Experimental and Theoretical Insights into Single Atoms, Dual Atoms, and Sub‐Nanocluster Catalysts for Electrochemical CO₂ Reduction (CO₂RR) to High‐Value Products

Abstract

Electrocatalytic carbon dioxide (CO₂) conversion into valuable chemicals paves the way for the realization of carbon recycling. Downsizing catalysts to single‐atom catalysts (SACs), dual‐atom catalysts (DACs), and sub‐nanocluster catalysts (SNCCs) has generated highly active and selective CO₂ transformation into highly reduced products. This is due to the introduction of numerous active sites, highly unsaturated coordination environments, efficient atom utilization, and confinement effect compared to their nanoparticle counterparts. Herein, recent Cu‐based SACs are first reviewed and the newly emerged DACs and SNCCs expanding the catalysis of SACs to electrocatalytic CO₂ reduction (CO₂RR) to high‐value products are discussed. Tandem Cu‐based SAC–nanocatalysts (NCs) (SAC–NCs) are also discussed for the CO₂RR to high‐value products. Then, the non‐Cu‐based SACs, DACs, SAC–NCs, and SNCCs and theoretical calculations of various transition‐metal catalysts for CO₂RR to high‐value products are summarized. Compared to previous achievements of less‐reduced products, this review focuses on the double objective of achieving full CO₂ reduction and increasing the selectivity and formation rate toward C–C coupled products with additional emphasis on the stability of the catalysts. Finally, through combined theoretical and experimental research, future outlooks are offered to further develop the CO₂RR into high‐value products over isolated atoms and sub‐nanometal clusters.

This review comprehensively summarizes the Cu‐based sub‐nanometric catalysts with critical insights for electrochemical CO₂ reduction (CO₂RR) to high‐value products. In addition, the non‐Cu‐based sub‐nanometric catalysts and theoretical calculations of different transition‐metal‐based catalysts are reviewed and detailed for CO₂RR to high‐value products. Finally, conclusions and future outlooks are provided for further development of the research domain

1. Introduction

The electrocatalytic CO₂ reduction reaction (CO₂RR) to valuable fuel products and chemical feedstocks illustrates the realization of both global carbon neutrality and chemical energy storage.^{[ 1 ]} In CO₂RR, electrocatalysts assisted with electricity promote the CO₂ conversion to various “high‐value” products involving the transfer of six or more electrons during the electrocatalytic reaction. C₁ products (e.g., methane (CH₄) and methanol (CH₃OH)), C₂ products (e.g., ethylene (C₂H₄), acetate (CH₃COO^‐), ethanol (C₂H₅OH)), and C₃ products (e.g., n‐propanol (C₃H₈O) and acetone (C₃H₅O)) are formed in such CO₂RRs.^{[ 2 ]} So far, the activity and selectivity of electrocatalysts toward reduction products have been mostly oriented to C₁ products, mainly CO and HCOO⁻ rather than highly reduced C₁ and multicarbon (C₂₊) products. Among the C₁ products, CH₄ and CH₃OH are attracting increasing attention from economic and environmental viewpoints.^{[ 3 ]} For instance, CH₃OH is the starting material for the synthesis of many chemicals and a source of energy and fuel,^{[ 3b ]} and CH₄ is used for various applications, such as energy carrier, owing to its great energy density, and the production of valuable chemicals and polymers due to the well‐established and existing infrastructure storages in consumption and distribution.^{[ 3a ]} The formation of carbon–carbon (C─C) bonds during the CO₂RR is of considerable value, because highly reduced multicarbon (C₂₊) products have greater economic value than C₁ products for myriads of applications in the chemical and fuel industries. For industrial applications, C₂₊ products from CO₂RR are preferable over C₁ products due to their high‐energy density, ease of storage and transportation, low toxicity, well‐suited chemistry, and the existence of infrastructures for processing.^{[ 4 ]} However, in addition to the inevitable competitive hydrogen evolution, the wide range of intermediates and CO₂ reduction products lead to low selectivity and formation rate of high‐value products. Key metrics, such as selectivity to a specific high‐value product, current density at applied potentials, and stability are among the key factors that need to be improved.

To date, copper (Cu)‐based catalysts are renowned catalysts for efficient electrocatalytic CO₂ gas transformation into various hydrocarbons and oxygenated products.^{[ 1} , ^{5 ]} Carbon‐based materials are designed via the incorporation of heteroatoms into their structure for CO₂RR to valuable fuels and chemicals,^{[ 6 ]} whereas there is a lack of reports toward high‐value products, particularly C₂₊ products. A plethora of theoretical studies recently revealed the greater possibility of CO₂ transformation to high‐value products using non‐copper transition metal‐based catalysts.^{[ 7 ]} The capability of converting CO₂ to high‐value products using non‐Cu transition metal‐based catalysts has also been demonstrated experimentally, as will be later discussed. Particularly, downsizing the catalytic materials of any type to the sub‐nanometric level (single atoms, diatomic, sub‐nanoclusters, and tandem single atoms and nanocatalysts) is attracting a lot of attention. This is due to the exposure of many active sites, highly unsaturated coordination environment, efficient atom utilization, and confinement effect. These trends improve the reactivity of the catalyst, due to the high electron density distribution and fast electron transfer.^{[ 8 ]} However, the stability of these sub‐nanometric catalysts is challenging under harsh reaction conditions. Remarkably, these sub‐nanometric catalysts can be stabilized by support over appropriate host nanomaterials such as MXene, covalent organic frameworks (COFs), metal–organic frameworks (MOFs), graphene, metal oxides, and metal sulfides.^{[ 2} , ^{9 ]} The interaction between the sub‐nanometric catalysts and crystalline supporting material may modify the electronic properties and geometric structures of these catalysts,^{[ 10 ]} and hence, the selectivity and activity of the catalysts during the CO₂ transformation. Photochemical CO₂ conversion to high‐value products can also be realized using sub‐nanometric catalysts supported on an appropriate host material.^{[ 11 ]} However, the low solar‐to‐chemical conversion efficiency, low density of active sites, quick recombination of photoexcited electron–hole pairs, and the susceptible false‐positive results from contaminants may hamper the development of this strategy for large‐scale application.^{[ 12 ]} Therefore, the electrochemical system is considered a suitable strategy for large‐scale CO₂ conversion due to its advantage of operating at large current densities and integrating with renewable energy sources.

Several reviews on the CO₂ conversion applications of single‐atom catalysts (SACs) have been reported to date.^{[ 8} , ⁹ , ¹² , ^{13 ]} However, most of them emphasize: 1) the general CO₂ conversion via photochemical, photo/electrochemical, and thermochemical processes, and less mechanistic insights toward high‐value product formation are provided, 2) mainly SACs with focus on C₁ products and less attention to C₂₊ high‐value products, 3) DACs for various catalytic applications (e.g., hydrogen evolution, oxygen evolution, oxygen reduction, CO₂ reduction). Less emphasis has been given to CO₂ reduction that has mainly limited been to CO and formate, and there is a lack of discussion on experimental and theoretical insights of Cu‐ and non‐Cu‐based catalysts toward high‐value products, or 4) structural sensitivity and interactions between SAC and support. Despite these efforts made in the study of SACs, a review is lacking on the electrochemical CO₂ reduction to high‐value products over Cu‐ and non‐Cu‐based dual‐atom catalysts (DACs), tandem SAC–nanocatalysts (SAC–NCs), and sub‐nanocluster catalysts (SNCCs) with experimental and theoretical insights.

SACs are mononuclear metal atoms that contain isolated metal atoms with no defined synergistic effect between neighboring atoms, whereas DACs are composed of two neighboring metal atoms either directly bonded to each other or atom‐bridged, both of which are involved in the electrocatalytic reactions to improve the performance due to the synergistic and complementary effects between the two single atoms. The tandem SAC–NCs contain one SAC and another nanocatalyst (nanocluster or nanoparticle), where the nanocatalyst active site can further improve the catalytic activity and selectivity in tandem or synergy with the SAC active sites. SNCCs consist of a few metal atoms (typically <10) that form a stable cluster on a host material. The differences between the SACs, DACs, tandem SAC–NCs, and SNCCs related to active sites and electronic and geometric properties are summarized in Table 1 that is further detailed in Table S1 of the Supporting Information. These sub‐nanometric catalysts (SACs, DACs, SAC–NCs, and SNCCs) are becoming new frontiers in the field of nanocatalysis due to their high surface area exposure, maximum atom utilization, confinement effects, and unique electronic and geometric properties compared to the conventional nanoparticles (NPs). In conventional NPs, only some portions of the metal atoms are exposed for catalytic applications, and they perform differently based on their shape, size, and composition. The advantages and limitations of comparisons of SACs, DACs, SAC–NCs, SNCCs, and nanoparticle catalysts are summarized in Table S2 of the Supporting Information. The sub‐nanometric catalysts with suitable host materials have attracted more attention for various catalytic applications due to their unique properties. They possess the advantages of heterogenous and homogenous catalysts, leading to superior reactivity and maximum atom utilization.^{[ 14 ]} To our knowledge, no comprehensive reviews are available treating both experimental and theoretical studies of CO₂ electroreduction to high‐value products over the sub‐nanometric Cu‐ or non‐Cu‐based catalysts. Here, we first review the key features required to realize CO₂RR toward high‐value products over Cu‐ and non‐Cu‐based SACs, DACs, tandem SAC–NCs, and SNCCs (Figure 1 ). Then, we briefly summarize the theoretical calculations of important intermediates toward high‐value products over Cu‐ and non‐Cu‐based transition‐metal SACs, DACs, tandem SAC–NCs, and SNCCs. Throughout, we provide insights into a mechanistic understanding of the catalysts and identify the main gaps existing today. We point out the explicit bottlenecks to further develop the CO₂RR over supported Cu‐ or non‐Cu catalysts, specifically toward increasing high‐value product formation.

Table 1.

Summary of the differences between SACs, DACs, tandem SAC–NCs, and SNCCs, related to their number of active sites and electronic and geometric properties (Table 1 is presented with additional details in Table S1 of the Supporting Information).

Catalyst	Number of active sites	Electronic properties	Geometric properties
SACs	One active site per atom, individual atoms dispersed on a host material; each isolated atom acts as an active site for the catalytic reaction.	Tailored by selecting specific elements; electronic alignment impacts interaction with reactants; SA–support electronic coupling influences the charge transfer and reaction kinetics.	Influenced by coordination environment, type and number of neighboring atoms, arrangement of support material or ligand around the active sites, etc.
DACs	Two active sites per catalyst; the two closely bonded, unbonded, or bridged atoms create a unique active site that facilitates the catalytic reaction.	Affected by electronic interaction and charge redistribution between the two elements, and the atom–support interaction also impacts the electronic properties.	Influenced by geometric arrangement (bond length, angle, and surface sites) between the two active sites and the arrangement of the support.
Tandem SAC–NCs	Two active sites per catalyst, SA and NC sites. The coexistence of SAs and NCs generates tandem or synergistic effects. SAs provide highly active and selective active sites, and NCs improve stability, mass transport, etc.	Merits of SACs and NCs; SAs show unique electronic configurations; NCs contribute to the overall electronic structures; this leads to tandem or synergistic effects, and their electronic interactions improve the reaction.	Affected by the geometric arrangement between the SACs and NCs (e.g., the proximity and location sites between the two sites), and their interaction with the supporting material all contribute to the reaction.
SNCCs	In one ensemble site, the number of active sites differs depending on the composition and size of the cluster, consisting of a small number of atoms (2 < x < 10 metal atoms).	Have quantum confinement and ligand effects that can modify the electronic structures; the cluster–support interaction influences the electronic properties.	Geometric arrangement of atoms (e.g., coordination of atoms), and ligand effects can impact the adsorption and activation of reactants, affecting the catalytic performance.

Overall, the geometric and electronic properties of SACs are intricately related and regulated for catalytic optimization. In DACs, the electronic and geometric properties are extremely specific to the precise combination of atoms and the preparation technique used. Thus, the properties considerably differ depending on the particular DAC system. The geometric and electronic properties of tandem SAC–NCs combine the unique features of both SAs and NCs, resulting in improved selectivity, catalytic activity, and stability, whereas SNCCs highly rely on their composition, size, and surface structure and are tailored for performance optimization.

Figure 1.

Schematic representation of the single‐atom catalysts (SACs), dual‐atom catalysts (DACs), tandem single‐atom catalyst–nanocatalysts (SAC–NCs), and sub‐nanocluster catalysts (SNCCs) for CO₂RR to high‐value products.

2. CO₂RR Performances of Cu‐Based SACs, DACs, Tandem SAC–NCs, and SNCCs

Under this section, special emphasis is given to review the experimental findings of Cu‐based SACs, DACs, and tandem SAC–NCs in mounting the catalysis of electrochemical CO₂ reduction to valuable fuels and chemicals, specifically to high‐value products. In addition, experimentally reported Cu‐based SNCCs (i.e. <10 metal atoms) are discussed for the electrochemical transformation of CO₂ to high‐value products. Figure 2 illustrates the distribution states of the abovementioned Cu‐based catalysts. The CO₂RR performance to high‐value products over the sub‐nanometric Cu‐based catalysts is summarized in Table 2 .

Figure 2.

Schematic representation of the distribution states of metallic species on single‐atom catalysts (SACs), homo‐ and heteronuclear dual atom catalysts (DACs), homo‐ and heteronuclear single‐atom catalyst–nanocatalysts (SAC–NCs), and sub‐nanocluster catalysts (SNCCs).

Table 2.

CO₂RR to high‐value products of Cu‐based SACs, DACs, SAC–NCs, and SNCCs.

Category	Material	Electrolyte	Type of cell	E [V vs RHE]	J [mA cm−2]	TOF	Stability [h]	Product FE [%]					Refs.
								CH4	CH3OH	C2H4	C2H5OH	C3H7OH
SACs	2Bn‐Cu@UiO‐67	1.0 m KOH	Flow	−1.50	340.2	16.3 s−1	……	81.0	……	≈2.0	≈3.0	……	[48]
	SA–Cu–MXene	0.1 m KHCO3	H‐type	−1.40	12.6	……	>30	……	59.10	……	≈5.0	……	[9e]
	Cu SAs/TCNFs	0.1 m KHCO3	H‐type	−0.90	≈45.5	……	50.0	……	44.00	……	……	……	[49]
	Cu/p‐Al2O3 SAC	1.0 m KOH	Flow	−1.20	94.8	5620 h−1	……	62.0	≈1.00	≈2.0	……	……	[23b]
	Cu–N2O2	0.5 m KHCO3	H‐type	−1.44	≈31.2	>2000 h−1	6.0	78.0	……	……	……	……	[50]
	Cu‐NC	0.1 m CsHCO3	Flow	−1.20	≈10.5	……	……	……	……	……	55.0	……	[51]
	Cu‐NC	0.5 m CsHCO3	H‐type	−1.10	35.6	3.458 h−1	>25.0	……	……	……	81.9	……	[27a]
	Cu‐CeO2‐4%	0.1 m KHCO3	H‐type	−1.80	≈32.5	……	2.5	58.0	……	≈15.0	≈2.0	……	[52]
	Cu‐SAs‐0.1	1.0 m KOH	Flow	−1.80	493.1	……	≈16.7	68.2	……	……	……	……	[53]
	Cu‐SAs/GDY	0.1 m KHCO3	H‐type	−1.30	24.0	2311 h−1	10.0	66.0	……	……	……	……	[54]
	Cu NWs/CoPc	1.0 m KOH	Flow	−1.40	203.9	……	20.0	……	……	58.3	≈10.0	……	[29]
	Cu‐SA/HGDY	1.0 m KOH	Flow	−1.10	166.3	2756 h−1	11.0	72.1	……	≈3.0	……	……	[16f]
	Cu‐TABQ	1.0 m KOH	Flow	−1.07	≈228.4	……	20.0	≈16.0	……	≈54.0	10.1	……	[28]
	Cu(OH)BTA	1.0 m KOH	Flow	−0.87	285.0	……	67.0 a)	……	……	57.0	11.0	1.0	[55]
	Cu‐SACs‐1N‐CQDs	0.1 m KHCO3	H‐type	−0.20	0.7	……	50.0	≈10.0	……	……	70.0	≈10.0	[56]
	Cu5Ce95O x	1.0 m KOH	Flow	−1.20	≈138.0	……	500.0	61.2	……	……	……	……	[57]
	Cu‐N1O3	1.0 m KHCO3	Flow	−1.80 b)	10.1	……	16.0 a)	……	……	33.0	42.0	……	[58]
DACs	CuI‐MAF‐2P	0.1 m KHCO3	H‐type	−1.50	5.7	……	8.0	≈56.0	……	≈21.0	……	……	[33b]
	CuI‐MAF‐2E	0.1 m KHCO3	H‐type	−1.30	5.7	……	10.0	≈20.0	……	51.2	……	……	[33b]
	PcCu‐Cu‐O	0.1 m KHCO3	H‐type	−1.20	≈3.7	……	4.0	≈13.0	……	50.0	……	……	[32a]
	Cu‐PzH	1.0 m KOH	Flow	−1.00	207.9	……	≈4.0	8.5	……	60.0	……	……	[33a]
	Cu‐PzI	1.0 m KOH	Flow	−1.00	149.5	……	≈4.0	52.0	……	16.0	……	……	[33a]
	Cu/In–N–C	0.5 m KHCO3	H‐type	−0.62	……	……	……	……	55.63	……	……	……	[35]
	Cu2/NC	0.5 m KHCO3	H‐type	−1.23	11.7	……	12.5	≈5.0	……	≈34.9	……	……	[59]
	CuBtz	0.1 m KHCO3	H‐type	−1.30	7.9	0.009 s−1	>20	……	……	44.0	21.0	4.0	[60]
	Cu2(obpy)2	0.1 m KHCO3	Flow	−1.40	73.80	……	100.0	82.0	……	≈3.0	……	……	[34a]
	Cuophen	1.0 m KOH	Flow	−1.40	319.0	……	>50	37.0	……	55.0	……	……	[61]
	SnCu‐HAB	1.0 m KOH	Flow	−0.57	38.08	0.0601 s−1	35.0	……	……	……	56.0	……	[34b]
	[Cu2(btec)(phen)2] n ·(H2O) n	0.1 m KHCO3	H‐type	−1.60	17.5	……	1.0	69.0	……	7.0	……	……	[62]
	Cu2	1.0 m KOH	Flow	−1.60	239.4	≈520 h−1	≈1.7 a)	≈12.0	……	51.0	≈5.0	……	[63]
SAC–NCs	PTF‐(Ni)/Cu	0.1 m KHCO3	H‐type	−1.10	3.1	……	11.0	≈5.0	……	57.3	……	……	[39b]
	Pdδ−–Cu3N	1.0 m KOH	Flow	−1.50	90.8	……	……	≈3.0	……	≈25.0	≈35.0	……	[39a]
	Cube‐Pd1Cu	0.5 m KHCO3	Flow	−1.10	118.0	……	5.0	≈15.0	……	≈62.0	……	……	[39c]
	Octa‐Pd1Cu	0.5 m KHCO3	Flow	−1.10	118.0	……	5.0	≈63.0	……	≈20.0	……	……	[39c]
	BiCu‐SAA	0.1 m KHCO3	H‐type	−1.10	……	……	11.0 a)	……	……	≈43.0	≈20.0	≈10.0	[64]
	AgCu SANP	1.0 KOH	Flow	−0.65	338.4	……	13.0	……	……	≈40.0	≈47.0	≈5.0	[65]
	Ni/Cu‐NP/CMK	1.0 m KHCO3	Flow	−1.10	294.0	……	15.0	……	……	72.3	……	……	[66]
	Ni‐SOD/NC + Cu NPs	0.5 m KHCO3	Flow	−0.72	100.0	……	20.0	……	……	62.5	……	……	[67]
	Ni SA/CuO	1.0 KOH	Flow	≈−0.80	811.5	……	3.0	……	……	54.1	28.8	……	[68]
	Fe‐TBTPP/Cucub	0.1 m KHCO3	H‐type	−1.05	3.0	……	11.0	……	……	36.0	≈5.0	≈3.0	[69]
	Ni SA/OM‐Cu	1.0 KOH	Flow	−0.76	91.4	……	50.0	……	……	15.0	63.0	……	[70]
	Cu/CoCuN x ‐C	0.1 m KHCO3	H‐type	−0.19	3.1	……	……	20.2	……	29.7	……	……	[71]
	Sb/CuO(VO)	0.1 m KHCO3	H‐type	−1.10	≈7.0	……	16.0	……	……	46.2	……	……	[72]
	Ni SAC/Cu‐R	1.0 m KHCO3	Flow	−1.25	≈370.0	……	14.0	……	……	60.0	……	……	[73]
SNCCs	Cu‐HKUST‐1	0.1 m KHCO3	H‐type	−0.95	≈15.0	……	6.0 a)	≈50.2	….….	≈30.3	….…	.….	[8a]
	Cu–C	0.1 m KHCO3	Flow	−0.70	1.1	0.035 s−1	>16	…….	…….	….	91.0	….	[45a]
	Hex‐2Cu‐O	0.1 m KHCO3	H‐type	−1.20	6.5	……	25.0	…….	…….	≈15.4	32.5	18.3	[74]
	Cu4 II‐MFU‐4l	0.5 MNaHCO3	H‐type	−1.20	9.8	0.070 s−1	24.0	92.0	…….	…….	…….	…….	[75]
	CuI‐NNU‐33(H)	1.0 m KOH	Flow	−0.90	320.6	……	5.0	82.0	…….	≈5.0	…….	…….	[46]
	Cutrz	0.1 m KHCO3	H‐type	−1.20	≈7.6	……	80.0	……	……	50.6	19.7	8.0	[44]
	Cu/N0.14C	0.1 m KHCO3	H‐type	−1.10	14.4	……	10.0	≈7.0	……	≈15.0	51.0	≈5.0	[76]
	Cu3‐Br	0.5 m KOH	Flow	−0.70	71.3	……	9.5 a)	……	……	55.0	……	……	[10]
	Cu4@Ti9	1.0 m KOH	Flow	−1.00	190.4	……	8.0	≈5.0	……	47.6	≈2.0	……	[77]
	Inz‐Cu3	1.0 m KOH	Flow	−0.90	≈200.0	……	2.2	≈10.0	……	35.3	31.5	……	[45e]
	CuBPZ	0.1 m KHCO3	H‐cell	−1.20	1.6	……	6.0	……	……	43.2	24.2	……	[78]
	Cu1Ni‐BDP	1.0 m KOH	Flow	−1.30	279.3	……	25.0	……	……	52.7	……	……	[79]
	N, OH‐Cu3	1.0 m KOH	Flow	−1.21	222.6	……	3.0	74.2	……	≈10.0	……	……	[80]
	Cu6(S001)	0.5 m KOH	Flow	−1.00	247.9	……	3.3	……	……	≈63.6	……	……	[81]
	SnCu1.5O3.5@MFI	0.1 m KHCO3	H‐cell	−1.55	14.3	……	25	66.6	……	……	……	……	[45c]
	Cu‐MOF‐20	0.5 m KHCO3	H‐cell	−1.00	7.4	……	10.0		……	……	82.5	……	[82]
	Cu6(MBD)6	1.0 m KOH	Flow	−1.40	119.	……	4.0	42.5	……	23.0	……	……	[83]

^a) Stability test not at the specified conditions; SAC ‐ single atom catalyst; DAC ‐ dual atom catalyst; SAC–NC ‐ tandem single‐atom catalyst–nanocatalyst; SNCC ‐ sub‐nanocluster catalyst; UiO‐67 family of zirconium metal–organic framework; TCNFs ‐ through‐hole carbon nanofibers; NC ‐ N‐doped carbon; V_o ‐ oxygen vacancy; GDY ‐ graphdiyne; MAF ‐ metal azolate framework; Pc ‐ phthalocyanine; Pz ‐ pyrazole; Btz ‐benzotriazole; PTF ‐ porphyrinic triazine framework; SAA ‐ single atom alloy; HKUST‐1 ‐ example of MOF based on Cu paddle wheels (─Cu₂(COO)₄, metal nodes), and 1,3,5‐benzene‐tricarboxylic acid (H₃BTC, organic link); Hex ‐ bicentric copper complexes; MFU ‐ Cu‐based metal–organic framework complex; NNU ‐ Cu‐based coordination polymer complex; AFW ‐ azolate framework; Hobpy‐1H‐[2,2′]bipyridinyl‐6‐one; Ti9 ‐ Ti₉O₉(BC)₁₈(OⁱPr)₃(^tBuC≡C)(CH₃CN) (BC, benzoic acid; ^tBuC≡CH, p‐tert‐butylphenylacetylene); MBD ‐ 2‐mercaptobenzimidazole; HAB ‐ hexaminobenzene; Inz‐1H‐Indazole; R ‐ reduced state; ophen‐1H‐[1,10]phenanthrolin‐2‐one; TABQ ‐ tetraminobenzoquinone; CMK ‐ cubic mesoporous carbon; SOD ‐ sodalite; TBTPP ‐ tetrabenzo‐tetraphenyl porphyrin; BDP ‐ 1,4‐benzendipyrazolate; N, OH‐Cu₃ ‐ nitrogen and hydroxyl‐immobilized Cu₃; J ‐ partial current density of the dominant product obtained by multiplying the current density with FE

^b) Applied potential versus Ag/AgCl.

2.1. Single‐Atom Catalysts

In heterogeneous catalysis, increasing the ratio of metal atoms boosts the catalytic performance. Theoretically, the atom‐to‐surface ratio can surge to 100% in heterogeneous metallic catalysts, which means that the active metal atoms are entirely exposed for catalytic application. Therefore, downsizing metallic catalysts to single atom, dual atom, or sub‐nanocluster (all atoms are surface atoms when n < 10) enables offering maximized active metal utilization and facilitates atomic economy.^{[ 15 ]} Single atoms are highly unstable, however, and they need a solid‐state supporting material. When a single atom is hosted by a solid carrier (metal oxides, carbon, zeolites, metal sulfide, MXene, and carbide), the metal single atom resides in a vacancy and functionalizes with bounded carbon atoms or heteroatoms, such as P, N, S, and O to form stable single‐atom‐sites (SASs).^{[ 10} , ^{16 ]} Many metal SASs, together with their supporting material, form a metal single‐atom‐site catalyst (SASC), and its arrangement affects the coordination environment of the catalytic center activity, selectivity, and stability.

The performance of SACs depends on the precise arrangement and coordination of the SAs on the host material. The catalytic applications of SACs have been expanded to numerous fields, such as oxygen evolution reaction,^{[ 17 ]} hydrogen evolution reaction (HER),^{[ 18 ]} oxygen reduction reaction,^{[ 19 ]} ammonia synthesis,^{[ 20 ]} urea synthesis,^{[ 21 ]} semihydrogenation,^{[ 22 ]} and CO₂RR.^{[ 12} , ^{16 ]} Although SACs mainly convert CO₂ to CO,^{[ 13k ]} high‐value products can also be obtained on these catalysts due to their dynamic structural sensitivity.^{[ 8b ]} Opposite to nanomaterials, active metal species in SACs are isolated and mononuclear, which helps maximize the utilization of the atom that possesses different catalytic behavior from atoms of nanomaterials.^{[ 19} , ^{23 ]} For instance, maximum CH₃OH Faradaic efficiency (FE) (59.1%) and higher current density (21.3 mA cm⁻²) were obtained from a Cu single atom immobilized on the surface of MXene (SA‐Cu‐MXene) at −1.4 V versus RHE, in comparison with that of Cu NPs on the same MXene (14.6% and 9.6 mA cm⁻²).^{[ 9e ]} The atomically well‐dispersed Cu sites in SA–Cu–MXene prevented C–C coupling from producing C₂₊ products by easing the binding energy barrier of the rate‐determining step, HCOOH* to *CHO, for CH₃OH formation as proved by density functional theory (DFT) calculation. The SA–Cu–MXene electrocatalyst offered a partial current density of ≈12.6 mA cm⁻² at −1.4 V versus RHE with greater stability for more than 30 h, where the CH₃OH FE remained >58%. However, the selectivity and formation rate of CH₃OH significantly declined at relatively high operating cathodic potential (−1.8 V vs RHE), which may hinder its compatibility for large‐scale applications. Similarly, the Zhang group synthesized Cu SAs coordinated with two carbon (Cu–C₂) atoms on hydrogenated graphene support (Cu‐SAs/HGDY) for CO₂ reduction to CH₄ with greater FE of 72.1% and a current density of 300 mA cm⁻² at −1.1 V versus RHE (Figure 3a).^{[ 16f ]} The Cu‐SAs/HGDY catalyst exhibited an outstanding CH₄ reaction rate with the respective partial current density and turnover frequency (TOF) of 230.7 mA cm⁻² and 2756 h⁻¹. The long‐term operation was carried out for 11 h at an industrially suitable current density (200 mA cm⁻²). Upon examination by various advanced characterization techniques, no formation of Cu NPs, clusters, or Cu─Cu bonds were found, indicating the absence of morphological, structural, or electronic changes. Although the catalyst exhibited outstanding stability during the 11 h electrolysis time, this reaction period is too short to provide any concluding remarks about the suitability of the catalyst for practical application. Benefiting from the effect of a Lewis acid, the Li group also demonstrated a greater transformation of CO₂ to CH₄ with 62% FE and higher catalytic activity (153 mA cm⁻²) at −1.2 V versus RHE from a Cu SA anchored on ultrathin porous Al₂O₃ support.^{[ 23b ]} The catalyst further exhibited a greater partial current density of 94.8 mA cm⁻² and a TOF of 5620 h⁻¹, indicating a higher reaction rate toward CH₄ formation. However, the stability of this catalyst for an extended reaction time remained largely unexplored. Likewise, a Cu SA catalyst encapsulated in N‐doped porous carbon showed a conversion of CO₂ to acetone with 36.7% FE at −0.36 V versus RHE.^{[ 24 ]} The theoretical investigation uncovered the coordination of Cu atoms with pyrrole‐N₄ atoms as the key active center by reducing the activation barrier of CO₂, eventually facilitating C–C coupling. Remarkably, a study by the Chen group demonstrated an efficient CO₂ conversion to acetate with FE and current density of 90.3% and 12.5 mA cm⁻² at −0.8 V versus RHE, respectively, using a Cu‐phthalocyanine (CuPc)‐based COF as an electrocatalyst.^{[ 25 ]} The in situ characterization and DFT calculation reveal the key role of a single CuPc active center and its electronic conditions to generate *CH₃, the vital reactant intermediate in the production of acetate, by circumventing the C–C dimerization pathway to produce C₂₊ products. The partial current density of the catalyst showed ≈11.3 mA cm⁻², indicating that almost all the total current is consumed for acetate formation. The catalyst further showed greater stability for more than 80 h, but at a lower current density (≈12 mA cm⁻²), much lower than the suitable current density for industrial applications. Cu SAs anchored to MXene (Ti₃C₂T _x ) nanosheets exhibited an efficient CO transformation to C₂H₄ with 71% FE at −0.7 V versus RHE due to the favorable Cu‐O₃ active center for the CO dimerization to OC–CO and lower the free energy for C₂H₄ generation, as unveiled by the theoretical study.^{[ 23a ]} Furthermore, Wang et al.^{[ 26 ]} also demonstrated the transformation of CO to C₂₊ products with a total $\mathrm{F}{\mathrm{E}}_{{\mathrm{C}}_{2+}}$ of 74% and acetate being the dominant product with FE ≈50% at −0.8 V versus RHE over Cu SAs anchored to nonstoichiometric carbon nitride support (Cu₁/n‐C₃N₄). According to the kinetic and DFT studies, the rate‐determining step is the hydrogenation of *CO to *CHO, promoting the *CO–*CHO dimerization on the Cu SAC through the Eley–Rideal mechanism, signifying the ability of SACs to transform CO₂ to C₂₊ products based on the rational design of catalysts.

Figure 3.

a) Schematic representation of the fabrication and characterization of a single Cu atom over hydrogenated graphene (Cu‐SA/HGDY), and its CO₂ reduction selectivity and reaction rate to produce CH₄. Reproduced with permission.^{[ 16f ]} Copyright 2023, Wiley‐VCH GmbH. b) Diagrammatic representation of the fabrication process and characterization of thin‐walled nanotube‐shaped TWN‐Cu_13.35‐600‐SACs electrocatalyst, and its electrochemical CO₂ reduction selectivity and reaction rate toward C₂H₅OH formation. Reproduced with permission.^{[ 27a ]} Copyright 2023, American Chemical Society.

The optimal metal loading in single‐atom catalysts is another challenging task that can potentially affect the catalytic performance.^{[ 23} , ^{27 ]} For example, Cu SAs anchored on ultrasmall CeO₂ clusters exhibited greater CO₂ conversion to CH₄ with maximum selectivity (67% FE) and reaction rate (364 mA cm⁻² partial current density) at industrially relevant current density, much higher than that of the catalyst loaded on commercial CeO₂ NPs ($\mathrm{F}{\mathrm{E}}_{\mathrm{C}{\mathrm{H}}_4}$ ≈ 42%), owing to the optimal loading, atomically precise structures, and loading locations on the ultrasmall cluster support.^{[ 27b ]} According to the experimental and theoretical studies, the ultrasmall CeO₂ clusters, ≈2.4 nm, displayed a truncated octahedral shape that offered favorable loading locations at the (100) facets and the edges between two (111) facets through coordination of Cu SAs with four oxygen atoms (Cu–O₄) with oxidation states close to Cu^II, which facilitates the *CO hydrogenation to *CHO instead of C–C coupling. Further increases in Cu SAs loading tend to agglomerate, and the C–C coupling pathway dominates to generate C₂H₄. Although the catalyst showed greater stability at industrially relevant current density, the reaction time was carried out only for ≈3 h, which is not enough to claim long‐term stability. On the other hand, Xia et al. recently synthesized ultrahigh Cu SA loading by a novel silica‐mediated hydrogen‐bonded organic framework (HOF)‐templated approach over thin‐walled N‐doped carbon nanotubes with Cu content reaching up to 13.35 w% (TWN‐Cu_13.35‐600‐SACs) (Figure 3b).^{[ 27a ]} The experimental findings showed that the Cu SAs have an oxidation state between Cu^I and Cu^II coordinated with three N atoms (Cu–N₃) to offer the best optimal intermediate binding energies that eventually promote the C–C dimerization between adsorbed formyl species, CO*, on adjacent Cu–N₃ sites. The catalyst demonstrated an outstanding CO₂ transformation to ethanol with ≈82% FE at −1.1 V versus RHE promoted by the increased adjacent Cu SAs loading (13.35 w%), more than twofold surge as compared to the catalyst with low Cu SAs loading (5.14 w%), indicating the advantage of single atom loading for high‐value product formation. The TWN‐Cu_13.35‐600‐SACs electrocatalyst exhibited a reaction rate (TOF) of 3.458 h⁻¹ and a partial current density of 35.6 mA cm⁻² toward ethanol production. The DFT calculation confirms that the adjacent Cu–N₃ active centers were the real active sites. Their greater loading was detrimental to the high ethanol FE due to the favorable C–C coupling at the adjacent Cu–N₃ active centers. The coordination of two neighboring Cu–N₃ sites synergistically facilitates the C–C coupling to generate ethanol. In addition, it showed excellent operational stability for more than 25 h at −1.1 V versus RHE without performance decay. However, the operational current density is much lower than the industrially relevant current densities, and the electronic and structural stability were not studied after long‐term electrolysis (25 h). Likewise, Zhang et al. prepared high‐density Cu SAs functionalized on tetramino‐benzoquinone (Cu‐TABQ), which exhibited greater activity and selectivity for CO₂RR to C₂ products with 63.2% $\mathrm{F}{\mathrm{E}}_{{\mathrm{C}}_2}$ ($\mathrm{F}{\mathrm{E}}_{{\mathrm{C}}_2}{\mathrm{H}}_4$ = 53.1%, $\mathrm{F}{\mathrm{E}}_{{\mathrm{C}}_2}{\mathrm{H}}_5\mathrm{OH}$ = 10.1%) at −0.97 V versus RHE.^{[ 28 ]} The catalyst exhibited a greater partial current density of ≈115 mA cm⁻² at this applied potential. The interaction between the C═O linker of TABQ and Cu^II was identified as the key contributor to the improved catalytic activity and selectivity. Because the electronic states of Cu^II can be regulated by introducing the C═O electron‐withdrawing component in ligands through d–π conjugation. The DFT study also unveils that the π orbitals of ligands and orbital hybridization of Cu atoms lead to a strong d–π conjugation, which broadens the distribution of the d‐band center and elevates the Cu energy states that reduce the reaction energy barrier of the key reaction steps. In addition, the TABQ linker (C═O group) substantially tailors the d‐band center of anchored Cu^II, which dictates the adsorption energy of the key intermediate, *CO, on the catalyst and promotes the C–C dimerization for C₂ products formation. However, the catalyst showed a rapid C₂₊ selectivity decline after ≈15 h electrolysis time due to the reduction of Cu‐TABQ to Cu⁰ and subsequent aggregation to form Cu NPs. A recent study demonstrated an efficient CO₂ reduction to C₂ products over CoPc‐decorated Cu nanowires (Cu NWs/CoPc) reaching the FE as high as 69.9% at −1.0 V versus RHE as compared to pristine Cu NWs (35.2%).^{[ 29 ]} It showed a maximum $\mathrm{F}{\mathrm{E}}_{{\mathrm{C}}_2}{\mathrm{H}}_4$ of 58.3% and partial current density of 204 mA cm⁻² at −1.4 V versus RHE, indicating a higher selectivity and reaction rate to form C₂H₄. The experimental studies and theoretical calculations revealed that the enhanced product selectivity is due to the electronic interaction between the Cu substrate and CoPc SAs that facilitate the CO₂ reduction intermediates to make bonds with both Co and Cu sites, resulting in a favorable C–C coupling for C₂ products formation. The transformation of CO₂ to C₂ products could be further improved by compactly engineering the SACs (CoPc) with the substrate (Cu NWs) to significantly enhance the electronic interaction at the interphase sites. The long‐term operational stability was measured at industrially suitable current density (≈420 mA cm⁻²) for 20 h reaction time that maintained the C₂₊ product selectivity at ≈70%, indicating greater performance stability. Although the catalyst exhibited stable performance, no structural and electronic properties were examined after the 20 h electrolysis time, making it difficult to claim stability solely based on that performance.

In summary, single atom catalysts can produce not only C₁ products but also have the potential to generate C₂₊ products with greater reaction rates and selectivities by using various suitable host materials, anchoring ligands, and increasing the SAC loading that generates a synergistic effect between adjacent atoms. The strong hybridization between the orbitals of heteroatoms of the supporting linker and orbitals of the SACs can facilitate the adsorption and activation of CO₂ molecules through electron delocalization. In addition, multiatom site catalysts can be generated when the distance between the SACs is further reduced, which facilitates the C–C dimer. However, the stability of the SACs reported so far remains unsatisfactory due to several reasons, including, the formation of NPs and clusters, and mainly the sloughing of active sites from the supporting electrode during extended reaction times. Therefore, the transformation of CO₂ to C₂₊ products can be more favorable when SACs are loaded onto more appropriate host materials, which will be the focus of discussion in the next sections (DACs, tandem SAC–NCs, and SNCCs). In addition, developing a strategy for synthesizing SACs directly on the supporting electrode (e.g., carbon paper, carbon cloth, etc.) would solve the issues of operational stability rather than depositing the ink suspension of the catalysts on the electrode.

2.2. Dual Atom Catalysts

Based on a concept related to that of SAC, the emerging DACs are catalysts composed of two isolated metal atoms/ions that exhibit synergistic effects in catalytic performance.^{[ 14c ]} Compared to SACs, DACs increase the metal loading and, much more importantly, bring about a completely new stereoelectronic configuration toward catalytic activation due to cooperative interaction between the two atoms and with the substrate.^{[ 30 ]} In this type of catalyst, the metal–metal interactions can change the electronic structures and chemical states via electronic rearrangement between adjacent single metal active sites.^{[ 31 ]} The DACs can be classified as homonuclear (the two active metal atoms are the same, M_A–M_A) DACs and heteronuclear (the two metallic active sites are different, M_A–M_B) DACs.^{[ 16a ]} The heteronuclear DACs further show exceptional atomic structure, chemical states, and coordination environment that will result in different influences from that of homonuclear DACs in CO₂RR due to the different adsorption energies of key intermediates on the two active centers.

However, the characterization evidence levels of DACs varied in most studies undertaken for electrochemical CO₂ reduction. For instance, studies provided evidence for the presence of dual active sites for CO₂RR to high‐value products based on X‐ray absorption spectroscopy (XAS), which are of two types: X‐ray absorption near‐edge structure (XANES) to acquire relevant information related to coordination and oxidation state, and extended X‐ray absorption fine structure (EXAFS) to get information about the quantitative data on coordination number, atomic spacing, and atoms concentration,^{[ 32 ]} conventional characterization tools, theoretical calculations,^{[ 33 ]} aberration‐corrected high‐angle annular dark field scanning transmission electron microscopy (HAADF‐STEM) and XAS,^{[ 34 ]} and merely based on HAADF‐STEM image.^{[ 35 ]} Some of the abovementioned works confirmed the presence, distribution, position, chemical states, coordination number, and atomic spacing of DACs. However, the density of dual‐sites, the ratio among heteronuclear dual‐sites, and their catalytic activity, selectivity, and reaction rate at high current density remain inadequately explored. In addition, despite some research efforts made in both homonuclear and heteronuclear DACs, the electrochemical CO₂ transformation to high‐value products using DACs is in its infancy, as discussed below. Therefore, discovering the M_A–M_A(B) interactions in DACs can provide an opportunity to realize the relationships between the catalytic properties and geometric structure in CO₂RR to high‐value products at an atomic level.

2.2.1. Homonuclear DACs

Recent research progress of homonuclear DACs with metal–metal interaction exhibited tremendous efforts to control the local coordination and atomic structural configuration that involve multiple adsorption sites, with promising CO₂ transformation to high‐value reduction products.^{[ 32} , ^{36 ]} For example, an MOF‐based catalyst (PcCu‐Cu‐O) composed of Cu‐based phthalocyanin ligands (PcCu‐(OH)₈ and CuO₄ nodes with square‐planar structure synthesized by solvothermal method showed a CO₂ transformation to C₂H₄ with greater selectivity (FE = 50%) and catalytic activity (J = 7.3 mA cm⁻²) at −1.2 V versus RHE (Figure 4a) in an H‐type cell.^{[ 32a ]} The DFT investigation revealed that the CO₂ to CO adsorption energy is higher on the PcCu component and lower on the CuO₄ unit due to the variation in coordination environments around the metal ions. Thus, the combination of both components facilitates the formation of CO on the CuO₄ unit that migrates to the PcCu unit for hydrogenation to *CHO, which eventually results in a reduced C–C coupling energy barrier. Moreover, the in situ attenuated total reflectance‐Fourier transform infrared (ATR‐FTIR) analysis also showed the generation of the most important intermediates, including *COOH, *COCHO, *CHO, and *CH₂, indicative of the favorable pathway to yield C₂H₄. According to the various ex situ and in situ characterization techniques, the catalyst exhibited no noticeable electronic and structural changes during the 4 h reaction time, whereas this electrolysis time is not long enough to claim stability. In addition, the ethylene partial current density (obtained by multiplying FE by J) is ≈3.7 mA cm⁻² at −1.2 V, suggesting a lower reaction rate to generate ethylene, about half of the total current density. Likewise, Wang et al. prepared a series of single‐chain homomorphic catalyst models, [Cu(4‐XPz)₂] _n . solvent, where Pz refers to pyrazole and X to Br, Cl, I, H, abbreviated Cu‐PzBr, Cu‐PzCl, Cu‐PzI, and Cu‐PzH, respectively, for CO₂RR.^{[ 33a ]} The substituted halogen atoms influence the structural and geometric properties, establishing different synergistic effects among the two Cu active sites, and thus, the product selectivity varies. High FEs of 60% and 52% toward C₂H₄ and CH₄ were obtained at −1.0 V versus RHE from Cu‐PzH and Cu‐PzI with greater current densities of 346.46 and 287.52 mA cm⁻² in a flow cell, respectively, due to the difference in dihedral angle and distance between the two Cu centers that are active (Figure 4b). However, evidence for the geometric structures and coordination microenvironments is based only on conventional techniques and theoretical calculations. In addition, a catalyst with Cl‐bridged Cu‐dimer (Cu₂) offered greater catalytic activity and selectivity toward C₂H₄ formation relative to its single‐atom counterpart due to the favorable site formation for C–C coupling.^{[ 32b ]} However, the selectivity toward C₂H₄ remains lower than those of many nanocatalyst counterparts due to the competing HER. The selectivity could be further enhanced by adjusting the distance between the two Cu active centers. In addition, the long‐term stability test for the catalyst showed a slight decline in current density and a mild fluctuation in C₂H₄ FE even within 5 h, indicating that the catalyst is not highly stable.

Figure 4.

a) Representation of the structure of metal–organic frameworks composed of CuO₄ square‐planar nodes and octahydroxyphthalocyanin) Cu^II (PcCu‐(OH)₈) ligands, abbreviated as PcCu‐Cu‐O), and their electrochemical CO₂ reduction performance to high‐value products along with the in situ ATR‐FTIR measurement during the CO₂RR. Reproduced with permission.^{[ 32a ]} Copyright 2021, American Chemical Society. b) Chain structure of Cu‐PzX (where X = H, Cl, Br, I) catalysts used for CO₂RR in a flow cell and FEs for C₂H₄ and CH₄ on Cu‐PzX electrocatalysts at different potentials. Reproduced with permission.^{[ 33a ]} Copyright 2021, Wiley‐VCH GmbH. c) The structure of Cu(I) triazolate frameworks functionalized with dinuclear copper with different sizes of side ligands for stable, efficient, and tunable CO₂ reduction to C₂H₄/CH₄ at −1.3 V versus RHE. Reproduced with permission.^{[ 33b ]} Copyright 2022, Wiley‐VCH GmbH. d) Schematic representation of a structural layer and CO₂RR to CH₄ on the catalyst [Cu₂(O‐o‐bpy)₂], in which HO‐o‐bpy = 1H‐[2,2′]bipyridinyl‐6‐one), and show higher selectivity and catalytic activity to CH₄ at −1.4 V versus RHE. Reproduced with permission.^{[ 34a ]} Copyright 2023, American Chemical Society.

In addition, a study by the Zhang group demonstrated selectivity control of CH₄ and C₂H₄ by altering the size of side ligands on Cu(I) triazolate frameworks (Figure 4c).^{[ 33b ]} The selectivity ratio of C₂H₄/CH₄ was tailored gradually and reversed from 11.8:1 to 1:2.6 by replacing the small‐sized ligand with a larger‐side ligand, providing the highest C₂H₄ and CH₄ products selectivities of 51% and 56%, current densities of ≈10.9 and 11.2 mA cm⁻², and formation rates of 48.2 and 81.1 µmol m⁻² s⁻¹ at −1.3 and −1.5 V versus RHE, respectively. The current densities and hydrocarbon product selectivity declined after 8 h of electrolysis. Although the authors claimed no noticeable structural and electronic changes after the stability test, the conventional characterization techniques may not be suitable for scrutinizing the structural evolution of DACs. A study was conducted on a di‐copper complex [Cu₂(O‐o‐bpy)₂], in which each tetrahedral Cu(I) atom is coordinated by the two N atoms of an o‐bpy ligand, this bipy being substituted by an ortho oxo group, and by the other Cu atom and its bipyridinyl oxo group; this ligand is derived from HO‐o‐bpy = 1H‐[2,2′]bipyridinyl‐6‐one). The CO₂RR on this complex provided a greater FE of 82% to CH₄ with ≈90 mA cm⁻² current density at −1.4 V versus RHE (Figure 4d).^{[ 34a ]} Remarkably, this electrocatalyst exhibited a greater reaction rate toward CH₄ with a partial current density of 73.8 mA cm⁻² and a formation rate of 956 µmol m⁻² s⁻¹, set among the top previously reported catalysts. Compared to single Cu atoms and completely exposed dicopper (I) sites, the confined catalyst exhibited a stronger affinity toward C₁ intermediates and facilitated hydrogenation to yield CH₄ by suppressing the C–C coupling. The electrochemical durability of this catalyst was also tested at −1.4 V versus RHE for 100 h, which showed a slight decline in selectivity and catalytic activity. According to the conventional characterization techniques and HAADF‐STEM, XANES, and EXAFS analysis, no deceptive changes were observed after long‐term electrolysis, indicating greater stability of the active site. The high catalytic stability may be due to its confinement within the layered space, and the authors performed long‐term stability of the control experiments to identify the main reason for the greater stability. Another study unveiled the effect of distance between two adjacent Cu atoms through Cu‐doping by pyrolysis at different temperatures and recognized that the close distance between neighboring Cu–N _x species is suitable for C–C coupling and C₂H₄ production, whereas the large distance between neighboring Cu–N _x exhibited high CH₄ formation.^{[ 36a ]} The DFT calculations further supported the formation of C₂H₄ through binding two CO reactant intermediates on the two neighboring Cu–N _x active centers, whereas the single Cu–N _x species facilitated CH₄ formation.

Thus, in homonuclear DACs, the different species surrounding the Cu center affect the activity and selectivity of products. In addition, the synergy effect between vicinal active centers bestowed by the difference in coordination environment of the distance and angle potentially influences the C–C coupling. The local coordination environment disparity and confinement effect further adjust the binding energy of important reaction intermediates, and thus, alter the final product. Therefore, modulation of electronic effects (electronic structure, charge transfer), optimization of geometric effects (coordination environment, adsorption configuration, and interatomic distance), and the choice of supporting materials are crucial to rationally design homonuclear DACs to fulfill the limitation of SACs and expand the utilization of SACs in tandem for efficient CO₂RR to high‐value products.

2.2.2. Heteronuclear DACs

In comparison with homonuclear DACs, DACs that are heteronuclear with two different neighboring active metals that contain M_A–M_B bonds influence the adsorption of key intermediates differently due to the difference in electronic properties of the two different metal atoms. The heteronuclear DACs sometimes also present isolated M_A and M_B without interaction between each other, but contribute a synergistic effect during the catalytic reaction.^{[ 16} , ^{37 ]} The existence of heterometal atoms interaction occurring inside the active metal centers diminishes the energy barrier of crucial steps and affects the selectivity of CO₂ reduction products. To date, a few works have reported Cu‐based heteronuclear DACs for CO₂/CORR to high‐value products.^{[ 34} , ^{35 ]} For instance, the Chen group recently demonstrated efficient heteronuclear DACs consisting of Cu and Sn atoms prepared on a novel metal–organic framework (CuSn‐HAB, HAB‐hexaiminobenzene) via a solvothermal method for CO₂ reduction to C₂H₅OH with about 56% FE and 68 mA cm⁻² current density at −0.57 V versus RHE.^{[ 34b ]} In addition, this catalyst provides greater reaction rates toward C₂H₅OH formation with a partial current density of >38 mA cm⁻² and a TOF of 0.0601 s⁻¹. In this catalyst, the four nitrogen atoms were coordinated to Cu (CuN₄), while the Sn atoms were bonded with two oxygen and two nitrogen atoms (SnN₂O₂) that exhibited greater oxygen affinity. Thus, the SnN₂O₂ site plays a key role in promoting the generation of the crucial intermediate, *OCH₂, and the CuN₄ site for *CO, leading to the *CO–*OCH₂ dimerization for ethanol formation (Figure 5a). However, the transformation of CO₂ to C₂H₅OH decreased from ≈56% at −0.57 V versus RHE (≈90 mA cm⁻²) to ≈10% at −0.87 V versus RHE (≈225 mA cm⁻²), indicating that the catalyst failed to realize the ethanol formation at industrially relevant current density. The performance of the catalyst slightly declined after a long‐term operation (35 h) at the best applied potential. The detailed postcharacterization analysis, however, indicated no morphological, structural, or oxidation state of the catalyst changes after the long‐term CO₂RR. Thus, the slight decrease in current density and FE could be due to the sloughing of active sites from the substrate.

Figure 5.

a) Schematic representation of the structures, HAADF‐STEM, EDS mapping, and ethanol FE of heteronuclear dual atom CuSn‐HAB catalyst. Reproduced with permission.^{[ 34b ]} Copyright 2023, American Chemical Society. b) TEM, EDS‐mapping, and HAADF‐STEM images of heteronuclear CuIn DACs and the FEs for electrochemical CO₂ conversion to methanol and other valuable products. Reproduced with permission.^{[ 35 ]} Copyright 2023, Elsevier B.V.

In addition, Cu–In dual atom catalysts were prepared by MOF‐derived approach for the electrocatalytic synthesis of highly reduced C₁ product (methanol) from CO₂ with 55.63% FE at −0.62 V versus RHE (Figure 5b).^{[ 35 ]} Benefiting from the synergistic effect between In and Cu, the dual Cu–In atomic catalysts showed ≈5.6 and 3.9 times higher selectivity and yield rate than nontandem Cu single atom catalysts toward methanol by suppressing the HER process, respectively. However, this study lacks detailed experimental studies and theoretical calculations. For example, no attempts were made to analyze which active site is responsible for stabilizing the key intermediate to provide methanol, and no long‐term operational test was carried out. Li et al. prepared DACs with two neighboring Cu–Ni or Cu–Cu atoms attached to N‐doped carbon frameworks for comparison between heteronuclear and homonuclear DACs for CO electroreduction to high‐value products.^{[ 34c ]} Both the Cu–Cu and Cu–Ni dual atomic catalysts exhibited different CO reduction selectivities owing to the synergy effect between neighboring metal active centers. The Cu–Cu DACs offer an effective CO transformation to C₂₊ products with $\mathrm{F}{\mathrm{E}}_{{\mathrm{C}}_{2+}}$ of ≈91% and partial current densities of 32 and 33 mA cm⁻² for ethylene and acetate at −1.66 V versus RHE, respectively. On the contrary, the Cu–Ni ones provide higher selectivity to CH₄ formation with FE as high as 53% at the same applied potential. The DFT calculations reveal that the dual Cu active centers enable the conversion of two molecules of CO to ethylene and acetate through suitable C–C coupling on both Cu atomic sites. However, after the replacement of a Cu atom by a Ni atom, the CO adsorption energy barrier became too strong, and only a single Cu atom was able to act as an active center for the favorable electroreduction of CO to CH₄. Although the authors claimed that the structural properties of the catalysts remained unchanged after the reaction based on X‐ray photoelectron spectroscopy (XPS) and XRD analysis, no clear information was provided related to reaction time and applied potential. In addition, replacing the Ni atom by less CO binding affinity elements, such as Ag, or Zn, could further modulate the product selectivity of heteronuclear DACs. Furthermore, a theoretical study demonstrated that CuMn/C₂N and CuCr/C₂N have outstanding activities for CO₂RR to CH₄ at low onset potentials,^{[ 38 ]} but experimental investigations are lacking.

The dual single‐atom electronic structures are beneficial in keeping the merits of single atoms and in promoting cooperation between adjacent metal sites. Thus, the synergistic effects of the dual hetero single‐atoms can potentially help to drive the catalytic CO₂/CO reduction reactions with remarkably different activity and selectivity compared to the dual homo single‐atoms and nontandem catalysts. So far, more reports have appeared on homonuclear DACs for CO₂ reduction with greater selectivity toward C₂₊ products relative to heteronuclear DACs. The studies on heteronuclear DACs for electrochemical CO₂ reduction to high‐value products are insufficient. Therefore, tremendous efforts are required to rationally design Cu‐based DACs for CO₂RR to high‐value products via compensating the limitation of SACs. The introduction of extra metal atoms adjacent to the SACs to form DACs enables owning the legacy of maximum atom utilization, atomic dispersity, superior selectivity and catalytic activity, and breaking the intrinsic properties of the SACs. Understanding and tailoring the electronic properties of the heteronuclear DACs through proper selection and arrangement of the two atoms makes it possible to modify the adsorption, activation, and subsequent transformation of CO₂ to high‐value products. In addition, optimizing the coordination environment and the distance between the two atoms can influence the adsorption configuration, activation, and reaction kinetics of CO₂. Overall, careful design and optimization of catalysts are crucial during the DAC preparation.

2.3. Tandem Single‐Atom Catalyst–Nanocatalyst

A tandem SAC–NC is a catalyst consisting of both an SAC and a nanocatalyst (nanocluster or nanoparticle, NP) involved in the catalytic process. This system combines the advantages of both a SAC dispersed on a suitable support and a nanocatalyst present on the same support material. The tandem SAC–NC can display dissimilar catalytic mechanisms and different product selectivities compared to those observed with the nontandem counterparts, and they have recently attracted increased attention. Like DACs, the combination of catalysts in SAC–NC is categorized as homonuclear or heteronuclear. In both cases, the single atoms in the SAC and the NC have different electronic and geometric properties, enabling them to adsorb different kinds of reactants and key intermediates in CO₂RR and synergistically lead to high‐value products.^{[ 39 ]} For example, the Sargent group employed a tandem catalyst consisting of CoPc for CO₂ to CO reduction and Cu NC (CoPc@HC/Cu) for the subsequent conversion of CO to C₂₊ products with greater selectivity ($\mathrm{F}{\mathrm{E}}_{{\mathrm{C}}_2}{\mathrm{H}}_4$ = 61%, $\mathrm{F}{\mathrm{E}}_{{\mathrm{C}}_{2+}}$ = 82%) at industrially relevant current density (800 mA cm⁻²).^{[ 39d ]} The CoPc@HC/Cu electrocatalyst offered partial current densities of ≈488 and 656 mA cm⁻² for C₂H₄ and C₂₊ production, respectively. The high selectivity and production rate were realized due to the enhanced local CO accessibility to promote the C–C coupling on the second catalyst (Cu NC). In this system, two crucial steps are involved: CO₂ to CO in the first step (CoPc) and subsequent conversion of CO to C₂₊ products in the second step (Cu NPs). This system, therefore, requires careful design for efficient performance on both steps, and the distance between the two active sites is key to controlling the mass transport limit of CO to Cu NPs for greater C₂₊ selectivity and activity. The tandem catalyst maintained the C₂H₄ product selectivity for 16 h at industrially relevant current density (800 mA cm⁻²), shedding light on the possible scalability of tandem catalysts for CO₂ reduction to high‐value products. However, this electrolysis time is too short, and the structural and electronic properties of the catalyst after the reaction time were not examined to understand the operational stability. Likewise, Meng et al. prepared a porphyrinic triazine framework (PTF)‐(Ni)/Cu tandem heteronuclear SAC–NC through uniform dispersion of Cu NPs onto the PTF bonded to atomically dispersed Ni–N centers for an improved electrocatalytic transformation of CO₂ to C₂H₄ at −1.1 V versus RHE with 57.3% FE, sixfold more as compared to PTF/Cu, the nontandem catalyst, that formed CH₄ as the main product (Figure 6a,b)^{[ 39b ]} The tandem catalyst offered a partial current density of 3.1 mA cm⁻² with well‐maintained selectivity and catalytic activity of C₂H₄ for over 11 h. The DFT calculations and operando characterizations revealed that the generated CO from PTF(Ni) induced C–C dimerization on the nearby Cu NPs, promoting the formation of C₂H₄ as the main product. In addition, a recent study demonstrated an efficient CO₂ conversion to C₂ products with 78.2% FE and partial current density of 90.8 mA cm⁻² at −1.5 V versus RHE by stabilizing the Cu^δ+ sites in Pd^δ−–Cu₃N electrocatalyst that contains a charge‐separated Pd^δ−–Cu^δ+ atom pair. This catalyst showed a 15‐fold increase in performance toward C₂ product formation compared to the nontandem catalyst (Cu₃N) (Figure 6c,d).^{[ 39a ]} The operando characterizations and DFT calculations reveal that the Pd^δ− site exhibits favorable CO binding energy with the nearby Cu^δ+ sites, synergistically promoting the C–C dimerization to C₂ products. However, the long‐term stability test of the catalyst to scrutinize the structural and electronic properties with product selectivity and reaction rate remained unexplored. Moreover, Hu et al.^{[ 40 ]} developed a tandem Ni SA‐encapsulated Cu NPs catalyst for an efficient CO₂ transformation to acetate with ≈45% FE at −0.5 V versus RHE in an H‐cell. The selectivity to acetate further improved to 60% at high current densities of 50 and 200 mA cm⁻² and >80% to C₂ (acetate and ethanol) at 75 and 200 mA cm⁻² in a flow cell. The formation rate of acetate and C₂ products reached ≈60 and 120 nmol s⁻¹ mg_Cu ⁻¹ at 200 and 150 mA cm⁻², respectively. The catalyst exhibited good stability for 10 h electrolysis time at 50 mA cm⁻², maintaining the selectivity of acetate >80% of its original activity. However, no evidence of the structural and electronic properties after the stability test was provided. According to the operando and DFT studies, the generated CO on Ni SA sites migrates to Cu NPs sites for further conversion to acetate through C–C dimerization. Therefore, in tandem SAC–NCs system, the activity and selectivity toward C₂₊ products could be boosted by overcoming the transport limit of CO from the generated site (e.g., Ni SAs) to the feeding site (e.g., Cu NPs) for suitable C–C coupling. The distance between the SACs and NCs in the tandem catalyst could play a key role in beating the mass‐transport limit.

Figure 6.

a,b) Synthesis of Cu NPs on the porphyrinic triazine framework attached with atomically dispersed Ni–N centers (PTF(Ni)/Cu) and CO₂RR performance comparisons between (PTF(Ni)/Cu) and (PTF/Cu) materials. Reproduced with permission.^{[ 39b ]} Copyright 2021, Wiley‐VCH GmbH. c,d) Schematics for the fabrication and characterization of Pd^δ−–Cu₃N and comparison of CO₂RR performance between Pd^δ−–Cu₃N and Cu₃N catalysts. Reproduced with permission.^{[ 39a ]} Copyright 2023, American Chemical Society.

On the other hand, an atomic homonuclear Cu dimer constructed in a Cu‐based MOF through a liquid strategy exhibits greater selectivity to C₂ products with 71% FE at an ampere level current density (0.9 A cm⁻²).^{[ 41 ]} This catalyst realized partial current densities of 388.8 and 639.0 mA cm⁻² toward C₂H₄ and C₂ (C₂H₄ and C₂H₅OH) product formation, respectively, indicating a greater reaction rate to produce high‐value products. The synchrotron studies and DFT calculations indicate that the O₃Cu₁···Cu₂O₂ dimers are key active sites with O₃···Cu₁ for acetate and Cu₂···O₂ for C₂H₄ formation due to the difference in charge density between the two sites that lead to different electron transfer capability from the active centers to *CO, resulting in a different CO–CO dimerization. The catalyst exhibited good operational stability under high current density (300 mA cm⁻²) for 24 h with a trend of structural reconstruction. However, detailed structural and electronic properties after the long‐term electrolysis at high current density remained uninvestigated. Similarly, the Han group demonstrated tandem moderately dispersed atomic Cu sites and Cu NPs (M‐Cu₁/Cu NPs) over an N‐doped carbon matrix for efficient CO₂ conversion to C₂₊ products with $\mathrm{F}{\mathrm{E}}_{{\mathrm{C}}_{2+}}$ of ≈75.4% and a partial current density of 289.2 mA cm⁻² at −0.6 V versus RHE.^{[ 42 ]} It also showed outstanding stability with a selectivity that maintained >70% to C₂₊ products at high current density (400 mA cm⁻²) for 40 h continuous electrolysis time. The electronic and structural postcharacterization analysis indicated no apparent morphological, chemical, or structural changes, suggesting that SAC–NCs can be promising candidates for industrial applications. The experimental and theoretical investigations unveiled that the atomically dispersed Cu centers enhance the H₂O dissociation to generate *H, while the Cu NPs promote the C–C dimerization step, i.e., the formed *H diffuses to Cu NP sites and regulate the coverage of *H on Cu NPs, facilitating the *CO hydrogenation to *CHO and subsequent coupling of *CHO results in the formation of the key intermediate, *OHCCHO*, for C₂H₄ and C₂H₅OH formation.

Therefore, the electronic interaction between the M_A site of SAC and the NC site through the regulation of electronic structures can improve the catalytic activity for CO₂ reduction by promoting electron transfer and facilitating the activation of CO₂. Additionally, the electronic configuration of the tandem catalyst can regulate the stability and reactivity of reaction intermediates involved in CO₂ conversion. The reaction pathways can be tailored by controlling the electronic states, forming desired products at a high rate. The geometric arrangement (the distance and bond angle between the SAs and NCs) of the active sites in the tandem catalysts can also significantly influence the adsorption configuration and accessibility of CO₂ and key intermediates. Thus, optimal spatial arrangement is required to boost the CO₂ reduction reaction rate and product selectivity. Moreover, in the tandem SAC–NCs, it is fascinating to increase the SA loading near the NCs or anchor DACs and NCs adjacently on appropriate support to significantly influence the coordination environments and electronic structures that potentially affect the binding energetics of reactant molecules and crucial intermediate species toward the formation of high‐value products.

2.4. Sub‐Nanocluster Catalysts

The structure‐activity relationships of SNCCs comprising a few atoms (typically <10 metal atoms) have multiple active sites distributed across the host material that make them distinct from those of their nanoparticle counterparts. In addition, the reactivities of atomic nanoclusters highly rely on the hosting materials or the ligands neighboring them due to the interaction between the porous environment and the catalytic materials through modulating the geometric structures and electronic properties of catalysts.^{[ 43 ]} The interaction becomes more profound when the size of the SNCC matches the cavity or is finely functionalized with the support. Thus, its geometric structure can be firmly attached to the porous material or display an easy manipulation in the void space that can potentially affect the adsorption energy of important intermediate species and the CO dimerization barrier. This latter step is crucial for producing high‐value products under these reaction circumstances.^{[ 44 ]}

The host material significantly affects the CO₂RR performance of SNCCs.^{[ 8} , ⁴⁴ , ^{45 ]} For example, recently, Wang et al. designed and fabricated acid and alkaline‐resistant Inz‐Cu₃ (Inz‐1H‐Indazole) SNCC that can switch the symmetry structure by changing the angle and distance between neighboring Cu active sites.^{[ 45e ]} It exhibited greater CO₂ reduction selectivity to C₂ (ethanol and ethylene) products with FEs of ≈42.2% and 66.79% in acidic (pH = 2, at −320 mA cm⁻² current density) and alkaline (1 m KOH, at −0.9 V vs RHE), respectively. The catalyst exhibited greater C₂ partial current densities of 135 and 300 mA cm⁻² in acidic and alkaline conditions, respectively, suggesting high reaction rates to generate C₂ products. The experimental studies and DFT calculations revealed that the neighboring asymmetric Cu active centers with proximity play the key role in stabilizing the important intermediate, *CHOHCH₃, enhancing the selective generation of asymmetric C₂ products. The Inz‐Cu₃ electrocatalyst also exhibited better selectivity to C₂ product with slight fluctuations in both acidic and alkaline electrolytes at industrially suitable current densities for more than 2 and 7 h of electrolysis time, respectively. Although the catalyst exhibited stable structural, chemical, and electronic properties, the activity showed significant fluctuations during these few hours of electrolysis. In addition, Cu electrocatalysts prepared by an amalgamated Cu–Li approach on carbon support (Figure 7a,b) exhibit an outstanding electrochemical CO₂ conversion to ethanol at −0.7 V versus RHE (FE = 91%), which is a benchmark toward a single oxygenated product on Cu‐based catalysts.^{[ 45a ]} The catalyst offered a current density of ≈1.2 mA cm⁻² and a TOF of 0.035 s⁻¹ at −0.7 V versus RHE, indicating a greater reaction rate even at a low current density. This is due to the reversible transformation of isolated Cu atoms to atomically defined Cu _n sub‐nanoclusters (n = 3 and 4) under the electrochemical reaction circumstances that favorably allow the C–C coupling to yield ethanol. Before the CO₂RR commencement, four oxygen atoms from water and hydroxyl group are anchored on the Cu ion that is subsequently reduced to metallic Cu (Cu⁰) through the transfer of electrons from the carbon support and form Cu₄ or Cu₃ SNCCs by amalgamating with nearby atomic Cu species. The Cu species are then connected by surface hydroxide groups, acting as the main active center for CO₂ adsorption from the solution and complete the CO₂ transformation to ethanol by multiple proton and electron transfers. In addition, the catalyst exhibited no apparent decline in catalytic activity and selectivity toward ethanol formation for over 16 h at −0.7 V versus RHE. However, it offered a very low current density of ≈1.2 mA cm⁻² at −0.7 V versus RHE, suggesting lower catalytic activity. Recently, an asymmetric low‐frequency pulse strategy (ALPS) has demonstrated a controllable selectivity of CO₂RR products using trinuclear Cu‐dimethyl pyrazole complex, Cu₃(DMPz)₃, through modulating the Cu oxidation states in the Cu sub‐nanoclusters.^{[ 45b ]} Two different ALPS profile regions were designed to produce CH₄ and C₂H₄ controllably, and both profiles were compared with the potentiostatic technique (Figure 7c). In the ALPS‐1 profile, anodic (E _a) and cathodic (E _c) potentials of 1.27 and −1.28 V versus RHE for 30 and 300 s electrolysis time, respectively, obtained a high product selectivity of CH₄ with 80.3% FE. Interestingly, under the ALPS‐2 profile of $E_{{\mathrm{C}}_1}$ = −0.58 V versus RHE for a duration of 30 s followed by 600 s reaction time at $E_{{\mathrm{C}}_2}$ = −1.08 V versus RHE, and switch to an E _a = 0.42 V versus RHE for 30 s, electrolysis provides a different product, C₂H₄, with 70.7% FE. According to the analysis, the disparity in product selectivity between the two conditions was due to the formation of various Cu oxidation states in Cu‐based SNCCs, with favorable Cu (I and II) in the ALPS‐1 condition to achieve CH₄, and Cu (0 and I) for the later to obtain C₂H₄ as a target product. These product selectivities are significantly higher than the potentiostatic condition that offered FEs of only 34.5% and 5.9% toward C₂H₄ and CH₄, respectively. In addition, the ALPS strategies exhibited greater operational stability (ALPS‐1 for 300 h, ALPS‐2 for 145 h) than the conventional potentiostatic condition (less than 1 h). The structural analysis revealed that the catalyst has reconstructed into well‐dispersed Cu‐based nanoclusters after the long‐term pulsed potential electrolysis, promoting catalytic stability by preventing cluster aggregation in the ALPS potential profiles. However, the relative product formation rates obtained from the ALPS and potentiostatic techniques are overlocked, limiting the versatile convenience of the ALPS method over the potentiostatic one.

Figure 7.

a,b) Synthesis of Cu SA on carbon–support and its electrochemical CO₂ reduction performance. Reproduced with permission.^{[ 45a ]} Copyright 2020, Springer Nature. c) Design and CO₂RR performance of Cu–dimethyl pyrazole complex, Cu₃(DMPz)₃, catalyst using asymmetric low‐frequency pulse frequency (ALPS) in different pulse durations and potensiostatic techniques. Reproduced with permission.^{[ 45b ]} Copyright 2023, American Chemical Society. d,e) Structures of [Cu₈] SNCC and the unit cell in NNU‐33(H) and NNU‐33(S), CO₂RR performance, in situ Raman before and under CO₂ reduction electrolysis potentials, and schematics of the NNU‐33(H) without and with the boosted cuprophilic interaction for CO₂RR to methane. Reproduced with permission.^{[ 46 ]} Copyright 2021, American Chemical Society.

Additionally, as indicated in Figure 7d,e, Zhang et al. prepared the Cu(I)‐based coordination polymer (NNU‐33(S) and NNU‐32) electrocatalysts for an efficient transformation of CO₂ to CH₄ owing to the intrinsic intramolecular cuprophilic interactions.^{[ 46 ]} The replacement of sulfate radicals with hydroxyl radicals under the electrolysis process leads to a dynamic in situ reconstruction from NNU‐33(S) to NNU‐33(H), which eventually resulted in high CH₄ selectivity (FE = 82%) at −0.9 V versus RHE due to the enhanced cuprophilic interaction inside the catalyst structure. In addition, this electrocatalyst exhibited a higher partial current density of ≈322 mA cm⁻² at −0.9 V versus RHE toward CH₄ formation due to the cuprophilic interaction that positively impacts the efficiency, catalytic activity, and reaction rate. According to the in situ investigations, the Cu(I) species in NNU‐33(H) remained unchanged, suggesting that the Cu(I) active sites were neither reduced to Cu(0) nor changed into oxide species. The catalyst showed stable selectivity and catalytic activity for 5 h electrolysis time, maintaining the $\mathrm{F}{\mathrm{E}}_{\mathrm{C}{\mathrm{H}}_4}$ and total current density at 75% and 350 mA cm⁻², respectively. The detailed characterization analysis suggested no variation in the structural and electronic properties of the catalyst before and after the stability test, while this electrolysis time is not enough (only 5 h) to claim long‐term operational stability. Likewise, a study unveiled the high CH₄ formation selectivity (FE = 66.6% ± 3.2%) and a current density of 21.7 mA cm⁻² at −1.55 V versus RHE from SnCu_1.5O_3.5 nanocluster encapsulated in MFI zeolite due to the generation of *CO intermediates in the supported zeolite channels, which provides a favorable condition to undergo a multistep protonation reaction further.^{[ 45c ]} The catalyst exhibited good catalytic performance for only 25 h and showed a significant fall in CO₂RR performance, when the electrolysis time exceeded this period due to the morphological changes.

The most important aspect to recall here is that different substrates on which the metallic SNCCs are supported are influenced by the structure type that can be selected in terms of pore sizes to match the catalyst, electronic properties to affect the binding energy of important intermediates, and chemical compositions that enable to anchor the active catalysts functionally. The composition and arrangement of atoms within the nanocluster can create active sites with specific electronic configurations that promote an efficient CO₂ conversion toward high‐value products by promoting the adsorption and activation of CO₂. Similarly, the arrangement of atoms can also create a cooperative interaction between adjacent metal atoms within the SNCC that influences the accessibility of active sites, adsorption energy of CO₂ molecules and key reaction intermediates, and subsequent product distribution. By and large, SNCCs can provide solutions for SACs and/or DACs related to low‐metal loading, stability specifically at industrially relevant current densities, and the absence of more ensemble active centers. Consequently, the rational fabrication of SNCCs with a well‐controlled number of atoms on a suitable support, tailoring the electronic structure, and optimizing the ensemble size and arrangement is crucial for an efficient transformation of CO₂ into high‐value products.

The experimental studies conducted with Cu‐based SACs, DACs, SAC–NCs, and SNCCs as electrocatalysts for CO₂RR to highly reduced C₁ (CH₄ and CH₃OH) and C₂₊ (C₂H₄, C₂H₅OH, and C₃H₇OH) products have been reviewed and the summary is presented in Table 2. The selectivity, reaction rate, and stability are among the key parameters for the scalability of the electrochemical CO₂ reduction technology. According to Table 2, the following trends can be drawn regarding selectivity, reaction rate, and stability.

At first, the trend shows that the selectivity in SAC systems is about twofold higher for C₁ products, mainly CH₄ than C₂ products, in a wide potential range, with the best selectivity reaching 81% at −1.5 V versus RHE. However, only one case reached a high C₂H₅OH selectivity of 81.9% at −1.1 V versus RHE. The transformation of CO₂ to C₂₊ products is realized with SACs by increasing the catalyst loading to minimize the proximity between adjacent active sites and promote the C–C coupling toward C₂₊ products. The trend of selectivity to C₂₊ products is higher than C₁ products on DACs with the highest FEs of 60% to C₂H₄ at −1.0 V versus RHE due to the modified structural microenvironments caused by the two metal nuclei that facilitate the C–C dimerization. However, a maximum CH₄ selectivity of 82% has been recorded from the DAC system because of the confinement effect. Interestingly, the selectivity toward C₂₊ products overtakes by 13 times that of C₁ products for the SAC–NC systems because of the presence of multinuclear sites: one SA site and another ensemble site residing neighboring the SAC site, together promoting the C–C coupling through the synergistic or tandem effect and offering the highest selectivity of 72.3% to C₂H₄ at −1.1 V versus RHE. Similarly, the trend in C₂₊ product selectivity observed from SNCCs is about 2.5‐fold higher than that for C₁ products with the maximum C₂H₅OH selectivity of 91% at a relatively lower potential of −0.7 V versus RHE. The SNCCs can also provide the highest CH₄ selectivity (92%), suggesting that this system depends on the electronic properties and chemical compositions. Interestingly, the selectivity toward a single high‐value product remains higher with these sub‐nanometric catalysts than with the nanoparticle counterparts. The results, generally, possess the CO₂ reduction selectively trend to C₁, e.g., CH₄ with SACs, and C₂₊, e.g., C₂H₄, with DACs, SAC–NCs, and SNCCs with a much higher tendency to C₂₊ product selectivity with SAC–NC systems, suggesting greater possibility for scaling up the CO₂RR technology.

Moreover, the reaction rates expressed in partial current density and TOF and the operational stability also showed disparities among the different sub‐nanometric level catalysts. Only a few studies on SACs, DACs, and SNCCs provide the reaction rates in partial current density and TOF, whereas most such studies offer only the partial current density as the reaction rate parameter to estimate the formation rate of a product. According to Table 2, the highest reaction rate observed among the high‐value CO₂ reduction products is CH₄ recorded from SAC with partial current density and TOF of 340.2 mA cm⁻² and 16.3 s⁻¹, respectively. Many CO₂RRs on these sub‐nanometric catalysts have been carried out at industrially relevant current densities that offered greater partial current densities and selectivities. However, the long‐term operational stability study on the sub‐nanometric catalysts for high‐value product formation at industrially relevant current density is in its infancy, mostly within a few hours of electrolysis, compared to the nanoparticle counterparts.^{[ 47 ]} Although the higher selectivity, greater reaction rate, suitability of being subordinated to renewable energy sources, and flexible electrolyzer design shed light on the possible scale, the lack of long‐term operational stability makes these sub‐nanometric catalysts incompatible with practical application. Thus, the aim of solving the stability issues should get a lot of attention, like the catalyst activity and selectivity, in order to attain large applications.

3. Non‐Cu‐Based SACs, DACs, Tandem SAC–NCs, and SNCCs

This section thoroughly explores experimentally reported non‐Cu‐based sub‐nanometric (SACs, DACs, tandem SAC–NCs, and SNCCs) electrocatalysts for CO₂RR to high‐value products. Among the transition metal‐based electrocatalysts reported so far for CO₂RR, only Cu showed a fairly high selectivity toward the formation of high‐value products, whereas most others exhibited CO and/or formate production. However, metallic Cu exhibited broad product selectivity headed over 16 various CO₂ reduction products, among which the 12 products were C₂ or C₃, suggesting the intricate nature of CO₂ reduction on this type of catalyst.^{[ 84 ]} The moderate adsorption of CO intermediate on the Cu surface promotes the generation of the C─C bond or *CHO through CO coupling, CO–CHO dimerization, or hydrogenation.^{[ 85 ]} Up to now, the regulation of C─C bond generation remains a challenging task due to the limited understanding of the C─C dimerization process over Cu‐based electrocatalysts,^{[ 86 ]} which has restricted the growth of highly selective electrocatalysts for the transformation of CO₂ to high‐value products. Therefore, more efforts must be devoted to developing Cu‐free electrocatalysts for an alternative C–C coupling platform and providing an approach for manipulating CO₂ conversion pathways to high‐value products.

Remarkably, downsizing the structures of non‐Cu‐based transition‐metal catalysts to sub‐nanometric dimensions allows the improvement of the hydrogenation activity and boosts the selectivity of CO₂RR to high‐value products.^{[ 87 ]} For example, a recent study demonstrated an efficient transformation of CO₂ to acetate using dispersed Sn atoms supported on carbon (Sn/C‐0.12) with 90% FE at −0.6 V versus RHE, which is among the best catalysts of both Cu‐ and non‐Cu‐based for CO₂ reduction to acetate.^{[ 87d ]} The catalyst exhibited an acetate partial current density of 0.66 mA cm⁻² at this applied potential, and the reaction rate increases with increasing the applied potential, but the main product switches from acetate to formate formation. The catalyst sustained the catalytic activity at different applied potentials, and the acetate selectivity remained nearly constant for 26 h reaction at −0.6 V versus RHE, indicating the greater operational stability of the catalyst. However, the high selectivity toward acetate is realized only at relatively low current density, making it challenging to enhance the formation rate of acetate at industrially suitable current densities. In addition, Han et al. prepared Zn SAs attached on microporous N‐doped carbon (SA‐Zn/MNC) toward efficient and selective CO₂ transformation to CH₄.^{[ 87a ]} This non‐Cu transition SA catalyst offered the respective high FE, partial current density, and yield rate of 85%, 31.8 mA cm⁻², and 158 ± 4 µmol h⁻¹ cm⁻² at −1.8 V versus SCE to CH₄, indicating much better performance compared with most Cu‐based catalysts for CO₂RR to CH₄ (Figure 8a,b). The DFT investigation showed that Zn SAs promoted CH₄ production by facilitating CO hydrogenation through an oxygen‐binding intermediate (*OCHO). The catalyst showed long‐term operational stability for 35 h at −1.8 versus SCE with the respective FE and partial current density of 84% and 33.9 mA cm⁻², indicating almost steady selectivity and reaction rate throughout the electrolysis time. Similarly, Wu et al. demonstrated that a cobalt phthalocyanine (CoPc) immobilized on carbon nanotubes catalyzed the transformation of CO₂ to methanol, exhibiting an appreciable catalytic activity with a high FE of more than 40% at −0.94 V versus RHE (Figure 8c).^{[ 87b ]} However, the methanol selectivity and reaction rate experienced a decreasing trend at relatively high cathodic potentials due to the competing hydrogen evolution. In addition, the long‐term stability test indicated that the catalyst is unstable and falls in selectivity from 44% in the first hour of electrolysis to ≈26% in the next 4 h reaction time, because of the ligand reduction. A recent report confirmed the maximum conversion of CO to methanol with a high FE of 65% at 30 mA cm⁻² using a Co SA catalyst.^{[ 88 ]} Lakshmanan et al. demonstrated an efficient CO₂ reduction to C₂₊ products with 63% FEs ($\mathrm{F}{\mathrm{E}}_{{\mathrm{C}}_2}{\mathrm{H}}_5\mathrm{OH}$ = 45%, $\mathrm{F}{\mathrm{E}}_{\mathrm{C}{\mathrm{H}}_3\mathrm{COOH}}$ = 18%) at −0.8 V versus RHE over Fe SAC prepared by ionic exchange strategies in Nafion coated functionalized CNTs (Fe‐n‐f‐CNTs).^{[ 89 ]} The catalyst exhibited a partial current density of 10.6 mA cm⁻² and a yield rate of 56.42 µmol h⁻¹ at −0.8 V versus RHE. This is because of the cooperative dual active sites between the Fe SACs that coordinated with three oxygen atoms (Fe–(O)₃) and the functionalized CNTs, dynamically boosting the transformation of CO₂ to C₂₊ products. In addition, the catalyst showed a promising long‐time stability test at −0.8 V versus RHE for 10 h with nearly constant current density. However, the product quantification was not analyzed during the 10 h electrolysis time which makes it vague to claim the stability of the catalyst in product selectivity. Similarly, Singh et al. recently demonstrated an efficient transformation of CO₂ to C₂H₅OH (FE = 66.8%) with 2.87 mA cm⁻² partial current density at −0.67 V versus RHE using Co chelated covalent organic framework (COF) (Co‐TAPA‐OPE, TAPA‐tri‐amino‐triphenyl amine, OPE‐Oligo‐(p‐phenyleneethynylenes)).^{[ 90 ]} The catalyst exhibited a greater intrinsic reaction rate of CO₂ to C₂H₅OH with a TOF of 0.04 s⁻¹ at −0.67 V versus RHE, higher than most Cu‐based sub‐nanometric catalysts. The in situ XAS study unveiled that the Co‐SA site in Co‐TAPA‐OPE COF transiently undergoes a reversible coordination environment and oxidation state during CO₂RR, creating a suitable environment for C₂H₅OH formation. The in situ FTIR measurement for real‐time reaction monitoring further showed the formation of *CHO species that can favorably couple with *CO for C–C dimerization toward C₂H₅OH generation. The long‐term stability test was performed at −0.67 V versus RHE for 24 h reaction time, which showed outstanding catalytic activity and selectivity. However, the recorded current density at this applied potential is insufficient, making it incompatible with scalability issues.

Figure 8.

a,b) Characterization, mechanism, and CO₂RR of Zn‐SAs anchored on microporous N‐doped carbon. Reproduced with permission.^{[ 87a ]} Copyright 2020, American Chemical Society. c) Characterization, mechanism, and electrochemical CO₂RR performance using Co‐phthalocyanine immobilized on carbon nanotubes. Reproduced with permission.^{[ 87b ]} Copyright 2019, Springer Nature. d) Adsorption energies of *H, *CO, and coadsorption of *H and *CO over Co‐phthalocyanine and Zn–nitrogen–carbon, the ratio of CH₄/CO production rate, and anticipated reaction mechanism of CO₂ transformation to CH₄ on the tandem CoPc/Zn–N–C catalyst. Reproduced with permission.^{[ 14d ]} Copyright 2020, Wiley‐VCH GmbH.

Furthermore, a catalyst prepared by anchoring Bi single atoms in the oxygen vacancy of CuO (Bi‐CuO(V_O)) showed an improved production rate in C₂H₄ and reached ≈27 µmol mg_cat ⁻¹ h⁻¹ at −1.05 V versus RHE, which is approximately twofold higher relative to the CuO catalyst (14.8 µmol mg_cat ⁻¹ h⁻¹).^{[ 91 ]} In addition, the C₂H₄ FE and partial current density reached 48% and 9.1 mA cm⁻² at −1.05 V versus RHE, outperforming the CuO nanoparticles counterpart. It further exhibited a maximum C₂H₄ selectivity of 60% at industrially suitable current density (400 mA cm⁻²). The improved selectivity toward C₂H₄ is due to the synergistic effect between the SA and the support, CuO(V_O). The DFT calculation indicates that introducing Bi SAs and V_O on CuO strengthens the adsorption of CO₂ and lowers the adsorption energy for the C–C dimerization, thus improving the production of C₂H₄. Moreover, the Bi‐CuO(V_O) electrocatalyst exhibited nearly constant C₂H₄ product selectivity and reaction rate for 20 h electrolysis time. The authors employed XRD and HAADF‐STEM techniques to investigate the structural changes after 1 h reaction time. However, it would be useful to examine the structural changes after the long‐term electrolysis time (20 h) to argue for the catalyst stability. In general, the electronic (e.g., electronic structure and charge transfer properties) and geometric (e.g., surface coordination) effects of SACs and supporting materials significantly influence the selectivity, catalytic activity, reaction rate, and stability. Understanding and adjusting these effects paves the way for developing efficient non‐Cu‐based SACs to transform CO₂ into high‐value products.

The non‐Cu‐based DACs can also transform CO₂ into high‐value products through a cooperative effect of two atoms. For instance, the Wang group fabricated the tandem CoPc and Zn–N–C heteronuclear electrocatalyst (Co₁Zn₁‐DACs) for CO₂ conversion to CH₄ (Figure 8d).^{[ 14d ]} It exhibited a maximum CH₄ FE of 18.3% ± 1.7% with a partial current density of 44 ± 7.3 mA cm⁻² at −1.24 V versus RHE, much higher than the nontandem counterparts. The tandem Co₁Zn₁ electrocatalyst exhibited a 100‐fold surge in the CH₄/CO ratio formation rate compared to that of the nontandem catalyst counterpart. The DFT calculations showed that the conversion of CO₂ to CO first occurred on CoPc, and subsequent migration occurred to the Zn–N₄ surface for CO to CH₄ conversion. This suggested the benefit of DACs in converting CO₂ to high‐value products through a two‐step reaction over non‐Cu catalysts. Thus, the first step requires an efficient catalyst for the successful transformation of CO₂ to CO. Then, the generated CO can further be reduced to high‐value products in the second single‐atom active site. To realize efficient conversion, the two active sites in the DACs should be compactly adjacent to one another to overcome the mass transport limit of CO from the generated site (first SAC) to the consumption step (second SAC) for the subsequent conversion to a high‐value product. The long‐term stability test of a catalyst is determinantal for practical application, whereas no report can be found related to the stability check in this study. In addition, although more DFT calculations recently predicted various transition metal‐based catalysts for CO₂RR to high‐value products at low overpotentials, as discussed below,^{[ 7} , ^{92 ]} the experimental investigations remained neither realized nor comprehensively understood. So far, the non‐Cu‐based DACs have scarcely been reported for CO₂RR to high‐value products. Thus, tremendous efforts are required to design the non‐Cu‐based DACs rationally and understand their electronic, geometric, and synergistic effects for CO₂RR to high‐value products.

Similar to the non‐Cu‐based DACs, there is a lack of reports of non‐Cu‐based tandem SAC–NCs for CO₂RR to high‐value products. Recently, a Sn‐based tandem electrocatalyst containing SnS₂ nanosheets (NSs) and Sn SAs anchored on 3D O‐rich carbon support that effectively coordinated the dispersed Sn atoms with the three O atoms (Sn₁‐O3G), denoted as SnS₂/S₁‐O3G was reported for electrochemical CO₂ transformation to ethanol in an H‐type cell.^{[ 93 ]} The tandem SnS₂/S₁‐O3G electrocatalyst offered FEs above 70% to ethanol over the potential range of −0.6 to −1.1 V versus RHE, with the best FE as high as 82.5% at −0.9 V versus RHE with 17.8 mA cm⁻² current density. Several control experiments and theoretical calculations show that the CO₂ reduction pathway to ethanol over the tandem catalyst follows a formyl–bicarbonate coupling mechanism. The SnS₂ NSs and Sn₁‐O3G components transform CO₂ to formate and CO, respectively, which subsequently coupled to provide a platform for C─C bond generation toward ethanol formation. Thus, the synergistic effect between the tandem SAC (for CO) and NCs (for formate) enables a strategy for manipulating the CO₂ conversion mechanism toward high‐value products. In addition, the SnS₂/S₁‐O3G electrocatalyst preserved ≈80% C₂H₅OH selectivity and 97% of the original catalytic activity during a 100 h continuous long‐term stability test at −0.9 V versus RHE. However, the detailed characterization analysis for the sample after the 100 h continuous reaction was not performed, which obstructs the claim of structural and electronic stabilities of the catalyst.

Like the abovementioned sub‐nanometric catalysts, there are unsatisfactory reports of non‐Cu‐based SNCCs for CO₂RR to high‐value products. The Liu group recently prepared carbon‐supported Sn dispersions and nanoclusters using an amalgamation process for controllable CO₂RR to various high‐value liquid products.^{[ 87d ]} The carbon‐supported Sn clusters (Sn/C‐1.2, 3–4 atoms) anchored by O at the exterior exhibited an outstanding CO₂ conversion to ethanol with an FE of 92% at −0.4 V versus RHE, which is a benchmark catalyst for CO₂‐to‐ethanol generation. However, the catalyst exhibited a low partial current density of 0.48 mA cm⁻² at −0.4 V versus RHE and caused the main product to switch to formate at high cathodic potential. Furthermore, an increase in Sn loading swung the main product from ethanol to formate, while acetate became the dominant product on a single‐atom Sn catalyst, suggesting that optimal catalyst synthesis through a novel approach was crucial. In addition, the Sn/C‐1.2 electrocatalyst maintained the catalytic activity and ethanol selectivity for 30 h continuous operation at −0.4 V versus RHE. However, the main product switched from ethanol to formate at a relatively high cathodic potential, indicating its incompatibility at industrially relevant current densities. To our knowledge, the reports on non‐Cu‐based SACs, DACs, tandem SAC–NCs, and SNCCs are insufficient for CO₂ reduction to high‐value products (Table 3 ). Thus, much effort is awaited on sub‐nanometric non‐Cu‐based catalysts to offer a C─C bond formation platform and provide an approach for operating the CO₂ reduction pathway to high‐value products. According to Table 3, the promising non‐Cu‐based transition metal catalysts for CO₂ reduction to high‐value products are mainly Co‐ and Sn‐based sub‐nanometric catalysts with almost similar selectivity and reaction rate toward a specific highly reduced product like that of Cu‐based catalysts. Although the non‐Cu‐based sub‐nanometric catalysts exhibited high selectivity, the operated current densities were too low, leading to a lower formation rate of the desired products. In addition, like the Cu‐based sub‐nanometric catalysts, the long‐term stability test of non‐Cu‐based sub‐nanometric catalysts for CO₂RR to high‐value products is in its initial stage.

Table 3.

CO₂RR to high‐value products of non‐Cu‐based SACs, DACs, and SAC–NCs.

Category	Material	Electrolyte	Type of cell	E [V vs RHE]	J [mA cm−2]	TOF	Stability [h]	Product FE [%]					Refs.
								CH4	CH3OH	C2H4	C2H5OH	C3H7OH
SACs	SA‐Zn/MNC	1.0 m KHCO3	H‐type	−1.80 a)	31.8	……	35.0	85.0	…….	……	…….	…….	[87a]
	CoPc/CNT	0.1 m KHCO3	H‐type	−0.94	10.6	1.05 s−1	12.0	…….	44.0	…….	…….	…….	[87b]
	Fe‐n‐f‐CNT	0.5 m KHCO3	H‐type	−0.80	18.0	……	10.0	…….	…….	…….	45.0	…….	[89]
	Co‐TAPA‐OPE	0.2 m KHCO3	H‐type	−0.67	2.87	0.04 s−1	24.0	……	…….	……	66.8	…….	[90]
	Sn‐THO	0.1 m KHCO3	Flow	−1.60	34.5	……	80.0	46.5	…….	…….	…….	…….	[94]
	CoPc/CNT	0.1 m KHCO3	H‐type	−0.95	12.0	……	3.0	…….	42.0	…….		…….	[95]
DACs	CoPc/Zn‐NC	1.0 m KOH	Flow	−1.24	44.3	……	……	18.3	…….	…….	…….	…….	[14d]
SAC–NCs	SnS2/Sn1‐O3G	0.1 m KHCO3	H‐type	−0.90	14.7	……	100	…….	…….	…….	82.5	…….	[93]
SNCCs	Sn/C‐1.2	0.1 m KHCO3	RDE	−0.40	0.48	……	30.0	…….	…….	…….	92.0	…….	[87d]

^a) The applied potential versus SCE; SA ‐ single atom, MNC ‐ microporous nitrogen‐doped carbon; CNT ‐ carbon nanotube; THO ‐ triphenylene‐2,3,6,7,10,11‐hexakis(olate); Pc ‐ phthalocyanine; NC ‐ nitrogen‐doped carbon; O3G – oxygen‐enriched 3D carbon support; TAPA ‐ tri‐amino‐triphenyl amine; OPE ‐ oligo‐(p‐phenyleneethynylenes); RDE ‐ rotating disk electrode in a single cell.

4. Theoretical Aspects of CO₂RR to High‐Value Products

Advancements in ab initio techniques, particularly DFT calculations, have empowered theoretical investigations of surface and electrochemical reactions.^{[ 86 ]} DFT calculation has become an essential instrument in catalysis as a whole, particularly in the investigation of CO₂RR, providing deep insights into diverse aspects of the electrocatalytic mechanism.^{[ 96 ]} DFT calculations enable a comprehensive understanding of reaction mechanisms by elucidating the intricate electronic and atomic‐scale processes involved in CO₂RR.^{[ 97 ]} They play a crucial role in predicting catalyst activity and selectivity, facilitating the identification of promising catalyst candidates through the calculation of adsorption and reaction energies.^{[ 98 ]} Moreover, DFT calculation‐driven catalyst screening expedites the exploration of a wide range of materials without the need for exhaustive experimental testing.^{[ 7} , ^{99 ]} Tailoring catalyst design for superior performance is also substantiated through insights from DFT calculation analyses. These investigations elucidate the intricate impact of factors such as composition, morphology, size, surface defects, as well as temperature and pressure on catalytic behavior.^{[ 100 ]} DFT calculations further elucidate reaction kinetics, revealing crucial information about reaction barriers and rate‐determining steps that guide the optimization of reaction conditions. As a bridge between theory and experiment, DFT calculations aid in the interpretation of experimental results, providing a theoretical framework to rationalize observed behaviors. Finally, DFT calculation‐guided fundamental studies unravel electronic structure changes, charge transfer processes, and reactivity trends at the atomic level, deepening our comprehension of the underlying principles governing CO₂RR and paving the way for future advancements in sustainable energy conversion.^{[ 101 ]} In the subsequent sections, we first delve into discussions concerning theoretical works reported on CO₂RR to high‐value products over Cu‐based isolated atoms and nanocluster catalysts. Then, the theoretical aspects of non‐Cu‐based sub‐nanometric catalysts for the transformation of CO₂ to high‐value products are discussed.

4.1. Theoretical Aspects of Cu‐Based SACs

SACs typically contain either metallic species anchored onto conductive matrices, such as porous carbon frameworks and metal oxides,^{[ 16} , ^{24 ]} or exist as single‐atom alloy (SAA) structures.^{[ 100} , ^{102 ]} For example, Shi et al. proposed graphdiyne (GDY)‐supported Cu single atoms (Cu SAs/GDY) as a catalytic material for CO₂RR to CH₄.^{[ 54 ]} The integration of Cu atoms with GDY, forming Cu─C bonds, induces changes in the electronic structure of Cu atoms, resulting in a higher valence state compared to their standard Cu⁰ state. This modification stabilizes isolated Cu atoms by preventing aggregation, and the altered electronic structure governs the modulation of reaction intermediates, thereby enhancing the activity and selectivity of Cu SAs/GDY for the CO₂ transformation to CH₄. The Cu SAs/GDY creates bonding with CO₂ with an optimal binding energy of −0.2 eV (Figure 9a). Conversely, the adsorption energy of H₂O is −0.05 eV, underscoring the Cu SAs/GDY superior affinity for CO₂ over H₂O. This pronounced CO₂ preference leads to concentrated CO₂ intermediates at the catalyst surface, with selective adsorption occurring predominantly at Cu sites. Consequently, this configuration promotes efficient mass transfer during CO₂ reduction reactions by suppressing the competing HER. As indicated in Figure 9b, the calculated free energy barrier for HER is 0.31 eV, indicating that HER is sluggish compared to the CO₂RR process on the Cu SAs/GDY catalyst. All the free energies in Figure 9b,e are obtained from DFT‐optimized structures without performing a transition‐state search. The DFT charge‐transfer investigation indicates that charges are transferred from the d orbital of Cu to the GDY framework. This makes the Cu atoms more conducive for electron coupling to the p orbital of the CO₂ adsorbate (Figure 9c). The PDOS analysis exposed the strong overlapping of the d orbitals of Cu and p orbitals of the GDY C, verifying the generation of Cu–C in Cu SAs/GDY (Figure 9d). Furthermore, the PDOS analysis shows the overlapping of the d orbital of Cu and the p orbital of *OCHO and *COOH intermediates. A full mechanistic study of the reaction from CO₂RR to CH₄ was performed using DFT calculations combined with a computational hydrogen electrode (CHE), and three possible pathways were proposed (Figure 9e). The subsequent step after CO₂ adsorption is either the generation of *OCHO or *COOH intermediates. However, the binding energy of *COOH is higher by 0.88 eV compared to *OCHO, indicating that the latter is the most plausible intermediate. The most thermodynamically stable intermediate following the first proton–electron transfer is identified as *OCHOH. This is sequentially succeeded by *OCH₂OH and *OCH₂ on the third and fourth proton–electron transfer, respectively. In path III, the only uphill step (potential limiting step) involves the hydrogenation of *OCHO to *OCHOH with a free energy of 0.34 eV. Consequently, based on these findings, pathway III in Figure 9e emerges as the most reliable reaction mechanism. On the Cu(211) surface, the potential limiting step is the protonation of *CO to *CHO with a free energy of 0.74 eV.^{[ 85a ]} Consequently, the Cu SAs/GDY catalyst, as proposed, exhibits superior performance compared to the pure Cu surface. However, this outcome is solely derived from thermodynamic analysis. Additional kinetic studies are necessary to establish a firm conclusion regarding the reaction rates of this catalyst. Similarly, combined experimental and theoretical studies were conducted to explore the performance of a Cu SAC anchored on N‐doped graphene for the CO₂RR to CH₄.^{[ 53 ]} Two different models (CuN₄ and CuC₄) were constructed, and the computed free energy values for *COOH generation are −0.11 and 0.11 eV on CuN₄ and CuC₄, respectively. This discrepancy signifies that the CuN₄ configuration is notably more favorable for the generation of *COOH intermediates. Moreover, the reaction‐free energy of *H at the N site in CuN₄ is 0.99 eV, whereas, at the C site in CuC₄, it is 1.24 eV, indicating the pivotal role of nitrogen moieties in enabling water dissociation and promoting the supply of adsorbed hydrogen species (*H). Consequently, the pronounced affinity of the N site for *H further accelerates the CO₂RR to CH₄, contributing to its overall efficiency.

Figure 9.

DFT calculations for the Cu‐SAs/GDY model catalyst a) the adsorption energy of H₂O and CO₂ on the model catalyst with insets of H₂O and CO₂ adsorption, respectively. White, gray, red, and orange colors refer to H, C, O, and Cu, respectively. b) HER free energy diagram. c) Difference in charge density before and after Cu‐atom bonding with GDY. d) Projected density of states (PDOS) in Cu‐SAs/GDY at the Cu–C atoms interface. e) Possible reaction mechanisms for the CO₂RR to CH₄ over Cu‐SAs/GDY. Reproduced with permission.^{[ 54 ]} Copyright 2022, Wiley‐VCH GmbH. f) Schematic representation of the 4 × 4 unit cells of pure Cu(111) and one dopant sites (the interfacial‐Cu sites, dopant‐related sites, and Cu sites are denoted by purple, red, and green triangles, respectively), and the selected 23 transition metal elements for screening the dopant catalyst with trapping energies on Cu(111) displayed inside the portion of the periodic table. g) Gibbs free energies for the C(O)CO and C(O)–CHO dimerization, and h) energy diagram and schematic illustration of the reaction mechanisms of CO₂ to C₂H₄ reduction via CO dimerization to C(O)CO on Sc@Cu(111). Reproduced with permission.^{[ 103 ]} Copyright 2023, American Chemical Society.

On the other hand, Zhang et al. studied Cu(111)‐based SAA for CO₂ conversion to ethylene.^{[ 103 ]} This study was conducted entirely through computational methods, using DFT and microkinetic modeling. The transition states (activation energies) were calculated using the climbing image nudged elastic band (CI‐NEB) method. The thermal stability was assessed using ab initio molecular dynamics (AIMD) simulations within the NVT ensemble at 800 K for 10 ps. The study investigated a couple of transition metals (23 types) as dopants. Figure 9f shows the model surfaces of pristine Cu(111), SAA‐Cu(111), and the stability of the SAA that was evaluated by systematically calculating the trapping energies of the single atoms under investigation on the Cu(111) surface. Apart from Sc, Pt, Y, and Pd, the other SAAs exhibit positive trapping energies, suggesting unfavorable SAA generations. The most suitable dopant metal, potentially enhancing both the activity and selectivity of C₂H₄, is primarily identified by assessing the viability of the C─C bond generation. Among the 23 candidates, only nine Cu‐based SAAs exhibited the capability to initiate C─C bond generation through CO dimerization to OC–CO (Figure 9g). Sc‐doped Cu(111) demonstrates the most favorable CO dimerization with the respective lowest reaction energy and kinetic barrier of 0.09 and 0.57 eV compared to the 22 other substrate alloying elements investigated. The reaction energies for CO dimerization exceed 0.4 eV for all other SAAs, resulting in higher kinetic barriers than on Sc@Cu(111). Kinetic barriers for CO hydrogenation to CHO were computed for the chosen SAAs. On Sc@Cu(111), the kinetic barrier is 0.83 eV, whereas, on Y@Cu(111), it is 0.64 eV, indicating a kinetic preference for the process on Y@Cu(111). Nevertheless, Sc@Cu(111) supports thermodynamically feasible *CHO and *COCHO formations. Upon comparing the reaction pathways involving COCO and COCHO formation on Sc@Cu(111), it becomes evident that the COCHO‐involved pathway exhibits a higher kinetic barrier of 1.91 eV, which is much higher than that in the COCO‐involved pathway. Thus, the production of C₂ intermediates is predominantly an outcome of CO dimerization on the Sc@Cu(111) surface. Figure 9h depicts the preferred reaction mechanism involving *C(O)CO, along with the corresponding reaction energies and kinetic barriers over Sc@Cu(111). The initial activation of the CO₂ molecule takes place through its adsorption on the Sc@Cu(111) surface via the Sc–O and Cu–C interactions. The hydrogen molecule undergoes dissociation at the interfacial‐Cu sites, accompanied by a kinetic barrier of 0.37 eV. The resulting dissociated H atom facilitates the breakdown of *CO₂ into *CO and *OH by attaching to the oxygen atom within CO₂. Importantly, it should be noted that this dissociation process must overcome a significant kinetic barrier of 1.57 eV, which is the rate‐determining step in the overall reaction process. In summary, the impressive performance in generating the desired C₂H₄ can be attributed to the cooperative influence of the Sc dopant and Cu(111) substrate. The largely unoccupied Sc d band effectively triggers CO₂ activation. Meanwhile, the Cu d band, which is a nearly filled orbital, facilitates the desorption of C₂H₄ and promotes the generation of C–C coupling. Although DFT‐based thermodynamic and kinetic analyses suggest Sc@Cu(111) as a promising catalyst, experimental verification would be crucial.

To overcome the dominance of CO evolution from SACs hosted by N‐doped graphene, metal‐supported SACs are employed, enabling the hydrogenation of CO₂ to high‐value products.^{[ 102a,b ]} For instance, a recent combination of DFT calculations and experimental work demonstrated that a copper‐supported iron‐single‐atom catalyst achieved a remarkable CO₂‐to‐methane conversion rate, with an FE of 64%.^{[ 102b ]} Remarkably, this activity surpasses that of pristine copper by 32 times under identical electrolyte and bias conditions. Regarding the reaction pathways, it is demonstrated that *CO is more likely to undergo hydrogenation to form *COH over *CHO on the Fe sites of Cu‐FeSA, particularly when solvation effects are considered. In contrast, pristine Cu prefers *CHO. Additionally, the research highlights that C–C coupling is energetically less favorable on Cu‐FeSA than on pristine Cu. Similarly, Zhao and Lu studied Cu‐based single‐atom catalysts for the electroreduction of CO₂ to CH₃OH using first‐principle methods. They performed extensive DFT calculations to screen promising SAA catalysts. A Cu surface atom was substituted with 26 transition metals as a dopant atom. The Co@Cu has been identified as a promising SAA catalyst. It exhibits the capability to efficiently convert CO₂ into liquid fuel CH₃OH,^{[ 102a ]} displaying both a low overpotential and high selectivity in the process. The study explored the reaction pathway of CH₃OH formation on Co@Cu and established that the most thermodynamically favorable sequence is

$$\begin{array}{ccc}\hfill \mathrm{C}{\mathrm{O}}_2& \to & \mathrm{COOH}\ast \to \mathrm{CO}\ast \to \mathrm{COH}\ast \hfill \\ & & \to \mathrm{CHOH}\ast \to \mathrm{C}{\mathrm{H}}_2\mathrm{OH}\ast \to \mathrm{C}{\mathrm{H}}_3\mathrm{OH}\hfill \end{array}$$ (1)

The reduction of COH* to CHOH* has been identified as the potential determining step with an energy barrier of 0.82 eV. The activation barriers are determined using a potential‐dependent transition‐state search technique.^{[ 104 ]} In another recent comprehensive DFT study on single‐atom copper alloy catalysts, the significance of the adsorption mode was highlighted, specifically, whether the adsorption site is adjacent to copper or directly bound on top of the doped metal.^{[ 105 ]} The findings revealed that the adjacent hollow site proves to be beneficial in enhancing the activity of the catalyst. Interestingly, DACs were also theoretically designed to break the persistent limitation of scaling relations on the CO₂RR catalytic performance.^{[ 38 ]} Twenty‐one types of DACs based on transition metals were functionalized with monolayer C₂N for an efficient CO₂RR to CH₄. Among all candidates, CuMn/C₂N and CuCr/C₂N were reported to be the most active and selective for the reduction of CO₂ to CH₄ with low limiting potentials of −0.32 and −0.37 V, respectively. Recently, a static and dynamic DFT study was conducted to evaluate C–C coupling in CO₂RR on single Cu‐modified covalent triazine frameworks (Cu‐CTFs).^{[ 106 ]} The dynamic DFT study employed constrained AIMD in the NVT ensemble using the Nosé–Hoover method, with solvent effects modeled through explicit solvation. Static DFT calculations, utilizing the climbing image nudged elastic band (CINEB) method, were used to determine the transition energy barrier for C–C coupling. The findings indicate that C–C coupling proceeds through *CO and *CHO coupling rather than *CO─*CO dimerization. The activation energy for *CO─*CHO coupling was calculated to be below −0.69 V versus CHE using CINEB, and less than 0.1 eV with AIMD.

Overall, many DFT studies have been reported on Cu‐SACs focusing on the formation of C₁ products, particularly CO and formate formation, and these topics have been covered in other review papers.^{[ 8} , ^{107 ]} While some reported works claim the formation of multicarbon products on Cu‐based SACs,^{[ 24} , ^{32 ]} much of the DFT calculations or combination of DFT calculations and experimental studies confirm the formation of C₁ products. Some argue that C₂₊ product formations could be due to reversible transient reconstruction/cluster formation under working conditions that are difficult to detect under ex situ conditions.^{[ 45} , ⁵¹ , ^{108 ]} A combination of detailed DFT calculations that consider experimental conditions alongside experimental operando and in situ techniques might be helpful to validate these assumptions.

4.2. Theoretical Aspects of Cu‐Based SNCCs

SNCCs exhibit significant potential in nanocatalysis due to their sensitivity to single‐atom changes that result in distinct electronic, geometric, and catalytic properties compared to their larger clusters or bulk counterparts.^{[ 109 ]} Although theoretical investigations into electrocatalytic CO₂ reduction on SNCCs are relatively limited, the reports on Cu‐based nanoclusters showed promising selectivity and activity toward the formation of high‐value products.^{[ 102} , ¹⁰³ , ^{110 ]} For instance, Raju et al. have investigated size‐selected Cu _n SNCCs activity (n = 3–6) for CO₂RR to CH₄ using the DFT and CHE model.^{[ 110 ]} The study revealed a similarity in the CO₂RR activity between sub‐nanoclusters with an odd electron number, namely, Cu₅ and Cu₃, and between sub‐nanoclusters with an even electron number, specifically Cu₆ and Cu₄. The reaction pathway involves a sequence of steps: * + CO₂ → COOH* → CO* + H₂O → CHO* → CH₂O* → CH₃O → O* + CH₄ → OH* → * + H₂O. These steps occur uniformly across all copper sub‐nanoclusters studied. In the case of Cu₄ and Cu₆ sub‐nanoclusters, *CO to *CHO hydrogenation is the rate‐determining step (RDS). In contrast, on the Cu₃ and Cu₅ sub‐nanoclusters, the removal of adsorbed OH* from the sub‐nanocluster surface (OH* → * + H₂O) is the RDS. Additionally, the competing HER process is more suppressed on sub‐nanoclusters with odd electron numbers than on sub‐nanoclusters with an even number of electrons. Moreover, the selectivity between the generation of CH₄ and CH₃OH depends on two competing reaction steps: CH₃O* → * + CH₃OH and CH₃O* → O* + CH₄. Note that all these energies are derived from DFT optimization, without considering activation barrier calculations. Overall, the findings suggest a higher preference for CH₄ over CH₃OH.

4.3. Theoretical Aspects of Non‐Cu‐Based Sub‐Nanometric Catalysts

Linear scaling relations (LSRs) have been crucial in elucidating the mechanism of CO₂RR and providing insights into reaction pathways and catalyst performance predictions of non‐Cu‐based catalysts.^{[ 98} , ^{111 ]} The origin of high CO₂RR overpotentials on metallic catalysts is ascribed to the LSR of the binding energies of the reaction intermediates.^{[ 98b ]} Breaking LSR is of great interest to realize product‐selective CO₂RR and various strategies to overcome the LSR are suggested.^{[ 112 ]} As discussed above, SACs provide a notable opportunity to break LSR due to their distinct atomic configurations and unique electronic structures compared to regular metal catalysts. Recently, a study investigated an electrochemical transformation of CO₂ to CH₄ and CH₃OH products using single‐atom anchored on defective 2D molybdenum borides (SA‐Mo₂B₂) catalysts.^{[ 92a ]} The study included the evaluation of 23 SA‐Mo₂B₂ SACs alongside nine pure noble metal catalysts. Initially, analysis of the noble metal catalysts revealed evidence of an LSR between the binding‐free energies of two key CO₂RR reaction intermediates, CO* and CHO* (Figure 10a). Subsequently, the same analysis was extended to the 23 SA‐Mo₂B₂ SACs, revealing a break of the LSR phenomenon on these SACs (Figure 10a). The noble metal atoms anchored onto the Mo₂B₂ substrate served as sites of carbon affinity, while the Mo atoms within the Mo₂B₂ matrix acted as sites of oxygen affinity. The investigation extended to analyze the free energy profiles of C₁ high‐value products (CH₄ and CH₃OH) on the two most promising SACs, Rh‐Mo₂B₂, and Ir‐Mo₂B₂ (Figure 10b). The findings of this investigation revealed the selective production of CH₄ on Rh‐Mo₂B₂ and CH₃OH on Ir‐Mo₂B₂ at the respective ultralow overpotentials of −0.32 and −0.27 V induced by breaking the LSRs. The free‐energy diagram depicting the CO₂ reduction to CH₄ over Cu(211) is also incorporated for comparative purposes. The solvation correction calculations were incorporated into the free‐energy estimates using an explicit solvation model. However, the free‐energy diagrams presented do not include activation energy barrier calculations. The thermodynamic stability of the SACs was assessed using AIMD simulations at 300 K with an NVT ensemble. Additionally, the electrochemical stability of the catalysts was evaluated by calculating their dissolution potentials. Similarly, 5 SACs, including Mn–N₄–C, Os–N₄O–C, Cr–N₄–C, Rh–N₄O–C, and Ru–N₄O–C, were identified for possible CO₂RR to CH₄ based on the influences of the electronegativity, the number of outmost d shell electrons, coordination number, and the bond length of the central atom and the nearest neighbor atom to correlate the catalytic activity (Figure 10c,d).^{[ 113 ]} A recent theoretical study by the Fu group also demonstrated Co@In₂Se₃ as the most promising non‐Cu‐based catalyst for an efficient CO₂RR to CH₄ at a low limiting potential of −0.385 V.^{[ 7b ]}

Figure 10.

a) Gibbs free energy of *CHO versus the adsorption free energy of *CO + H⁺ + e⁻ on nine pristine metal surfaces. Twenty‐three SA‐Mo₂B₂ single atom catalysts at 0 V versus RHE. The limiting fitting of G[*CHO] versus G[*CO+ H⁺ + e⁻] on the pristine metals is indicated as a gray line. Adsorption free energy difference (ΔG) = G[*CHO] – G[*CO + H⁺ + e⁻] for the reaction mechanism *CO + H⁺ + e⁻→*CHO on Rh(111), Cu(111), and Rh‐Mo₂B₂ is revealed as inset and on the iso‐energy line (cyan), ΔG is 0. b) Diagram of ΔG for CO₂RR to CH₃OH and CH₄ on Ir‐Mo₂B₂ and Rh‐Mo₂B₂ SACs relative to the production of CH₄ on Cu(211) with atomic structures of the reaction intermediates depicted in the inset (white, red, gray, violet, and dark yellow for H, O, C, Mo, and SA, respectively). Reproduced with permission.^{[ 92a ]} Copyright 2023, American Chemical Society. The CO₂RR limiting potential of c) M–N₄–C, and d) M–N₄O–C versus the Gibbs free energies of *OH and *CHO at 0 V versus SCE. Reproduced with permission.^{[ 113 ]} Copyright 2022, American Chemical Society.

Moreover, carbon‐based SACs, including graphene, carbon nitride, graphdiyne, and MOF‐based carbon materials, are among the most promising catalysts for CO₂RR to high‐value products.^{[ 105} , ^{114 ]} For example, the Fe atoms supported on graphdiyne (Fe/GDY) exhibited the transformation of CO₂ to CH₄ and C₂H₅OH computationally through the *HCO hydrogenation for the former and the coupling of *HCO with CO for the latter.^{[ 114b ]} Complete linear synchronous transit and quadratic synchronous transit computations were performed to calculate the C–C coupling barrier. For the formation of ethanol, the low kinetic barrier for C–C coupling occurs between HCO* and CO*, with a value of only 0.54 eV. This barrier is lower than that for the direct dimerization of the adsorbed CO*. In addition, single transition metal‐doped g‐C₃N₄ materials are employed as electrocatalysts for the CO₂RR, aiming to produce C₁ products, including CO, HCOOH, CH₃OH, and CH₄.^{[ 115 ]} These electrocatalysts are assessed using DFT calculations. A total of 28 SA transition metal‐doped g‐C₃N₄ (TM‐g‐C₃N₄) materials are screened based on stability, activity, and selectivity descriptors. Among these materials, Ti‐g‐C₃N₄ and Sc‐g‐C₃N₄ stand out as active and selective catalysts to produce CH₃OH and CH₄. A DFT study explored the electrochemical reduction mechanism of CO₂ using non‐noble metal–nitrogen–carbon (M–N–C, M = Fe, Co, Ni) single‐atom catalysts. The study focuses on analyzing reaction pathways, efficiency, and selectivity for producing various products (CO, HCOOH, HCHO, CH₃OH, and CH₄), revealing that Fe–N₄ and Co–N₄ embedded in graphene exhibit low overpotentials, making them promising candidates for high‐efficiency CO₂ reduction to CH₄.^{[ 116 ]} DFT investigations were conducted on a series of single transition metal atoms anchored on a nitrogen‐doped graphene/graphdiyne heterostructure (TM‐N₄@GRA/GDY) for CO₂RR. Among these catalysts, Co–N₄@GRA/GDY exhibited superior catalytic activity and selectivity toward CH₄, with a low limiting potential of −0.567 V.^{[ 116 ]} Some combined experimental and theoretical studies have been conducted on non‐Cu‐based SACs to realize the transformation of CO₂ to high‐value products.^{[ 87} , ⁸⁸ , ⁸⁹ , ^{95 ]} For instance, a study by Lakshmanan et al. demonstrated the transformation of CO₂ to ethanol using a Fe‐n‐f‐CNTs catalyst.^{[ 89 ]} The CO₂ molecule was first reduced to CO on the Fe SA site and subsequently transformed into ethanol on the CNTs. The DFT calculations unveiled that the Fe–(O)₃ configuration in the Fe‐n‐f‐CNTs catalyst was distorted under CO₂ reduction conditions, and the carboxylic functional group on the CNTs played a pivotal role in stabilizing Fe SAs through an electrostatic interaction established between oxygen in a carboxylic functional group and Fe sites (Figure 11a). The free energies were calculated at the PBE level, and the energies are based on optimized structures. Transition‐state search calculations were not included. The CO₂ molecule was first converted to CO through the protonation of adsorbed CO (*CO) either to *OCHO or *COOH pathway on the Fe SA sites for a subsequent reduction to ethanol on the CNTs site.

Figure 11.

The DFT calculations of a) free energy diagram for CO₂ reduction on Fe single atoms sites of nafion‐coated functionalized CNT (Fe‐n‐f‐CNTs) and nafion‐coated CNTs (Fe‐n‐CNTs), and coordination information of H on SO₃ of Fe‐n‐CNTs. Reproduced with permission.^{[ 89 ]} Copyright 2022, Wiley‐VCH GmbH. b) Free energy profile of CO₂ reduction to CO (pathway i) and subsequent electroreduction of CO to CH₄ on CoPc and ZnN₄ DACs (pathway ii). Reproduced with permission.^{[ 14d ]} Copyright 2020, Wiley‐VCH GmbH. c) Free energy profiles and corresponding key intermediate structures for CO₂ reduction to ethanol over Sn₁‐O3G via a formyl–bicarbonate coupling pathway. Reproduced with permission.^{[ 93 ]} Copyright 2023, Springer Nature.

Recently, there has been a growing interest in DACs alongside SACs for catalytic applications.^{[ 30} , ^{117 ]} Compared to SACs, there is a limited number of theoretically reported works on DACs. In the following, we provide a summary of the key theoretical studies on non‐Cu‐based DACs. Xiao et al. systematically investigated the potential of homonuclear SACs, DACs, and triple‐atom catalysts (TACs) for CO₂RR.^{[ 118 ]} They explored 24 transition metals (TM = Ti, V, Cr, Mn, Fe, Co, Ni, Cu, Zn, Zr, Nb, Mo, Ru, Rh, Pd, Ag, Hf, Ta, W, Re, Os, Ir, Pt, Au) to construct single‐, double‐, and triple‐metal active centers on a 2D molybdenum carbide (MXene) substrate functionalized by oxygen groups (Mo₂CO₂). Through DFT screenings, they identified one DAC (Rh) and three TACs (Rh, Os, Ir) as highly active and selective catalysts for CH₄ and C₂H₅OH production, exhibiting ultralow overpotentials ranging from −0.15 to −0.48 V. Notably, DACs and TACs demonstrated superior performance compared to SACs in the CO₂RR, attributed to multisite adsorption of key reaction intermediates (*HCOO, *CH, and *CH₂), lowering limiting potentials for the CO₂ → *HCOO and *CHOH → *CH steps and resulting in low overall overpotentials. Additionally, TACs facilitated C–C coupling reactions for C₂ products due to their ability to accommodate two intermediates simultaneously at multinuclear sites. The activation energy barrier for the formation of the C─C bond was calculated using the CI‐NEB method and solvation corrections are included for the free energy of the reaction intermediates. The other work by Yu et al. used a combination of DFT calculations and machine learning (ML) techniques to investigate the potential of DACs with an inverse sandwich structure anchored on defective graphene (gra) for catalyzing the CO₂RR to generate C₁ products.^{[ 119 ]} The thermal stability of the structures was assessed using AIMD simulations in the NVT ensemble at 300 K. The Poisson–Boltzmann model for implicit solvation was used to simulate an H₂O solvent environment. Free‐energy calculations of the intermediates were performed without considering transition‐state searches. This work explored five homonuclear M₂@gra (M = Co, Ni, Rh, Ir, and Pt) DACs and 127 heteronuclear MM@gra (M = Co, Ni, Rh, Ir, and Pt, M = Sc–Au). The DFT calculations revealed that the Rh₂@gra DAC exhibited superior performance compared to the other homonuclear DACs, as well as single‐ and double‐atom catalysts with noninverse sandwich structures. Among the 127 heteronuclear DACs, 14 were found to be stable and demonstrated good catalytic performance. Specifically, RhIr@gra and RhPt@gra demonstrated remarkably low limiting potentials of −0.36 and −0.26 V, respectively, for CH₃OH production. Furthermore, the ML model successfully predicted 154 stable and high‐performance DACs out of a pool of 784 DACs with inverse sandwich configurations, showcasing their potential for CO₂RR. In general, CH₃OH and CH₄ were identified as the preferred end products over most DACs, as the *CO, *HCOOH, and *CH₂O intermediates exhibited strong binding, leading to further hydrogenation during the CO₂RR process. The FeCo–NC DACs were investigated using DFT for CO₂ reduction to C1 products and were compared with Fe–NC/Co–NC SACs.^{[ 120 ]} The synergy between Fe and Co dual atoms balances structural stability and catalytic activity, resulting in stable CO₂ adsorption and enhancing multielectron transfer capabilities. The CO intermediate serves as an electronic and geometric modifier, optimizing second CO adsorption and facilitating CO₂ reduction pathways. FeCo–NC demonstrates higher CO₂RR activity for CH₃OH and CH₄ production, with CO intermediate assistance lowering limiting potentials, whereas CO is the only product on Fe–NC/Co–NC. Overall, this work identifies FeCo–NC as a high‐performance CO₂RR catalyst, unveiling a dual‐atom synergistic effect through CO intermediate assistance. In summary, DACs are integral in providing both top and bridge sites to enhance the adsorption of complex intermediates in catalytic processes.^{[ 121 ]} While DACs inherit the advantages of SACs in efficiently reducing CO₂ to CO, they also synergistically modify the electronic structures, influencing the subsequent reactions. A combined experimental and theoretical study of the CoPc/Zn–N–C DAC catalyst showed greater CO₂ reduction selectivity to CH₄.^{[ 14d ]} The DFT calculations revealed that the CO₂ molecule is reduced to CO on the CoPc site much more easily than on the Zn–N₄ site (Figure 11b, pathway i). Then, the in situ formed CO migrated to the Zn–N₄ site for the subsequent hydrogenation with energetically more favorable and plausible reaction pathways on the Zn–N–C site to yield CH₄ (Figure 11b, pathway ii), in line with the experimental findings.

In addition, a detailed experimental and theoretical study by the Sargent group has shown a successful conversion of CO₂ to ethanol over a tandem SAC–NC (SnS₂/Sn₁‐O3G).^{[ 93 ]} The authors employed several theoretical models to understand the underlying reaction mechanism. Two key reaction pathways were identified for the conversion of CO₂ to ethanol. One of them was the conversion of CO₂ to HCOOH on the Sn NPs sites of SnS₂, and the other was CO₂ reduction to CO occurring on the Sn SA site of Sn₁‐O3G. Through the proton‐coupled electron transfer (PCET) mechanism, the Sn‐CHO intermediate formed from the dehydration of the carbonyl group in HCOOH*, a product released from the Sn NPs as a liquid product and subsequently adsorbed onto the Sn SA site. This promoted coupling of *CHO and *CO(OH) on the Sn and adjacent O sites to generate the C–C dimerization with a lower free energy barrier of 0.13 eV, respectively, acting Sn and O atoms as dual active sites in the Sn₁‐O3G. Through a couple of PCET steps and necessary bond cleavages, CH₃CH₂OH (ethanol) was released as a final product (Figure 11c). Recently, a theoretical investigation demonstrated the effective transformation of CO₂ to high‐value products over Fe₃/MoS_{2− x} and Ru₃/MoS_{2− x} non‐Cu‐based SNCCs.^{[ 7b ]} Although more and more theoretical studies show the potential to convert CO₂ to high‐value products over non‐Cu‐based transition metals, corresponding experimental reports are rare. Thus, combined experimental and theoretical investigations are highly suggested to realize the transformation of CO₂ to high‐value products on non‐Cu‐based catalysts.

In summary, there has been a recent surge in DFT studies focused on sub‐nanometric catalysts for the formation of C₁ and C₂ products. However, most reported works are related to C₁ products. The DFT calculations play a crucial role in understanding the intricate mechanisms governing the formation of these products on various sub‐nanometric catalysts (SACs, SAAs, DACs, and SNCCs) employed in CO₂ reduction reactions. They also explain why some catalysts cannot provide the desired product. For instance, a comprehensive review of experimental reports coupled with DFT calculations and microkinetic modeling investigated why C–C coupling is difficult on DACs.^{[ 122 ]} The main findings, particularly on Fe/Fe–N₆–C, Fe/Ni–N₆–C, and Ni/Ni–N₆–C DACs, demonstrated that C–C coupling is not favorable in terms of thermodynamics and kinetics due to the preferential adsorption of CO on the bridge site that prevented subsequent C–C dimerization, consolidating the prevalence of CO as the primary product in the DACs under investigation. Interestingly, as discussed above, DFT calculations also show a way in which DACs can offer high‐value products by facilitating the dimerization of *CO to form a C─C bond or protonate to offer *COH for further reduction to CH₄ and CH₃OH. The principal focus areas in DFT calculation aspect include i) electronic structure and adsorption properties, ii) reaction mechanisms, iii) catalytic activity and selectivity, and iv) understanding structure–activity relationships. The application of DFT calculations has demonstrated a significant impact, and it is anticipated that such approaches will persist in the foreseeable future of the catalysis field. The emergence of high‐throughput DFT calculations and machine learning screening of extensive datasets of catalyst materials will offer theoretical guidance for the rational selection and optimization of catalysts in the future. However, pure DFT calculations cannot fully account for the experimental operational conditions and often show some discrepancies with the experimental results. Specifically, the kinetic aspects and the analysis of different adsorbate coverages under realistic electrochemical conditions are often lacking.^{[ 123 ]} Validating predictions by DFT calculations using experimental data facilitates the identification of potential candidate catalysts. Therefore, combining DFT calculations and experimental approaches would always be the preferred approach in the rational design of catalysts. The general summary of the active sites and electronic and geometric effects of SACs, DACs, tandem SAC–NCs, and SNCCs on the electrochemical CO₂ reduction activity and selectivity are provided in Table 4 and Table S3 (Supporting Information).

Table 4.

The summary on the active sites, electronic, and geometric effects of SACs, DACs, tandem SAC–NCs, and SNCCs on CO₂ reduction activity and selectivity.

Catalyst	Active sites	Electronic effects	Geometric effects
SACs	One active site for CO2 adsorption and activation. Difficulty to adjust the adsorption of every intermediate on a single site concurrently (the scaling relationship limit, SRL).	The electronic structure of the active site determines its ability to interact with CO2 and affects the bonding strength of intermediates. The existence of unoccupied orbitals promotes adsorption/activation of CO2. Charge transfer occurs between the catalyst and support, and between the catalyst and CO2, facilitating the adsorption/activation of CO2.	The coordination environment and arrangement of SA substantially influence the adsorption and reactivity of CO2, and the spatial arrangement of the SA impacts the orientation and binding strength of CO2 that influence the selectivity and reaction pathway.
DACs	Two possible active sites for CO2 adsorption and activation. The additional active site changes the adsorption structure that breaks the SRLs based on SACs. There are possible synergistic effects between the atoms.	The charge transfer and redistribution of electrons between the two atoms alters the electron density and reactivity of the atoms that impact the adsorption and activation of CO2, a synergistic effect is generated from the cooperative interaction of the two atoms, and the combined electronic properties of both atoms create a unique electronic environment that improves the catalytic activity.	The coordination and spatial arrangement of the two atoms and atoms with the support impact the adsorption, activation, and reaction kinetics of CO2. The specific configuration of the two atoms forms unique active centers, the geometric arrangement of the two atoms can determine the accessibility of active sites for CO2 reduction, and different coordination environments can offer distinct electronic structures and binding strengths.
Tandem SAC–NCs	Two possible active sites (SAs and NCs) for CO2 adsorption/activation. The presence of NCs breaks the SRLs, and there is tandem or synergistic effect between the SA and NC.	The interaction between the SAs, NCs, and support improves the catalytic activity. The SAs modify the electronic structure and create active sites for CO2. The NCs offer a high surface area for enhanced contact with CO2 that promotes efficient adsorption and reaction kinetics. The tandem catalysts tune the energy landscape, leading to changes in the adsorption/activation barrier for CO2 reduction.	Both the SACs and NCs have unique geometric properties that influence their catalytic activity. In addition to the geometric impact of SAs, the NCs are influenced by different factors like shape, particle size, and arrangement. The combined geometric effects of SAs and NCs generate a tandem or synergistic effect that leads to enhanced CO2 reduction performance.
SNCCs	The active sites for CO2 adsorption can vary depending on the size and composition of the SNCCs.	As the size of the SNCCs decreases, the electronic structure changes, leading to the advent of confinement effects that modify the reactivity of the cluster. The electronic effects from the supporting material can also modulate the charge transfer between the support and the cluster, affecting the binding energies of CO2 and reaction intermediates.	The geometric effects depend on the size, shape, and arrangement of SNCCs on the support. The arrangement (distance and angle) or dispersion of SNCCs on the support affects the number of active site accessibility, promoting the adsorption/activation of CO2. The supporting material can also impact the spacing and coordination environment of SNCCs.

In summary, the electronic effects of these sub‐nanometric catalysts are characterized by various techniques: X‐ray photoelectron spectroscopy (XPS), X‐ray absorption spectroscopy (XAS), and in situ/operando spectroscopy. These methods offer insights into the electronic structures and charge transfer dynamics during CO₂RR. The geometric configurations are characterized by transmission electron microscopy (TEM), scanning tunneling microscopy (STM), Fourier transform infrared (FTIR), and X‐ray scattering methods at atomic resolutions. These techniques offer insights into the catalyst surface coordination during CO₂RR. Computational modeling and density functional theory (DFT) calculations are crucial tools to understand the electronic and geometric properties of these sub‐nanometric catalysts. In general, the electronic effects (e.g., electronic structure and charge transfer properties) and geometric effects (e.g., surface coordination and spatial arrangements) substantially influence the efficiency and selectivity of CO₂RRs. Understanding and tailoring the electronic and geometric properties of these sub‐nanometric catalysts can, therefore, improve the adsorption, activation, and subsequent transformation of CO₂, leading to different product selectivity and overall reduction performances.

5. Conclusion and Outlook

In summary, the experimental and theoretical studies conducted on Cu‐ and non‐Cu‐based sub‐nanometric (SACs, DACs, SAC–NCs, and SNCCs) electrocatalysts for CO₂RR to high‐value products have been reviewed and discussed. As shown in Figure 12 , the sub‐nanometric catalysts are promising candidates for constructing efficient electrocatalytic systems due to their highly unsaturated coordination environment, efficient atom utilization, and confinement effect. In spite of the high activities and selectivities recorded, the studies made in these sub‐nanometric catalysts for CO₂RR to high‐value products are still limited, and there are remaining significant challenges in the practical utilization of these electrocatalysts, including how to achieve high catalyst loading and long‐term operational stability. Moreover, our understanding of the geometric structures and electronic properties of the catalysts and the undergoing dynamics during the real reaction conditions is insufficient. Some trends in selectivity, research gaps, and future research directions are proposed to overcome the following existing challenges.

1. Although the Cu‐based SACs have the potential to offer high‐value products (mainly CH₄) with fairly great selectivity and activity, the mononuclearity of the metal sites and the absence of ensemble sites in SACs make it difficult to convert CO₂ to C₂₊ products. However, the advent of DACs alters the structural and stereoelectronic environments by involving two metal nuclei, and consequently, the homonuclear DACs show a better tendency toward C₂₊ product selectivity. Furthermore and most interestingly, SAC–NCs‐based systems demonstrate greater CO₂RR selectivity toward C₂₊ products, reflecting that tandem catalysts are promising candidates for CO₂ transformation to C₂₊ products. The CO₂RR mechanisms on DACs and SAC–NCs can be carried out through 1) the MA site effectively reduces CO₂ to CO with a weak CO adsorption energy, whereas the MB (in DACs) or ensemble (in SAC–NCs) sites, with a moderate CO binding adsorption, have a good capacity to couple adsorbed CO. Thus, the generated CO from MA diffuses to MB or ensemble sites to promote the CO coupling to C(O)–CO, yielding C₂₊ products or 2) the synergistic effects of both active centers facilitate the C–C dimerization to yield C₂₊ products. On the other hand, the product selectivity obtained from Cu‐based heteronuclear DACs shows CH₄ and CH₃OH. The product selectivity trends observed from SNCCs are mainly C₂₊ products rather than C₁ products. In addition, SNCCs also offer the highest CH₄ selectivity, indicating that this system is mainly influenced by the electronic properties and chemical compositions. In general, the selectivity trends highlight the dominance of CO₂ reduction to C₁ products, mainly CH₄, from Cu‐based SACs and heteronuclear DACs, and to C₂₊ products, mainly C₂H₄, from Cu‐based homonuclear DACs, SAC–SNCCs, and SNCCs. Moreover, these sub‐nanometric catalysts exhibit greater reaction rates to single high‐value products at industrially relevant current densities. Although similar trends in product selectivity can be seen from non‐Cu‐based SACs, DACs (heteronuclear), SAC–NCs, and SNCCs, the number of reports on these sub‐nanometric non‐Cu‐based catalysts is still insufficient, particularly the utilization of these sub‐nanometric catalysts with modifications through rational design. For instance, combining the SACs with NCs to construct tandem SAC–NCs can potentially modulate its selectivity from C₁ to C₂₊ products. The careful engineering of SAC–NCs proximity can help tackle the mass transport limit of CO from the generated site (e.g., SA site) to the consumption site (e.g., NCs site) for favorable C–C coupling.

2. The preparation of the sub‐nanometric catalysts on suitable host materials with copious surface binding sites (by introducing adequate anchoring heteroatom sites, e.g., O, P, N, and S into the support) to coordinate the metal atoms can maximize catalyst loading, which may modulate the electronic properties (e.g., coordination number and oxidation state) and geometric structures of neighboring atoms and eventually boost the selectivity to C₂₊ products. Apart from the commonly used O or N atoms‐based supports as coordinating units, S, P, B, and graphdiyne containing host materials can be used to effectively coordinate the small‐size metal catalysts to adjust the electronic and geometric structures of metal centers. In addition, two or more heteroatoms can be used at once (e.g., Cu SAs anchored on N, B‐codoped carbon support) to significantly influence the electronic and geometric structures of the active centers. However, a too large number of catalysts loading may result in agglomeration, leading to clusters or NPs formation and ultimately affecting the activity, selectivity, and reaction rate. Hence, the structures of the host materials (e.g., defect, ligand coordination, space confinement, and crystallinity) must be well‐controlled to circumvent the abovementioned limitations. In addition, much effort should be devoted to the combined effects between two adjacent active species and synergistic effects with the support deploying catalytic behaviors, together with addressing insights into the electrocatalytic mechanisms. Moreover, it is highly suggested to employ suitable CO₂ reduction systems to efficiently modulate the transient microenvironments and dynamics of the *CO/*H surface coverage to promote C–C coupling.^{[ 124 ]} Thus, examining the relative coverage of the competitive *CO and *H over sub‐nanometric active sites is essential to boost the generation of C₂₊ products.

3. The relatively low operational stability of sub‐nanometric catalysts at a high current density is another challenge that deserves the same emphasis as that of the catalytic performance (e.g., activity, selectivity, and reaction rate). The operational stability trend observed for the sub‐nanometric catalysts to high‐value product formation is at its initial stage, only a few hours at high current densities. The deactivation of active sites due to the strong adsorption of carbonaceous intermediates or by‐products, transformation of reactive centers into clusters and/or NPs, leaching of metal centers, and the sloughing of active sites are the key reasons for the instability of the sub‐nanometric catalysts. Therefore, increasing the interaction between the metal and the host material and adjusting the electrocatalytic reaction under suitable circumstances may hinder the aggregation of reactive sites. In addition, the rational design of preparing these sub‐nanometric catalysts directly on the working electrode rather than the deposition of ink suspensions onto the electrode can avoid the issues related to the sloughing of active catalysts.

4. Tracking the true instantaneous intermediate formation and the actual structural evolution of active sites during the electrocatalytic reactions helps to deepen mechanistic understanding of the sub‐nanometric catalysts and usher the rational design of active and selective catalysts. Suitable operando/in situ characterization methods, such as X‐ray absorption spectroscopy, X‐ray diffraction, infrared spectroscopy, and Raman analysis have been used to monitor the structural evolution of electrocatalysts and identify the key carbonaceous intermediates during the real‐reaction time. However, the in situ spectroscopic techniques used today show a lack of accuracy with high measurement errors in identifying the coordination number, bond length, geometric structure, and oxidation state of catalysts.^{[ 125 ]} The structural evolution process, formation of new active sites, and generation of key intermediates during the CO₂RR are instantaneous, commonly in the order of nano‐ to picoseconds, while the speed for data acquisition of the operando characterization technologies used today (e.g., XAS) ranges from seconds to milliseconds. This makes it challenging to identify the real‐active sites and recognize the instantly formed intermediates that ultimately hamper the mechanistic understanding under the real‐time reaction condition. Therefore, more accurate and ultrafast operando spectroscopic techniques should be developed to monitor the instantly generated intermediates and the dynamically formed structures during the real‐time CO₂RR over the sub‐nanometric catalysts.

5. So far, Cu‐based catalysts are renowned for the conversion of CO₂ to many hydrocarbon and oxygenated products, whereas product selectivity remains a great challenge. When the size of Cu is reduced to atomic and SNCC levels, the selectivity, activity, and reaction rate radically vary compared to the NP counterparts due to the changes, mainly in electronic properties and geometric structures. Thanks to the emerging atomic level catalysts, sub‐nanometric non‐Cu‐based catalysts are also becoming capable of transforming CO₂ into high‐value products, which helps to expand the catalyst exploration toward wide transition‐metal catalysts. Recently, a study on tin‐based tandem (Sn single atom and SnS₂ nanosheets on carbon foam, SnS₂/Sn₁‐O3G) electrocatalyst exhibited an outstanding CO₂ reduction to ethanol with selectivity ≈83%,^{[ 93 ]} outperforming most reported Cu‐based catalysts. This achievement is realized due to the synergistic effect between the SnS₂ nanosheets, Sn single atoms, and the 3D O‐rich carbon support. Therefore, the search for more efficient non‐Cu‐based electrocatalysts with dual‐active sites should attract more attention through experimental and high‐throughput theoretical calculations. Moreover, a deeper investigation of factors that influence the coverage of carbonaceous intermediates and the interaction between the support and the active catalyst species would offer an understanding of the C–C dimerization process.

6. Currently, the experimental results combined with theoretical calculations are becoming an increasingly popular and efficient approach to investigate the electronic properties of electrocatalysts and adsorption energetics of important intermediates in CO₂RR, unveiling to some extent understanding the pathway of sub‐nanometric electrocatalysis. Considering the dynamic changes of active centers in sub‐nanometric catalysts under real operating conditions, the theoretical predictions have limitations in providing rational perceptions of the electrocatalytic reactions. Thus, other simulation techniques must be used to examine the evolution of real active sites during the reaction. Moreover, the shift of catalyst design from trial‐ and error‐based strategies to theoretically guided design approaches is an essential alternative to be effective in the quest for electrocatalytic CO₂ reduction to the desired products.^{[ 126 ]} Additionally, combining machine learning and high‐throughput theoretical calculations allows envisaging the identification of potential sub‐nanometric transition‐metal‐based electrocatalysts for CO₂ transformation to high‐value products.

7. The scalability and techno‐economic development of CO₂RR to valuable products are other dubious issues. The efforts made so far shed light on the scalability of this field due to mainly the suitable integration of CO₂RR with renewable energy sources and the promising progress in catalytic activity, product selectivity, and formation rate. High‐value CO₂ reduction products, particularly C₂₊ products, are attracting much more attention than less‐reduced products (CO and formate) due to their enormous benefits in economic values, applications, high‐energy density, toxicity, storage, and transportation. To date, the consumption of chemicals surges globally with time, even those achievable from CO₂ reduction. For example, ≈158 million tons of ethylene (C₂H₄) was used in 2020, and this number was predicted to become 207 million tons by 2027, increasing the demand by 4.5% every year. Additionally, the market value of these chemicals even increases. For instance, the price of C₂H₄ is ≈1300 US$ per ton, which is much higher than that of formate (740 US$ per ton) and CO (600 US$ per ton).^{[ 127 ]} Surprisingly, CO₂ can be transformed into those chemicals at low cost by subordinating the electrocatalyst system with renewable energy sources. However, the progress of CO₂ reduction to high‐value products over the sub‐nanometric catalysts (SACs, DACs, tandem SAC–NCs, and SNCCs) still is in its infancy, which urges continuous efforts to solve the stability, selectivity, and reaction rate of the active catalysts at industrially relevant current densities. In addition, the issues related to the synthesis and characterization of the sub‐nanometric catalysts can also potentially hamper their scalability.

8. Finally, due to their unique electronic and geometric properties, sub‐nanometric catalysts are getting more attention from the catalysis and energy communities beyond CO₂ reduction. These types of catalysts own highly unsaturated surfaces, numerous active sites, confinement effect, and greater atom utilization efficiency, making them promising candidates for various catalytic applications ranging from two‐electron‐transfer reactions to sophisticated coupling reactions involving multi‐electron‐transfer, such as water splitting,^{[ 128 ]} oxygen reduction reaction,^{[ 73} , ^{102 ]} ammonia synthesis,^{[ 102} , ^{129 ]} and urea formation.^{[ 130 ]} Thus, we reckon that more efficient SACs, DACs, SAC–NCs, and SNCCs will be nurtured in the near future for myriads of catalytic applications.

Figure 12.

Schematic illustration of future development of sub‐nanometric catalysts for CO₂RR to high‐value products.

This review provides a comprehensive summary and insights into both the experimental and theoretical studies of Cu‐ and non‐Cu‐based SACs, DACs, tandem SAC–NCs, and SNCCs for the electroreduction of CO₂ to high‐value products, with particular emphasis on research optimizing the production of C₂₊ products. Despite the abovementioned limitations, the exceptional performances of SACs, DACs, SAC–SNCCs, and SNCCs in CO₂ reduction catalysis are informative for huge future development, particularly in C₂₊ product formation.

Conflict of Interest

The authors declare no conflict of interest.

Supporting information

Supporting Information

Biographies

Abebe Reda Woldu received his Ph.D. in Physical Chemistry from University of Chinese Academy of Sciences, China, in 2020. He then worked as an assistant professor (Jan. 2020‐Aug. 2021) at Bahir Dar University, Ethiopia. He subsequently worked as a postdoctoral researcher at Shantou University, and was selected as an Excellent Talent of Shantou University Outstanding Talents Funding Program. Currently, he holds an assistant professor position at Shantou University. His research interest focuses on nanomaterials synthesis and strategies toward performance enhancement for catalytic applications.

Asfaw G. Yohannes obtained his Ph.D. in chemistry from the Karlsruhe Institute of Technology, Germany, in 2019. He previously worked as a lecturer (2013‐2016) and assistant professor of physical chemistry (2020‐2021) at Bahir Dar University, Ethiopia. He is currently a postdoctoral associate at the University of Calgary, Canada. His research focuses on theoretical and experimental CO₂ conversion and the production of synthetic fuels from CO₂ and CO.

Zanling Huang received her BS degree in chemistry from Guangzhou University in 2019 and her PhD degree in biology from Shantou University in 2024. Her research focuses on electrolyte microenvironment tailoring for electrochemical water splitting and glucose sensor, nanomaterials for photo degradation.

Pierre Kennepohl is physical inorganic chemist whose research focuses on the use of spectroscopy and computational methods to study the electronic structure of complex systems, ranging from molecular catalysts to functional materials. He is currently Associate Dean for Community and Innovation in the Faculty of Science at the University of Calgary, where he works to connect scholars with external partners to maximize the impact of academic research.

Didier Astruc, born in Versailles, studied in Rennes where he defended his PhD with René Dabard before NATO post‐doctoral Fellowship at MIT, Cambridge, Mass with Richard Schrock (2006 Nobel Laureate) and sabbatical leave at UC Berkeley with Peter Vollhardt. He has been Professor of Chemistry at the University of Bordeaux since 1984 and Institut Universitaire de France since 1995. Member of the French, German (Leopoldina) and several European Academies and Fellow of the Royal Society of Chemistry and ChemPubSoc Europe, he has authored 600 papers including several books. His present interests, with his research group, are in nanocatalysis and nanobiomedicine.

Liangsheng Hu received his PhD in chemistry from The Hong Kong Polytechnic University in 2018, then became a faculty member of Shantou University, and is now an associate professor in chemistry. His research interests focus on the synthesis and application of functional nanomaterials in electrochemical and photochemical catalysis. He has co‐authored over 80 peer‐reviewed papers in top scientific journals, such as Chem. Rev., Coordin. Chem. Rev., Angew. Chem. Int. Ed., Nat. Commun., Adv. Energy Mater., and Nano Energy.

Xiao‐Chun Huang is currently a professor at the Department of Chemistry in Shantou University. She received her PhD degree from Sun Yat‐sen University in 2004 under the supervision of Prof. Xiao‐Ming Chen. Her main research interest is the structural design and application of new porous materials in particular CO₂ utilization. To date, she has co‐authored over 100 peer‐reviewed papers, with a total citation number of over 7000 and h‐index of 43.

Woldu A. R., Yohannes A. G., Huang Z., Kennepohl P., Astruc D., Hu L., Huang X.‐C., Experimental and Theoretical Insights into Single Atoms, Dual Atoms, and Sub‐Nanocluster Catalysts for Electrochemical CO₂ Reduction (CO₂RR) to High‐Value Products. Adv. Mater. 2024, 36, 2414169. 10.1002/adma.202414169