P-block metal-based (Sn, In, Bi, Pb) electrocatalysts for selective reduction of CO₂ to formate

Since the industrial revolution in the nineteenth century, fossil fuels such as coal, petroleum, and natural gas have been exploited as the main energy source to sustain our economy and society. Extensive consumption of fossil fuels has caused an increase in the concentration of CO₂ in the atmosphere from 278 ppm at the beginning of the industrial revolution to 410 ppm today, which has brought about environmental problems associated with the greenhouse effect of CO₂.^1,2 To alleviate these negative effects and close carbon recycle, conversion of CO₂ into value-added chemicals and fuels is proposed as a promising strategy because this process not only mitigates CO₂ emissions into the Earth’s atmosphere but also produces commodity chemicals that can be used either as fuels or as precursors in many industrial chemical processes.^3–5 A variety of routes for chemical conversion of CO₂ have been explored, including thermocatalytic reduction,^6–8 photocatalytic reduction,^9–14 electrochemical reduction,^15–18 and photoelectrochemical reduction^19–22. Among these strategies, electrochemical reduction of CO₂ to form C₁ feedstocks (such as CO) and useful fuels (e.g., formate, methanol, and ethylene) driven by electricity from renewable energy sources, such as solar and wind, has been widely proposed as a very promising sustainable solution. This is largely because the cost of renewable energy sources is projected to decrease as the conversion technologies improve. For example, the cost of solar power has plunged from 3 USD W⁻¹ in 2008 to 0.5 USD W⁻¹ in 2020 and is projected to further reduce below 0.05 USD W⁻¹ by 2030.² A major milestone for the development of electrochemical CO₂ reduction technologies will be the reproducible preparation of efficient and robust electrocatalysts to accelerate the sluggish dynamic processes of the electrochemical CO₂ reduction. High selectivity toward specific products is also a very important factor due to the high cost for product separation.

Recently, p-block metal-based catalysts, including Sn,^23–26 In^27–30, Bi,^31–33 and Pb,^34,35 have been shown to drive the electrochemical CO₂ reduction with high selectivity toward C₁ products, in particular, formate or formic acid with a selectivity of up ∼100%. Metal-free carbon-based catalysts such as N-doped carbon nanotubes and N-doped fullerenes have also been shown to drive the CO₂ reduction preferentially toward formate.^36–38 Although the reported selectivities are generally lower than those of p-block metal-based catalysts, robustness and environmental friendliness are the desirable properties of these nonmetal catalysts. These are materials of interest because formate (formic acid) is a highly valued product for many industrial processes. Among the common organic acids, formic acid is the most corrosive because of its low pK_a, ∼3.75, being about 10 times more ionized or acidic than acetic acid. Besides having acidic properties, formic acid behaves as an aldehyde and, hence, has reducing properties. Furthermore, formic acid can also be used as the chemical fuel for fuel cells. This has triggered great efforts to understand the underlying reaction mechanism and the structure–activity relationship that may provide rational guidelines for further improving the activity, selectivity, and stability of these types of electrocatalysts.

In this article, we will review recent developments and understanding of p-block metal-based (e.g., Sn, In, Pb, and Bi) electrocatalysts for the electrochemical CO₂ reduction reaction (CO₂RR). Although there are already a few comprehensive reviews on the electrocatalysts for the electrochemical CO₂RR,^{16,17,39–43} our main focus will be specifically on the p-block metal-based electrocatalysts with high selectivity for formate production. In the following, we will start with a brief introduction to the fundamentals and challenges of CO₂ reduction, followed by classification of electrocatalysts and possible reaction pathways. We then review in detail Sn, In, Bi, and Pb based electrocatalysts, in terms of the fundamental understanding of the surface chemistry, active sites, reaction mechanism, and structure–activity relationships for the CO₂RR. Strategies to enhance the activity including nanostructures, defect engineering, doping, and alloying will also be briefly discussed. Finally, we summarize the existing challenges and present perspectives for the future development of this exciting field.

A basic electrochemical CO₂ reduction system is composed of an anode, a cathode, an electrolyte containing dissolved CO₂, and a proton exchange membrane, as shown in Fig. 1. The cathode supplies catalytic active sites for the electrochemical CO₂RR, and the anode promotes oxidation reactions, e.g., oxygen evolution reaction (OER). The membrane separates oxidation products from reduction products while allowing the exchange of protons (H⁺) to keep charge balance. The electrolyte serves as a medium to transfer charged species (e⁻/H⁺) and also as a medium to dissolve CO₂. Obviously, the CO₂RR at the electrocatalyst is the most crucial process for the overall electrochemical CO₂ reduction system. It is generally believed that the overall electrochemical CO₂RR process includes CO₂ solvation from the gas phase, adsorption, and activation of CO₂ molecules on the electrocatalyst surface, followed by multiple-step proton–electron coupled transfer reactions to form hydrocarbons and desorption. Because of the complex chemical and physical processes, the efficiency and selectivity of the electrochemical CO₂RR confront many fundamental challenges, which can be grouped into the following three aspects:

• Low CO₂ solubility. The solubility of CO₂ in aqueous electrolytes is only 34 mM under the standard condition.⁴⁴ This low solubility limits the CO₂ mass transport rate and the reduction rate on the electrocatalysts and, therefore, limits the current densities to a few tens of mA cm⁻². In order to reach an industrially viable CO₂ reduction rate, some cell design schemes that increase the solubility and mass transport to the electrodes have been recently proposed, such as flow cells or gas diffusion electrodes.^45–49 Furthermore, ionic liquid and organic electrolytes are also developed to increase the CO₂ solubility and reaction kinetics.^50–53 

• Large overpotential. The thermodynamic cost of reducing CO₂ is comparable to that for the hydrogen evolution reaction (HER). For example, thermodynamic equilibrium potentials [versus reversible hydrogen electrode (RHE)] for the reduction of CO₂ to CO and formic acid are −0.11 V and −0.20 V, respectively.⁵⁴ However, a large activation energy barrier is required to activate CO₂ molecules, which results in large overpotential requirements to drive the overall reaction. Therefore, a key factor to drive the CO₂RR at lower overpotentials lies at the activation of CO₂ molecules, which is fundamentally difficult because of the inherent chemical stability of CO₂ associated with a linear centrosymmetric molecular structure and large bonding energy of C=O double bond (∼750 kJ mol⁻¹).⁵⁵ Furthermore, the CO₂ reduction process may involve complicated proton-coupled multi-step electron transfer reactions that result in the sluggish reaction kinetics. As a result, a high reduction potential of −1.49 V vs RHE is required to form the key intermediate CO₂⁻ radicals through the initial one-electron transfer process, making the reaction energetically prohibitive to take place.^56,57

• Poor selectivity. The large overpotentials required to activate CO₂ lead to the formation of highly reactive species, i.e., bent CO₂⁻ radicals. Slight potential differences result in divergent reaction pathways via 2, 4, 6, 8, 12, or even more electron transfer and produce a large spectrum of products, including CO, formate, methane, methanol, and ethylene (see Table I).^58,59 Furthermore, due to the large cathodic potential required to drive the CO₂ reduction, the HER becomes a competitive side reaction in aqueous solutions, which detrimentally further impacts the selectivity for CO₂ reduction.⁶⁰ The separation of complicated products would impose additional costs for electrochemical CO₂RR technologies. Hence, developing electrodes with high selectivity to a specific product is highly desirable in this field (Table II).

Since the pioneering works by Hori et al. in the 1980s and 1990s,^61–63 great efforts have been devoted to the development and understanding of electrocatalysts for electrochemical CO₂ reduction, especially within the past decade.^{17,43,64–67} According to the major products of CO₂ reduction (i.e., the selectivity), electrocatalysts can be classified into four groups, as graphically shown in Fig. 2, including group I, transition metals, such as Fe, Ni, Pd, and Pt, which catalyze mostly the HER; group II, metals, such as Ag, Au, and Zn whose major product is CO; group III, Cu and its derived oxides, which promote the formation of hydrocarbons or oxygenates; and group IV, p-block metals (Sn, In, Bi, and Pb) and their oxides, primarily producing formate or formic acid.⁶¹ 

Over the past few decades, the reaction pathways on different electrocatalysts have been studied both experimentally and theoretically.^17,64,68,69 It is generally agreed that the rate-determining step (RDS) of CO₂RR is the first electron transfer to surface-adsorbed ^*CO₂ (where ^* denotes the adsorption site) to form ^*CO₂^•− radicals [Fig. 3(a)]. The next protonation step is determined by which atom binds to the electrode surface. If the carbon atom of ^*CO₂^•− binds strongly to the surface, the oxygen atom would be protonated to form ^*COOH, while if the oxygen atoms bind to the surface, the carbon atom would be protonated to form ^*OCHO. The selectivity and activity toward the CO₂RR of the above-mentioned groups are essentially determined by the relative binding strength of the reaction intermediates (^*OCHO, ^*COOH, ^*CO, and ^*H) on the surfaces.^70,71 For instance, a very strong interaction between ^*CO and late transition metals such as Pt and Ni in group I prevents the adsorbed ^*CO to desorb as CO or its further reduction, leading to H₂ as the main product. On the other hand, more balanced interaction between CO₂ reduction intermediates and catalysts in groups II to IV leads to the formation of various CO₂ reduction products. In the case of group II, including Au, Ag, and Zn, a relatively weak binding strength of ^*CO with the catalyst surface benefits the desorbed ^*CO and formed CO. Cu and Cu oxides bind ^*CO with the moderate strength so that further reduction and reactions (e.g., C—C coupling) may occur to form a wider product distribution, such as methane, methanol, ethylene, and other multi-carbon products, via ^*CHO/^*COH intermediates. Finally, post-transition metals in group IV possess a particularly high selectivity toward CO₂ reduction against the HER because ^*H binds weakly to these metal surfaces, therefore inhibiting the HER. Additionally, the characteristic oxophilicity of p-block metals favors the formation of oxygen-bond intermediates such as ^*OCHO (over carbon-bond ^*COOH), which have been shown to be key intermediates for the production of formate or formic acid as the major product [Fig. 3(a)].

To summarize, the binding strength of key intermediates on the metal electrode surfaces not only explains the selectivity trends observed in different groups but also guides the possible reaction mechanisms proposed for product formation. It provides guidance for rationally designing electrocatalysts for CO₂ reduction. In the following, we will briefly discuss the different electrochemical CO₂RR behaviors on group II, III, and IV electrocatalysts.

Group II metals, Au,^72–76 Ag,^77–81 and Zn,^82–85 show the highest selectivity for CO production. CO is an important feedstock for the water–gas shift reaction and Fischer–Tropsch synthesis.^86–89 To form CO, the reaction pathway begins with chemical adsorption of CO₂ on the catalyst surface, followed by the formation of the COOH^* intermediate [shown in Fig. 3(c)]. The adsorbed COOH^* intermediate is further reduced by the reaction with another proton and electron, yielding CO and H₂O. Based on this reaction pathway, suitable catalysts for CO production should have a strong binding to COOH^* and a weak binding to ^*CO. However, according to the density functional theory (DFT) calculations by Nørskov, the adsorption energies of ^*CO and ^*COOH on these metal surfaces usually vary parallelly; it is hard to induce an increase in the binding strength of ^*COOH to the surface without increasing the ^*CO adsorption energy at the same time.⁶⁴ 

Many attempts have been done to optimize the activity/selectivity of Au, Ag, and Zn for CO. Surface morphology,^90,91 catalyst size,^92,93 and structure^94,95 play important roles in the activity and faradaic efficiency (FE) toward the reduction of CO₂ to CO because these factors determine the type and number of surface sites exposed as active sites. For example, nanostructured Ag can improve FE of CO up to 92%.⁹⁶ The intrinsic activity of nanostructured Ag has also been shown to increase by 20 times as compared with polycrystalline bulk metals under an overpotential of 0.5 V, which is attributed to the highly curved nanostructure surface because it can stabilize CO₂⁻ intermediates and decrease the activation barrier. Furthermore, electronic effects, modulated by alloying,^97,98 doping,⁹⁹ and forming defects,^54,85 and reaction conditions, such as temperature,^100,101 pressure,¹⁰² and electrolytes,^77,103,104 also have a significant influence on electrochemical CO₂RR performances.

Cu and its derivative compounds are unique catalysts that can directly convert CO₂ into a wide variety of hydrocarbon and oxygenate products (CH₄, CH₃OH, C₂, and C₂₊).^105,106 The selective conversion of CO₂ into C₂H₄ or C₂₊ is highly valuable because such products can be directly used in the existing chemical industry. However, large kinetic barriers for producing C₂H₄ or C₂₊ on Cu surfaces lead to unsatisfactory current density and uncompetitive FE at a low overpotential. Great efforts have been devoted to understanding the reaction mechanism and to modulating the structure and electronic properties of the catalysts to increase the activity and selectivity,^107,108 for instance, by nanostructuring,^109,110 alloying,^111–115 controlling crystal facets,^116–118 and grain boundaries (GBs).¹¹⁹ 

• CH₄ and CH₃OH. ^*CO is likely the common intermediate for the production of CH₄, CH₃OH, and C₂₊. For CH₃OH, the ^*CO intermediates need to undergo a series of hydrogenation reactions to form ^*OCH₃ intermediates. Since protons could combine with a carbon atom instead of oxygen atoms, the next proton–electron coupled transfer process results in a competition between methanol and methane formation. Back et al. introduced the OH binding energy as a selectivity determining descriptor to evaluate the possibility of producing these two products.¹²⁰ The catalyst with a weak OH binding energy is selective to the methanol formation. On the other hand, a strong OH binding energy leads to C—O dissociation coupled with C—H bond formation and therefore a much higher selectivity toward CH₄.

• C₂ and C₂₊ products. The sufficient coverage of various intermediates such as ^*CO, ^*CH₂, and ^*CH₃ and the C—C coupling reaction is the necessary condition to obtain C₂ and C₂₊ products.¹²¹ Provided that the kinetic barrier of C—C bond formation is much larger than that of C—H and C—O, the formation of C₂ products such as acetaldehyde, ethylene, and ethanol requires larger overpotentials and usually these products are obtained with a lower FE in comparison to that of simple C₁ products. It is difficult to identify the formation mechanism of multi-carbon products, because of the multi-step proton–electron coupled transfer occurring during the process. As a result, there is an open debate on whether C₂H₆ is formed by the dimerization of ^*CH₃ species or results from further hydrogenation of C₂H₄/^*OCH₂CH₃ related intermediates.¹²² 

P-block metals such as Sn, In, Bi, Pb, and their oxides catalyze CO₂ reduction to formate or formic acid as major products with FE greater than 90% in CO₂-saturated aqueous solution.⁶¹ Compared with others, formate is a highly desirable and profitable product shown in Fig. 3(b).¹²³ It is one of the attractive candidates as a liquid fuel for the hydrogen economy because of its high energy density, and it is considered as an important raw material for the production of various organic agents.^124,125 Compared to CH₃OH and CH₄ with six- and eight-electron reactions, respectively, the two-electron transfer route for formate formation has more opportunity to achieve the high FE, and it seems more commercially viable at the present stage.^126–128 

The reaction pathways for these types of electrocatalysts are shown in Fig. 3(c). There are two pathways to form formate or formic acid (HCOOH/HCOO⁻): (i) via direct carbon-bond ^*COOH intermediates from stabilized ^*CO₂^•− radicals, as described by Hori et al.,⁶¹ and (ii) through a proton–electron coupled transfer process via ^*OHCO (^*OCHO^* or HCO^*O^*) intermediates.¹²⁹ DFT calculation by Yoo et al. suggested that there is a strong linear correlation between the free energies of COOH^* and H^*.¹²⁹ If the formation of HCOOH proceeds via the COOH^* intermediate, it is inevitable to face with the competitive HER. Therefore, there is a growing agreement that the reaction pathway via ^*OHCO intermediates is more plausible for these highly selective electrocatalysts.

• Formate formation on nonmetallic electrocatalysts. Although the catalytic activity of pure carbon materials toward the electrochemical reduction of CO₂ can be neglected, heteroatom-doped carbon electrocatalysts exhibit high catalytic activity because the charge distribution of the carbon materials is modified, which modulates the intermediate formation energies.^15,36 The highest formate selectivity by carbon-based materials has been achieved upon N-doping, forming pyridinic N active sites.^37,38,130 The reaction mechanisms proposed for nonmetallic electrocatalysts are more elusive than those proposed for metallic electrocatalysts, possibly due to a fewer amount of studies carried out with these types of electrocatalysts. On one hand, some experimental studies and DFT calculations propose the direct carbon-bond ^*COOH intermediate mechanism as plausible for the CO₂ reduction on N-doped carbon electrodes.^37,130 Although N doping does not favor the adsorption of ^*COOH, it seems to favor the desorption of HCOOH, which is a key factor for high formate selectivity.¹³⁰ On the other hand, Wang et al. proposed that pyridinic N provides basicity to adjacent C, which enhances the CO₂ adsorption. Subsequently, the adsorbed CO₂ molecule is reduced to a stabilized COO^* radical instead of ^*COOH intermediates. A second rapid electron transfer and protonation lead to the formation of formate as the main product.¹³¹ 

Sn-based materials have been identified as attractive electrocatalysts for selectively reducing CO₂ to formate due to their good activity, high selectivity, non-toxicity, and abundance in the Earth’s crust for large scale applications.^127,132,133 Numerous Sn-based electrocatalysts have been investigated in the past few years.^134–137 In 1994, Hori et al. first reported that metallic Sn was active for electrochemical CO₂ reduction with high formate selectivity (FE ≈ 88%) in 0.1M KHCO₃ aqueous solution.⁶¹ Following Hori’s seminal work, large research efforts have devoted to studying the active site and reaction mechanism of the CO₂RR on Sn-based electrocatalysts. In the following, we will provide a brief review on relevant studies of the active site and strategies for improving the catalytic performance, including (i) increasing the number of active sites by morphology control and nanostructuring and (ii) increasing the intrinsic activity by electronic modulation via doping, alloying, and defects.

It is known that a pure Sn metal surface tends to be oxidized to SnO or SnO₂ when exposed to air. However, according to the Pourbaix diagram (Fig. 4), under CO₂RR conditions, the oxidized states of Sn are unlikely to exist on the Sn surface because the large cathodic potential needed for formate production is much more negative than the standard reduction potential for Sn oxides.¹³⁸ Therefore, early works generally assumed that the bare metallic Sn surface was the active site for CO₂RR (see Refs. 61, 139, and 140). The large overpotential for CO₂ was attributed to the large kinetic barrier for electron transfer from Sn to CO₂ to form the initial CO₂⁻ intermediate.¹⁴¹ This mechanism is also commonly invoked for many other metal electrodes.⁵⁹ 

However, it should be noted that the Pourbaix diagram is based on thermodynamics considerations and ignore kinetics.¹⁴² Even though the potential for the CO₂RR past the reduction potentials for metal oxides, metastable metal oxides were known to persist on electrode surfaces during cathodic reactions.^143–146 In addition, the reported CO₂RR activities of Sn electrodes are strongly dependent on the pretreatment conditions for the Sn electrodes used in different studies.^133,139,147 Based on these considerations, Chen and Kanan carefully compared the CO₂RR activities of Sn electrodes subjected to different pretreatments and demonstrated that SnO_x is essential for the CO₂RR.¹⁴¹ In particular, they showed that a Sn electrode with a native SnO_x layer exhibited a potential-dependent CO₂ reduction activity consistent with previously reported activity by Hori et al. On the other hand, etching off the SnO_x layer from the electrode leads to negligible activity for the CO₂RR and higher HER selectivity. However, an electrode deliberately prepared with a mixed phase of Sn and SnO_x resulted in much higher CO₂RR activity than that of the pristine Sn foil, showing an eightfold increase in the current density and fourfold increase in formate FE (Fig. 5). Although it was not clear whether the CO₂RR occurs on the SnO_x surface or at the interfaces between Sn⁰ and SnO_x, the authors proposed that the presence of SnO_x may provide a facile pathway for electron transfer to the stabilized CO₂⁻ intermediate.^141,148,149 Nevertheless, insights from this seminal work provide an important guideline to prepare metal/metal oxide composite electrocatalysts with superior activity. Bocarsly and co-workers used in situ attenuated total reflection infrared spectroscopy (ATR-IR) to study the mechanism of CO₂ reduction on Sn electrodes. Their experiments further proved that a surface SnO_x layer is stable under the reducing potential conditions for the CO₂RR and is crucial for formate formation.¹⁴⁸ Furthermore, their study also suggests that surface-bound Sn carbonates are crucial reaction intermediates for transforming CO₂ into formate. As shown in Fig. 6, it was proposed that the reduction of CO₂ is preceded by a two-electron reduction of the electrode from a native SnO₂ to a Sn^II oxyhydroxide. The Sn^II species then react with CO₂ to form the surface-bound Sn carbonates, as evidenced by their in situ ATR-IR. Tin carbonate undergoes the necessary transfers of two electrons and a proton to form formate, which quickly desorbed, and the surface returns to the Sn^II oxyhydroxide surface.

In operando spectroscopic methods such as x-ray absorption spectroscopy (XAS) and Raman spectroscopy have also been employed to monitor the structural changes of electrocatalysts during the reaction process.¹⁵⁰ As shown in Fig. 7, concentrations of the three oxidation states (Sn⁴⁺ in SnO₂, Sn²⁺ in SnO, and Sn⁰) were found to dynamically change in a different range of potentials. A large amount of SnO starts to form at −0.8 V vs Ag/AgCl and reaches the maximum at −1.1 V vs Ag/AgCl. In the moderate potential window, SnO_x (1 < x < 2) with mixed oxidation states shows a maximal FE for formate production. When the potential is more negative than −1.5 V vs Ag/AgCl, the SnO₂ nanoparticles are basically reduced to Sn⁰, and at the same time, the FE for formate is significantly decreased. In operando Raman spectroscopy was also used to monitor the oxidation state changes of SnO₂ during the CO₂RR process. The efficiency of formate production was significantly decreased after SnO₂ was reduced to metallic Sn. Experiments and DFT calculations by Deng et al. also suggested that the SnO_x layer plays a pivotal role in boosting CO₂ adsorption via the formation of optimal coverage of hydroxyl groups on the electrode surface.¹⁵¹ These results provided strong evidence of the importance of Sn oxides for the CO₂RR to formate.

Although the presence of a SnO_x surface layer is crucial for the electrochemical CO₂RR, its thickness should be optimized. Too thick SnO_x layer may cause a negative effect. Zhou and co-workers studied the dependence of formate selectivity on the thickness of SnO_x on Sn nanoparticles with a size of ∼100 nm.¹⁴⁹ They found that nanoparticles with ∼3.5 nm thick SnO_x exhibited the optimal formate selectivity of 64% at −1.2 V vs Ag/AgCl, and a further increase in SnO_x thickness decreased the formate selectivity but led to an increase in CO and H₂ formation.

The mechanistic studies of the CO₂RR on Sn metal or its metal oxides have been mainly based on DFT calculations.^136,152,153 The strong scaling relationship between the free energies of COOH^* and H^* indicates that the formation of formate with high selectivity is unlikely to involve the COOH^* intermediate.¹²⁹ Instead, the free energy of ^*OCHO formed by the proton–electron transfer process decouples with that of H^*, and therefore, it is likely the intermediate for producing HCOOH or formate.¹²⁹ Jaramillo’s group combined experiments and DFT calculations for CO₂RR on a series of metal catalysts including Sn, Au, Ag, Cu, Zn, Pt, and Ni to understand the reaction mechanism and key intermediates for formate production.¹⁴⁷ Interestingly, a clear volcano curve (Fig. 8) is observed with Sn sitting at the top of this volcano when the binding energy of ^*OCHO is used as a descriptor for CO₂ reduction to formate, i.e., Sn exhibits the optimal ^*OCHO binding energy for the highest activity toward formate production.^147,152 However, it should be noted that the catalyst surfaces (i.e., metallic Sn) used for DFT calculations are different from the real surface (Sn/SnO_x) observed experimentally.¹⁴⁵ Furthermore, the highest activity of metallic Sn predicted by DFT is contradictory to the experimental results by Chen and Kanan that showed pure metallic Sn having a low CO₂RR activity. Experimental results suggest that the surface-bound Sn carbonate is the crucial reaction intermediate for transforming CO₂ into formate. Thus, DFT calculations based on the “real” surface (Sn/SnO_x) is necessary to address this discrepancy.

Interestingly, besides O, sulfur on the surface may also enhance the CO₂RR. Similar to SnO_x, SnS_x could also be electrochemically reduced to metallic Sn with residual sulfide on the surface facilitating the CO₂RR. Zheng and co-workers used atomic layer deposition of SnS_x on Au substrates, followed by an electrochemical reduction process to synthesize sulfur-modulated tin [Sn(S)] catalysts.¹⁵⁴ XAS studies reveal a higher oxidation state of Sn in Sn(S) compared with Sn in Sn nanoparticles. The Sn(S)/Au accelerates the CO₂RR at geometric current densities of 55 mA cm⁻² with a FE of 93% for formate formation at −0.75 V vs RHE in 0.1M KHCO₃. Furthermore, Sn(S) catalysts show excellent stability for more than 40 h of operation. The authors suggested that the presence of sulfur atoms at the surface promoted uncoordinated sites and favored the selective reduction of CO₂ to formate, as supported by DFT calculations. Li et al. prepared SnS₂ nanosheets supported on reduced graphene oxide (SnS₂/rGO).¹⁵⁵ Under working potentials, the SnS₂/rGO precatalysts partially transformed to metallic Sn forming Sn/SnS₂/rGO, which showed a maximum formate FE of 85% and a current density of ∼14 mA cm⁻² at −1.4 V vs Ag/AgCl in 0.5M NaHCO₃.

There are essentially two ways to improve the activity of Sn-based electrocatalysts: (i) morphology and nanostructure engineering to increase the number of active sites and (ii) modulating the electronic structures by doping and alloying defects or strains. Increasing the number of active sites can be achieved by controlling the particle sizes,^156–158 types of supports (e.g., carbon black, graphene, and MOF),^159–162 and nanostructure engineering (e.g., nanoparticles,^158,163,164 nanowires,^165–167 nanosheets,^155,168,169 and core–shell structure^167,170). For example, Zhang et al. synthesized SnO₂ nanocrystals with sizes ranging from 3 nm to 200 nm using a hydrothermal method.¹⁵⁷ To maximize the specific surface area, the nanosized catalysts were loaded on high surface area carbon supports such as carbon black and graphene. It was shown that the SnO₂ nanocrystals were quickly reduced to the metallic phase at an onset potential of ∼1 V [vs saturated calomel electrode (SCE)] in aqueous NaHCO₃ solutions. The FE for formate production exhibits a notable size dependence with 5 nm SnO₂ nanocrystals supported on graphene showing a FE over 93% and current densities >10 mA cm⁻²¹⁵⁷ The authors proposed the notable reactivity toward CO₂ reduction results from a compromise between the strength of the interaction between CO₂⁻ and the nanoscale Sn surface and subsequent kinetic activation toward protonation and further reduction. Won et al. synthesized hierarchical SnO_x/Sn dendrites with a multi-branched conifer-like structure and demonstrated that the formate selectivity is strongly proportional to the oxygen content in the SnO_x surface layer.¹³³ The unique surface morphology with a high content of oxidized Sn contributes to a large surface area and the stabilization of intermediates more effectively than the clean Sn surface. The rate of formate formation was 228.6 μmol h⁻¹ cm⁻² at −1.36 V vs RHE. Gu et al. synthesized ultra-small size of SnO nanoparticles supported on carbon black.¹⁵⁸ The ultra-small SnO nanoparticles exhibit remarkably high activity and selectivity for CO formation, which was attributed to the large surface area of the ultra-small SnO nanoparticle (numbers of active sites for activity) and the local pH effect arising from the dense packing of nanoparticles in the carbon black matrix (selectivity).

It should be noted that the change of morphology or nanostructuring not only affects the surface area but also significantly changes the numbers of surface atoms, crystal edge sites, corners sites, GBs, etc., which usually have different coordination environments compared to its bulk. This would result in different binding strengths of intermediate ^*COOH and ^*CO during the CO₂RR process. Lei and co-workers successfully synthesized ultrathin Sn 2-dimensional (2D) sheets with an averaged thickness of 1.4 nm confined in few-layered graphene.¹⁶² The higher electrochemical active area of the ultrathin Sn sheets affords larger amounts of active sites to efficiently adsorb CO₂, as confirmed by the 9 times higher CO₂ adsorption capacity compared with that of the bulk counterpart. More interestingly, the x-ray absorption fine structure spectroscopy (XAFS) study indicates that the Sn—Sn coordination numbers in the Sn ultrathin sheets are reduced to 1.4 and 2.7, in comparison with 2 and 4 in its bulk counterpart. The ultrathin Sn sheets display a current density of 21.1 mA cm⁻² at −1.8 V vs SCE and a maximum FE of 89%, which is roughly 13 times larger than those of bulk Sn. Recently, single-atom catalysts (SACs) have been largely exploited due to their coordinatively unsaturated sites with unique electronic and geometric structures for the unexpected electrocatalytic performance.¹⁷¹ Ni, Co, Sn, and Cu based SACs have been synthesized and exhibited high current densities and selectivity to target products.^172–177 Xie and co-workers synthesized kilogram-scale single-atom Sn dispersed on nitrogen-doped graphene using a quick freeze–vacuum drying–calcination method [Fig. 9(a)].¹⁷⁷ Remarkably, the Sn^δ+ SAC exhibited a low onset overpotential (60 mV), a large turnover frequency (11 930 h⁻¹), and a long-stability (200 h) for reducing CO₂ to formate [Fig. 9(b)]. XAS and transmission electron microscopy (TEM) demonstrated that the atomically dispersed Sn atoms are positively charged (Sn^δ+), because of the charge transfer from Sn to nitrogen-doped graphene, forming Sn—C and Sn—N bonds. The positively charged (Sn^δ+) enables CO₂ activation and protonation to proceed spontaneously by stabilizing CO₂^−* and HCOO^−*, as further verified by in situ Fourier transform infrared spectra and DFT calculations. Furthermore, N-doping facilitates the rate-limiting formate desorption step, as verified by the decreased desorption energy from 2.16 eV to 1.01 eV and the elongated Sn—HCOO⁻ bond length.

In addition, grain boundary (GB) engineering is also reported as an effective way to increase the CO₂ or CO electrochemical reduction.^{94,119,165,166} GBs break the local symmetry and introduce disorder structure and unsaturated sites, which largely change the bonding strength of intermediates.¹⁷⁸ Fan and co-workers fabricated a 1-dimensional (1D) SnO₂ nanofiber with a wire-in-tube (WIT) structure, which is composed of nanoparticles interconnected through GBs.¹⁶⁶ The WIT SnO₂ nanofibers showed superior selectivity and stability for HCOOH production, with a FE of greater than 90%. The excellent catalytic activity was proposed to result from following aspects: (i) the large surface area of the WIT SnO₂ nanofibers, for example, is 10 times larger than that of SnO₂ nanoparticles, which may introduce more active sites for CO₂⁻ absorption, and (ii) the high-density of GBs could reform the bonding strengths between adsorbate and the catalyst surface to stabilize the catalytically active intermediates. Kumar et al. reported that SnO₂ porous nanowires with a high density of GBs exhibit a formate FE of 80% at −0.8 V vs RHE, which is 1.8 times higher than that of SnO₂ nanowires with a low density of GBs.¹⁶⁵ Their results suggested that GBs not only offer a larger number of active sites but also optimize the adsorption energy for intermediates.

Although increasing the number of active sites can improve the performance of electrocatalysts, the loading capacity and the stability of electrodes are still limitations of electrocatalytic behavior.⁷¹ Hence, it is crucial to enhance the intrinsic activity of electrocatalysts. The intrinsic activity is determined by the binding strength of reaction intermediates with the electrocatalyst surfaces.^179,180 Therefore, tuning the surface electronic structure is the key to the development of highly efficient electrocatalysts. Strategies include doping,^181,182 alloying,^{113,161,183,184} defect engineering,^185,186 and strain engineering.^187,188

Heteroatom doping is the most common method. A series of N, Cu, S, and Bi doped Sn oxides for CO₂RR have been reported in recent years.^{154,168,189–191} Doping may change the electronic properties of Sn-based materials, such as the bandgap, work function, and density of states around the Fermi level.^192,193 Tailoring these properties is an effective approach to enhance the conversion of CO₂ to HCOOH.^191,194,195 For example, An et al. reported that after introducing Bi atoms into SnO, the hybridization of Bi 6s and Bi 6p with the O 2p orbital leads to the formation of an additional electronic state, which helps to stabilize the ^*OCHO intermediate on the SnO(001) surface (see Fig. 10).¹⁶⁸ The Bi-doped SnO presented an excellent FE of ∼93% at −1.1 V vs RHE and superior stability over 30 h. The authors also claimed that Bi doping can impede the reduction of Sn²⁺ during CO₂ electrochemical reduction progress.

Alloying is an attractive approach to decouple the adsorption energy of the reaction intermediates. Therefore, alloying has been shown to have a great impact on the product selectivity of CO₂RR. Many metals such as Cu,^196–198 Ag,^170,199 Au,²⁰⁰ In,^201,202 Bi,²⁰³ and Pb²⁰⁴ have been alloyed with Sn to form bimetallic or ternary catalysts. Among these bimetallic catalysts, Cu—Sn has drawn much attention. Zheng et al. performed thermodynamic analysis of the reaction energetics using DFT calculations and found that Cu—Sn alloys could suppress the production of H₂ and CO to achieve high formate selectivity (Fig. 11).¹⁹⁸ They subsequently prepared Cu—Sn alloys with different Cu to Sn ratios in which they found that the CuSn₃ showed a formate FE of 95% at −0.5 V vs RHE and excellent stability after 50 h. A Bader analysis, in situ Sn L₃-edge, and Cu K-edge XAS suggest that electrons are donated from Sn to Cu, leading to a higher oxidation state of Sn^δ+ in CuSn₃. This is essentially consistent with the study of single-atom Sn^δ+ dispersion on the N-doped graphene mentioned above.¹⁷⁷ Bai et al. prepared Sn, Pd, and Pd_xSn alloys with different x values ranging from 0.25 to 4.²⁴ They found that the electrocatalytic activity and selectivity for HCOOH and CO were highly dependent on the surface electronic structure of the alloys. A remarkable FE of nearly 100% for HCOOH was achieved over the PdSn catalyst at the lowest overpotential of 0.26 V, while both CO formation and H₂ formation were completely suppressed. The changes in the FE for HCOOH and overpotential were in parallel with the ratio of Pd⁰/Pd²⁺ in Pd_xSn, indicating that the activity for producing HCOOH is sensitive to the surface oxide species. DFT calculations suggested that the formation of the key reaction intermediate of ^*OCHO, and the product formic acid, was the most favorable over the PdSn alloy surface.

Luc et al. synthesized the Ag-doped Sn/SnO_x electrocatalyst, consisting of an Ag—Sn bimetallic core and an ultrathin shell (∼1.7 nm) of partially oxidized SnO_x.¹⁷⁰ There is a volcano-type correlation between Sn concentration of AgSn/SnO_x catalysts and FE of CO₂ reduction to formate at −0.8 V vs RHE with a maximum FE found at 25 at. % of the Sn content [Fig. 12(a)]. The increase in Sn content in the AgSn/SnO_x alloy leads to lattice expansion [Fig. 12(b)], which stabilizes the ^*OCHO intermediates and facilitates the formation of formate. However, a thick SnO_x layer would form when the Sn content is over 25%, causing the decreased electrical conductivity and lower overall performance. Li and co-workers reported the tuning of Sn-based catalysts for the electrochemical CO₂RR selectively toward CO or formate by controlling the thickness of the SnO₂ layer coated on Cu nanoparticles: a thicker (1.8 nm) SnO₂ layer shows Sn-like activity to generate formate, whereas the thinner (0.8 nm) shell is selective to the formation of CO with the conversion FE reaching 93% at −0.7 V vs RHE.²³ DFT calculations suggest that the 0.8 nm SnO₂ shell-like alloys with a trace of Cu, causing the SnO₂ lattice to be uniaxially compressed and favors the production of CO over formate. The active SnO₂/Cu nanoparticle catalysts have potentials to replace the noble metal catalysts (Au and Pd) for the electrochemical CO₂RR to CO.

The scenario of In-based electrocatalysts for the CO₂RR is quite similar to that of Sn.^61,205–207 For example, Hori et al. reported that the In electrode could achieve a current density of 5 mA cm⁻² with a FE of 94.9% for formate at −1.55 V vs SHE in 0.1M KHCO₃.⁶¹ The FE of In is even higher than that of Sn (88.4%) under the same conditions. Later, comparative experiments by Bocarsly’s group indicated that the FE of the In electrode toward formate largely depends on the surface conditions; an etched electrode with removal of In oxides shows low FE, while an anodized electrode with In oxides shows significantly improved a formate selectivity of >80% at −1.55 V vs SCE in CO₂-saturated 0.5M K₂SO₄.³⁰ Based on in operando ATR-IR measurements, they proposed that the formation of a redox-active In₂O₃ surface layer is essential for the high selectivity of the In electrode for formate. The In₂O₃ could react with water to form In(OH)₃ during the CO₂RR process, and In(OH)₃ can subsequently react with dissolved CO₂ to form a surface confined In—CO₃ species, which is further reduced to generate formate at a low overpotential. This reaction mechanism is consistent with the well-established CO₂ uptake by alkali hydroxides to form carbonate and water.²⁰⁸ The importance of In₂O₃ in the CO₂RR was further proved by the achievement of ∼100% FE for formate at −1.3 V vs Ag/AgCl on deliberately synthesized In(OH)₃ and In₂O₃ nanoparticles.²⁰⁹ 

Besides O, sulfur on the In surface may also enhance the CO₂RR performance, as recently reported by Ma et al.²¹⁰ In this study, S-doped In₂O₃ nanoparticles supported on carbon fibers enabled the selective CO₂RR to formate with a high FE of >85% in a broad range of current densities (25–100 mA cm⁻²). Based on DFT calculations for the CO₂RR to HCOOH and CO on pure indium and sulfur-doped indium surfaces, the authors proposed that sulfur species could accelerate the activation of H₂O and stabilize the adsorption of the ^*OCHO intermediate and, therefore, give rise to the excellent selectivity and activity.

Morphology control or nanostructuring is the most straightforward method to further improve the activity of In-based electrocatalysts.^{27,205,211,212} For example, Luo et al. prepared 3-dimensional (3D) hierarchical porous In electrodes by template-free electrodeposition on Cu meshes.²⁰⁵ The porous structure provided a large surface area and exhibited a high FE for formate (∼90%) in the potential range of −1.0 to −1.2 V vs RHE and reached an unprecedented formate production rate of 1.14 mmol cm⁻² h⁻¹ at −1.2 V. More interestingly, a further electrochemical study indicated that the highly nanoporous structure of the In catalysts induced a high local pH in the vicinity of the electrode and therefore suppressed the HER and enhanced the selectivity for the CO₂RR to formate. Xia et al. electrodeposited dendritic In foams from Cl⁻-containing aqueous solution, and the dendritic In electrocatalyst exhibited an enhanced FE of 86% for formate and a current density of 5.8 mA cm⁻² at −0.86 V vs RHE.²⁷ Dong et al. prepared an oxide-derived nanoporous In—Sn alloy with controllable morphology and size by electroreduction of indium tin oxide nanobranches.²⁰² When the nanopore size decreased from 1176 nm to 65 nm, the FE to formate increased from 42.4% to 78.6% at −1.1 V vs RHE.

Doping and alloying with other elements have also been pursued to tailor the electronic properties and hence the selectivity and activity of In-based electrocatalysts. However, different from Sn-based electrocatalysts, for most cases, In alloying with other transition metals enhanced the selectivity of CO₂RR toward CO.^213–217 In particular, Rasul and co-workers prepared the Cu—In alloy through electrochemical reduction of the thermally oxidized Cu metal foil in InSO₄ solution.²¹⁸ HR-TEM and EDS mapping images [Figs. 13(a) and 13(b)] suggested that In atoms preferentially locate at the surface of the Cu foil. The Cu—In electrode catalyzed the CO₂RR to CO with a FE of 90% at −0.5 V vs RHE, while the pure Cu electrode generated mixed products of CO (FE of ∼45%), H₂ (FE of ∼45%), and HCOOH (FE of ∼10%). DFT calculations suggested that In energetically prefers to locate at the edge site of the Cu surface and modifies the local geometric and electronic structure of the Cu surface for the CO₂RR [see Figs. 13(c)–13(e)]. The presence of In weakens the adsorption of H and improves the stability of ^*COOH, while the ^*CO adsorption energy on top of the Cu atoms remains unchanged, thus explaining the suppression of the HER and COOH formation. Hoffman et al. prepared dendritic Cu—In alloys of various compositions and found that the products of CO₂RR depend on the alloy compositions: with an In fraction of 80 at. %, 62% FE for formate is obtained at −1.0 V vs RHE, while for 40 at. % In, formate with 49% FE and an optimal ratio of 2.6:1 (H₂:CO) syngas is produced.²¹⁶ It was suggested that the presence of two different elements at the surface, one d metal (Cu) and the other sp metal (In), may enable the modulation of the adsorption strength of ^*COOH and ^*CO intermediates and changes their trend with the applied potential, which ultimately leads to the observed product distribution.^215,218

He and co-workers reported a particularly interesting work on the high-throughput synthesis of In_xM_1−x (M = Fe, Co, Ni, Cu, Zn; 0 ≤ x ≤ 1) alloys for CO₂ reduction.²¹⁹ The selection of compositions was based on the relationship of primary products with M—CO bonding strengths [Fig. 14(a)], i.e., a high CO heat of adsorption (ΔH_CO) to the metal surface tends to form H₂ and hydrocarbon, while a low ΔH_CO value supports for HCOOH production (p-block metals).^59,220–222 The metal with a 10 kcal mol⁻¹ of ΔH_CO is optimal for CO formation.⁵⁹ In their study, In was selected because it weakly binds with hydrogen or CO. Although In has a general character of weak binding with hydrogen or CO, alloying In with late 3d transition metals could lead to stronger M—CO bonding interactions, hence promoting CO production over the HER. This conjecture was verified experimentally for experimentally synthesized series of In_xM_1-x alloys [see Figs. 14(b) and 14(c)]. In appears to suppress the HER for all the alloys, and the selectivity for CO is higher than those of the pure metal electrodes. Cu_0.8In_0.2 shows the highest CO evolution partial current. This method indicates that the relative bond enthalpies can be used as a “descriptor” for predicting and rationally designing highly active catalysts.

Using the strong chemical interaction between electrocatalysts and supporters (e.g., reduced graphene oxide, rGO) is also a useful way to enhance conductivity and improve the activity of electrochemical CO₂RR.^223,224 Zhang et al. reported that the formate FE and specific current density of In₂O₃–rGO hybrid catalysts are 1.4-fold and 3.6-fold higher than those of the physical mixture In₂O₃/rGO at −1.2 V vs RHE (Fig. 15).²²⁵ DFT calculations were further performed to simulate the change of the electronic structure and adsorption ability of intermediates. The results demonstrated that the strong chemical coupling interaction between In₂O₃ and rGO leads to faster electron transfers from rGO to In₂O₃ and a higher degree of electron delocalization in In₂O₃–rGO. The electron-rich structure has a low energy barrier to activate CO₂ and hence stabilizes HCOO^−* and promotes the formation of formate.

Besides the electrochemical CO₂RR, In-based catalysts have been extensively explored as thermocatalysts for the hydrogenation of CO₂ to synthesize methanol and C₅₊ liquid fuels with a selectivity of 78.6%,^{217,226–228} as well as photocatalysts for CO₂ photoreduction.^28,229–232 Interestingly, similar to the electrocatalysis, the presence of oxide species is important for a high catalytic activity of these systems. Finally, it should be noted that one disadvantage of In-based electrocatalysts is that the price of In metal is much higher (∼10 times) than that of Sn because In is largely used as the key material for transparent electrodes in the flat panel display.^233,234

The first report of Bi-based CO₂ reduction electrodes was presented by Komatsu et al. in 1995.²³⁵ They reported 100% formate FE and current densities about 10 mA cm⁻² in the potential region of −1.4 V to −1.7 V (vs SCE) in CO₂-saturated KHCO₃ solutions. However, Bi-based electrocatalysts did not receive as much attention as Sn. Research on Bi-based electrocatalysts was revisited in 2013 by Rosenthal et al., whose work was concerned with CO₂ reduction in ionic liquids, in which CO was the major reduction product.^53,236 More recently, Bi-based electrocatalysts have been shown to electrochemically reduce CO₂ to formate with overall performance comparable to Sn- and In-based materials.^237–241 Bi-based catalysts prepared via electrodeposition on carbon paper have achieved a current density of 15.2 mA cm⁻² with a formate FE of 96.4% at −1.8 V (vs SCE) in 0.5M NaHCO₃ saturated with CO₂.²³⁹ In addition, Liu et al. recently reported Bi-based electrocatalysts prepared by growing Bi₂O₃ nanosheets on a conductive multiple channel carbon matrix using hydrothermal synthesis whose current density and formate FE were shown to reach about 17 mA cm⁻² and 93.8% at −1.23 V vs RHE in 0.1M KHCO₃ electrolyte saturated with CO₂.²³⁸ 

Different from Sn-based and In-based electrocatalysts for which the importance and role of metal oxide species in the CO₂RR performance is a matter of general agreement, literature on the active sites and mechanism for Bi-based electrocatalysts are conflicted about whether BiO_x species play a role in the reaction. On one hand, Pander et al. analyzed the in situ ATR-IR spectra on the Bi surface during CO₂ electrochemical reduction and found that the amount of oxide layer on the Bi surface remains unchanged after initial reduction and no reaction intermediates such as metal-carbonate species (readily observed on Sn- and In-based electrodes) could be detected under working conditions.²⁴² Based on this spectroscopic evidence, they classified the Bi-electrode as an oxide-independent CO₂RR electrocatalyst.²⁴³ On the other hand, many recent experimental and theoretical reports highlighted the importance of Bi—O interaction for the performance of Bi-based CO₂RR electrocatalysts. DFT calculations showed that the Bi(111) surface is inert to the adsorption of most species except O₂.²⁴⁴ Bi—O bonds have stronger stability than Bi—C bonds due to a more effective hybridization between Bi 6s, 6p and O 2p. Therefore, the adsorbates with O—Bi bonds are systematically more stable than those with C—Bi bonds, which results in the stabilization of formate intermediates (^*OCHO) over carboxyl intermediates (^*COOH), leading to the characteristics of the selective formation of HCOOH over CO during the CO₂RR. Some recent experimental and theoretical calculation reports have proposed that Bi³⁺ lone pair electrons in Bi oxides are able to facilitate adsorption and interaction of the catalysts with CO₂.^245–247 The Bi 6s, 6p–O 2p hybridized electronic states make a significant contribution to the top of the valence band in Bi oxides; the energy level and spatial distribution of such electronic states result in the formation of stereochemically active lone pairs that promote electron donation to acidic adsorbed species such as CO₂. In this sense, Deng et al. found that the Bi—O structure of bismuth oxides can also accelerate the adsorption and activation of CO₂, switching at the same time the rate-determining step from the initial electron transfer process (to form CO₂⁻ radicals) to the subsequent protonation step.²³⁷ 

Electrochemical deposition is one of the most common techniques for the preparation of Bi electrodes with different morphologies. Zhong et al. studied the effect of addition of KBr into the electrolyte during the electrochemical deposition process.²³⁹ They found that it alters the morphology, particle size, and chemical state of the Bi catalysts, leading to the preparation of dendrite-structured electrocatalysts with large electrochemical surface areas. A maximum FE of 96.4% and a partial current density of 15.2 mA cm⁻² were achieved at −1.8 V vs SCE in CO₂-saturated 0.5M NaHCO₃. Similarly, Koh et al. prepared hierarchical Bi dendrites by electrodeposition of the Bi³⁺ precursor in ethylene glycol.²⁴⁸ The Bi dendrites had superior intrinsic catalytic properties owing to the exposure of high index surfaces with a large amount of undercoordinated sites that were believed to effectively stabilize the ^*OCHO intermediate. As a result, it exhibited the maximum formate FE of about 89% at −0.74 V vs RHE.

2D Bi nanosheets are widely studied catalyst structures because it often leads to improved performances and there are numerous synthetic techniques to obtain such a structure. Bulk Bi has a layered structure isomorph like black phosphorus that can be exfoliated to form 2D (nanosheets) structures by the so-called liquid-phase exfoliation. Zhang et al. used this technique to prepare ultrathin 2D Bi nanosheets, which were shown to have improved performances as compared with the bulk Bi electrode. Exfoliated nanosheets presented the enhanced formate FE of 86% and a current density of 16.5 mA cm⁻² at −1.1 V vs RHE in CO₂-saturated 0.1M KHCO₃.²⁴⁹ DFT calculations revealed that the edge sites can more effectively stabilize the formation of the ^*OCHO intermediate and are, therefore, the preferred active sites. Ultra-thin Bi nanosheets (about 10 nm) with numerous low-coordination sites have been fabricated through hydrothermal treatment of the electrodeposited Bi-electrode.²⁵⁰ The performance of the nanosheet structured electrocatalysts featured a formate FE higher than 90% in a potential range from −0.9 V to −1.2 V vs RHE, achieving current densities around 14 mA cm⁻². The authors argued that the improved catalytic performance was attributed to a higher partial density of states of the Bi p band around the Fermi level, which usually facilitates the orbital hybridization of the catalyst with adsorbed reactants.

Metallic Bi nanosheets can also be prepared through topotactic transformation of Bi compounds such as layered Bi oxyhalides (BiOI²⁵¹ and BiOBr²⁵²) and layer carbonate (Bi₂O₂CO₃).³³ For example, Li and co-workers prepared BiOI nanosheets and found that the nanosheet structure was maintained when BiOI was cathodically reduced to metallic Bi. The Bi nanosheets can catalyze the CO₂RR to formate with an excellent selectivity of >90% and a large partial current density of 24 mA cm⁻² at −1.74 V vs SCE. Although, in principle, it is possible to prepare Bi monolayers, it has been theoretically demonstrated that for Bi nanosheets, it is not always “the thinner the better.”²⁵³ Due to the existence of a sizable bandgap, the Bi monolayer exhibits lower catalytic activity than thicker metallic Bi nanosheets in a microkinetic model of CO₂ conversion to HCOOH.

Introducing structure disorders or defects can also provide local electronic states, which may have a positive impact on the stabilization of reaction intermediates.^254,255 For example, Li and co-workers prepared Bi nanoparticles/Bi₂O₃ nanosheets with abundant GBs through a controlled hydrothermal route.²⁵⁶ A large partial current density for formate (24.4 mA cm⁻²) and high FE (>90%) for formate under a wide potential range have been achieved, which was ascribed to the intermediate stabilization effects associated with GBs. In another work, Gong et al. prepared defective Bi/Bi₂O₃ nanotubes, which showed 60 mA cm⁻¹ of formate partial current density at −1.05 V vs RHE in 0.5M CO₂-saturated KHCO₃ solution.²⁵⁴ DFT calculations suggested that structural defects (e.g., Bi vacancies) can boost the electrocatalytic performance by stabilizing ^*OCHO intermediates. Moreover, this catalyst can achieve a large current density of 288 mA cm⁻¹ at −0.61 V vs RHE in 1M KOH when used in a flow cell reactor. The large current density is larger than commercial standards (>200 mA cm⁻¹).

The intrinsic CO₂ reduction activity of Bi electrodes has been shown to benefit from the formation of alloys with other metals and advanced composite structures.^257,258 Hoffman et al. studied the electrochemical reduction of CO₂ at dendritic Cu, Bi, and Cu—Bi alloys with varying compositions.²⁵⁸ Due to the crystal lattice and symmetry mismatch between Cu and Bi, the dendritic structures of the alloys displayed a high density of defect sites, which serve as active catalytic sites. The catalytic properties of bimetallic Cu—Bi catalysts could be effectively tuned upon varying the composition, reaching a maximum of 90% formate FE at around −0.85 V vs RHE for 40–60 at. % Bi concentration. The authors argued that the modulation of catalytic properties originated from the adjustment of the relative adsorption strength of reaction intermediates.

Bi—Sn metallic composites have also been shown to yield high selectivity and stability for CO₂ reduction. Owing to the formation of active sites through favorable orbital interactions at the Sn—Bi interface, a Bi—Sn bimetallic catalyst converted CO₂ to formate with 96% FE and production rate of 0.74 mmol h⁻¹ cm⁻² at −1.1 V vs RHE.²⁰³ DFT simulation showed that at the Bi—Sn interface, p and d orbitals of Sn electron states upshift away from the Fermi level, leading to the shifting of the electron density from more electronegative O atoms to the p and d orbitals of Sn atoms. This increases the stability of ^*OCHO intermediates on Bi—Sn surfaces compared with pure Sn or Bi, resulting in better catalytic properties (Fig. 16).²⁰³ 

Moreover, Bi-graphene composites have been shown to improve the CO₂ adsorption capacity, which alleviates problems related to the weak adsorption of CO₂ and mass transport limitations, characteristic of the CO₂ reduction process. Chen et al. synthesized a composite structure of Bi₂O₃ nanosheets and N-doped graphene quantum dots (NGQDs) by a simple solvothermal method (Fig. 17).²⁵⁹ They showed that these materials are able to promote the CO₂ reduction at an average formate FE of 95.6% in the range of −0.9 V to −1.2 V vs RHE. DFT calculations revealed that increased adsorption energy of CO₂ and the OCHO^* intermediate, promoted by NGQDs, results in the enhanced performance of nanostructured Bi₂O₃ catalysts.

Despite its well-known toxicity, the mechanism of Pb-based electrocatalysts for CO₂ reduction to formate is worth exploring. Although the overpotentials required to drive the reaction on Pb electrodes are usually higher than those required with other p-block electrodes, high formate FE (∼95%) is usually obtained.^62,260–262 Early mechanistic studies based on DFT calculations suggested that the origin of the characteristic high selectivity of the Pb electrocatalyst for HCOOH production (over CO and H₂) lies in the strong O-affinitive and weak C-, H-affinitive characteristics of Pb, which leads to the involvement of the ^*OCHO species as a key intermediate to produce HCOOH exclusively, preventing unwanted H₂ production at the same time.^129,263 However, later DFT calculations considered the effects of the formation of the solvent and cations in the electrolyte and suggested that the reaction may preferentially proceed via direct carbon-bond ^*COOH intermediates, which have the same high formate selectivity.²⁶⁴ 

In fact, experimental mechanistic studies based on in situ ATR-IR carried out by Pander et al. showed that the CO₂ reduction on Pb electrodes did not proceed through the formation of Pb carbonates, which suggests that the reaction occurs on pure metallic Pb surfaces via direct carbon-bond ^*COOH intermediates.²⁴² However, the author reported that the total faradaic yield for formate production increased upon anodization of the electrode surface, suggesting that surface oxide species improve the performance of Pb electrodes. Lee and Kanan reported a similar observation on nanocrystalline Pb films prepared by reducing PbO₂ precursors had a significant higher formate FE than that of polycrystalline Pb foil electrodes.²⁶⁰ 

Despite Pb-based electrocatalysts have excellent selectivity for formate formation, the intrinsic toxicity and concomitant environmental unfriendliness make this material not ideal for electrochemical CO₂ reduction, and as a result, it has gained less attention than other p-block metals. There is only a small amount of studies on the nanostructuring and doping of Pb-based electrodes such as plate, nanowires, nanoparticles, sulfide-derived Pb, and oxide-derived Pb.^{34,35,265–267} Their catalytic properties, especially FE, have a close relationship with nanostructures, morphologies, crystalline structures, and composition effects of Pb electrodes. For example, the Pb electrodes with the wafer structures show the enhanced selectivity of formate.²⁶⁶ Fan et al. fabricated lead electrodes with a honeycomb primary structure and a dendrite-like secondary structure grown along the [100]-axis. Such a structure was shown to supply abundant electrode/electrolyte contact surface area and strong structural stability for the CO₂RR.²⁶¹ Additionally, the tip closed by [111] planes showed local high fields, which results in a high concentration of CO₂ on the tip surface, which was beneficial to improve the reaction rate and decrease the overpotential. An FE of 97% and a partial current density of −7.5 mA cm⁻¹ for formate production were achieved under the potential of −0.99 V vs RHE that was stable for more than 6 h.

In terms of the influence of composition, Choi et al. notably showed that the Sn—Pb alloy surface led to higher electrical conductivity and stronger electron-donating ability compared to the single metal electrode. These effects were ascribed to the formation of SnO_x and Pb⁰ on the alloy surface.²⁰⁴ It was shown that the presence of Pb (15–35 at. %) had the ability to stabilize SnO₂.²⁶⁵ At the same time, the alloy composition may supply synergistic catalytic effects on selective adsorption sites and electronic interactions.^265,268

In this article, we provide a review on the recent developments and understanding of the p-block metal-based (Sn, In, Bi, and Pb) electrocatalysts for electrochemical CO₂ reduction. This unique group of electrocatalysts has been shown to drive CO₂ reduction to formate with a particularly high FE well above 90%. This is fundamentally governed by the unique surface physicochemical properties of this group of materials, which favorably stabilize the ^*OCHO intermediate to form formate, and at the same time, they are unfavorable for the formation of ^*COOH (intermediate to CO or CH₄) or ^*H (intermediate to H₂). We focus on the fundamental understanding of surface chemistry, active sites, and structure–activity relationships. Strategies to enhance the activity including morphology control, nanostructuring, grain boundaries, doping, alloying, and defect engineering are briefly reviewed.

Regarding the active sites for the CO₂RR, a key observation is that the presence of metal oxide species on the catalyst surfaces under the electrochemical CO₂RR conditions is crucial for the activity and selectivity of Sn and In based electrocatalysts. The persistence of oxide species can be rationalized on the basis of the oxophilicity of p-block metals, which is considered to be moderate in a quantitative scale of oxophilicity.²⁶⁹ Inferred from in situ spectroscopic studies, the p-block metal oxides might provide surface basicity for the chemisorption of acidic CO₂ molecules to form surface bound metal carbonates. Furthermore, the oxide species may also facilitate electron transfer to the stabilized CO₂⁻ intermediate, which is regarded as the RDS for the CO₂RR. Now, it is interesting to ask the question: how the CO₂ molecule or other reaction intermediates interact with the catalyst surface. Previous studies on the surface chemistry of p-block metal oxides indicated that the p-block metal is dominated by the group oxidation state N and a lower N-2 oxidation state, which is associated with the occupation of a metal s² lone pair, as found in compounds of Sn²⁺, Bi³⁺, Pb²⁺, and In⁺ on the surface of In₂O₃.²⁸⁷ Lone pairs project out of the surface, hence providing localized surface electron density where for the chemisorption of CO₂.^247,270,271 Therefore, more detailed research on the role of surface oxides may lead to important design guidelines to maximize both the intrinsic and overall performances of p-block metal catalysts for CO₂ reduction.

Mechanistic studies have been largely based on DFT calculations. It has been found that there is a strong correlation between the DFT calculated ^*OCHO binding energies and the experimentally reported activities of different electrocatalysts, which have led to the conclusion that ^*OCHO, the key intermediate for formate formation. However, it should be noted that the catalyst surfaces (i.e., metallic Sn) used for DFT calculations are different from the real surfaces with oxide species observed experimentally. In operando characterizations suggest that the surface-bound metal carbonate is a crucial reaction intermediate for transforming CO₂ into formate. The situation is further complicated by the dynamic behavior of the catalyst composition during the reduction reaction conditions with multiple oxidation states present under different reaction conditions. At present, it is still not well understood the role of oxide species in the reaction pathway and mechanism for the CO₂RR to formate. Therefore, combining theoretical calculations with “actual” reaction intermediates evidenced from in operando characterizations, such as infrared spectroscopy, Raman spectroscopy, and XAS under working conditions, is particularly important in order to elucidate the reaction mechanism.

Significant progress has been made for the development of new electrocatalysts with better selectivity, activity, and stability. Many strategies including morphology control, surface modification, nanostructuring, grain boundaries, doping, alloying, and defect engineering have been adopted to synthesize a variety of materials. The goals of these methods are to increase their active site density and/or to promote their site-specific activity, but at this stage they are still based on a trial-and-error approach. It is straightforward to understand that shrinking the dimensions and sizes of electrocatalysts can create more active sites on the surface for electrochemical reactions. Meanwhile, nanostructuring may also introduce new reaction sites, such as preferential exposure of different crystal facets, more atoms with coordinately unsaturated sites at steps and edges, and structural defects such as vacancies and GBs that may afford unexpected electrocatalytic activities. However, how these new reaction sites impact the electrocatalytic performance (i.e., structure–activity relation), which is essential for rational design of better catalysts, remains to be answered.

On the other hand, a standard procedure for measurements and comparisons is still needed. The nanostructured materials prepared by different scientific groups usually have different surface areas, which complicates the comparison of a fair evaluation of relevance of the particular material properties for the CO₂RR. Additionally, the CO₂RR performance assessment is further complicated by the complex methods of measurement, which can be altered by even the trace amount of impurity in the electrolyte. In this sense, compared with ill-defined porous electrodes commonly used in CO₂RR studies, single crystal thin films permit a high degree of control over crystal facets, defects, and compositions, which could be used as a model catalyst to gain a definitive structure–property relationship at a molecular level.

Doping and alloying with foreign atoms can induce a significant change in the electronic properties and are the most effective ways to develop a new electrocatalyst. Electronic structure parameters such as the density of state near the Fermi level dictate the binding strength with reaction intermediates and hence the activation barrier for the reaction.^{247,272,273,288} However, so far there is still a lack of understanding of how doping, alloying, or defects influence the electronic structure and its relationship with CO₂RR selectivity and activity. Moreover, alloying to form bimetallic or ternary catalysts could offer synergistic effects from different active sites and hold the potential to significantly alter the selectivity and activity. For example, as discussed above, alloying Sn with Cu and Ag can enhance the selectivity for formate, while alloying In with Cu alter the selectivity to CO. Furthermore, theoretical studies indicated that scaling relations that exist between adsorption energies of different reaction intermediates can be tailored by alloying or doping. Therefore, a deeper insight into catalytic reaction pathways involving scaling relations among reactive intermediates and structure–property relationships is highly needed as “descriptors” for the rational design of highly active catalysts with better selectivity and stability.

K. H. L. Zhang is grateful for funding from the National Natural Science Foundation of China (Grant No. 21872116) and funding for the “single-crystal electrochemistry” project from the State Key Laboratory of Physical Chemistry of Solid Surfaces.