M KHCO₃ through in situ ATR-SEIRAS and were easily hydrogenated to produce CH₄ due to lower Cu–Cu coordination numbers [227], whereas the coexistence of CO_B and CO_L are apt to promote the formation of C₂H₄ (Fig. 7a, b). In addition, Xu’s group investigate different sites available on surfaces of Cu-poly and OD-Cu using operando ATR-SEIRAS during CORR [291]. They identify that the CO_L located at 2 058 cm⁻¹ on OD-Cu was different from that on Cu-poly (2 073 cm⁻¹) in Fig. 7c but resembled CO_L on Cu(100) observed by Hori et al. [292]. Combined reactivity studies and operando ATR-SEIRAS data, the distinct CO_L was most likely responsible for the OD-Cu’s ability of facilitating C–C coupling, which was consistent with the conclusions of Hwang’s group [289].

Fig. 7

In situ ATR-FTIR study of a MP-Cu and b EP-Cu. (a, b) Reproduced with permission from Ref. [227]. Copyright © 2021, Elsevier. c Operando ATR-SEIRAS spectra showing CO binding sites. Reproduced with permission from Ref. [291]. Copyright © 2019, American Chemical Society. d Schematic structures of adsorbed intermediates on Cu(100) and their calculated vibrational frequencies. Potential-dependent absorbance spectra for Cu(100) in the presence of e ¹³CO and with f D₂O as electrolyte. (d–f) Reproduced with permission from Ref. [56]. Copyright © 2017, Wiley

Beyond *CO, *CHO, *COOH, *HCO, *OCH₃ *OC₂H₅, *HCOO, *OCCOH, and *OCCOH were observed on various catalysts to demonstrate preferred reaction pathways through in situ ATR-SEIRAS [56, 62, 289, 290, 293]. For example, *COOH and bidentate COO⁻ species on Cu atoms were observed and possibly separated in the subsequent pathways to produce CO or HCOOH, respectively [293]. The CHO species from *CO hydrogenation at around ~ 1 754 cm⁻¹ was the key intermediate for C–C coupling and replaced *O*CCO on the F–Cu(111) facets as the RDS based on DFT calculations [62]. In addition, Xiong et al. reported that *CHO served as not only the key intermediate for C–C coupling to form *OCCHO, but also the precursor of CH₄ at more negative bias based on in situ ATR-SEIRAS and DFT calculations [290]. Furthermore, in situ ATR-SEIRAS coupled with DFT calculations showed that two bands at 1 191 and 1 584 cm⁻¹ are detected in Fig. 7d–f, corresponding to the C–O–H and C–O stretching of *OCCOH during CORR on Cu(100) electrodes at low overpotentials [56]. This observation directly confirmed for the first time the hypothesis that the C–C coupling to C₂ products on Cu(100) takes place through the CO–CO dimerization mechanism and the vibration characteristics of CO dimer cannot be observed on Cu(111). Together with ATR-SEIRAS, ambient-pressure XPS, and DFT calculation, Shao’s group elucidated the mechanism of CO₂RR [294]. This work revealed that CH₄ and CH₃OH produced via *OCH₃ obtained from O-bound CO₃, while the formation of C₂ hydrocarbon products was suggested to result from the C–C coupling between *CO and *CHO. In addition, *OCCO was observed for the first time and identified as an intermediate of C₂ products pathway by using in situ ATR-SEIRAS [289]. Through the time-resolved IR analysis, CO dimerization occurred concurrently with CO adsorption (~ 5 s), while the conversion of *CO-to-*CHO has slower kinetics (~ 30 s), clearly indicating that C₁ path and C₂ path have different kinetics.

4.3 X-Ray Absorption Spectrum

The XAS spectra using synchrotron radiation sources is a powerful tool to probe the atom-specific geometry, valence states, and electronic structure of catalysts, thereby providing comprehensive understanding of the properties of the catalytic active sites and providing a valuable way for real-time reaction kinetics during CO₂RR [29, 295]. XAS spectra include X-ray absorption near-edge structures (XANES) and EXAFS. The former could quickly identify element species and the corresponding local structures. In addition, it is sensitive to the average oxidation state of elements, the unoccupied electron state, and charge transfer between occupied and unoccupied states. The latter provides local structural information including the bond distance and coordination number. Thus, operando XAS has been widely utilized to probe the active sites of electrodes and provide the real-time reaction dynamics under working conditions during CO₂RR.

The oxidation state of Cu in the process of CO₂RR was generally regarded as a key factor in determining the activity and selectivity of Cu-based catalysts, especially in promoting C–C coupling, which was well-characterized by operating XAS under continuously varying cathode potential bias. However, there has been a long-lasting debate about the effect of (sub)surface oxygen due to the challenges in precisely and quantitatively distinguishing the Cu⁺ species during CO₂RR. The presence of Cu⁺ on Cu-based catalysts has been demonstrated by the evolution of the Cu K-edge. For instance, Yu and co-works’ research showed that the content of Cu⁺ species in multi-hollow Cu₂O was 32.1% after 20 min of reaction through linear-combination fitting of Cu K-edge spectra (Fig. 8a, b), which was consistent with the Raman measurements [268]. The retained Cu⁺ species during CO₂RR were considered to be the key factor for the high C₂₊ selectivity, which was consistent with previous research (Fig. 8c) [184]. On the other hand, other reports have showed that the initial Cu⁺ species could not affect the selectivity of C₂₊ products and were reduced to metallic Cu at the operating potential. Chang et al. observed that the reoxidation Cu may be the paramount key of products profile by employing in situ XAS and SERS [233]. Moreover, it was found that the amount of Cu⁺ decreases with time due to the slow Cu⁺  → Cu⁰ reduction process. It should be noted that the XAFS signal may be not sensitive to detect the trace amounts Cu⁺ species on the surface of catalysts. For example, Mistry reported that the features of Cu and Cu₂O were still detected in the first 0.25 h [35], whereas only the characteristics of metallic Cu could be discernable after 1 h. Although the Cu⁺ content of about 25%–28% on the surface of catalyst was detected by STEM–EDS, the XAFS signal was dominated by the signal for bulk metallic Cu.

Fig. 8

a Cu K-edge XAS spectra of the multi-hollow Cu₂O catalyst at − 0.61 V_RHE. b Change of the Cu⁺/Cu⁰ ratio with the reaction time at − 0.61 V_RHE. (a, b) Reproduced with permission from Ref. [268]. Copyright © 2020, American Chemical Society. c Calculated the Cu⁺/Cu⁰ ratio with respect to time. Reproduced with permission from Ref. [184]. Copyright © 2020, Springer Nature. d Average oxidation state of Cu in Cu(B). Reproduced with permission from Ref. [53]. Copyright © 2018, Springer Nature. e Time-resolved in situ Cu K-edge XANES spectra of the Cu_2−xS nanocrystals and f corresponding FT-EXAFS spectra. (e, f) Reproduced with permission from Ref. [312]. Copyright © 2021, Springer Nature. g The K-edge XANES of Cu_0.5NC. h Fourier transform of the EXAFS spectra of Cu_0.5NC. (g, h) Reproduced with permission from Ref. [145]. Copyright © 2021, Wiley. CNs of the first i Cu–O and j Cu–Cu. (i, j) Reproduced with permission from Ref. [40]. Copyright © 2021, Springer Nature

Different from oxide-derived Cu-based catalysts, nonmetal elements (e.g., B, N, P, F, and S) substitution has been demonstrated to regulate the oxidation state of Cu-based catalysts [22, 53, 62, 309, 310]. During CO₂RR, in situ XANES was usually used to detect the average oxidation state of Cu. For instance, Zhou et al. prepared a series of B-doped Cu catalysts and used B as the dopant to tune the ratio of Cu⁺ to Cu⁰, which achieved a maximum FE of ~ 80% for C₂ hydrocarbons. Due to the changes of surface geometry and the interaction between B and Cu, electrons transfer from Cu to B, resulting in the formation of Cu⁺. In situ XANES results showed that the average oxidation states of Cu increased with the increase in B content during CO₂RR (Fig. 8d). DFT calculations indicated that the B dopant not only improved the reaction energy of the *CO + *H → *CHO step to inhibit the C₁ pathway, but also decreased the reaction energy of the C–C coupling step to enhance the C₂ pathway. Their theoretical and experimental results indicated that the Cu⁺ sites played a key role in promoting the conversion of CO₂ to C₂ products. In addition, the average oxidation states of Cu in the F–Cu, Cu₂S–Cu–V, and Cu₃N catalysts were between 0 and + 1 during CO₂RR by using in situ XANES [22, 62, 309], which is beneficial to increase the FE of C₂₊ products. It was noteworthy that the Cu₂S and Cu₃N phases could be fully reduced into metallic Cu during CO₂RR [311, 312]. In particular, Xiong and coworkers reported polycrystalline Cu NPs prepared by the electrochemical desulfurization and surface reconstruction of Cu_2−xS [312], which achieved the FE of 68.6% toward C₂H₄ production at − 1.2 V_RHE. In situ XAS and OH⁻ adsorption characterizations indicated that Cu_2−xS was converted to the metal Cu with high-index facets during the continuous electroreduction reaction (Fig. 8e, f), which were the active sites for the C–C coupling.

In addition to detecting the average oxidation state of metal atoms, XAS can also be used to explore the coordination number of metal atoms. Molecular catalysts containing M–N₄ sites are also highly active and selective for CO₂RR as well, in which a reversible change between the atomically dispersed M–N₄ sites in M–N–C and NPs was confirmed with operando XAS during the CO₂RR [295, 306, 313]. The major products of molecular catalysts with a well-defined M–N₄ structure (MPcs: M = Mn, Fe, Co, and Ni) were CO [313], while that of CuPc were CH₄ [306]. Importantly, this phenomenon has also been reported on SACs [145], in which metallic Cu on Cu NPs are the active sites for CO₂ reduction to C₂ products. K. Dilan et al. reported that the isolated Cu sites in Cu–N–C material transiently and reversibly convert into metallic Cu NPs during electrolysis by operando XANES and EXAFS (Fig. 8g, h) [145]. Cu NPs were formed after prolonged electrolysis at −1.2 V_RHE. A variety of products (CO, C₂H₄, EtOH, CH₄, etc.), especially higher FE of EtOH (55%), were detected due to the coexistence of the isolated Cu sites and Cu NPs. Moreover, Cu NPs have also been detected by ex situ XPS and are responsible for CO₂RR into C₂ products.

In addition, operando XAS has also been utilized to monitor the coordination number of a variety of Cu-based catalysts. The coordination number of Cu can affect catalytic activity of Cu-based catalysts, which can be subtle or significant, depending on the nature of the catalyst. It is worth noting that lowering the coordination number by introducing doping, defects, and grain boundary is also a strategy to enhance the adsorption of reaction intermediates and reduce the energy barrier of C–C coupling, allowing improving the selectivity of C₂₊ products during CO₂RR. In Table 3, the decrease in Cu–Cu coordination number showing enhanced CO₂RR performance has been observed. To prepare highly efficient CO₂RR electrocatalysts for the formation of C₂₊ products, several catalyst design strategies, such as O-D Cu, alloy, and halogen incorporation, have been demonstrated to lower the coordination number of Cu. For example, Peter and coworkers reported that CuO NS catalysts were completely chemically reduced to metallic Cu after 130 min at −0.84 V_RHE during CO₂RR by using operando XAS [40]. Linear combination fittings revealed that the metallic Cu accounted for 99% and the remaining 1% was indeterminate. The CN value of Cu–Cu gradually increased during the reduction process and tended to be stable at 120 min (Fig. 8i–l). The obtained CN value of Cu (~10) was smaller than that of the bulk of Cu (12), indicating the presence of the undercoordinated Cu atoms.

Table 3 Summary of CN for recently reported CORR/CO₂RR catalysts fitted by EXAFS spectrum

In summary, operando XAS provides the information of average valence states, coordination numbers, and the evolution of catalysts under operating conditions, thus elucidating reaction active sites. However, the processes of CO₂RR occur at the electrode/electrolyte interface, and operando XAS measures the average structural information of samples. Therefore, operando XAS is very suitable for probing catalysts with a single active site. In addition, operando XAS was not sensitive for detecting Cu⁺ species. To date, the oxidation state of Cu as the catalytic active center is controversial during CO₂RR, but the unsaturated state of Cu with higher C₂₊ selectivity has been fully confirmed. The OD-Cu and the nanostructured Cu catalysts show a lower CN through fitting the EXAFS curve, which may be the real reason for the increased catalytic activity of the OD-Cu.

4.4 Isotope Labeling

So far, Cu-based catalysts, as the most promising CO₂RR electrocatalysts, have great potential to directly convert CO₂ into valuable hydrocarbons and oxygenates at a reasonable current density to reduce carbon emissions. Although the products of carbon dioxide reduction are complex, up to 18 different possible products, it is commonly accepted that, except for HCOOH, other products are initiated by reducing CO₂ into *CO. Further protonation of *CO or *CO dimerization leads to a variety of products.

Since it is difficult to control the selectivity of a given product, in order to further improve the selectivity of catalysts, it is essential to first identify and understand the active sites and reaction pathways of products on Cu-based catalysts. It is noting that isotope labeling is an effective method to identify active sites and determine reaction pathways, including ¹⁸O, ¹³C, and deuterium (D) isotope labeling. ¹⁸O isotope labeling can not only confirm the residual O in OD-Cu during CO₂RR, but also determine the source of O in the products. For instances, Lum et al. synthesized ¹⁸O enriched O-derived Cu by oxidation/reduction cycling in H₂¹⁸O [197]. Secondary-ion mass spectrometry showed that only a small fraction (< 1%) of ¹⁸O content remains on the ¹⁸O enriched O-D Cu in Fig. 9a, indicating the presence of rapid reoxidation during CO₂RR. Importantly, they determined for the first time that the O in EtOH originated from the solvent water rather than CO by reducing of C¹⁶O in H₂¹⁸O electrolytes (Fig. 9b) [20]. They proposed a new intermediate called Grotthuss chain ethynyl (C–CH) concerted hydrolysis to explain their experimental observations. The pathway involved C–CH, then rapidly reduced to EtOH as: CO $\to$ *COCO $\to$ *COCOH $\to$ ${\text{*COHCOH}}$ $\to$ *CCOH $\to$ *CHCOH $\to$ *CCH $\to$ CHCHOH $\to$ ${*}$ CH₂CHOH $\to$ *CH₃CHOH $\to$ CH₃CHOH. However, Bell and coworkers clarified that the explanation of this observation did not require a novel mechanism of “C–CH” [148], which may be likely the result of isotopic competition between transiently produced AcH and solvent H₂O due to the formation of EtOH via the reduction of AcH during CORR or CO₂RR. This analysis was further confirmed by adding 0.2% of AcH to 98% H₂¹⁸O [99]. In addition, Jouny et al. reported the detected acetic acid with a signal of 62 amu (1 amu = 1.660 5 × 10⁻²⁷ kg) when labeled C¹⁸O was used [99], in which one O came from the labeled C¹⁸O and one O originated from OH⁻ ions in the electrolyte (Fig. 9c), implying that the high acetate selectivity was attributable to the higher local pH at the triple-phase boundary. O exchange was not observed between acetic acid and H₂¹⁸O, which indicated that *COCH₂ or *COCH₃ could be the intermediate of acetic acid. Further isotopic labeling studies clarified that *CO initially dimerized and further protonated to *CCOH and acetic acid was formed through direct OH attack of a ketene intermediate [314].

Fig. 9

a ¹⁸O content in OD¹⁶ Cu 1 h, OD¹⁶ Cu with CO₂RR for 3 min in H₂¹⁸O electrolyte, and OD¹⁶ Cu that was soaked in H₂¹⁸O for 3 min. Reproduced with permission from Ref. [197]. Copyright © 2018, Wiley. b Chart showing products with an ¹⁸O pathway and products without an ¹⁸O pathway. Reproduced with permission from Ref. [20]. Copyright © 2018, American Chemical Society. c AcH in mixture of unlabeled H₂O and labeled H₂¹⁸O. Reproduced with permission from Ref. [99]. Copyright © 2018, Springer Nature. d Hypothetical scenario in which the reduction of a mixture of ¹³CO and ¹²CO₂ is carried out on a catalyst with two types of active sites. Reproduced with permission from Ref. [200]. Copyright © 2019, Springer Nature. e Time-dependent absolute product formation rates for C₂H₄. f Comparison of DEMS ion current sweeps over time for C₂H₄-related molecular fragment. (e, f) Reproduced with permission from Ref. [318]. Copyright © 2019, Springer Nature. g Reaction formulae for the cross-coupling between CO and AcH and the self-coupling during CORR. h Reaction formulas for cross-coupling and self-coupling. (g, h) Reproduced with permission from Ref. [81]. Copyright © 2020, American Chemical Society. i KIE of H/D in CO₂RR to C₂H₄ and HER performance. Reproduced with permission from Ref. [62]. Copyright © 2019, Springer Nature

For ¹³C isotopic labeling, the rapid equilibrium exchange between bicarbonate anions and CO₂ molecules was found by using in situ ATR-SEIRAS and isotopic labeling [315], which was further confirmed [46, 316]. In addition, by combining ATR-SEIRAS and isotopic labeling, Shao and coworkers found that the *CO was found to be consistent with the bicarbonate anion rather than the CO₂ in the solution, indicating that the CO₂ reactant during CO₂RR came from the CO₂ molecule in equilibrium with the bicarbonate anion [81]. It was also worth noting that the carbon source of the CO₂RR product was indeed from the CO₂ molecule rather than the electrolyte or catalysts through the experiment of ¹³CO₂ isotope labeling [317]. More importantly, isotopic effects can be used to distinguish active sites based on the difference in the isotopic composition of the products [200, 318]. For example, by using isotopic labeling (Fig. 9d), OD-Cu catalysts have three types of active centers, which were used to selectively produce C₂H₄, EtOH/HAc, and PrOH, respectively [200]. In contrast, this phenomenon was not observed on the polycrystalline, (100), or (111) Cu catalysts. Strasser and coworkers found an unexpected increase of nearly 50% for C₂H₄ production under CO₂/CO co-feeds (Fig. 9e), which corresponded to a cross-coupled ¹²CO₂-¹³CO reactive pathway through kinetic isotope belling experiments and operando DEMS capillary (Fig. 9f) [318]. They concluded that non-interfering, reactant-specific surface locations of CO₂ and CO were present. In addition, Xu and coworkers reported that the formation of PrOH came from self-coupling of CO and cross-coupling between CO and AcH by using isotope labeling [81], in which CO attacked the carbonyl group of AcH (Fig. 9g, h).

Similar to the isotopic effects of ¹³C, the kinetic isotopic effect (KIE) was carried out by replacing the H₂O solvent with D₂O in the electrolyte during CO₂RR. Proton transfer occurring in RDS is usually affected by the KIE because the proton is too light. Therefore, KIE can be used to explore the role of proton in RDS. During CO₂RR, as the value of KIE approaches 1, the impact on CO₂ reduction is negligible, while that is usually greater than 10 for HER [319]. This suggests that the KIE could be considerable when the RDS involved proton transfer, and vice versa. Wang and coworkers reported that the KIE value of H/D over the S₂-In catalyst was 1.9, which provided evidence that H₂O dissociation was involved in the RDS of CO₂RR to HCOOH [320]. Interestingly, by using DFT calculations, they found that the reaction from *CO-to-*CHO was the RDS for the CO₂RR to C₂H₄ [62]. To verify whether the RDS in CO₂RR to C₂₊ products involved proton transfer, they measured the KIE of catalysts. The KIEs over Cu, I–Cu, Br–Cu, Cl–Cu, and F–Cu catalysts, defined as the ratio of C₂H₄ formation rates in H₂O and D₂O, are 2, 1.8, 1.5, 1.3, and 1.2 (Fig. 9i), respectively. This results not only indicated that H₂O dissociation was involved in the RDS, but also implied that the presence of F accelerated the H₂O activation. In general, C₂H₄ has two different formation pathways, including the protonation of *CO and CO–CO coupling, which may be distinguished by direct experimental evidence provided through the KIE effect.

4.5 DFT Calculations

Due to the complex reaction networks, the corresponding reaction barrier, and cation effects, fully understanding the pathway of complex products and well establishing the structure-to-performance relation during CO₂RR are particularly challenging. In the past decades, DFT calculations with computational hydrogen electrode (CHE) model have been widely employed to study the reaction intermediates, verify experimental observations, and predict possible structure of catalysts [321, 322]. DFT calculations applied to the electrode/electrolyte interface are a huge challenge, but great progress has been made to describe reactions more accurately by improving models, including a few, one layer, and multiple layers of explicit water molecules [323,324,325,326]. A recent detailed review of the foundations, advanced computational models, H-shuttling approach, scaling relations, and suitable descriptor can be found elsewhere [96, 101, 327, 328]. In this part, we will focus on computational modeling of CO₂ and CO reduction to C₂₊ products, which generally includes three important processes: the protonation of *CO (e.g., *CHO and *COH), C–C coupling, and post-C–C-coupling steps.

The pathways for the formation of C₂₊ products are complex and controversial among DFT calculations, but it is generally accepted that *CO exists on Cu during CO₂RR or CORR. This intermediate can be further hydrogenated to *COH and *CHO by adding of H⁺/e⁻ through the generation of an O–H or C–H bond, respectively. Based on DFT calculations with the CHE model, the *CHO species on Cu (211) was first proposed by Peterson and coworkers [57], but without considering reaction barriers and solvent. By building structure-sensitive maps of reaction intermediates, Calle-Vallejo and coworkers proposed that the stability of *COH and *CHO depended on the coordination number of Cu [329]. Cu tended to form * CHO more easily as the coordination number decreased. Therefore, it is fine to assume that CO reduction proceeds via *CHO on the surface of Cu. For Cu (100), Luo et al. concluded that the formation of *CHO was more favorable than that of *COH (TS barrier of 0.64 eV vs. 0.92 eV) [323], which was calculated by using the two-water solvation model. By building the model of full solvation, Cheng et al. also reported that *CHO formation was favored [330]. For Cu (111), Nie et al. found that the TS barrier of *CHO was higher than that of *COH with the two-water solvation model [331], which was consistent with the research of Xiao et al. [97]. In addition, by using explicit water with a full H coverage, Hussain et al. also reported that *COH formation was favored on Cu (111) [332]. Compared to Cu (111), *CHO becomes more stabilized relative to CO* on Cu₅₅ [333]. Furthermore, the adsorption energy of intermediates increased with the decrease in the size of Cu NPs [334], but *CHO has always been stabler than *COH (Fig. 10a), which was consistent with others research [335]. In general, *CO mainly forms *CHO intermediate on the surface of Cu (100), Cu (211), and nanoclusters, while *COH intermediate on Cu (111) surface.

Fig. 10

a The binding energies of intermediates at various sizes of Cu NPs. Reproduced with permission from Ref. [334]. Copyright © 2019, Elsevier. b ¹⁸O reaction energy diagram for CO₂RR to C₂H₄ on Cu(111) and F-Cu(111) facet. Reproduced with permission from Ref. [62]. Copyright © 2020, Springer Nature. c Lowest energy pathways for the electroreduction of CO to C₂H₄, MeCHO/EtOH. Reproduced with permission from Ref. [21]. Copyright © 2013, Wiley. Reaction energy diagram from the adsorbed C₂H₃O intermediate to C₂H₄ and EtOH for d pristine Cu and e Cu with Cu vacancy and subsurface S. (d, e) Reproduced with permission from Ref. [22]. Copyright © 2018, Springer Nature. f Reaction paths for C₂H₄ vs. EtOH on Cu(111). g Energy profiles of all intermediates toward C₂H₄ and EtOH on Cu and Ag/Cu catalysts. (f, g) Reproduced with permission from Ref. [23]. Copyright © 2019, American Chemical Society

Apart from the effect of crystal facets, the formation of TS barrier of *COH and *CHO is affected by the proton source such as *H, *H₂O, and H₂O [90, 330, 336, 337]. For instance, Goodpaster et al. reported that the TS barrier of *CHO via reaction *CO with *H through a Tafel process was 0.75 eV, while that with water by a Heyrovsky reaction was 0.6 eV [90]. Furthermore, Cheng et al. calculated three alternative reactions to generate COH*, in which the protons came from *H, H₂O, or *H₂O. The corresponding TS barrier were 1.45, 0.70, and 0.74 eV, respectively. In addition, the effect of different explicit solvation models should be noted. For the Cu (111) surface [324], *COH was more stabilized (0.10 eV) than *CHO upon the addition of 1 H₂O with the presence of a H₂O bilayer, and relative stability of *COH further increased (relative to the no-water model) by 0.29 eV. *CHO was more favored over *COH on the Cu (100) surface by using explicit or implicit solvation models, which was consistent with recent research [338].

The formation of C–C bonds by coupling of two C₁ intermediates is a key step to produce C₂₊ products during CO₂RR. As shown in Scheme 2, C–C bonds can be formed from the coupling of various C₁ intermediates via a L–H mechanism. First of all, the *CO–*CO coupling is in competition with the protonation of *CO (e.g., *CHO and *COH) due to its high formation kinetic barrier. For example, by using the display solvent model, Cheng et al. reported that the energy barrier of C–C bonds formed by coupling of two *CO is 0.69 eV on the Cu (100) surface [336], which is much lower than that of *CHO (0.96 eV). Furthermore, Martin and coworkers used a model that included the effects of electrochemical potentials, solvents, and electrolytes to investigate into the pathways by which C–C bonds [90]. Their results suggested that the formation of C–C bonds proceeded via formation of a CO dimer at low overpotentials on Cu (100) surfaces, whereas at high overpotentials, the C–C bond was formed by coupling of *CO and *CHO. However, the electrochemical reduction of *CO followed a thermodynamics-controlled *COH or *CHO pathway instead of *CO–*CO coupling on Cu (100), Cu (110), and Cu (111) surfaces when using an implicit solvation model or two-water model [62, 97, 323, 338].

Scheme 2

Separation of C₁ and C₂ pathways from the *CO intermediate

By combining experiment with DFT calculations, the direct dimerization of *CO was proposed as the key step toward C₂₊ products at low overpotentials on the Cu electrode. The energy barrier of *CO–*CO coupling was fully explored by using theoretical calculations with different solvent models [19, 44, 45, 78, 264, 303, 338,339,340], in which that on Cu (100) surface was lowest in different Cu crystal planes [e.g., Cu (211), Cu (110) and Cu (111)]. The energy barrier of C–C bonds by coupling of various C₁ intermediates (e.g., *CO, *CHO, and *COH) was investigated in detail, which included OC–CO, OC–COH, OHC–CO, HOC–COH, and OHC–CHO. On the whole, the energy barrier of *CO–*CO coupling was higher than other C–C coupling on the surface of different Cu crystal planes [62, 90, 97, 323, 336, 340, 341]. In particular, Wang and coworkers reported that the formation of *OHCCH*O was thermodynamically favorable (Fig. 10b) [62]. In addition, Luo et al. reported that the energy barrier of *CO–*CO coupling was 1.22 eV on the Cu (100) surface, which was higher than that of CHO* (0.64 eV) and the dimerization of CHO* (0.22 eV) [323]. They proposed that the major route of C₂H₄/EtOH at low potential may be *CHO and subsequent the dimerization of *CHO, rather than the previously reported *CO–*CO coupling mechanism.

Due to the presence of the repulsion interaction among the *CO adsorbates, the increasing of *CO coverage not only reduces the binding energy of *CO, but also affects the energy barrier of *CO dimerization and hydrogenation. It is generally accepted that with the increase in *CO coverage, the binding energy of *CO and the energy barrier of *CO dimerization decrease significantly [62, 218, 339, 340, 342]. For the *CO hydrogenation, Ou et al. reported that the increase in *CO coverage may be easier to form *CHO on Cu (111), but vice versa on Cu (100) [44]. Furthermore, Huang and coworkers found that with the enrichment of *CO concentration on Cu surface, the energy barrier of *OC–CHO decreased from 0.44 to 0.30 eV, while that of *CO hydrogenation increased from 0.36 to 0.54 eV [158]. Their DFT calculations clarified that the formation of CH₄ was thermodynamically favorable at low CO coverage and *CHO becomes a common intermediate for the generation of CH₄ and C₂H₄ at high CO coverage, explaining the high FE of C₂H₄ and the inhibition of CH₄ at higher potentials. However, Xu et al. proposed that there was no quantitative relationship between CO partial pressure and surface CO coverage [343], and the hydrogenation of CO via an E–R mechanism is the RDS in the CORR rather than CO–CO coupling by using CO adsorption isotherms on Cu measured by in situ SERS.

Although understanding the initial C–C coupling steps has made much progress, only a few studies have addressed the basic steps of post-C–C coupling. However, some research groups and some researchers have begun to summarize the post-C–C coupling steps [26, 96]. On Cu-based catalysts, the distribution of C₂₊ products is complex and diverse. Since C₂H₄ and EtOH are the main C₂₊ products, recent DFT calculations have focused on providing an understanding of the key intermediates that control the C₂H₄ vs. EtOH selectivity. Therefore, we address recent progress on the theoretical understanding of their formation in the following sections. Based on the detailed analysis of the current research, we found that the relative stability of the bifurcated intermediate is the key to regulating the selectivity of C₂H₄ and EtOH.

It is generally accepted that the formation of C₂H₄ was thermodynamically favorable compared to EtOH on the regular Cu surface [21, 22, 80, 241, 323]. For instance, Koper and coworkers calculated the pathways for the CO reduction to C₂H₄, AcH, and EtOH [21]. DFT calculations clarified that the favorability of the *CH₂CHO hydrogenation step was inclined toward C₂H₄ formation by approximately 0.2 eV (Fig. 10c). Therefore, regulating the stability of shared intermediates may help to develop strategies tuning the proportion between C₂H₄ and EtOH. For example, Zhang et al. investigated the effects of energetics seen by the adsorbed bifurcated intermediates on the selectivity of C₂H₄ and EtOH by using DFT because they share a common intermediate state (*OCHCH₂) [22]. Their DFT results found that the intermediate of *OCHCH₃ was stabler than that of *O + C₂H₄ on a Cu₂S–Cu core–shell model with an atomic vacancy, whereas the opposite calculation results were presented on a pristine Cu slab (Fig. 10d, e). Therefore, the relative stability of *OCHCH₃ shifts the balance to enhance the selectivity of EtOH by suppressing C₂H₄ production, which was consistent with relevant reports [73, 241, 344]. Different from the scheme of *OCHCH₃, Luo et al. found that the breaking of C–O bond in ethylene oxide (CH₂–CH₂O*) required a higher barrier (1.01 eV) than the generation of ethylene hydroxide (CH₂–CH₂OH*) (0.85 eV) [323], from which C₂H₄ formation has a lower barrier in comparison with EtOH formation. Furthermore, Calle-Vallejo and coworkers provide a computational–experimental study of EOR to C₂H₄ and EtOH [345]. The adsorption energy of *OCH₂CH₂ was coordination number dependent, which became stabler as the coordination number decreases, thus promoting the production of EtOH.

In addition to the share intermediate of *OCHCH₂, the selectivity of between C₂H₄ and EtOH is determined by that of *CHCOH (denoted IM) rather close to C–C coupling [23, 218, 336, 346,347,348]. Specifically, the hydroxyl group of IM can be deoxidized to form *CCH (IM-C), leading to produce C₂H₄, while the hydrogenation of IM to *CHCHOH (IM-O) instead enhanced the formation of EtOH. For example, Cheng et al. reported that the reaction energy barrier of C₂H₄ including IM to IM-C via the E–R mechanism was 0.61 eV on Cu (100) [336], which was lower than that of EtOH including IM to IM-O (1.14 eV). They find that as the CO coverage increased, the barriers to both IM-C and IM-O pathways increased. At low CO coverage, the IM-C pathway was kinetically superior to the IM-O pathway, while the IM-O pathway was more favorable at 3/9 ML. In addition, they also reported that the relative stability of IM-C versus IM-O can be tuned by introducing diverse binding sites [23]. DFT calculations showed that multiple bind sites on the Ag/Cu surface had weaker bonding ability for IM-C relative to IM-O, indicating that the EtOH pathway became more favorable compared that on pure Cu (111) (Fig. 10f, g). It should be noted that the O in EtOH generated by the IM-O pathway must have come from CO.

5 Summary and Outlook

Although CO₂RR powered by renewable electricity is important in achieving carbon cycling and utilizing intermittent renewable energy sources, the technology is not yet commercially available due to the limitations of products selectivity, energy efficiency, and catalytic stability. In the past decade, the rapid development of advanced electrocatalysts has attracted a lot of attention and greatly improved the efficiency and selectivity of CO₂RR. However, C–C coupling of C₂₊ products through CO₂RR is far from ready for practical application, and the dominant factors that really dominate the reaction pathway are still controversial. In this progress report, we provide an overview of recent engineering strategies for enhancing the structure and composition of CO₂RR catalysts for C₂₊ products, and review advanced tools in understanding CO₂RR reactions, with a focus on the active sites and product generation pathways. These research issues and future research efforts to further understand the mechanism of CO₂RR are listed as below.

• 1. Further identification of true active sites. The performance of CO₂RR electrocatalysts and the understanding of the active sites have made great progress and been discussed in depth. Although the role of Cu⁺ species in promoting CO₂RR selectivity toward a specific product has been highlighted in some studies, recent studies have shown that Cu oxide was completely reduced under neutral conditions and Cu–O species were unlikely to be the active sites facilitating the formation of C₂₊ products during CO₂RR. Furthermore, in situ/operando spectroscopic characterizations can be used not only to explore the active phase of the dynamic catalyst but also to distinguish the active sites during CO₂RR. Further understanding active phases and active sites of structural evolution of Cu-based electrocatalysts under operando reaction conditions is of great interest, which may help to establish the structure–activity relationship and accelerate the design of more advanced catalytic materials.

• 2. Further determining the possible pathways of C₂₊ products. Completely understanding and determining the possible pathways of C₂₊ products on Cu-based electrocatalysts during CO₂RR is critical for designing useful principles for the design of highly selective and active catalytic materials for C₂₊ products. Despite a great deal of research has been reported in the past decades, the reaction mechanism of CO₂RR, especially the formation pathway of C₂₊ products, has not been fully revealed due to the extremely complex reaction pathway network. It is one of the most challenging tasks to accurately determine pathway of C₂₊ products. Combining advanced spectroscopic characterizations and isotope labeling can provide direct evidence of the intermediate states of the products, which is expected to speed up the research journey. Future work should focus on correctly identifying the intermediate states and improving the temporal and spatial resolution of the operational/in situ techniques to capture key intermediate states not detected in previous experiments.

• 3. Understanding the mechanism of CO₂RR with the aid of DFT calculations. Currently, DFT calculations are typically used to provide atomic-scale insights based on the calculated energy barrier of different product pathways over a variety of Cu-based catalysts. However, in most studies, computational models are ideal and simplified atomic configurations based on the initial structure of the catalyst, neglecting the dynamic changes of the catalyst structure during the reaction process, which may result in inappropriate reaction centers and is not conducive to providing correct theoretical guidance. In addition, DFT calculations can only provide the possible reaction pathway due to the complex pathways and intermediate states. Therefore, a computational framework should consider both structural changes and intermediate states. In situ/operando spectroscopy and isotope labeling can not only provide the evolution of catalyst structure, but also determine the adsorption of the key intermediate states. DFT Calculations are able to accurately interpret the spectral data of the catalyst, which makes it possible to establish a link between experimental spectral characteristics and theoretical models. Importantly, ML can use an ever-growing data set to understand the adsorption energies of different reaction intermediates to train the model to predict higher selective CO₂RR catalysts. Hence, DFT calculations based on the advanced experimental tools and intelligent ML should be an important research direction of CO₂RR in the future, which can provide a correct understanding of the catalytic mechanism to accelerate the design of high activity and high selectivity C₂₊ product catalysts.