Design strategies for dual-atom and multi-atom catalysts: Unlocking synergistic interactions in carbon-based electrocatalysis.

Skip to main content An official website of the United States government Here's how you know Here's how you know Official websites use .gov A .gov website belongs to an official government organization in the United States. Secure .gov websites use HTTPS A lock ( Lock Locked padlock icon ) or https:// means you've safely connected to the .gov website. Share sensitive information only on official, secure websites. Search Log in Dashboard Publications Account settings Log out Search… Search NCBI Primary site navigation Search Logged in as: Dashboard Publications Account settings Log in Search PMC Full-Text Archive Search in PMC Journal List User Guide PERMALINK Copy As a library, NLM provides access to scientific literature. Inclusion in an NLM database does not imply endorsement of, or agreement with, the contents by NLM or the National Institutes of Health. Learn more: PMC Disclaimer | PMC Copyright Notice Fundam Res . 2025 Sep 30;6(2):765–789. doi: 10.1016/j.fmre.2025.09.016 Search in PMC Search in PubMed View in NLM Catalog Add to search Design strategies for dual-atom and multi-atom catalysts: Unlocking synergistic interactions in carbon-based electrocatalysis Xuanni Lin Xuanni Lin a School of Biological and Chemical Engineering, Ningbo Tech University, Ningbo 315100, China b Key Laboratory of Biomass Chemical Engineering of Ministry of Education, College of Chemical and Biological Engineering, Zhejiang University, Hangzhou 310027, China Find articles by Xuanni Lin a, b , Zhengfei Chen Zhengfei Chen a School of Biological and Chemical Engineering, Ningbo Tech University, Ningbo 315100, China Find articles by Zhengfei Chen a, ⁎ , Zhongjian Li Zhongjian Li b Key Laboratory of Biomass Chemical Engineering of Ministry of Education, College of Chemical and Biological Engineering, Zhejiang University, Hangzhou 310027, China Find articles by Zhongjian Li b , Bin Yang Bin Yang b Key Laboratory of Biomass Chemical Engineering of Ministry of Education, College of Chemical and Biological Engineering, Zhejiang University, Hangzhou 310027, China Find articles by Bin Yang b , Qinghua Zhang Qinghua Zhang b Key Laboratory of Biomass Chemical Engineering of Ministry of Education, College of Chemical and Biological Engineering, Zhejiang University, Hangzhou 310027, China Find articles by Qinghua Zhang b , Jianguo Lv Jianguo Lv d State Key Laboratory of Silicon Materials, Key Laboratory for Biomedical Engineering of Ministry of Education, School of Materials Science and Engineering, Zhejiang University, Hangzhou 310027, China Find articles by Jianguo Lv d , Lecheng Lei Lecheng Lei b Key Laboratory of Biomass Chemical Engineering of Ministry of Education, College of Chemical and Biological Engineering, Zhejiang University, Hangzhou 310027, China Find articles by Lecheng Lei b , Yuanyuan Li Yuanyuan Li e Wallenberg Wood Science Center, Department of Fiber and Polymer Technology, KTH Royal Institute of Technology, Stockholm 10044, Sweden Find articles by Yuanyuan Li e , Raul D Rodriguez Raul D Rodriguez f Tomsk Polytechnic University, Tomsk 634050, Russia Find articles by Raul D Rodriguez f , Yang Hou Yang Hou a School of Biological and Chemical Engineering, Ningbo Tech University, Ningbo 315100, China b Key Laboratory of Biomass Chemical Engineering of Ministry of Education, College of Chemical and Biological Engineering, Zhejiang University, Hangzhou 310027, China c Hydrogen Energy Institute, Zhejiang University, Hangzhou 310027, China Find articles by Yang Hou a, b, c, ⁎ Author information Article notes Copyright and License information a School of Biological and Chemical Engineering, Ningbo Tech University, Ningbo 315100, China b Key Laboratory of Biomass Chemical Engineering of Ministry of Education, College of Chemical and Biological Engineering, Zhejiang University, Hangzhou 310027, China c Hydrogen Energy Institute, Zhejiang University, Hangzhou 310027, China d State Key Laboratory of Silicon Materials, Key Laboratory for Biomedical Engineering of Ministry of Education, School of Materials Science and Engineering, Zhejiang University, Hangzhou 310027, China e Wallenberg Wood Science Center, Department of Fiber and Polymer Technology, KTH Royal Institute of Technology, Stockholm 10044, Sweden f Tomsk Polytechnic University, Tomsk 634050, Russia ⁎ Corresponding authors. [email protected] [email protected] Received 2025 Jun 15; Revised 2025 Sep 21; Accepted 2025 Sep 25; Collection date 2026 Mar. © 2025 The Authors. Publishing Services by Elsevier B.V. on behalf of KeAi Communications Co. Ltd. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/). PMC Copyright notice PMCID: PMC13069863  PMID: 41971799 Abstract Carbon-based heteronuclear diatomic and multi-atomic catalysts (e.g., tri-atomic, quadri-atomic, and penta-atomic systems) have emerged as a promising class of materials capable of overcoming the scaling relationship limitations inherent to single-atom catalysts. These advanced catalysts exhibit unique advantages in catalyzing complex reactions involving multi-intermediate processes and proton-coupled electron transfer, offering enhanced activity, selectivity, and tunability. However, the fundamental interaction mechanisms between heteronuclear sites in dual-atom and multi-atom systems remain poorly understood, hindering their rational design. Moreover, conventional synthesis methods often lead to the aggregation of heteronuclear metal atoms, posing significant challenges for the precise structural control required for electrocatalytic applications in the future. This review provides a comprehensive analysis of recent breakthroughs in the field, focusing on the synergistic coupling interactions between diatomic and multi-atomic sites, emerging catalytic mechanism research methods, innovative synthesis strategies for heteronuclear catalysts, and the integration of high-throughput screening and machine learning with theoretical calculations to accelerate catalyst discovery. By elucidating the underlying principles governing these systems, we aim to establish robust design guidelines for heteronuclear diatomic and multi-atomic catalysts in energy conversion and environmental remediation. Furthermore, this review highlights future directions for unraveling catalytic mechanisms and developing scalable fabrication methods, paving the way for the next generation of advanced electrocatalysts. Keywords: Carbon-based heteronuclear metal atom catalysts, Electrocatalysts, Interaction mechanism, Synthesis strategy, High-throughput screening, Machine learning Graphical abstract This review comprehensively summarizes the latest research advancements in carbon-based heteronuclear metal atom catalysts for electrocatalysis, encompassing the intricate interactions between metal atom sites, catalytic mechanism research methods, innovative synthesis methodologies, and the application of cutting-edge technologies for catalyst design. Open in a new tab 1. Introduction Since their inaugural conceptualization in 2011, single-atom catalysts (SACs) have garnered substantial attention across diverse scientific and industrial domains, particularly in energy conversion and environmental remediation. Their unparalleled atomic efficiency, coupled with exceptional catalytic selectivity, positions SACs as a transformative class of materials. SACs are typically characterized by isolated metal atoms, often coordinated with surrounding heteroatoms, which are uniformly dispersed on various substrates, including carbon, metal oxides, metals/alloys, nitrides, carbides, and microporous materials [ [1] , [2] , [3] , [4] , [5] ]. Among these, carbon-based substrates, owing to their earth abundance, cost-effectiveness, environmental compatibility, and multifunctional properties, have emerged as a particularly promising platform for SACs [ 6 ]. Consequently, the precise engineering of single-atom sites on carbon matrices has become a focal point of research, driving significant advancements in electrocatalysis and beyond [ 7 , 8 ]. SACs face a fundamental limitation in their ability to simultaneously optimize the adsorption of all reaction intermediates due to the constraints imposed by linear scaling relationships in thermodynamics, where the adsorption strength of reactants can be too weak or too strong for SACs [ 9 , 10 ]. This inherent drawback significantly restricts their applicability in catalyzing complex electrochemical reactions that require the synergistic activation of multiple molecules and intermediates as well as multistep redox processes, such as the oxygen reduction reaction (ORR), carbon dioxide reduction reaction (CO 2 RR), and nitrate reduction reaction (NO 3 RR), owing to the geometric limitations of isolated metal centers [ 11 , 12 ]. To address this challenge, the emergence and rapid development of diatomic catalysts (DACs) and multi-atomic catalysts (MACs), including tri-atomic catalysts (TACs), quadri-atomic catalysts (QACs), and penta-atomic catalysts (PACs), have provided a promising solution to overcome the limitations of SACs ( Fig. 1 ). Fig. 1. Open in a new tab Chronology of development from carbon-based metal single-atom, dual-atom to multi-atom catalysts. Schedule of representative literatures for the above catalysts in the field of electrocatalysis from 2011 to 2024 [ 4 , 5 , [13] , [14] , [15] , [16] ]. Copyright 2011, Springer Nature; Copyright 2013, Springer Nature; Copyright 2023, Springer Nature; Copyright 2024, Wiley; Copyright 2023, Wiley; Copyright 2017, American Chemical Society. The strategic introduction of foreign metal atoms in proximity to the primary metal center confers significant advantages in targeted electrochemical processes, particularly in facilitating the activation of inert molecular species and enabling efficient C—C bond coupling [ 17 , 18 ]. Crucially, the local electronic environment of the catalytic metal center can be precisely engineered through synergistic interactions between bimetallic or multimetallic sites, which is beneficial to complex multi-electron transfer reactions. Moreover, the charge transfer in heteronuclear bimetallic or multimetallic sites optimizes the binding and release energetics of diverse reaction intermediates, thereby markedly enhancing catalytic efficiency and selectivity [ 19 ]. For example, the mononuclear CuN 4 catalyst requires a single Cu site to simultaneously mediate both CO 2 activation and water dissociation, demonstrating a high energy barrier for the COOH*→CO conversion. While Te sites demonstrate exceptional CO 2 activation capability, their poor water dissociation activity creates a kinetic bottleneck through insufficient proton supply. The TeN 2— CuN 3 diatomic catalyst overcomes these challenges through a synergistic mechanism where the Te site facilitates CO 2 activation while the Cu site specifically promotes water dissociation. This synergistic division of labor dramatically reduces the energy barrier of the key steps and improves the reaction kinetics [ 20 ]. Furthermore, DACs and MACs can effectively suppress competing side reactions in electrocatalytic processes, significantly enhancing the selectivity toward desired products [ 12 , 21 ]. Notably, in certain electrochemical reactions, DACs and MACs also contribute to improved catalytic stability, further underscoring their potential for practical applications. For example, the incorporation of Fenton-inactive metals adjacent to Fe sites induces a synergistic effect, substantially enhancing catalytic durability [ 22 , 23 ]. Such advancements underscore the potential of multi-atomic configurations to overcome intrinsic limitations of single-atom systems, offering a robust platform for next-generation electrocatalysts. Despite these advancements, fundamental scientific challenges in DACs/MACs research remain to be uncovered. Firstly, the dynamic interplay between heterometallic sites (e.g., site-distance effect, synergistic effect) during electrocatalysis lacks atomic-level understanding, hindering rational catalyst optimization. However, as the number of metal atoms increases to form MACs, the interaction mechanisms among individual atomic sites become increasingly intricate and less well-defined for specific electrocatalytic reactions. Therefore, cutting-edge characterization techniques must be further developed to resolve current limitations of the dynamic coordination environments of multi-metal centers and unclear transient reaction intermediates, which is critical for achieving atomic-level understanding of catalytic mechanisms. Further studies have provided detailed insights into the interaction mechanisms between dual metal sites across a range of electrolysis reactions, particularly in complex catalytic processes involving multiple proton-coupled electron transfer steps [ 24 , 25 ]. Secondly, precise construction of heteronuclear diatomic/multi-atomic sites requires overcoming metal aggregation and stabilizing heteronuclear metal, which demands advanced synthetic strategies. Simultaneously, the synthesis of MACs with uniformly dispersed single metal sites, while preventing metal agglomeration during high-temperature pyrolysis, often necessitates reducing metal loading and introducing substrate defects to anchor additional metal atoms [ 26 , 27 ]. These challenges underscore the urgent need for the development of robust synthetic strategies tailored to multi-atomic systems. Furthermore, the vast combinatorial space of metal pairs for DACs and MACs renders exhaustive experimental exploration impractical. To address this limitation, first-principles calculations have emerged as a powerful complement to experimental screening. In recent years, advanced computational techniques such as high-throughput screening and machine learning (ML) have gained considerable traction. When integrated with theoretical modeling, these approaches enable the high-precision prediction of carbon-based heteronuclear metal atom electrocatalysts for ORR, CO 2 RR, NO 3 RR, and other electrocatalytic processes [ [28] , [29] , [30] ]. This synergy between computational and experimental methodologies offers a transformative pathway for expediting the discovery and refinement of next-generation electrocatalysts, enabling the development of more efficient energy conversion technologies. In this review, we provide a comprehensive and systematic analysis of the interactions between metal sites in carbon-based heteronuclear diatomic and multi-atomic catalysts. By integrating representative case studies from the literature, we elucidate the fundamental mechanisms governing the interplay of distinct metal sites, offering atomic-level insights into the structure-activity relationships that underpin electrocatalytic performance. Meanwhile, state-of-the-art characterization techniques enabling electrocatalytic mechanistic insights are systematically summarized. Drawing on these foundational principles, we present a strategic framework for the controlled synthesis of DACs and MACs, outlining tailored preparation methodologies to expand their utility in electrocatalytic applications. Furthermore, we emphasize the transformative role of high-throughput screening and ML technologies in expediting the discovery and optimization of multi-metallic SACs. Finally, we explore the potential of these advanced computational tools to revolutionize the design and synthesis of multi-metallic SACs, enabling their deployment in high-performance applications such as energy conversion, environmental sustainability, and related electrochemical processes. This review not only bridges fundamental understanding with practical innovation but also sets the stage for the development of next-generation electrocatalysts with tailored functionalities and enhanced efficiency. 2. Interaction between heteronuclear atomic sites The rapid advancement and widespread implementation of in situ spectroscopic characterization techniques, such as in situ X-ray absorption spectroscopy (XAS), in situ attenuated total reflection Fourier-transform infrared (ATR-FTIR) spectroscopy, and in situ Raman spectroscopy, have profoundly enhanced our understanding of the intricate interactions between heteronuclear metal atomic sites in DACs, TACs, and PACs catalysts [ 31 , 32 ]. Five central concepts and regulatory approaches have been established to elucidate the interactions in DACs: (1) the electronic effect, which describes the modulation of the electronic structure of metal sites induced by neighboring atoms, influencing adsorption energetics and reaction pathways; (2) the synergistic effect, arising from the cooperative interaction between dual metal sites, which enhances catalytic properties beyond those achievable with SACs; (3) the site-distance effect, which highlights the critical role of spatial arrangement and interatomic distance in governing electronic coupling and reactant accessibility; (4) the spin effect, which (through spin-dependent interactions) adjusts metal spin states, Fermi levels, and reaction pathways by influencing magnetism, electron spin moments, and oxidation states at active sites; (5) the orbital hybridization, which establishes strategic hybridization of atomic orbitals to strengthen metal-metal interactions, preventing the leaching of d/p-block metals and optimizing catalytic performance by tailoring surface electronic structures. These effects collectively dictate the catalytic activity, selectivity, and stability of DACs. Moreover, the interactions between metal atomic sites in TACs and PACs have garnered significant attention in recent years [ 33 ]. Understanding these interactions is essential for tailoring the electronic structure and spatial configuration of catalytic sites, which are critical for achieving superior performance in energy conversion and environmental remediation applications. In this section, we summarize recent research findings to comprehensively elucidate these interaction mechanisms, providing a foundation for the rational design and synthesis of dual-atom and multi-atom electrocatalysts. By leveraging these insights, researchers can develop advanced catalysts with precisely tuned properties, ultimately improving the efficiency of energy conversion processes and addressing pressing environmental challenges. 2.1. Binuclear atomic structure 2.1.1. Electronic effect In contrast to SACs, DACs offer a compelling strategy to circumvent the intrinsic limitations imposed by adsorption energy scaling relationships, enabling precise optimization of key intermediate adsorption in complex multi-molecular and multi-step reactions, such as the ORR, CO 2 RR, and NO 3 RR. A defining feature of DACs is their electronic effect, wherein one metal center acts as the primary active site for adsorbing reaction intermediates, while the adjacent metal center functions as an electron regulator. This synergistic configuration allows for fine-tuning of the electronic environment at the active site, thereby enhancing catalytic efficiency and breaking traditional activity-selectivity trade-offs. For instance, Pan’s group proposed a proximity electronic effect of Ni/Co diatomic sites for boosting ORR and hydrogen evolution reaction (HER) performances [ 34 ]. The synthesis of Ni/Co diatomic sites anchored on a nitrogen-doped carbon substrate (NiCo DASs/N—C) was achieved through a metal-organic frameworks (MOFs)-assisted host-guest strategy. Advanced in situ characterization techniques, including in situ Raman spectroscopy and in situ attenuated total reflectance-surface enhanced infrared absorption spectroscopy (ATR-SEIRAS), revealed that the *OOH and *OH intermediates were preferentially adsorbed on the Co active site. Notably, the single-atomic Co site demonstrated strong interactions with the reactant O 2 molecule ( Fig. 2 a, b), unequivocally identifying the Co-N 4 configuration as the primary active site for the ORR. Complementary density functional theory (DFT) calculations and ab initio molecular dynamics (AIMD) simulations further elucidated the electronic effect in NiCo DASs/N—C. These studies demonstrated that the adjacent Ni-N 4 site functions as an electronic modulator, effectively fine-tuning the electronic localization of the single-atomic Co-N 4 site. Benefiting from the modulation effect of adjacent Ni-N 4 moiety, the *OH adsorption became weaker and the *H adsorption was enhanced on the Co-N 4 site. As a consequence, the NiCo DASs/N—C catalyst exhibited exceptional ORR and HER performance, characterized by high intrinsic activity and remarkable long-term durability ( Fig. 2 c, d). Fig. 2. Open in a new tab Electronic effect in binuclear atomic structure. In situ ATR-SEIRAS spectra of NiCo DASs/N—C in (a) 0.1 M KOH and (b) 0.5 M H 2 SO 4 at varying potentials during the ORR process. Corresponding free energy diagrams for (c) ORR and (d) HER on Ni SAs/N—C, Co SAs/N—C, and NiCo DASs/N—C [ 34 ]. Copyright 2023, Wiley. (e) Structural and adsorption configurations of critical intermediates on the Co site of Co-N-Ni bimetallic centers (C: brown, H: white, O: red, N: blue, Ni: cyan, and Co: green). (f) Free energy diagrams for CO 2 RR on different sites. (g) FEs of CO for the Co-N/NPCNSs, Co-N-Ni/NPCNSs, and Ni-N/NPCNSs catalysts [ 35 ]. Copyright 2021, Royal Society of Chemistry. Similarly, the electronic effect plays a pivotal role in the CO 2 RR, as it can significantly lower the energy barrier of the rate-determining step, thereby facilitating efficient CO 2 conversion. Zhuang and colleagues demonstrated this principle by designing N-bridged Co-N-Ni bimetallic sites dispersed on nitrogen-doped porous carbon nanosheets (Co-N-Ni/NPCNSs) [ 35 ]. In situ XAS measurements revealed dynamic changes in the oxidation states of the Co and Ni centers, as well as variations in the bond lengths of the bridging N atoms between Co and Ni. Additionally, a redistribution of electrons in the Co/Ni 3d orbitals was observed during the CO 2 RR process. These findings strongly suggest that the Co-N-Ni bimetallic sites act as the active centers for catalyzing CO 2 RR. Theoretical calculations further corroborated that the N-bridged bimetallic sites facilitate the formation of the critical *COOH intermediate and enhance CO desorption, which are essential steps for efficient CO 2 RR ( Fig. 2 e, f). With the tailored bimetallic effect between N-bridged Co-N-Ni sites, the Co-N-Ni/NPCNSs catalyst achieved remarkable performance, exhibiting a high Faradaic efficiency for CO (FE CO ) of > 90% and an exceptional turnover frequency of 2049 h -1 at a low overpotential of 370 mV, which outperforms counterparts ( Fig. 2 g). 2.1.2. Synergistic effect In addition to inheriting the intrinsic catalytic properties of SACs, DACs offer a unique advantage by providing bridging adsorption sites that enable synergistic modulation of the electronic structure, thereby enhancing electrocatalytic performance. When the distance between heteronuclear metal diatomic centers is sufficiently reduced, a strong chemical interaction arises, allowing reactants to be simultaneously adsorbed on both metal sites. This dual-site adsorption facilitates the establishment of interactions between the diatomic active centers and the reactants, effectively balancing catalytic activity and stability. This phenomenon is commonly referred to the synergistic effect of DACs. For instance, Lu and colleagues demonstrated the synergistic effects of dual metal atoms for high-efficiency NO 3 RR [ 36 ]. At the core of their findings, the NO 3 intermediates exhibited strong adsorption at the bridge sites of the diatomic centers, while the moderate interactions with most reaction intermediates at the Fe/Cu diatomic sites collectively reduced the overall energy barriers for NO 3 RR. This synergistic interplay resulted in exceptional catalytic activity and selectivity for NH 3 production. The Fe/Cu dual-atom system was designed to form a unique “N 2 Fe-CuN 2 ” coordination structure, featuring a metal-metal dimer where each metal atom is coordinated with two nitrogen atoms. This configuration was anchored at the holey edge sites of nitrogen-doped graphene, providing a stable and highly active catalytic environment. Computational and theoretical analyses revealed that the two oxygen atoms of the NO 3 - intermediate were strongly attracted to the Fe/Cu dual-atom sites. The binding energy of NO 3 - on the Fe/Cu dual sites was significantly stronger than that on single-atom Fe or Cu sites, effectively lowering the energy barrier of the initial anionic adsorption step, which is the rate-determining step in the NO 3 RR. The heteronuclear Fe/Cu dual-atom structure also facilitated the weakening of N—O bonds, thereby reducing the energy barriers for subsequent reaction steps and enabling a more efficient overall reaction pathway ( Fig. 3 a, b). Accordingly, the system demonstrated a remarkable activity of ∼38.5 mA cm -2 at −0.3 V and an impressive NH 3 selectivity of 92.51%. Furthermore, the catalyst achieved a high NH 3 yield rate of 1.08 mmol h -1 mg -1 at −0.5 V, underscoring its superior efficiency for NO 3 RR ( Fig. 3 c, d). Fig. 3. Open in a new tab Synergistic effect in binuclear atomic structure. (a) Reaction mechanism of NO 3 − RR and adsorption models of intermediates. (b) Free energy diagrams for intermediate states on Fe/Cu bimetallic sites at -0.3 V. (c) Polarization curves for Fe/Cu-HNG, Fe-HNG, Cu-HNG, and HNG in 1.0 M KOH with 0.1 M KNO 3 . (d) FEs of NH 3 for Fe/Cu-HNG at various potentials [ 36 ]. Copyright 2023, Springer Nature. (e) CO 2 adsorption configurations on FeCo-NC, Fe-NC, and Co-NC after geometry optimization. (f) Spin densities of CO assisted FeCo-NC. The yellow and cyan regions stand for spin up and spin down, respectively. (g) Charge density difference maps for CO adsorption on FeCo-NC and CO-assisted FeCo-NC, where yellow regions denote electron accumulation and cyan regions denote electron depletion. (h) The free energy diagrams of the eight-electron CO 2 RR pathway to CH 4 on Fe-NC, Co-NC, FeCo-NC, and CO-assisted FeCo-NC at 0 V. (i) The free energy diagrams of the six-electron CO 2 RR pathway to CH 3 OH on Fe-NC, Co-NC, FeCo-NC, and CO-assisted FeCo-NC at 0 V [ 37 ]. Copyright 2023, Wiley. Furthermore, the detailed mechanism underlying the synergistic effect of diatomic sites in electrocatalysis has been extensively investigated and analyzed through first-principles calculations. Using DACs such as FeCo-NC and SACs such as Fe-NC and Co-NC as representative examples, the advantages of DACs in the CO 2 RR have been clearly elucidated [ 37 ]. As illustrated in Fig. 3 e, the adsorption of CO 2 molecules on Fe-NC and Co-NC is relatively weak, maintaining a linear configuration, whereas FeCo-NC exhibits strong chemical adsorption, bending the CO 2 molecule and facilitating its activation. The diatomic centers of Fe and Co collaboratively provide bridge adsorption sites, enabling effective activation of CO 2 . In addition, the Fe-Co dual-atom center possesses surplus electrons and exhibits greater spin polarization compared to single-atom Fe or Co centers. This electronic configuration contributes to the stable adsorption of CO 2 and the efficient adsorption and activation of the *CO intermediate ( Fig. 3 f, g). As a result, the stable adsorption of *CO intermediates promote further progression of CO 2 RR along multi-electron pathways, leading to the generation of C1 products such as methanol (CH 3 OH) and methane (CH 4 ) ( Fig. 3 h, i). Besides, the synergistic effect in heteronuclear DACs engineers’ catalytic sites for multi-step reactions, offers a promising strategy for high-energy-density lithium-sulfur (Li-S) batteries and lithium-oxygen (Li-O 2 ) batteries. Zhang and coworkers designed DACs with Fe-Co dual sites anchored on nitrogen-doped hollow carbon spheres for Li-S batteries [ 38 ]. The unique dual-atom synergy significantly improves reaction kinetics, as evidenced by a low overpotential (0.136 V) and high-rate performance (688 mAh g - ¹ at 5 C). In situ Raman spectroscopy revealed rapid LiPSs conversion on Fe-Co DACs, with effective suppression of shuttle effects. DFT calculations elucidated that the interaction between Fe-Co diatomic sites accelerate the electron transfer in the kinetic transformation process, promote the nucleation and decomposition of LiPSs, and lower the Gibbs free energy for LiPSs reduction. Notably, the DACs-enabled cells achieved exceptional cycling stability (0.018% capacity decay per cycle over 1000 cycles at 1.0 C), outperforming single-atom counterparts (Fe/Co SACs) and physical mixtures, underscoring the efficient dual-function catalytic role of atomic-pair synergy during charging and discharging processes simultaneously. As for Li-O 2 battery, Ryu’s group demonstrated the exceptional catalytic performance of heteronuclear Fe-Ni dual-atom sites anchored on nitrogen-doped carbon nanotubes (NCNTs), achieving stability and activity in Li-O 2 systems [ 39 ]. Particularly, the sample loaded with Fe first (Ni-Fe-NCNTs) achieves a remarkable lifespan of 200 cycles and maintains a consistent ORR/oxygen evolution reaction (OER) overpotentials. The sequential loading of Fe and Ni on mesoporous defect sites of NCNTs not only enhances metal loading but also optimizes the synergistic interaction between the two metals. As revealed by DFT calculations, the Ni-Fe-NCNT configuration showed the lowest overpotentials ( η ORR : 0.71 V, η OER : 1.74 V) due to efficient charge redistribution, where DACs with Fe loaded first plays the maximized role to maintain a stable balance between NO 2 - mediation and DAC-associated reactions. Additionally, the Ni-Fe-NCNT showed enhanced capabilities in stabilizing NO 2 and reducing overpotential compared with homogeneous Fe-Fe DACs, indicating the significance of the catalyst combination of Ni and Fe. 2.1.3. Site-distance effect As the distance between heteronuclear diatomic sites varies, the interaction between neighboring metal single-atom sites either intensifies or weakens, giving rise to the site-distance effect. This effect plays a critical role in modulating the electronic and catalytic properties of DACs [ 40 , 41 ]. Recent studies have explored innovative strategies to harness the site-distance effect for fine-tuning electrocatalytic activity. For instance, by increasing the site density to a specific threshold, the distance between metal atoms is reduced, leading to the formation of adjacent diatomic sites that exhibit mutual coordination and electronic coupling [ 41 ]. Fig. 4 a illustrates five optimized models with varying atomic distances between Fe and Rh atoms, constructed using DFT calculations, as reported by Yao and colleagues [ 41 ]. Combined with theoretical calculations, electrostatic potential (ESP) analysis, and Bader charge analysis, the study revealed that different Fe-Rh atomic distances ( d Fe-Rh ) can modulate the electronic structure of Fe-Rh dual-atom catalysts supported on nitrogen-doped graphene (Fe-Rh x @NC) through the site-distance effect. Specifically, the FeRh-N 6 configuration exhibited a stronger Fe-Rh interaction compared to the Fe-N 4 /Rh-N 4 -1 configuration ( Fig. 4 a). This enhanced interaction led to a slight decrease in the ESP signal of the Fe atom in the FeRh-N 6 configuration, weakening the adsorption of nucleophiles. Furthermore, Bader charge analysis demonstrated that electrons were transferred from Rh to Fe in the FeRh-N 6 configuration. This electron transfer caused the valence electrons of the Fe atom to approach saturation, making it more challenging for nucleophiles to adsorb onto the Fe site ( Fig. 4 b). Consequently, variations in the d Fe-Rh distance significantly influence the adsorption capacity of Fe atoms, directly impacting the ORR performance. By employing a spatial confinement strategy to tune metal loading and vary site density, a series of Fe-Rh x @NC were designed and synthesized. Characterization results revealed that Fe-Rh 2 @NC possesses the optimal d Fe-Rh distance, which maximizes the site-distance effect. This configuration exhibited exceptional intrinsic ORR activity, achieving a half-wave potential (E 1/2 ) of 0.91 V, significantly surpassing that of commercial Pt/C (0.86 V). Fig. 4. Open in a new tab Site-distance effect in binuclear atomic structure. (a) The optimized structure of different atomic distances of Fe-N 4 , FeRh-N 6 , Fe-N 4 /Rh-N 4 -1, Fe-N 4 /Rh-N 4 -2, and Fe-N 4 /Rh-N 4 -3. (b) The O 2 adsorption energy of calculation models [ 41 ]. Copyright 2024, Springer. (c) Schematic illustration of O 2 adsorption configuration on M-M-N x porphyrin catalysts. (d) Correlation between distance effect and O—O bond cleavage. (e) Free energy profiles for Fe-Zn@SNC models with varying Fe-Zn distances [ 40 ]. Copyright 2023, Wiley. In-situ SR-IR contour map of the (f) Ir-NC catalyst and (g) Ir–Co-7.9 Å catalyst. (h) Reaction pathway of FAOR [ 42 ]. Copyright 2024, American Chemical Society. As another compelling illustration of the site-distance effect in ORR catalysis, a bi-functional ligand-assisted synthesis strategy was developed to precisely control the atomic spacing of Fe and Zn diatomic pairs anchored on a carbon substrate (Fe-Zn@SNC) [ 40 ]. Experimental evidence demonstrated that the interatomic distance between heteronuclear metal centers plays a pivotal role in achieving exceptional ORR performance. DFT calculations revealed that Fe-Zn sites with an interatomic distance of approximately 3 Å exhibit an optimal charge distribution, facilitating strong interactions with the antibonding orbitals of oxygen molecules ( Fig. 4 c–e). This finely tuned site-distance effect streamlines the dissociative pathway and substantially lowers the energy barriers for O—O bond cleavage, thereby enhancing ORR activity. Consequently, the Fe-Zn@SNC catalyst exhibited remarkable performance, achieving a E 1/2 of 0.86 V in acidic electrolyte. Moreover, by integrating ex-situ infrared spectroscopy with in situ synchrotron infrared spectroscopy (SR-IR) techniques, Yao and colleagues elucidated the heterogeneous dual-site spatial distance effect on synergistic catalysis [ 42 ]. By systematically varying the Ir-Co pair distances from 9.8 Å, 7.9 Å, and 5.8 Å to 3.6 Å, they observed that the electrocatalytic activity for the formic acid oxidation reaction (FAOR) followed a volcano-shaped trend. In situ X-ray absorption fine structure (XAFS) and in situ SR-IR analyses revealed that the oxyphilic metal Co accelerates hydrolysis to generate *OH species, while interfacial *OH facilitates the deprotonation process of formic acid. This mechanism is highly dependent on the spatial distance between the dual-site coupling catalysts. When the Ir-Co pair distance was optimized to 7.9 Å, the production and consumption of *OH reached a dynamic balance, which is critical for achieving high-efficiency synergism ( Fig. 4 f–h). This optimal distance avoids spatial inaccessibility and prevents *OH accumulation, which typically occurs at either excessively long or short distances due to resistance between *OH and reaction intermediates. As a result, the Ir-Co dual-site catalyst with a 7.9 Å spacing demonstrated superior FAOR performance, highlighting the importance of precise spatial control in DACs. 2.1.4. Spin effect The electronic structure and catalytic performance of magnetic-ion-containing catalysts can be modulated by spin-dependent interactions. In DACs, introduction of two adjacent metal atoms can form interacting orbitals or shared electron pairs and generate adjustable spin effect, achieving the regulation of metallic spin states and Fermi levels at metal sites. Recent studies found that the spin effect is reflected in the change of magnetism, electronic spin moments, and oxidation states of metal sites, enabling optimization of reaction pathways to concurrently enhance both catalytic activity and selectivity [ 43 , 44 ]. In the context of oxygen electrocatalysis, Guo and coworkers designed an atomically dispersed Fe,Cu/N—C catalyst, leading to new advances of electronic spin state modulation [ 43 ]. Zero-field cooling (ZFC) temperature-dependent magnetic susceptibility (M-T) tests revealed that the Fe sites in the Fe/N—C catalyst, featuring an FeN 4 moiety, adopt a lower spin state compared to those in the Fe,Cu/N—C catalyst, leading to a higher population of unpaired electrons in the latter ( Fig. 5 a). Complementary 57 Fe Mössbauer spectroscopy confirmed that the spin state of Fe in Fe,Cu/N—C is modulated by short-range electron interactions upon coupling FeN 4 with CuN 4 ( Fig. 5 b). Thereafter, the single d z² electrons of medium-spin Fe sites in Fe,Cu/N—C readily interact with the π*-orbitals of O₂ molecules, facilitating ORR kinetics ( Fig. 5 c). This results in a remarkable E 1/2 of 0.86 V, outperforming Fe/N—C (0.79 V) and Cu/N—C (0.77 V). Additionally, on the basis of theoretical screening, a unique half-metallic electronic structure induced by Fe and Zn atoms for ORR catalysis is proposed by Li and colleagues [ 45 ]. After the introduction of Zn, a transition from semiconductor to half-metal was also observed in the Fe-N-C and Fe/Zn-N-C catalysts ( Fig. 5 d, e), with spontaneous spin-polarized electrons filling at the Fermi level. Owing to the inherently narrow 3d band of Fe-N-C broadens upon Zn incorporation, the flow electron is positive to facilitate enhanced charge transfer from the catalyst surface to reaction intermediates. Furthermore, the spin-polarized catalyst generates a stray magnetic field that aligns the spins of the two unpaired electrons in the antibonding π* orbitals of paramagnetic O 2 , promoting stronger interaction with the above fully spin-polarized flowing electrons. This effect improves the capture of O 2 molecules. Consequently, the Fe/Zn-N-C demonstrates outstanding ORR performance in both alkaline and acidic media, achieving E 1/2 of 0.906 V and 0.806 V, respectively. Fig. 5. Open in a new tab Spin effect in binuclear atomic structure. (a) M-T curves and the corresponding calculated number of unpaired 3d electrons for all samples. (b) 57 Fe Mössbauer spectrum of Fe,Cu/N—C catalyst. (c) The transformation and orientation of the five d orbitals [ 43 ]. Copyright 2024, Elsevier. Proposed electronic structures of (d) Fe/Zn-N-C and (e) Fe-N-C active sites [ 45 ]. Copyright 2022, Royal Society of Chemistry. Magnetic susceptibility measurements of (f) FePPc and (g) FeMoPPc [ 46 ]. Copyright 2021, Wiley. Spin effect in DACs for complex multistep reactions, such as the nitrogen reduction reaction (NRR), has also promoted more efficient electrocatalysis by atomically regulating the spin state of metal centers. Zhang’s group fabricated a Fe-Mo h-SA electrocatalyst (FeMoPPc), where FeN 4 cooperated with MoN 4 active sites are co-anchored in polyphthalocyanine organic framework [ 46 ]. ZFC temperature-dependent magnetic susceptibility tests revealed a spin-state transition of the Fe centers from high-spin to medium-spin and a decreased of the effective magnetic moment for the Fe sites from 2.43 to 1.89 μeff after adding Mo ( Fig. 5 f, g). 57 Fe Mössbauer spectroscopy validated the dominant presence of Fe II in a medium-spin state within the FeMoPPc. Therefore, this electronic configuration characterized by partially vacant d orbitals and separate d electron facilitates the acceptance of lone-pair electrons from N 2 and promotes effective orbital overlap with the N 2p states. Consequently, the N Created by potrace 1.16, written by Peter Selinger 2001-2019 N triple bond is weakened, lowering the kinetic barrier for the initial hydrogenation of N 2 . Thus, the FeMoPPc exhibited outstanding electrocatalytic NRR performance, achieving faradaic efficiencies 2.0- and 9.0-fold higher, and NH 3 yields 2.0 and 17.2 times greater than those of FePPc and MoPPc, respectively. 2.1.5. Orbital hybridization Orbital hybridization strategies in hetero-diatom catalysts also demonstrate significant regulation effect by establishing strong interaction. These strategies effectively prevent leaching of lattice d-block and p-block metals while enhancing ORR performance through optimized surface electronic structures [ 47 ]. Constructing paired metal atom sites (e.g., Fe-Co, Fe-Mn, Fe-Al) through the incorporation of d-block and p-block elements enables precise tuning of the transition metal center's local coordination environment in DACs [ 48 , 49 ]. This approach allows for strategic modification of key properties including d-band electronic property and surface chemisorption behavior. Chen and co-workers explored the phenomenon of p-d orbital coupling between the Zn-Sn diatomic sites [ 50 ]. DFT calculations were conducted to investigate the effect of p-d orbital coupling over asymmetrical Zn-Sn dimer in Zn 1 Sn 1 /SNC model for CO 2 RR performance. Partial density of states (PDOS) and crystal orbital overlap population (COHP) analyses confirm strong p-d hybridization between Zn-3d and Sn-5p orbitals near the Fermi level, with an integrated COHP (ICOHP) value of −0.27. For the Sn-S/N coordination, PDOS highlights broad energy superpositions between Sn-5s/5p and S-3p (or N-2p) orbitals in the −10 to 0 eV range (ICOHP = −1.60 for Zn₁Sn₁/SNC vs. −1.39 for Sn₁/SNC and −1.33 for Sn₁/NC). Zn₁Sn₁/SNC exhibits the strongest Sn-S binding, thus favoring the stabilization of charge states at the Sn site. PDOS/COHP analyses reveal Sn-5s/5p and O-2p orbital hybridization forms bonding states (−10 to −3 eV) and antibonding states (−3  eV to Fermi level) of Sn-O bond. The higher intensity of antibonding states in Sn₁/NC and Sn₁/SNC results in weaker Sn-O binding, while Zn₁Sn₁/SNC demonstrates the most robust Sn-O bond strength for HCOO* adsorption ( Fig. 6 a–c). Fig. 6 d presents the relationship between ΔG PDS relative to U L (CO 2 )-U L (H 2 ) and the p band center (E p ), showing Zn 1 Sn 1 /SNC delivers superior CO 2 RR performance with both the smallest ΔG PDS value of 0.37 eV and the largest U L (CO 2 )-U L (H 2 ) value of 0.61 V for selective HCOOH production. Fig. 6. Open in a new tab Orbital hybridization in binuclear atomic structure. (a–c) pCOHP analysis of Sn and O atoms in the HCOO* species adsorbed on Sn 1 /NC, Sn 1 /SNC, and Zn 1 Sn 1 /SNC catalysts. (d) Comparison of ΔG PDS and U L (CO 2 )-U L (H 2 ) (V vs. RHE) for Sn 1 /NC, Sn 1 /SNC, and Zn 1 Sn 1 /SNC [ 50 ]. Copyright 2025, Springer Nature. (e) Gibbs free energy profiles of ORR intermediates absorbed on FeSn-C 2 N, Fe-C 2 N, and Sn-C 2 N. (f) Gibbs free energy diagram for the ORR pathway on FeSn-C 2 N, Fe-C 2 N, and Sn-C 2 N. (g) The dimer structure of active FeSn dual atom sites. (h) ORR polarization curves of C 2 N, Fe-C 2 N, Sn-C 2 N, FeSn-C 2 N, and Pt/C at 1600 rpm [ 51 ]. Copyright 2024, American Chemical Society. Similarly, Chen’s group also revealed that p-d orbital hybridization induced by the asymmetric coordination geometry of FeSn dual-atom sites accelerates the ORR [ 51 ]. The spherical symmetry of p orbitals and the more complex shape of d orbitals facilitate orbital hybridization between p-block metals and 3d transition metals. Since p orbitals typically lie at higher energies than d orbitals, p-d hybridization offers a more effective strategy for electronic structure modulation compared to conventional d-d hybridization [ 51 ]. First, the free energy (ΔG) analysis reveals that the potential-determining step of the ORR on Fe-C 2 N, Sn-C 2 N, and FeSn-C₂N is consistently the *OH → H 2 O step ( Fig. 6 e). Notably, *OH adsorption strength on FeSn-C 2 N (0.66 eV) is weaker than that on Sn-C 2 N (0.19 eV) and Fe-C 2 N (−0.18 eV) ( Fig. 6 f). This reduced *OH binding energy facilitates the subsequent desorption process, promoting water formation and thereby improving ORR kinetics. Moreover, the asymmetric FeSn dimer structure induces unique p-d orbital hybridization, which contributes to the outstanding ORR performance of as-synthesized FeSn-C 2 N DACs, as evidenced by a high E 1₁/2 of 0.914 V in experimental measurements ( Fig. 6 g, h). These findings highlight the role of orbital hybridization in tuning active-site electronics for efficient and selective CO 2 and O 2 electroreduction. In summary, the incorporation of a second single-atom site in DACs introduces unique chemical interactions between adjacent atomic centers while simultaneously modulating the electronic structure of the catalytic sites. The electronic effect, synergistic effect, site-distance effect, spin effect and orbital hybridization collectively govern the interactions between heteronuclear diatomic sites, influencing the adsorption and desorption dynamics of reactants and intermediates, thereby enhancing catalytic efficiency. Experimental studies have demonstrated that DACs not only provide top-site adsorption but also enable bridge-site adsorption, which facilitates the stabilization of complex reaction intermediates and ultimately improves electrocatalytic performance. These insights underscore the importance of understanding the interplay between bimetallic atomic sites in DACs, offering a pathway to optimize catalytic activity for a wide range of applications in energy conversion and environmental remediation. 2.2. Tri-nuclear atomic structure The aforementioned DACs have showcased significant advantages in various electrocatalytic reactions due to their highly tunable electronic structures and exceptional atomic utilization efficiency. Recent studies suggest that introducing a heterogeneous metal atom to create asymmetric heteronuclear trimetallic site catalysts can further optimize the adsorption energies of reaction intermediates, thereby enhancing electrocatalytic performance [ 52 , 53 ]. For instance, Zeng and colleagues reported a CoFeCu tri-atomic catalyst (CoFeCu-TAC) featuring a trinuclear active structure, which demonstrated remarkable and stable oxygen electrocatalysis performance [ 54 ]. When employed as a cathode in aqueous Zn-air batteries, the TAC-based system exhibited exceptional rechargeable stability and ultra-robust cycling performance. It achieved 1970 h over 3940 cycles at 2 mA cm -2 and 65 h over 130 cycles at 10 mA cm -2 , with negligible voltage decay. Based on XANE analysis and HAADF-STEM results, triangular CoFeCu-TAC models featuring CoN 4 -FeN 4— CuN 4 sites with an average metal-metal distance of 5.30 Å were constructed, which exhibited the lowest formation energy among all configurations ( Fig. 7 a). Comprehensive experimental and theoretical analyses revealed that the symmetry-breaking CoN 4 configuration in the TAC serves as the primary catalytic site. The asymmetric coordination geometry, induced by Fe/Cu incorporation, breaks the original CoN 4 symmetry and triggers orbital reconfiguration, thus leading to energy level modulation and charge redistribution ( Fig. 7 b, c). The heterometallic interactions cause downward shifts of d band center in Co 3d orbitals in CoFeCu-TAC compared to Co-SAC, weakening oxygen-containing intermediate adsorption. Bader charge analysis reveals electron accumulation at the Co sites in CoFeCu-TAC compared to monometallic analogs, facilitating O₂ activation. As a result, orbital hybridization collectively reduces ORR/OER overpotentials by optimizing the potential-determining steps, demonstrating d-orbital engineering in multiatomic systems can surpass single-atom catalytic limitations. Fig. 7. Open in a new tab Interactions in tri-nuclear atomic structure. (a) HAADF-STEM image of CoFeCu-TAC. (b) Calculated differential charge density. Co-N bond lengths in CoFeCu-TAC and Co-SAC. (c) Co 3d-orbital energy levels in CoFeCu-TAC and Co-SAC [ 54 ]. Copyright 2024, American Chemical Society. (d) Schematic atomic interface model of ZnN 3 CoN 3 FeN 2 S site in ZnCoFe-TAC/SNC derived from the EXAFS results. (e) The calculated total theoretical overpotential volcano map with ΔG O* − ΔG OH* and ΔG OH* as descriptors [ 55 ]. Copyright 2024, Royal Society of Chemistry. (f) Charge density differences of FCN-TM/NC. (g) Fe 3d-orbital energy levels in FCN-TM/NC and Fe-N-C [ 56 ]. Copyright 2025, Wiley. Furthermore, a ternary atomic ZnCoFe catalyst anchored in sulfur/nitrogen-doped carbon (ZnCoFe-TAC/SNC) was synthesized using a two-step wet chemistry strategy [ 55 ]. A unique ZnN 3 CoN 3 FeN 2 S configuration was established, characterized by N-coordinated single Co and Zn atoms positioned adjacent to S/N dual-coordinated Fe atoms ( Fig. 7 d), as confirmed by aberration-corrected high-angle annular dark-field scanning transmission electron microscopy (AC—HAADF-STEM) and XAS analysis. In situ XAS results revealed that Co serves as the primary active site, while Fe acts as a co-catalytic site. Additionally, the Zn atom functions as an “electron regulator” modulating the electronic structures around the catalytic sites. The heteronuclear multi-atomic configuration, enabled by the interactions among Fe, Co, and Zn atoms, optimizes the adsorption energies of oxygen intermediates by fine-tuning the 3d electronic structure. However, due to the high free energy of single metal atoms, their content is inherently limited, making optimal charge redistribution challenging. The incorporation of S and N heteroatoms further effectively modulates charge redistribution at the metal centers. The synergistic effects of metal-metal and metal-heteroatom electron coupling collectively enhance oxygen reaction performance. Theoretical calculations further elucidated the synergistic mechanism, revealing that the Co atom is the most favorable site for binding oxygen intermediates. Meanwhile, the presence of Fe atoms weakens the Co-O binding energy, lowering the d-band center of the Co atom. Additionally, the strong adsorption of *OH on the Zn site enhances OER activity, ultimately improving overall oxygen electrocatalytic performance ( Fig. 7 e). As a result, the synthesized ZnCoFe-TAC/SNC exhibited exceptional performance in both the ORR and OER, surpassing that of monometallic or bimetallic single-atom catalysts, as well as benchmark Pt/C catalysts. Besides, the spin effect in ternary systems featuring magnetic atoms also plays a unique role in electronic modulation for electrocatalysis. Zeng and coworkers reported a hetero-trimetallic FeCoNi single atom catalyst (FCN-TM/NC) featuring an asymmetric FeN 4 coordination ( Fig. 7 f) [ 56 ]. The incorporation of Co and Ni induces significant electron redistribution, which concomitantly modulates the spin state of Fe. As revealed in Fig. 7 g, Fe 3d orbital energy levels in FCN-TM/NC undergo a pronounced shift relative to those in Fe-N-C. Combined DFT calculations, in situ Raman, and in situ XAFS analyses reveal that Fe serves as the primary active site for *OH adsorption, while Co and Ni atoms induce a long-range electron-withdrawing/donating effect on the neighboring Fe center, modulating its electronic properties and resulting in weaker interaction of Fe site with *OH. The optimized adsorption and desorption energies of the intermediates are reflected in reduced energy barrier for both O 2 reduction and evolution catalysis. The FCN-TM/NC delivers exceptional performance, exhibiting E 1/2 of 0.941 V for ORR and requiring only 270 mV overpotential to achieve 10 mA cm -2 for OER, resulting in a low potential gap of 0.615 V. In conclusion, the primary advantage of TACs lies in the intricate interactions among tri-nuclear metal centers and their further reduced symmetry compared to DACs and SACs. This unique configuration endows TACs with distinctive electrocatalytic properties, particularly in dual- or multi-functional electrocatalytic reactions. The TACs provide more regulatory possibilities for electronic structure of SACs. However, the introduction of a third metal atom introduces new challenges, including a non-negligible metal aggregation effect and a more complex synthesis process, which often results in lower metal loading. The controllable design and preparation of asymmetric coordination heteronuclear trimetallic sites remain a significant challenge, requiring advanced synthetic strategies and precise control over atomic configurations. These challenges inspire the development of novel design philosophies for fabricating hetero-trimetallic single-atom site catalysts. 2.3. Quadr-nuclear and Penta-nuclear atomic structures In contrast to TACs, quadri-nuclear and penta-nuclear atomic structures, collectively referred to as high-entropy single-atom catalysts (HESACs), have garnered significant attention. These materials not only overcome the scaling relationship limitations inherent to SACs but also elevate the performance boundaries of carbon-based SACs to unprecedented levels. A seminal study by Tang and colleagues in 2024 demonstrated the incorporation of five distinct metals (Fe, Mn, Co, Ni, and Cu) into a carbon substrate to construct a π-network-destabilized high-entropy catalytic environment, elucidating its direct correlation with enhanced electrocatalytic activity [ 13 ]. DFT calculations revealed that the high-entropy catalytic environment reduces system symmetry, leading to a lower aromaticity value ( Fig. 8 a) [ 57 ]. This reduction in symmetry contributes to a more chaotic spatial distribution of electrons, which, in turn, enhances catalytic performance. The entropy-driven structural destabilization, combined with strong interactions between the single atoms and the carbon support, confers exceptional stability and activity to the HESACs ( Fig. 8 b). As a result, the HESACs exhibited remarkable improvements in both the ORR and OER. The ORR activity achieved a E 1/2 of 0.87 V, while the OER activity demonstrated an exceptionally low overpotential ( η 10 ) of 270 mV at 10 mA cm -2 . These performance metrics surpass those of state-of-the-art mixed noble metal catalysts and carbon-based metal catalysts, respectively. Fig. 8. Open in a new tab Interactions in quadr-nuclear and penta-nuclear atomic structure. (a) Aromaticity value of HESAC and single-atom metal. (b) Contour map of the limiting potential as a function of the Gibbs free energies ΔG 1 and ΔG 4 [ 13 ]. Copyright 2023, Springer Nature. (c) Lateral and side views of the optimized geometric configurations of FeCoNiRu-HESAC, with an averaged intermetallic distance of 6.11 Å. (d) Free energy change of ORR intermediates catalyzed by FeCoNiRu-HESAC at varying potentials. The inset illustrates the spin and charge distribution of FeCoNiRu-HESAC after *OH adsorption [ 14 ]. Copyright 2024, Wiley. In addition, an entropy-enhanced structure of HESACs incorporating four different metals (Fe, Co, Ni, and Ru) was proposed, exhibiting 5.2 times higher entropy compared to conventional SACs [ 14 ]. The nitrogen-doped graphene-supported HESAC (FeCoNiRu-HESAC) was synthesized and evaluated for its ORR performance ( Fig. 8 c). DFT calculations and ML were employed to predict the ORR activity of the FeCoNiRu-HESAC. The results revealed that the active Fe sites in the HESAC exhibited an ORR overpotential of 0.44 V, lower than the overpotentials of Fe, Co, Ni, and Ru SACs, which were 0.56 V, 0.55 V, 1.37 V, and 1.06 V, respectively. This demonstrates that the non-bonding interactions among Fe, Co, Ni, and Ru atoms synergistically modulate the catalytic activity of the Fe site within the HESAC system. Guided by ML and DFT analysis, the experimental ORR performance of the FeCoNiRu-HESAC was evaluated, demonstrating a low experimental overpotential and high activity comparable to that of Pt/C under both acidic and alkaline conditions ( Fig. 8 d). Furthermore, when applied in a zinc-air battery, the FeCoNiRu-HESAC exhibited an open-circuit potential of 1.39 V and a power density of 0.16 W cm -2 . However, research on quadri-atom and penta-atom catalysts is still in its nascent stages, with numerous challenges yet to be overcome. A critical issue lies in suppressing the agglomeration of disparate metal species during catalyst synthesis, which can compromise the uniformity and stability of the active sites. Additionally, the presence of multiple metals introduces structural complexity, making it difficult to elucidate the underlying catalytic mechanisms and interactions among the metal centers. Despite these challenges, the recent rise of ML technology offers a promising and efficient approach to studying HESACs. By integrating ML with DFT calculations, researchers can accelerate the discovery and optimization of HESACs. This ML- and DFT-guided strategy provides a new direction for the rational design of HESACs with enhanced activity for the ORR and other electrocatalytic reactions. In the sub-sections, the key interaction mechanisms and electrocatalysis performances of ORR, OER, CO 2 RR, NO 3 RR and other electrocatalysis for representative DAC/MACs are summarized in Table 1 , Table 2 , Table 3 , Table 4 , Table 5 . Table 1. The key interaction mechanisms of electrochemical reaction for representative DAC/MACs . Interaction mechanisms Catalyst Type Electrocatalysis References Electronic effect NiCo DASs/N—C DACs ORR&HER [ 34 ] FeCo-NPC DACs ORR [ 58 ] FeCoNiRu-HESAC QACs ORR [ 14 ] Co-N-Ni/NPCNSs DACs CO 2 RR [ 35 ] FeZnNC DACs CO 2 RR [ 59 ] Synergistic effect FeCoN X /C DACs ORR [ 60 ] Ni/Cu-N-C DACs CO 2 RR [ 61 ] FeNi-NSC DACs CO 2 RR [ 62 ] Fe/Cu-NC DACs CO 2 RR [ 63 ] FeCo-NC DACs CO 2 RR [ 37 ] Fe/Cu-HNG DACs NO 3 RR [ 36 ] Site-distance effect Fe-Zn@SNC DACs ORR [ 40 ] Fe-Rh x @NC DACs ORR [ 41 ] Cu/Ni-NC DACs CO 2 RR [ 64 ] NiFe-N bridge DACs CO 2 RR [ 65 ] Ir Co pair DACs FAOR [ 42 ] Spin effect Fe,Cu/N—C DACs ORR [ 43 ] Fe/Zn-N-C DACs ORR [ 45 ] Mg/Fe-N-C DACs OER [ 44 ] FCN-TM/NC TACs ORR&OER [ 56 ] FeMoPPc DACs NRR [ 46 ] Orbital hybridization FeAl-RNC DACs ORR [ 49 ] FeMn DSA / d NC DACs ORR [ 66 ] FeSn-C 2 N DACs ORR [ 51 ] FeCo-N 3 O 3 @C DACs ORR [ 67 ] CoFeCu-TAC TACs ORR&OER [ 54 ] Zn 1 Sn 1 /SNC DACs CO 2 RR [ 50 ] Open in a new tab Table 2. The electrocatalysis performances of ORR for DAC/MACs . Catalyst Type Electrolyte E 1/2 (V vs. RHE) References NiCo DASs/N—C DACs 0.1 M KOH 0.880 [ 34 ] CoFe-NC-900 DACs 0.1 M KOH 0.940 [ 68 ] Fe,Cu/N—C DACs 0.1 M KOH 0.860 [ 43 ] Fe/Ni-N-C DACs 0.1 M KOH 0.861 [ 69 ] FeCo-N 3 O 3 @C DACs 0.1 M KOH 0.936 [ 67 ] Fe/Zn-N-C DACs 0.1 M KOH 0.906 [ 45 ] 0.1 M HClO 4 0.808 Fe,Mn/N—C DACs 0.1 M KOH 0.928 [ 70 ] 0.1 M HClO 4 0.804 FeMo-N-C DACs 0.1 M HClO 4 0.840 [ 71 ] FeCoN X /C DACs 0.1 M HClO 4 0.860 [ 60 ] FCN-TM/NC TACs 0.1 M KOH 0.941 [ 56 ] CoFeCu-TAC TACs 0.1 M KOH 0.891 [ 54 ] ZnCoFe-TAC/SNC TACs 0.1 M KOH 0.901 [ 55 ] ZnCoFe-N-C TACs 0.1 M KOH 0.878 [ 72 ] PtFeCo/NC TACs 0.1 M KOH 0.870 [ 73 ] CoN 4— C-NiN 4— C-FeN 4 TSAs TACs 0.1 M KOH 0.880 [ 15 ] PtCuZn/SNC TACs 0.1 M HClO 4 0.859 [ 74 ] HESA PACs 0.1 M KOH 0.870 [ 13 ] Open in a new tab Table 3. The electrocatalysis performances of OER for DAC/MACs . Catalyst Type Electrolyte Overpotential @10 mA cm -2 (mV) References Co/Fe-SNC800 DACs 1.0 M KOH 240 [ 75 ] a-NiCo/NC DACs 1.0 M KOH 252 [ 76 ] Ni-Ir DAC DACs 1.0 M KOH 240 [ 77 ] FeCo-N-C-1.25 DACs 1.0 M KOH 295 [ 78 ] DA-FC—NG DACs 1.0 M KOH 268 [ 48 ] Cu,Co/NSC2 DACs 1.0 M KOH 339 [ 79 ] Fe-Ni-N-P-C DACs 0.1 M KOH 337 [ 80 ] 1.0 M KOH 250 Fe/Ni-N-C DACs 0.1 M KOH 322 [ 69 ] FeCo-N 3 O 3 @C DACs 0.1 M KOH 298 [ 67 ] FCN-TM/NC TACs 1.0 M KOH 270 [ 56 ] CoFeCu-TAC TACs 1.0 M KOH 310 [ 54 ] ZnCoFe-N-C TACs 1.0 M KOH 370 [ 72 ] PtFeCo/NC TACs 0.1 M KOH 270 [ 73 ] CoN 4— C-NiN 4— C-FeN 4 TSAs TACs 0.1 M KOH 393 [ 15 ] HESA PACs 1.0 M KOH 270 [ 13 ] Open in a new tab Table 4. The electrocatalysis performances of CO 2 RR for DAC/MACs . Catalyst Type Electrolyte CO FE (%) | J CO | (mA cm -2 ) References Co-N-Ni/NPCNSs DACs 0.1 M KHCO 3 96.4 3.2 [ 35 ] Cu-S-Ni/SNC DACs 0.1 M KHCO 3 98.1 24.5 [ 81 ] FeZnNC DACs 0.5 M KHCO 3 92 48.1 [ 59 ] NiFe-N bridge DACs 0.5 M KHCO 3 83 9.2 [ 65 ] FeNi-NSC DACs 0.5 M KHCO 3 96.1 40.1 [ 62 ] Fe/Cu-NC DACs 0.5 M KHCO 3 99 10.8 [ 63 ] CuCo-DSAC DACs 0.5 M KHCO 3 ∼97 – [ 82 ] Zn 1 Mn 1 -SNC DACs 0.1 M KHCO 3 97 10.2 [ 83 ] MoFe-N-C DACs 0.5 M KHCO 3 95.96 ∼11.72 [ 84 ] CoCu-DASC DACs 0.5 M KHCO 3 99.1 – [ 85 ] ZnNi-TACs TACs 1.0 M KHCO 3 98.3 42.2 [ 86 ] Open in a new tab Table 5. The electrocatalysis performances of NO 3 RR for DAC . Catalyst Type Electrolyte NH 3 FE (%) NH 3 yield rate References Fe/Cu-HNG DACs 0.1 M KNO 3 , 1.0 M KOH ∼92.51 1.08 mmol h -1 mg -1 [ 36 ] Cu-Fe-N-C DACs 0.1 M KNO 3 , 1.0  M KOH 95.08 1.22 mmol h -1 cm -2 [ 24 ] Cu/Ni-NC DACs 100 ppm NaNO 3 , 0.5 M Na 2 SO 4 97.28 0.324 mmol h -1 cm -2 [ 87 ] RuCu DAs/NGA DACs 0.1 M KNO 3 , 0.1 M KOH 97.2 1.63 mg h -1 cm -2 [ 88 ] CuNi DSACs DACs 0.1 M KNO 3 , 1.0 M KOH ∼94.09 ∼18.30 mg h -1 cm -2 [ 89 ] Cu/Co-CN DACs 0.1 M KNO 3 , 1.0 M KOH 93 7.238 mg h -1 ·mg -1 [ 90 ] FeCu-NC DACs 0.1 M KNO 3 , 0.1 M KOH 95 6.13 mg h -1 mg -1 [ 91 ] FeCu-NPCS DACs 0.1 M KNO 3 , 1.0 M KOH 97.42 82.72 mg h -1 mg -1 [ 92 ] Open in a new tab 3. Catalytic mechanism research methods of heteronuclear atom catalysts In situ experimental characterizations and deep theoretical calculations are crucial for revealing the catalytic mechanism. With the emergence of emerging in-situ experimental characterization and computational techniques, it provides important parameters or direct evidence to reveal catalytic mechanisms and structure-activity relationship. Operando/in situ synchrotron radiation XAS, in situ surface-enhanced Raman spectroscopy (SERS), in situ ATR-SEIRAS, in situ differential electrochemical mass spectrometry (DEMS), in situ electron paramagnetic resonance (EPR), operando Mössbauer spectroscopy, and in situ environmental TEM/STEM (E-TEM/STEM) combined with DFT calculations are of paramount importance in elucidating the geometry, transition adsorption states and reaction paths of DACs/MACs in electrocatalytic process. 3.1. In situ experimental characterizations 3.1.1. Operando/In situ synchrotron radiation XAS Operando/in situ synchrotron radiation XAS is an advanced technology based on synchrotron radiation light source, which is used to study the local atomic and electronic structure of materials. Nowadays, it has emerged as a cutting-edge technique in electrocatalysis, playing a pivotal role in elucidating the dynamic evolution of active sites, probing electronic structures and coordination environments of catalysts, identifying reaction intermediates, and assessing catalyst stability under operational conditions [ 93 , 94 ]. During electrocatalytic reactions, atomic structure of DACs/MACs often undergoes dynamic transformations. In situ XAS spectra proves indispensable for tracking the real-time evolution of metal atomic sites, including metal-metal interactions and the adsorption/desorption behavior of reactants or intermediates [ 95 , 96 ]. For instance, local structures of Ni/Fe atoms and the Fe incorporation dynamics during activation were probed by Operando XAFS analysis, indicating Fe progressively coordinate with Ni atoms via oxygen bridges and ultimately form a triatomic configuration [ 97 ]. Additionally, in situ EXAFS analysis reveals potential-driven bond reorganization directly enabling oxygen intermediate adsorption. As applied potential increases, the Se-N bond contracts from 1.55 to 1.35 Å, the Fe-N bond shortens from 1.50 to 1.48 Å, while the Fe-Se bond elongates from 2.24 to 2.30 Å. The concerted changes involving contracting metal-N bonds alongside weakening metal-chalcogen bonds precisely modulate d-band centers to optimize oxygen species binding, confirming structural adaptation for intermediate stabilization during electrocatalysis [ 95 ]. Furthermore, in situ XAS measurements unravels the dynamic structural evolution and catalytic mechanism of bimetallic sites during electrochemical reactions. The Fe-Co bimetallic centers function as voltage-dependent active sites, exhibiting small positive-energy shifts in the absorption edge positions of the Fe K-edge XANES spectra during ORR and the Co K-edge XANES spectra during OER ( Fig. 9 a). EXAFS analysis further confirms that Fe and Co serve as the primary active sites for the ORR and OER, respectively, directly evidencing strong synergistic interactions between the dual-metal sites for ORR and OER [ 67 ]. Fig. 9. Open in a new tab In situ experimental characterization for catalytic mechanism research. (a) XANES spectra of Fe K-edge and Co K-edge at different potentials [ 67 ]. Copyright 2024, Springer Nature. (b) In situ FTIR spectra of DAP-(Cr, Rh)/CN [ 98 ]. Copyright 2025, Wiley. (c) In situ Raman spectra of Co/CoMn-NC [ 99 ]. Copyright 2024, Wiley. (d) In situ DEMS of DCDO electrooxidation in 1.0 M KOH on Cu x O@CF [ 100 ]. Copyright 2023, Elsevier. (e) EPR analysis of aqueous ·OH generation in SA-Fe-N 5 vs Fe-N 5 after 2000 cyclic voltammetry cycles [ 101 ]. Copyright 2025, American Chemical Society. (f) Operando 119 Sn Mössbauer spectra of Sn 1 –CuO recorded at −0.8 V [ 102 ]. Copyright 2023, American Chemical Society. (g–j) Temperature-dependent HRTEM imaging reveals dynamic structural evolution during in situ heating TEM [ 103 ]. Copyright 2024, Springer Nature. 3.1.2. In situ ATR-SEIRAS In-situ ATR-SEIRAS operates on the absorption of infrared light by specific chemical bonds in a sample. This absorption excites transitions between molecular vibrational energy levels, producing characteristic absorption peaks. By monitoring the spectral position and intensity of these fingerprint signatures, this technique enables precise tracking of chemical bond transformations and functional group dynamics in materials under investigation, providing fundamental insights into their physicochemical behavior. Thus, in situ ATR-SEIRAS can serve as a powerful tool for tracking electrocatalytic reaction intermediates and their adsorption behavior on catalyst surfaces, providing critical insights into reaction mechanisms and pathways. Utilizing this technique, Xiong and co-workers successfully identified adsorbed HCOO ad species during the FAOR, unequivocally confirming the formate pathway as the direct reaction route ( Fig. 9 b) [ 98 ]. Notably, the HCOO ad vibrational band in DACs exhibits a distinct positive shift compared to SACs in Operando FTIR spectra, revealing fundamentally different adsorbed path of HCOO ad at dual-metal sites. In another work of FAOR, the interaction mechanism between dual atomic sites was revealed. It is discovered that adsorbed *OH species on monometallic Ir sites are replaced by formic acid with the voltage increases [ 42 ]. For comparison, Ir-Co-x DACs maintain persistent *OH signals at high potentials, demonstrating Co’s function as an *OH reservoir that regulates the reaction microenvironment. 3.1.3. In situ SERS In situ SERS spectroscopy probes molecular structure, chemical bonding, adsorbed species, and phase transitions through laser excitation of molecular vibration and measurement of scattered light frequency change (Raman shift). This powerful technique enables real-time monitoring of reactant and intermediate structural evolution during electrochemical processes, while simultaneously identifying active sites and providing key evidence for determining catalytic reaction pathways. Taking the ORR as an example, in situ SERS spectroscopy serves as a powerful tool for probing the intermediate steps and paths of the reaction by directly monitoring key intermediates including *OH and *OOH species, determining the intermediate steps and paths of the reaction. A representative study by Hu’s group on Co/CoMn-NC catalysts demonstrated capability of the technique where characteristic Raman peaks emerged at 500 and 605 cm -1 with the decrease of the cathode potential corresponding to Co(OH) 2 and CoOOH formation ( Fig. 9 c) [ 99 ]. Notably the complete absence of Mn-related characteristic Raman peaks provided definitive evidence that the Mn site is not the active centers in this catalytic system. In situ SERS spectroscopy further enables monitoring of carbon structure stability and intermediate transformation process. As reported by Pan and coworkers, the consistent I D /I G ratio throughout testing confirmed high structural stability [ 34 ]. The technique captured the potential-dependent evolution of reaction intermediate, revealing transformation of CoOOH to Co(OH) 2 in the ORR process with the applied voltage decreasing. Notably, the progressive intensification of the 1150 cm -1 Raman band (O—O stretching vibrations of superoxide O 2 - ) with decreasing potential provided direct spectroscopic evidence for strong interactions between Co-N 4 active sites and superoxide intermediates. 3.1.4. In situ DEMS In situ DEMS technology represents a powerful analytical platform that integrates electrochemical systems with mass spectrometry, enabling real-time detection and quantification of electrochemical reaction gases, volatile intermediates and gas products during potential cycling. This advanced technique provides millisecond-resolution tracking of product evolution, allowing precise determination of gas-phase species type, formation rate and Faraday efficiency. Furthermore, through strategic isotope labeling experiments, DEMS offers unique insights into product selectivity, reaction pathways, and underlying gas evolution mechanisms. For instance, in situ DEMS has proven indispensable for elucidating reaction mechanisms in formaldehyde oxidation, particularly for tracing hydrogen origins in H 2 evolution. As demonstrated in Fig. 9 d, intermittent potential application in KOH electrolyte containing deuterated formaldehyde (DCDO) yields exclusively D 2 (m/z = 4), confirming the organic substrate as the sole deuterium source [ 100 ]. The potential-dependent emergence of mass spectrometry signals further establishes the electrochemical nature of D 2 generation, requiring direct electrical energy input. These collective findings provide definitive evidence that H 2 production during formaldehyde oxidation derives entirely from organic reactant H atoms. 3.1.5. In situ EPR In situ EPR allows real-time monitoring of the behavior and changes of unpaired electrons in materials under specific experimental conditions. Coupling with spin-trapping agents, it enables real-time monitoring of reactive intermediates (e.g., hydroxyl radicals, superoxide radicals) generated during electrocatalytic processes, providing crucial mechanistic insights into reaction pathways [ 101 , 104 ]. As shown in Fig. 9 e, in situ EPR spectra were performed to contrast the electrochemical Fenton-like effect of CuN 5 -SAzyme-assisted Fe-N 5 Catalysts (SA-Fe-N 5 ) and Fe-N 5 in ORR [ 101 ]. EPR spectroscopy following multiple cyclic voltammetry cycles under ORR conditions with 5,5-Dimethyl-1-pyrroline N-oxide trapping was conducted to detect ·OH free radical generation. Notably, after 2000 cycles, no ·OH signal was detected for SA-Fe-N 5 , in sharp contrast to the Fe-N 5 counterpart. This observation confirms that Cu incorporation effectively suppresses the electro-Fenton effect of Fe-N 5 . Furthermore, advances in in-situ EPR provide critical insights into the correlation between electronic structure evolution and performance in electrode materials. By resolving electron spin states, this technique enables explicit identification of charge transfer stages, thereby distinguishing redox steps [ 105 ]. Its successful application in probing reaction kinetics and elementary steps in battery systems demonstrates great potential to deepen mechanistic understanding in electrocatalysis research for DACs and MACs. 3.1.6. Operando Mössbauer spectroscopy Operando Mössbauer spectroscopy is suitable for studying the dynamic behavior of materials containing specific elements such as iron and tin during reaction processes. It can be conducted to quantitatively monitor and analyze the operando-formed catalytic intermediates during electrocatalysis. Liu and co-workers employed operando 57 Fe Mössbauer spectroscopy to unambiguously demonstrate the reversible catalytic cycling of electronic states and coordination structures in single-atom Fe species during the ORR [ 106 ]. At onset potentials, high-spin FeN 5 converts to low-spin O 2 ⁻-Fe II N 5 , while applied bias (0.5 V and 0.7 V) triggers decrease of low-spin FeN 4 and formation of high-spin O 2 -Fe II N 4 . All configurations reversibly restore to initial states upon bias removal, demonstrating dynamic interconversion. The metastable state represents the most catalytically active configuration among various catalyst states. Identifying these metastable species remains exceptionally challenging due to the continuous dynamic evolution of active sites during catalytic processes. Combined with DFT calculation, operando 119 Sn Mössbauer spectroscopy successfully tracked and identified in situ produced Sn 4+ -O 3— Cu + species in CO 2 RR ( Fig. 9 f) [ 102 ]. 3.1.7. In situ E-TEM/STEM In-situ E-TEM/STEM imaging plays a key role in the future study of the catalytic mechanism for DACs/MACs, enabling real-time visualization of existence form and distribution of the metal atoms in the catalyst and the migration, aggregation or exfoliation of metal sites [ 103 , 107 ]. Notably, Wang and co-workers demonstrated the power of in situ ETEM imaging in precisely identifying active sites, where two distinct protrusions on the catalyst surface were resolved as water molecules [ 108 ]. Subsequent experiments in a mixed gas environment revealed the dynamic evolution of these protrusions, directly linking the adsorbed hydroxyl groups to CO conversion and unambiguously confirming the catalytically active sites. Furthermore, Sun and colleagues employed in situ heating TEM to track the dynamic phase evolution of iron species during precursor pyrolysis, capturing the complete transformation pathway from nanoparticle expansion and fragmentation to atomic migration, ultimately leading to single-atom site formation ( Fig. 9 g–j) [ 103 ]. We hope to utilize in situ E-TEM/STEM technique to identify the adsorption configurations of reaction intermediates at metal active sites in DACs/MACs combined with electron energy loss spectroscopy (EELS) and energy dispersive spectrometer (EDX) under electrochemical reaction and reveal the reaction mechanism in the future. 3.2. Theoretical calculation 3.2.1. Reaction paths DFT calculations serve as a powerful computational technique for determining rate-limiting steps and the overpotential of the reaction through free energy analysis of reaction pathways to evaluate catalytic activity. These calculations provide critical complementary evidence to experimental studies by (1) elucidating the origin of high intrinsic activity in catalytic systems, (2) identifying optimal active site configurations, and (3) guiding the rational design of metal combinations for DACs/MACs during preliminary catalyst development. For example, Luo and colleagues demonstrated that non-bonding interactions between heterometallic atoms (Fe-M) can cooperatively modulate the electronic structure and catalytic activity of Fe centers in both FeMN 6 -DAC and FeMN 8 -DAC systems [ 109 ]. DFT-derived ORR volcano plots confirmed FeNiN 6 -DAC and FeNiN 8 -DAC at the vertex of volcanic graph, exhibiting the lowest ORR overpotential ( Fig. 10 a). The Gibbs free energy analysis indicates that the formation of H 2 O from *OH for FeNiN 6 -DAC is the rate-determining step of ORR. Guided by these theoretical predictions, the team synthesized FeCoN 8 -DAC, FeNiN 8 -DAC and N-doped graphene. Experimental results showed superior performance of FeNiN 8 -DAC with an exceptionally low overpotential of 0.39 V, validating the computational aided design strategy. Fig. 10. Open in a new tab Theoretical calculation for catalytic mechanism research. (a) Volcano plot correlating the theoretical overpotential (η ORR ) with the d-electron count of the counterpart metal for FeMN 6 -DAC and FeMN 8 -DAC structures [ 109 ]. Copyright 2023, Elsevier. AIMD calculations of the structure of the (b) pCOHP between the metal active site center and O in the *OH adsorption (cyan: bonding contributions; purple: antibonding contributions). (c) The linear relationship between ICOHP and ΔE ads *OH [ 34 ]. Copyright 2023, Wiley. Evolution of M-C bond lengths during the catalytic process for (d) *COOH and (e) *CO intermediate steps [ 110 ]. Copyright 2021, Wiley. 3.2.2. Transition adsorption states In electrocatalytic reactions, a comprehensive understanding of reaction mechanisms necessitates detailed analysis of transition adsorption states through multiple physicochemical descriptors and electronic structure properties. As dictated by the Sabatier principle, adsorption energy stands as a critical parameter governing intermediate binding strength on catalyst surfaces. In addition, further insight can be gained through PDOS analysis of both surface metal atoms and adsorbed species before and after interaction. It not only reveals the orbital interaction and identify bonding/anti-bonding state formation, but also tracking shifts in the transition metal d-band center position, both of which are essential factors to determine the evolution of adsorption strength. The COHP analysis serves as a powerful complementary tool for electronic structure characterization, overcoming the limitations of conventional band theory in chemical bonding analysis by quantitatively evaluating the contribution of interatomic interactions to chemical bonds. Fig. 10 b, c shows the ICOHP from minus infinity to the Fermi level exhibits a linear correlation with ΔE ads *OH, as revealed by Pan and coworkers [ 34 ]. Notably, Co-N 4 sites in Ni-N 4 /Co-N 4 configurations show enhanced *OH desorption capability compared to isolated Co-N 4 moieties. This electronic structure modification leads to facilitated OH - generation and reduced overpotential, confirming the superior ORR performance of NiCo DASs/N—C catalysts. Complementary to electronic structure calculations, Bader charge analysis and charge density difference mapping provide quantitative and qualitative insights into interfacial charge transfer between reaction intermediates and catalyst surfaces. For mechanistic understanding, AIMD simulations enable atomic-scale tracking of electrocatalytic processes, revealing the dynamic evolution from reactants to products through transition states while monitoring the surface structure of the catalyst. Lu’ group conducted AIMD simulation to track the thermostability of the adsorbed complexes under ambient conditions (298 K) during CO 2 RR progress ( Fig. 10 d, e) [ 110 ]. The results showed the pronounced differences corresponded to the *COOH adsorption/desorption thermodynamic behavior on the NiZn-N 6 —C, Ni-N 4 —C, and Zn-N 4 —C catalysts, further confirming the synergistic effect of Ni-Zn dual-metal sites in the kinetic pathway. 4. Synthesis strategy of heteronuclear atom catalysts 4.1. Bi-metallic SACs synthesis The construction of carbon-based metal SACs hinges on maintaining isolated metal atoms throughout the synthesis process, which is typically achieved by reducing metal loading amounts and increasing anchoring defects on the carbon substrate. In contrast, the synthesis of carbon-based catalysts with diatomic sites presents a significantly greater challenge, as it requires precise anchoring of two metal atoms in adjacent defects [ [111] , [112] , [113] ]. To address this complexity, various synthesis strategies have been developed to prepare well-distributed heteronuclear DACs, including bottom-up methods, top-down methods, and other unique synthetic approaches tailored to specific reaction systems [ 114 , 115 ]. Below, we provide a detailed introduction to these synthesis methods, illustrated with representative examples. 4.1.1. Bottom-up method The bottom-up method is a widely used approach for constructing heteronuclear diatomic centers dispersed on carbon substrates, typically involving the pyrolysis of binuclear metal complex precursors [ 93 , 116 ]. In addition to traditional pyrolysis, several advanced synthesis techniques, such as atomic layer deposition (ALD) [ 116 , 117 ], impregnation [ 38 ] and ion-exchange [ 47 , 118 ], have been integrated with pyrolysis to achieve precise synthesis of DACs. These methods enable precise control over the atomic configuration and coordination environment of the metal centers, which are critical for optimizing catalytic performance. Traditional pyrolysis methods leverage the structural advantages of precursors, such as zeolitic imidazolate frameworks (ZIFs), to anchor metal atoms. For example, Li and colleagues pioneered the synthesis of diatomic (Fe, Co)/N—C catalysts using ZIFs, which are rich in nitrogen, to anchor Fe and Co sites [ 16 ]. Through a one-pot high-temperature pyrolysis process, atomically dispersed Fe-Co dual sites were formed on nitrogen-doped porous carbon. Similarly, Dai’s group developed a sulfur-ligand exchange-assisted MOFs calcination strategy to prepare Fe-Co-N bimetallic centers modulated by spatial sulfur-bridge ligands [ 119 ]. This involved encapsulating Fe(acac) 3 molecules into Co-ZIF-8, followed by ligand exchange with 1-methylimidazolidine-2-thione (S-mIm) to introduce sulfur-containing complexes. After pyrolysis, a porous carbon-based catalyst with Fe-Co-N bimetallic sites adjacent to spatial sulfur-bridge ligands (Spa-S-Fe,Co/NC) was obtained ( Fig. 11 a). Fig. 11. Open in a new tab Bottom-up method for synthesizing bi-metallic SACs. (a) Scheme of S-ligand exchange-assisted MOFs calcined strategy for preparing Spa-S-Fe,Co/NC [ 119 ]. Copyright 2023, American Association for the Advancement of Science. (b) Illustration of ALD synthesis of Pt-Ru dimers on NCNTs [ 117 ]. Copyright 2019, Springer Nature. ALD technology enables the precise and controllable synthesis of carbon-based metal catalysts through alternating injections of precursor vapors, resulting in self-limiting reactions at the molecular level. By adjusting the number of deposition cycles, order, and type of precursors, researchers can achieve atomic-level control over catalyst synthesis. For instance, Fan and colleagues successfully prepared Pt-Ru bimetallic dimers on NCNTs using a two-step ALD process [ 117 ]. The first ALD step deposited isolated Pt atoms on NCNTs, while the second step introduced Ru atoms to form Pt-Ru dimers ( Fig. 11 b). This approach highlights the potential of ALD for constructing well-defined diatomic structures with tailored properties. The impregnation-adsorption method is another effective strategy for synthesizing DACs, allowing precise control over metal loading and coordination environments. Zhang and co-workers employed a two-step solvent impregnation method to produce FeCoN 6 supported by nitrogen-doped hollow carbon spheres (Fe-Co DACs) [ 38 ]. Using SiO 2 as a template, polydopamine (PDA) was self-polymerized on its surface, followed by the adsorption of Co ions. A second impregnation step introduced Fe species, which were chemically adsorbed onto the Co single-atom sites. After annealing in an NH 3 atmosphere and removing the SiO 2 template, the Fe-Co DACs catalyst was obtained. Ion-exchange combined with pyrolysis is a powerful strategy for constructing heteronuclear bimetallic sites, often utilizing MOFs as carriers to introduce additional metal ions. Chen and colleagues demonstrated this approach by introducing Co ions into a zinc coordination polymer (Co/HMT-Zn) via ion-exchange [ 120 ]. PDA was then coated onto the surface to anchor Fe ions, resulting in paired Co-Fe DACs after high-temperature pyrolysis. The final catalyst exhibited 0.97 wt. % Fe loading and 1.11 wt. % Co loading, showcasing the effectiveness of this method for synthesizing well-defined diatomic structures. 4.1.2. Top-down method In contrast to the bottom-up method, the top-down approach typically utilizes external energy sources, such as mechanical energy, to break down larger metal bulk structures into single atoms or to achieve better dispersion of metal single atoms. When combined with high-temperature pyrolysis, this method enables the controllable preparation of heterogeneous DACs on carbon substrates. The ball-milling method, a mechanochemical approach, is a representative top-down strategy for constructing both homo- and heteronuclear DACs. Compared to traditional wet chemical methods, ball-milling offers several advantages, including reduced solvent usage and the ability to synthesize kilogram-scale products in a single batch. Ji’s group demonstrated the effectiveness of this method by preparing DACs with precisely controlled interatomic distances [ 121 ]. The key to their synthesis lies in the selection of appropriate metal precursors, such as bimetallic cryptates and bis(diphenylphosphine)ferrocene metal complexes (MCl 2 dppf, where M = Pd, Ni). These precursors ensure that each pair of metal atoms is confined at a specific atomic distance during pyrolysis, ultimately resulting in well-distributed DACs on carbon supports. 4.1.3. Other synthetic methods The two afore-mentioned synthetic methods for preparing DACs, bottom-up and top-down approaches, each possess unique strengths but also face inherent challenges. These challenges include the need for precise and complex control over multi-step reaction processes, as well as the potential use of toxic or environmentally harmful chemical reagents. To address these limitations, researchers have developed innovative synthesis methods that combine the advantages of both approaches while minimizing their drawbacks. Recently, Du’s group introduced a novel method termed “entropy-engineered middle-in synthesis of metal dual-atom compounds (EEMIS-DSAC)” [ 89 ]. This approach integrates the benefits of both bottom-up and top-down methods, enabling the direct fabrication of various DACs. The process begins with the preparation of a “sandwich”-structured nanofiber membrane using electrospinning technology. The outer layers of the membrane consist of pure polyacrylonitrile (PAN) fibers, while the central layer is composed of PAN fibers doped with metal salts (e.g., Cu and Ni salts). During pyrolysis, the metal salts are converted into nanoparticles via bottom-up carbothermal reduction. Subsequently, these nanoparticles are precisely transformed into Cu and Ni single-atom structures using top-down chemical vapor deposition techniques. The synthesis is driven by an increase in system entropy, allowing the two methods to synergistically produce a diverse range of DACs ( Fig. 12 a). This innovative strategy offers a scalable and efficient route for synthesizing DACs with precise atomic control. Fig. 12. Open in a new tab Other synthetic methods for synthesizing bi-metallic SACs. (a) Schematic illustration of EEMIS-DSAC strategy [ 89 ]. Copyright 2024, American Chemical Society. (b) Schematic synthetic procedure of graphitic CN-supported PtCo DACs via a photoinduced anchoring method [ 122 ]. Copyright 2024, American Chemical Society. Xie’s group developed a groundbreaking photoinduced anchoring strategy for synthesizing PtCo DACs on graphitic carbon nitride (CN) [ 122 ]. In this method, Co ions are first anchored onto the graphitic CN through the lone-pair electrons of N atoms under light irradiation, forming Co-SACs. Subsequently, photoelectrons generated at the Co sites capture nearby Pt ions, enabling their precise anchoring to form heteronuclear PtCo DACs ( Fig. 12 b). This light-driven approach leverages the unique electronic properties of graphitic CN and the photoelectric effect to achieve highly controlled and efficient synthesis of DACs. The precise construction of heteronuclear diatomic structures is the primary challenge in synthesizing heteronuclear diatomic catalysts. The rational design of composite precursors and regulation of ligands, including the use of composite precursors containing targeted metal sources (e.g., bimetallic MOFs, supramolecular carrier loaded with bimetallic Prussian blue analogues) or the introduction of bidentate or multidentate ligands (such as diethylenetriamine pentaacetic acid, porphyrin) to stabilize heteronuclear metal pairs, followed by controlled pyrolysis represents one of the effective strategies for fabricating diatomic catalysts with well-defined heteronuclear active sites [ 123 , 124 ]. Moreover, by selecting and modifying support materials such as high-defect substrates containing pyridinic/pyrrolic nitrogen, vacancies, or other doped heteroatoms (S, P), different metal species can be effectively stabilized [ 125 ]. Post-modification methods like polymer coating help prevent metal atom agglomeration during high-temperature pyrolysis [ 126 ]. Furthermore, synthesis strategies optimization, such as sequential metal deposition strategies, where one metal is first anchored followed by introducing a second metal via ALD method, provides an effective solution to competitive adsorption while enabling precise heteronuclear site construction [ 117 ]. 4.2. Tri-metallic SACs synthesis Heterogeneous TACs represent a cutting-edge class of materials that have gained significant attention in recent years. These catalysts, which feature three distinct metal species anchored at the single-atom level, offer unique opportunities for optimizing catalytic performance through synergistic interactions among the metal centers. Various fabrication methods have been developed to simultaneously downscale three metal species to the single-atom level, leveraging defect-rich porous carbon supports to stabilize the metal atoms. These supports are often synthesized through the pyrolysis of nitrogen-rich MOFs or other well-defined ligand frameworks, as illustrated in the examples below. 4.2.1. MOF derivative synthesis method MOFs are highly versatile materials due to their diverse three-dimensional network structures, which allow for the adjustment of organic ligand lengths or geometric configurations [ 127 ]. This flexibility enables the directional synthesis and functionalization of MOFs, making them ideal precursors for constructing stable TACs. Among MOFs, ZIF-8 is particularly advantageous due to its abundant organic ligands and microporous structure, which facilitate the introduction of different metal atoms or metal centers through simple chemical or thermal treatments, thereby enabling precise regulation of electrocatalytic performance. Qiu and colleagues designed a tri-metal MOF precursor, which, after a one-step pyrolysis process, yielded an accordion-structured ZnCoFe-N-C TAC with a high density of triatomic active sites ( Fig. 13 a) [ 72 ]. This synthesis method allows for the adjustment of precursor types and supported metals to form catalysts with desired tri-nuclear atomic structures. Wu and associates reported a M-N/C TAC with a trinuclear active structure by pyrolyzing a Mn-ion-adsorbed ZnCo-ZIF precursor [ 128 ]. The resulting catalyst featured an unusual trinuclear site with a nitrogen-coordinated Mn single atom adjacent to two Co atoms (Co 2 MnN 8 ), anchored in an N-doped carbon matrix. With low metal loadings (1.45 wt. % Co and 0.29 wt. % Mn), the Co 2 MnN 8 TAC sites were uniformly distributed on the MOF-derived carbon substrate, demonstrating the effectiveness of this approach. Fig. 13. Open in a new tab Synthesis strategy of tri-metallic SACs. (a) Schematic synthetic procedure of accordion-structured ZnCoFe-N-C [ 72 ]. Copyright 2023, Wiley. (b) Schematic illustration of the preparation of CoN 4— C-NiN 4— C-FeN 4 TACs [ 15 ]. Copyright 2023, Wiley. 4.2.2. Other synthetic methods Beyond MOFs, other nitrogen-rich organic ligands and polymers can serve as favorable frameworks to establish suitable coordination environments for TACs. These materials offer unique advantages, such as improved dispersion of metal single atoms and enhanced stability [ 47 ]. Graphitic C 3 N 4 polymers, with their π-conjugated units, are excellent substrates for dispersing metal single atoms [ 129 ]. Inspired by this, Tian’s group synthesized various metal-C 3 N 4 precursors and wrapped them with polydopamine (PDA) [ 73 ]. Subsequent calcination in an argon atmosphere at 810 °C resulted in the formation of N-doped porous carbon-supported TACs with uniformly dispersed Pt, Fe, and Co single atoms. Lu and colleagues utilized g-C 3 N 4 as a carbon substrate to support linear tri-atom structures, forming unique linear TACs [ 130 ]. Using ALD, atomic Cu grippers were first fabricated on g-C 3 N 4 . A second NiO x ALD step produced Ni y Cu 1 /g-C 3 N 4 catalysts (where y represents the atomic ratio of Ni to Cu). The Ni atoms were located in the cavities of two adjacent Cu-N 2 coordinated sites, forming a chain-like Cu-OH—Ni-OH—Cu structure with high metal loadings (8.1 wt. % Cu and 3.1 wt. % Ni). Qiu’s group developed a metal encapsulation-segregation-overlay strategy to prepare heterogeneous M 1 N 4— C-M 2 N 4— C-M 3 N 4 TACs ( Fig. 13 b) [ 15 ]. This approach involves the triple-decker coupling of spatially separated M 1 N 4 , M 2 N 4 , and M 3 N 4 single atoms, achieving a three-fold enhancement in single-atom loading. For instance, IrPtCu TACs achieved a remarkable metal loading of up to 21.24 wt. %, demonstrating the potential of this method for synthesizing high-performance TACs. 4.3. Multi-metallic SACs synthesis The emergence of multi-metallic SACs (high-entropy SACs) has significantly advanced the field of catalysis, introducing both new opportunities and challenges. Preparing these materials requires not only preventing the aggregation of different metal species during pyrolysis but also identifying the optimal combination of metal atoms to achieve desired catalytic properties. This often necessitates the use of DFT calculations and ML to screen for ideal metal compositions [ 131 , 132 ]. Below, we highlight several innovative synthesis methods developed to address these challenges and enable the fabrication of high-entropy SACs. Tian and colleagues developed a general movable type printing method to synthesize multi-metallic SACs [ 27 ]. This innovative approach directly transfers dispersed metal single atoms from printing templates to nitrogen-doped carbon (NC) supports in a point-to-point manner, effectively addressing the issue of metal agglomeration commonly encountered in traditional synthesis methods ( Fig. 14 a). The strategy provides a versatile pathway for the rational design of multi-metallic SACs, enabling the loading of dissimilar metallic elements ranging from 5 to 11. Fig. 14. Open in a new tab Synthesis strategy of multi-metallic SACs. (a) Schematic illustration of movable type printing method to prepare HESACs [ 27 ]. Copyright 2022, Springer Nature. (b) Illustration for preparing ultrahigh-loading SACs (UHL-SACs) [ 133 ]. Copyright 2024, Springer Nature. (c) Schematic synthetic procedure of UHDM-N-C via the proposed ultrahigh-N-doping engineering assisted method [ 26 ]. Copyright 2024, Springer Nature. Qu and associates introduced a universal negative pressure annealing approach to construct SACs with high metal areal density ( Fig. 14 b) [ 133 ]. This method successfully synthesized 13 different metal SACs supported on polymeric carbon nitride (PCN) with ultrahigh metal loadings ranging from 27.3 to 44.8 wt. %. The authors also demonstrated the adaptability of this approach to other substrates, such as NC derived from guanine pyrolysis. Using this substrate, they prepared ultrahigh-loading SACs with five metals (Pt, Fe, Co, Ni, and Cu) as N-coordinated isolated sites, achieving metal loadings of 15.6, 3.1, 4.1, 2.3, and 7.3 wt. %, respectively. This method highlights the potential for scalable synthesis of high-entropy SACs with diverse metal compositions. Moreover, Lu and colleagues proposed a universal synthesis approach for directly transforming metal-sulfide nanoparticles into ultrahigh-density metal-nitrogen-carbon (UHDM-N-C) SACs ( Fig. 14 c) [ 26 ]. For example, a mixture of carbon dots (CDs), thiourea, urea, and NiCl 2 ·6H 2 O was heated to 600 °C under an inert atmosphere to form NiS nanoparticles supported on ultrahigh-N-doped carbon. Further heating to 600–700 °C caused the NiS nanoparticles to decompose, releasing Ni atoms that were captured by the N-doped carbon support. The surface functional groups of the CDs acted as inorganic ligands to chelate Ni ions, while the cross-linking of CDs provided a stable framework for high N-doping during pyrolysis. This strategy was extended to prepare quinary high-entropy SACs containing Ni, Cu, In, La, and Ce with a total metal loading of 25 wt. %, as well as vicenary high-entropy SACs with 20 metal elements and a total metal content of 18.1 wt. %. This approach demonstrates the potential for synthesizing ultrahigh-density SACs with complex multi-metallic compositions. In summary, current synthetic routes for DACs/MACs remain confined to laboratory-scale demonstrations, lacking industrial-scale validation. The primary costs associated with synthesizing carbon-based heteronuclear DACs/MACs stem from raw materials, equipment, and energy consumption. Precursor synthesis typically relies on diverse chemical reagents, which contribute substantially to overall expenses. However, emerging synthetic strategies, such as mechanochemical ball milling, dramatically minimize solvent usage, thereby offering a cost-effective alternative. Meanwhile, fabricating DACs/MACs demands advanced instrumentation, such as high-temperature pyrolysis furnaces, chemical vapor deposition (CVD) systems, and ALD equipment. These high-precision tools entail substantial capital investment and maintenance overhead, further elevating production costs. Consequently, there is a pressing need to develop more efficient and scalable synthetic process to streamline manufacturing processes while maintaining atomic-level precision. The synthesis process also requires substantial energy input in both electrical and thermal forms. Conventional methods like high-temperature pyrolysis and CVD typically demand extended processing times, resulting in low time-efficiency. Developing effective synthetic approaches that can achieve high metal precursor capture efficiency within shorter timeframes is crucial. For industrial-scale production, continuous flow reactors show particular promise by enabling accurate control of precursor mixing and reaction conditions. This technology effectively addresses the common issue of heterogeneous site formation where single atoms coexist with nanoparticles. Implementing advanced in-situ monitoring systems further allows for real-time process adjustments, ensuring consistent catalyst quality during large-scale manufacturing. 5. Exploration and prediction of heteronuclear diatomic and multi-atomic metal atom catalysts With the rapid advancement of high-throughput screening and ML, the traditional trial-and-error approach to catalyst discovery is being replaced by more efficient and predictive computational methods [ [134] , [135] , [136] ]. While DFT calculations have long been used to verify the theoretical performance of materials, they are limited in their ability to predict unknown chemical structures. High-throughput in silico first-principles calculations, combined with data-driven ML approaches, have emerged as powerful tools to overcome these limitations, enabling the prediction of promising catalysts with diatomic, triatomic, and multi-atomic sites for various electrochemical reactions. Below, we explore specific examples of these approaches in action. 5.1. High-throughput screening High-throughput computation is an efficient strategy for guiding the design of heteronuclear metal atom catalysts. This approach involves constructing numerous models with diverse potential active sites based on specific coordination structures, followed by evaluating their electrochemical performance and stability to establish structure-activity and structure-stability relationships [ 137 , 138 ]. Key descriptors, such as free energy [ 139 ], d-band center, spin polarization degree of the active center [ 69 ] and valence and electronegativity-dependent descriptors [ 140 ], play a critical role in determining electrocatalytic activity. These descriptors are used to build predictive models, which serve as universal screening criteria for designing high-performance and stable electrocatalysts. In a study aimed at identifying catalysts for the electrochemical NO 3 RR, a series of graphitic carbon nitride (g-CN)-supported heterogeneous DACs, denoted as M 1 M 2 @g-CN, were investigated using high-throughput screening [ 141 ]. Among 21 candidates, FeMo@g-CN and CrMo@g-CN were identified as the most promising, exhibiting low limiting potentials of −0.34 V and −0.39 V, respectively. [ 127 ]. Electronic structure analysis revealed that the activation of NO 3 RR was driven by a synergistic effect arising from the coupling of the d orbitals of the M 1 M 2 dimers with the antibonding orbitals of NO 3 - . Similarly, Cao and colleagues developed a volcano plot-based high-throughput screening method to evaluate the ORR performance of DACs [ 142 ]. Using ΔG OH* and ΔG O2* as activity descriptors, they screened 267 DACs containing 3d, 4d, or 5d metals and identified 38 promising candidates ( Fig. 15 a–d). Experimental validation confirmed that the theoretically screened MnCoN 6 DAC exhibited high ORR catalytic activity compared to commercial Pt/C, demonstrating the efficiency and accuracy of the high-throughput screening approach. Fig. 15. Open in a new tab High-throughput screening of heteronuclear diatomic and multi-atomic metal atom catalysts. (a) The correlations between ΔG OOH* and ΔG OH* on DACs. (b) ORR activity volcano plot (log(-j k ) at U = 0.84 V) as a function of ΔG(O 2 *) an ΔG(OH*). The dashed line stands for the ORR activity of Pt (111). (c) The phase diagram of rate-determination step as a function of ΔG OH* and ΔG O2* . (d) High-throughput screening workflow for DACs with promising ORR activity reported in references [ 142 ]. Copyright 2024, Wiley. 5.2. Machine learning ML, leveraging a variety of algorithms, has emerged as a powerful tool for constructing predictive models to estimate unknown catalyst activity and identify materials with optimal catalytic performance in theory [ [143] , [144] , [145] ]. Concurrently, several studies have experimentally validated the activity of catalysts screened through ML, further demonstrating the accuracy and practicality of this technology [ 142 , 146 , 147 ]. Unlike high-throughput screening methods, ML places greater emphasis on data collection and analysis [ 148 , 149 ]. As the number of data samples increases, the predictions and screenings of catalytic materials conducted through ML become more accurate [ 150 , 151 ]. The ML models in electrocatalysis were trained on a combination of experimental and computational datasets, including DFT calculations, electrochemical performance data, material characterization datasets, and literature-curated databases [ 152 , 153 ]. The two predominant methods for data generation involve manual extraction from literature and DFT simulations [ 150 ]. Below, we introduce the processes and methodologies employed in the screening of heteronuclear metal atom catalysts using ML, along with specific examples. Recently, Li’s group utilized ML to accelerate the calculation of intermediate adsorption energies in the CO 2 RR, enabling rapid screening of diatomic metals and carbon-based supports [ 154 ]. They investigated DACs consisting of 17 transition metals supported on carbon-based g-C 2 N and g-C 9 N 4 substrates with varying C-to-N ratios. By training an ML model on DFT calculations of 30 diatomic metal catalysts, they predicted the optimal adsorption configurations of intermediates, generating a complete dataset of 918 adsorption energy values. This approach demonstrated a 15-fold speed enhancement compared to traditional DFT calculations, significantly reducing computational costs. Similarly, Cao and colleagues combined DFT and ML to systematically investigate the CO 2 RR performance of DACs with Si-transition metal (Si-TM) dual-atom active sites [ 155 ]. They constructed 27 SiTMN 6 DACs and 336 SiTMN 5 An DACs (where A represents B or C elements, and n denotes the positions of N atoms embedded in N-doped graphene). Using ML, they screened 3 SiTMN 6 DACs and 6 SiTMN 5 An DACs with high selectivity for converting CO 2 to CH 4 or CH 3 OH. The study revealed that the p z band distribution, influenced by p-d orbital coupling and coordination environments, plays a critical role in optimizing intermediate adsorption and reaction activity. Gong and co-workers proposed a universal and interpretable descriptor model to unify activity and selectivity predictions for multiple electrocatalytic reactions, including O 2 /CO 2 /N 2 reduction and OER [ 152 ]. The descriptor incorporates easily obtainable properties of DACs, such as atomic properties, reactant interactions, synergistic effects, and coordination environments. This approach eliminated the need for over 50,000 DFT calculations, enabling the rapid identification of optimal catalysts for various reactions ( Fig. 16 ). The universality of this model was validated through experiments and supported by numerous published studies. Fig. 16. Open in a new tab General ML technology for screening DACs flow diagram [ 152 ] . Copyright 2024, Springer Nature. As a category of emerging intelligent technology, high-throughput screening and ML combined with DFT calculation have shown great application value in the intelligently synthesis of catalytic materials for specific reaction. While high-throughput screening and ML have shown great promise in catalyst discovery, they also face inherent challenges as below. (1) High computational costs: high-throughput screening methods often require substantial computing power, leading to high costs [ 156 ]. (2) Data scarcity: the accuracy of ML predictions is constrained by the availability of large, high-quality datasets [ 157 ]. (3) Interpretability: the “black-box” nature of ML models necessitates careful selection of descriptors and their physical interpretation to ensure meaningful insights [ 158 ]. Despite these challenges, the integration of high-throughput screening and ML with DFT calculations is expected to significantly reduce the repetitive manual labor involved in catalyst discovery, accelerating the development of new catalytic materials. 5.3. Choice of algorithms Artificial neural networks (ANNs), convolutional neural networks (CNNs), gaussian process regression, kernel ridge regression, random forests (RFs), support vector machines (SVMs), clustering and multiple linear regression are common algorithms, applying for material design, binding energies prediction, adsorption energy prediction, molecular orbital energy prediction, determining the descriptors, catalytic activity prediction and simplification of DFT calculations [ 150 , 159 ]. Selecting the optimal algorithm for a given task presents notable challenges, requiring careful consideration of both the specific task and data availability. ML algorithms play a pivotal role in accelerating the discovery and optimization of electrocatalysts. Among the most widely used algorithms, SVMs and RFs are favored for their robustness in handling high-dimensional data and capturing nonlinear relationships. SVMs excel in classification and regression tasks by maximizing the margin between data classes, while RFs, an ensemble method, improve predictive accuracy by aggregating multiple decision trees, reducing overfitting risks. RFs have emerged as the most popular algorithm across diverse catalytic subdisciplines, outperforming even widely used ANNs. This preference stems from the inherent characteristics of catalytic datasets, which are typically structured in numerical, tabular formats and often fall within the small-data regime (< 1000 data points). RFs exhibit superior performance compared to neural networks in handling such datasets [ 150 ]. Neural networks, particularly deep learning architectures like CNNs, are employed for complex pattern recognition in catalyst structures and properties. Their ability to automatically extract hierarchical features from raw data makes them invaluable for large-scale screening. For smaller datasets, Gradient Boosting Methods (e.g., XGBoost) are preferred due to their efficiency and interpretability. Clustering algorithms, such as k-means, Density-Based Spatial Clustering of Applications with Noise (DBSCAN) enable unsupervised exploration of catalyst datasets, revealing intrinsic groupings or structural motifs. For outlier detection, density-based methods (e.g., DBSCAN) or specialized algorithms are often preferred. Meanwhile, Genetic algorithms (GAs) optimize catalyst compositions by mimicking evolutionary selection, guided by ML-predicted fitness functions. The selection of each algorithm depends on data size, dimensionality, and the specific catalytic property targeted. Hybrid approaches, combining multiple algorithms, often yield the best performance, balancing accuracy and computational cost. Future advancements may focus on integrating domain knowledge into these models to enhance interpretability and generalizability. 6. Conclusions and outlooks Research on DACs, TACs, QACs, and PACs catalysts has shown significant promise, yet it remains in its early stages. The scale-up preparation of carbon-based heteronuclear multi-atom catalysts for industrial production and practical applications continues to present substantial challenges. However, understanding the interactions between individual metal centers within DACs, TACs, QACs, and PACs with advanced characterization techniques provides valuable insights for the precise design of efficient electrocatalysts tailored for various electrocatalytic applications. Appropriate synthesis methods and emerging theoretical technologies offer promising directions for future industrial applications, reducing labor and material costs compared to traditional experimental approaches. In this review, we summarize recent advancements in the interaction mechanisms, catalytic mechanism research methods, synthesis strategies, and emerging theoretical techniques for carbon-based heteronuclear metal atom catalysts. a) Understanding the interaction mechanisms between metal single sites in DACs, TACs, QACs, and PACs is crucial for guiding their design and preparation for broader practical applications. The development of in situ electrochemical-spectroscopy technologies, combined with theoretical calculations, has significantly advanced the exploration and analysis of these interactions. Various cases have been summarized, highlighting the role of these mechanisms in different electrocatalytic reactions, such as the ORR, CO 2 RR, and NO 3 RR. These insights provide a foundation for optimizing catalytic performance through precise control of electronic and structural properties. b) The catalytic mechanisms of heteronuclear atomic catalysts (DACs/MACs) are most effectively elucidated through the integration of advanced in situ characterization techniques and theoretical calculations. Cutting-edge methods such as operando synchrotron XAS, in situ ATR-SEIRAS, in situ SERS, in situ DEMS, in situ EPR, operando Mössbauer spectroscopy, and in situ E-TEM/STEM provide real-time atomic- and electronic-scale insights into dynamic active-site evolution, intermediate adsorption, and reaction pathways. For instance, in situ XAS spectra reveals voltage-dependent coordination changes in bimetallic sites, while in situ ATR-SEIRAS spectra identifies key intermediates and their distinct behavior at dual-metal sites. Complementarily, DFT calculations and AIMD simulations decode electronic structure modulation (e.g., d-band center shifts, ICOHP trends) and kinetic pathways, linking interaction mechanisms to enhanced activity. c) The development of universal synthesis methods is essential for achieving the large-scale preparation of electrocatalysts and their industrial application. We introduce two main synthesis methods for DACs: bottom-up and top-down approaches. For MACs, synthesis methods are still in the early stages of development, but several innovative strategies have been proposed. These include MOF-derived synthesis, template-assisted methods, and entropy-driven approaches. These methods aim to achieve high metal loadings while preventing the formation of metal clusters or particles during pyrolysis. d) High-throughput screening and ML technologies have revolutionized the discovery of highly active and stable catalytic materials. These methods can handle large datasets and tasks simultaneously, significantly reducing the time and cost associated with traditional DFT calculations. Through several case studies, the importance of selecting appropriate descriptors, such as free energy, d-band center, and coordination environment, has been highlighted. Moreover, judicious selection of algorithms, along with integration of multiple approaches, is critical to achieving an optimal balance between accuracy and computational efficiency. ML models trained on limited datasets can predict catalyst performance across a broad range of compositions and structures, offering a powerful tool for accelerating catalyst discovery. To this end, continued research and following outlook in this field are of great challenges and value: a) While considerable advances have been achieved in elucidating the interactions among metal single sites in DACs and MACs, existing structural characterization methodologies remain constrained in their ability to precisely resolve the coordination environments of heteronuclear metal centers and transient reaction intermediates. There is an urgent need for more precise characterization methods, such as fast in situ XAS, to clarify reaction pathways and establish reliable structure-activity relationships. These developments will facilitate a more profound elucidation of catalytic mechanisms and provide a robust foundation for the rational design of highly efficient catalysts. b) While certain progress has been made in probing the interactions within heteronuclear atomic catalysts (DACs/MACs), current characterization techniques still face limitations in resolving the dynamic coordination environments of multi-metal centers and transient intermediates. To address these challenges, future research should prioritize the development of multimodal Operando platforms that synergize complementary techniques. For instance, integrating in situ electrochemical with Operando XAS spectra and atomic-resolution STEM imaging, bridging the gap between electronic structure analysis and real-space atomic arrangements and offering a multidimensional perspective on reaction pathways. c) The synthesis methods for carbon-based heteronuclear multi-atom catalysts are typically suitable for laboratory-scale experiments but have not yet been validated for large-scale industrial production. Developing scalable synthesis strategies that ensure high metal loadings and prevent metal clustering is essential. Additionally, the high-temperature pyrolysis process used in industrial synthesis faces challenges such as complex production processes, non-scalability of traditional methods, and high raw material costs. Future research should focus on designing pyrolysis systems tailored for industrial applications, facilitating the transition from experimental research to practical use. d) The compositional flexibility of DACs, TACs, and PACs, including element composition, coordination environment, and chemical bonding types, presents a vast design space that traditional high-throughput screening methods struggle to explore efficiently. ML technology, while currently limited by data scarcity and accuracy, offers a promising solution by leveraging limited datasets to predict catalyst performance across a broad range. As more catalytic activity descriptors are established and sample data accumulates, ML methods are expected to replace manual labor in the intelligent screening of catalytic materials. The integration of high-throughput screening and ML will further enhance the efficiency and accuracy of catalyst discovery. CRediT authorship contribution statement Xuanni Lin: Writing – original draft, Investigation, Formal analysis, Data curation. Zhengfei Chen: Writing – review & editing, Supervision, Project administration. Zhongjian Li: Supervision, Project administration. Bin Yang: Supervision, Project administration. Qinghua Zhang: Supervision, Project administration. Jianguo Lv: Supervision, Project administration. Lecheng Lei: Supervision, Project administration. Yuanyuan Li: Supervision, Project administration. Raul D. Rodriguez: Supervision, Project administration. Yang Hou: Writing – review & editing, Supervision, Funding acquisition, Conceptualization. Acknowledgments Declaration of competing interest The authors declare that they have no conflicts of interest in this work. Acknowledgments The authors acknowledge funding from the National Key Research and Development Program of China (2022YFB4002100), the National Natural Science Foundation of China (22278364, 22208296, 22425805, U22A20432, 22211530045, 22178308), the development project of Zhejiang Province's “Jianbing” and “Lingyan” (2023C01226), the Fundamental Research Funds for the Central Universities (226-2024-00060), the Key Technology Breakthrough Program of Ningbo “Science and Innovation Yongjiang 2035” (2024H024), and the Fundamental Research Funds for the Zhejiang Provincial Universities (226-2025-00224). Biographies Xuanni Lin ( BRID: 07728.00.25001 ) is currently a postdoctoral researcher at Zhejiang University working with Professor Yang Hou. She received her Ph.D. degree from Beijing University of Chemical Technology in 2024. Her research focuses on the electrocatalyst design for high-efficiency clean energy conversion and environment-related electrochemical processes. Zhengfei Chen ( BRID: 06066.00.32759 ) is a professor at the School of Biological and Chemical Engineering, Ningbo Tech University, China. He was awarded a Ph.D. in materials chemistry at the University of Melbourne, Australia in 2012. His current research interests are focused on synthesis of carbon related materials for catalysis, energy devices, and ionic liquids as green solvents for materials synthesis and CO 2 capture. Yang Hou ( BRID: 06268.00.60512 ) is a professor at the College of Chemical and Biological Engineering at Zhejiang University. He received his Ph.D. degree from the Dalian University of Technology in 2011, and then did postdoctoral research at the University of California, Riverside, University of Wisconsin-Milwaukee, and Technische Universität Dresden. His current research interests focus on the design and synthesis of low-dimensional materials for energy and environmental applications. Footnotes Peer review under the responsibility of Editorial Board of Fundamental Research. Contributor Information Zhengfei Chen, Email: [email protected]. Yang Hou, Email: [email protected]. References 1. Yin S.H., Yang S.L., Li G., et al. Seizing gaseous Fe2+ to densify O2-accessible Fe–N4 sites for high-performance proton exchange membrane fuel cells. Energy Environ. Sci. 2022;15:3033–3040. [ Google Scholar ] 2. Lin X., Shi L., Liu F., et al. Large-scale production of holey carbon nanosheets implanted with atomically dispersed Fe sites for boosting oxygen reduction electrocatalysis. Nano Res. 2022;15:1926–1933. [ Google Scholar ] 3. Bavykina A., Kolobov N., Khan I.S., et al. Metal–organic frameworks in heterogeneous catalysis: Recent progress, new trends, and future perspectives. Chem. Rev. 2020;120:8468–8535. doi: 10.1021/acs.chemrev.9b00685. [ DOI ] [ PubMed ] [ Google Scholar ] 4. Qiao B., Wang A., Yang X., et al. Single-atom catalysis of CO oxidation using Pt1/FeOx. Nat. Chem. 2011;3:634–641. doi: 10.1038/nchem.1095. [ DOI ] [ PubMed ] [ Google Scholar ] 5. Zhang X., Guo J., Guan P., et al. Catalytically active single-atom niobium in graphitic layers. Nat. Commun. 2013;4:1924. doi: 10.1038/ncomms2929. [ DOI ] [ PubMed ] [ Google Scholar ] 6. Ye F., Gong L., Long Y., et al. Topological defect-rich carbon as a metal-free cathode catalyst for high-performance Li-CO2 batteries. Adv. Energy Mater. 2021;11 [ Google Scholar ] 7. Chen J., Li Z., Wang X., et al. Promoting CO2 electroreduction kinetics on atomically dispersed monovalent ZnI sites by rationally engineering proton-feeding centers. Angew. Chem. Int. Ed. 2022;61 doi: 10.1002/anie.202111683. [ DOI ] [ PubMed ] [ Google Scholar ] 8. Hu X., Liu Y., Cui W., et al. Boosting industrial-level CO2 electroreduction of N-doped carbon nanofibers with confined tin-nitrogen active sites via accelerating proton transport kinetics. Adv. Funct. Mater. 2023;33 [ Google Scholar ] 9. Sun Q., Yue X., Yu L., et al. Well-defined Co2 dual-atom catalyst breaks scaling relations of oxygen reduction reaction. J. Am. Chem. Soc. 2024;146:35295–35304. doi: 10.1021/jacs.4c12705. [ DOI ] [ PubMed ] [ Google Scholar ] 10. Ouyang Y., Shi L., Bai X., et al. Breaking scaling relations for efficient CO2 electrochemical reduction through dual-atom catalysts. Chem. Sci. 2020;11:1807–1813. doi: 10.1039/c9sc05236d. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 11. Li X., Mitchell S., Fang Y., et al. Advances in heterogeneous single-cluster catalysis. Nat. Rev. Chem. 2023;7:754–767. doi: 10.1038/s41570-023-00540-8. [ DOI ] [ PubMed ] [ Google Scholar ] 12. Tang T., Bai X., Wang Z., et al. Structural engineering of atomic catalysts for electrocatalysis. Chem. Sci. 2024;15:5082–5112. doi: 10.1039/d4sc00569d. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 13. Lei X., Tang Q., Zheng Y., et al. High-entropy single-atom activated carbon catalysts for sustainable oxygen electrocatalysis. Nat. Sustain. 2023;6:816–826. [ Google Scholar ] 14. Tamtaji M., Kim M.G., WANG J., et al. A high-entropy single-atom catalyst toward oxygen reduction reaction in acidic and alkaline conditions. Adv. Sci. 2024;11 doi: 10.1002/advs.202309883. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 15. Yang A., Su K., Lei W., et al. Ternary synergism of heterogeneous M1N4-C-M2N4-C-M3N4 single-atom sites to manipulate the electrocatalytic pathway for zn-air battery and water splitting. Adv. Energy Mater. 2023;13 [ Google Scholar ] 16. Wang J., Huang Z., Liu W., et al. Design of N-coordinated dual-metal sites: A stable and active Pt-free catalyst for acidic oxygen reduction reaction. J. Am. Chem. Soc. 2017;139:17281–17284. doi: 10.1021/jacs.7b10385. [ DOI ] [ PubMed ] [ Google Scholar ] 17. Guan H., Zhang Y., Fan W., et al. Regulated Cu diatomic distance promoting carbon−carbon coupling during CO2 electroreduction. Small. 2025;21 doi: 10.1002/smll.202406605. [ DOI ] [ PubMed ] [ Google Scholar ] 18. Yang W., Jia Z., Zhou B., et al. Why is C–C coupling in CO2 reduction still difficult on dual-atom electrocatalysts? ACS Catal. 2023;13:9695–9705. [ Google Scholar ] 19. Wang X., Xu L., Li C., et al. Developing a class of dual atom materials for multifunctional catalytic reactions. Nat. Commun. 2023;14:7210. doi: 10.1038/s41467-023-42756-8. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 20. Jiao J., Yuan Q., Tan M., et al. Constructing asymmetric double-atomic sites for synergistic catalysis of electrochemical CO2 reduction. Nat. Commun. 2023;14:6164. doi: 10.1038/s41467-023-41863-w. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 21. Wei Z., He J., Yang Y., et al. Fe, V-co-doped C2N for electrocatalytic N2-to-NH3 conversion. J. Energy Chem. 2021;53:303–308. [ Google Scholar ] 22. Sui R., Liu B., Chen C., et al. Constructing asymmetric Fe–Nb diatomic sites to enhance ORR activity and durability. J. Am. Chem. Soc. 2024;146:26442–26453. doi: 10.1021/jacs.4c09642. [ DOI ] [ PubMed ] [ Google Scholar ] 23. Xia F., Li B., An B., et al. Cooperative atomically dispersed Fe–N4 and Sn–Nx moieties for durable and more active oxygen electroreduction in fuel cells. J. Am. Chem. Soc. 2024;146:33569–33578. doi: 10.1021/jacs.4c11121. [ DOI ] [ PubMed ] [ Google Scholar ] 24. Zhang X., Liu X., Huang Z.F., et al. Regulating intermediate adsorption and H2O dissociation on a diatomic catalyst to promote electrocatalytic nitrate reduction to ammonia. Energy Environ. Sci. 2024;17:6717–6727. [ Google Scholar ] 25. Huang S., Lin F., Wang S., et al. Asymmetric microenvironment tailoring strategies of atomically dispersed dual-site catalysts for oxygen reduction and CO2 reduction reactions. Adv. Mater. 2024;36 doi: 10.1002/adma.202407974. [ DOI ] [ PubMed ] [ Google Scholar ] 26. Chang J., Jing W., Yong X., et al. Synthesis of ultrahigh-metal-density single-atom catalysts via metal sulfide-mediated atomic trapping. Nat. Synth. 2024;3:1427–1438. [ Google Scholar ] 27. Rao P., Deng Y., Fan W., et al. Movable type printing method to synthesize high-entropy single-atom catalysts. Nat. Commun. 2022;13:5071. doi: 10.1038/s41467-022-32850-8. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 28. Jiao D., Li X., Sun M., et al. Machine learning driven rational design of dual atom catalysts on graphene for carbon dioxide electroreduction. Nano Res. 2024;18 [ Google Scholar ] 29. Wu L., Guo T., Li T. Data-driven high-throughput rational design of double-atom catalysts for oxygen evolution and reduction. Adv. Funct. Mater. 2022;32 [ Google Scholar ] 30. Zhang L., Guo X., Zhang S., et al. Building up the “Genome” of bi-atom catalysts toward efficient HER/OER/ORR. J. Mater. Chem. A. 2022;10:11600–11612. [ Google Scholar ] 31. Sarma B.B., Maurer F., Doronkin D.E., et al. Design of single-atom catalysts and tracking their fate using operando and advanced X-ray spectroscopic tools. Chem. Rev. 2023;123:379–444. doi: 10.1021/acs.chemrev.2c00495. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 32. Liang Z., Song L., Sun M., et al. Tunable CO/H2 ratios of electrochemical reduction of CO2 through the Zn-Ln dual atomic catalysts. Sci. Adv. 2021;7:eabl4915. doi: 10.1126/sciadv.abl4915. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 33. Chen L., Tang C., Zheng Y., et al. Triple-atom catalysts for one-step N–C–N coupling toward urea synthesis: A DFT study. Sci. China Mater. 2023;66:2346–2353. [ Google Scholar ] 34. Li M., Zhu H., Yuan Q., et al. Proximity electronic effect of Ni/Co diatomic sites for synergistic promotion of electrocatalytic oxygen reduction and hydrogen evolution. Adv. Funct. Mater. 2023;33 [ Google Scholar ] 35. Pei J., Wang T., Sui R., et al. N-Bridged Co–N–Ni: New bimetallic sites for promoting electrochemical CO2 reduction. Energy Environ. Sci. 2021;14:3019–3028. [ Google Scholar ] 36. Zhang S., Wu J., Zheng M., et al. Fe/Cu diatomic catalysts for electrochemical nitrate reduction to ammonia. Nat. Commun. 2023;14:3634. doi: 10.1038/s41467-023-39366-9. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 37. Cao S., Zhou S., Chen H., et al. How can the dual-atom catalyst FeCo–NC surpass single-atom catalysts Fe–NC/Co–NC in CO2RR? – CO Intermediate assisted promotion via a synergistic effect. Energy Environ. Mater. 2023;6 [ Google Scholar ] 38. Sun X., Qiu Y., Jiang B., et al. Isolated Fe-Co heteronuclear diatomic sites as efficient bifunctional catalysts for high-performance lithium-sulfur batteries. Nat. Commun. 2023;14:291. doi: 10.1038/s41467-022-35736-x. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 39. Lim Y., Chang H., Kim H., et al. Sequential element control of non-precious dual atom catalysts on mesoporous carbon nanotubes for high performance lithium–oxygen batteries. J. Mater. Chem. A. 2024;12:28953–28964. [ Google Scholar ] 40. Xie Y., Chen X., Sun K., et al. Direct oxygen-oxygen cleavage through optimizing interatomic distances in dual single-atom electrocatalysts for efficient oxygen reduction reaction. Angew. Chem. Int. Ed. 2023;62 doi: 10.1002/anie.202301833. [ DOI ] [ PubMed ] [ Google Scholar ] 41. Liu T., Chen Y., Xu A., et al. Regulating atomic Fe-Rh site distance for efficient oxygen reduction reaction. Sci. China Chem. 2024;67:1352–1359. [ Google Scholar ] 42. Jiang S., Xue J., Liu T., et al. Visualization of the distance-dependent synergistic interaction in heterogeneous dual-site catalysis. J. Am. Chem. Soc. 2024;146:29084–29093. doi: 10.1021/jacs.4c10613. [ DOI ] [ PubMed ] [ Google Scholar ] 43. Wang L., An Q., Sheng X., et al. Modulation of electronic spin state and construction of dual-atomic tandem reaction for enhanced pH-universal oxygen reduction. Appl. Catal. B Environ. Energy. 2024;343 [ Google Scholar ] 44. Zhang J., Zhao Y., Zhao W., et al. Improving electrocatalytic oxygen evolution through local field distortion in Mg/Fe dual-site catalysts. Angew. Chem. Int. Ed. 2023;62 doi: 10.1002/anie.202314303. [ DOI ] [ PubMed ] [ Google Scholar ] 45. Li H., Di S., Niu P., et al. A durable half-metallic diatomic catalyst for efficient oxygen reduction. Energy Environ. Sci. 2022;15:1601–1610. [ Google Scholar ] 46. Wang Y., Cheng W., Yuan P., et al. Boosting nitrogen reduction to ammonia on FeN4 sites by atomic spin regulation. Adv. Sci. 2021;8 doi: 10.1002/advs.202102915. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 47. Zhang L., Dong Y., Li L., et al. Concurrently boosting activity and stability of oxygen reduction reaction catalysts via judiciously crafting Fe–Mn dual atoms for fuel cells. Nano-Micro Lett. 2024;17:88. doi: 10.1007/s40820-024-01580-5. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 48. Tang T., Han J., Wang Z., et al. Diatomic Fe-Co catalysts synergistically catalyze oxygen evolution reaction. Nano Res. 2024;17:3794–3800. [ Google Scholar ] 49. Liu Y., Li J., Lv Z., et al. Efficient proton-exchange membrane fuel cell performance of atomic Fe sites via p–d hybridization with Al dopants. J. Am. Chem. Soc. 2024;146:12636–12644. doi: 10.1021/jacs.4c01598. [ DOI ] [ PubMed ] [ Google Scholar ] 50. Peng B., She H., Wei Z., et al. Sulfur-doping tunes p-d orbital coupling over asymmetric Zn-Sn dual-atom for boosting CO2 electroreduction to formate. Nat. Commun. 2025;16:2217. doi: 10.1038/s41467-025-57573-4. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 51. Wang X., Zhang N., Guo S., et al. p-d orbital hybridization induced by asymmetrical FeSn dual atom sites promotes the oxygen reduction reaction. J. Am. Chem. Soc. 2024;146:21357–21366. doi: 10.1021/jacs.4c03576. [ DOI ] [ PubMed ] [ Google Scholar ] 52. Liu J., Xu H., Zhu J., et al. Understanding the pathway switch of the oxygen reduction reaction from single- to double-/triple-atom catalysts: A dual channel for electron acceptance–backdonation. J. Am. Chem. Soc. 2023;3:3031–3044. doi: 10.1021/jacsau.3c00432. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 53. Ren Y., Sun X., Jing H., et al. Efficient electrochemical reduction of CO2 on g-C3N4 monolayer-supported metal trimer catalysts: A DFT study. Chem. Asian J. 2023;18 doi: 10.1002/asia.202201232. [ DOI ] [ PubMed ] [ Google Scholar ] 54. Zhong J., Liang Z., Liu N., et al. Engineering symmetry-breaking centers and d-orbital modulation in triatomic catalysts for zinc-air batteries. ACS Nano. 2024;18:5258–5269. doi: 10.1021/acsnano.3c08839. [ DOI ] [ PubMed ] [ Google Scholar ] 55. Chen C., Chai J., Sun M., et al. An asymmetrically coordinated ZnCoFe hetero-trimetallic atom catalyst enhances the electrocatalytic oxygen reaction. Energy Environ. Sci. 2024;17:2298–2308. [ Google Scholar ] 56. Liu W., Liang Z., Jing S., et al. Asymmetrical triatomic sites with long-range electron coupling for ultra-durable and extreme-low-temperature zinc–air batteries. Angew. Chem. Int. Ed. 2025;64 doi: 10.1002/anie.202503493. [ DOI ] [ PubMed ] [ Google Scholar ] 57. Poater J., Duran M., Solà M., et al. Theoretical evaluation of electron delocalization in aromatic molecules by means of atoms in molecules (AIM) and electron localization function (ELF) topological approaches. Chem. Rev. 2005;105:3911–3947. doi: 10.1021/cr030085x. [ DOI ] [ PubMed ] [ Google Scholar ] 58. Pei Z., Zhang H., Guo Y., et al. Atomically dispersed Fe sites regulated by adjacent single Co atoms anchored on N-P Co-doped carbon structures for highly efficient oxygen reduction reaction. Adv. Mater. 2024;36 doi: 10.1002/adma.202306047. [ DOI ] [ PubMed ] [ Google Scholar ] 59. Tang Q., Hao Q., Zhu Q., et al. Intrinsic electron transfer in heteronuclear dual-atom sites facilitates selective electrocatalytic carbon dioxide reduction. Adv. Energy Mater. 2025;15 [ Google Scholar ] 60. Xiao M., Chen Y., Zhu J., et al. Climbing the apex of the ORR volcano plot via binuclear site construction: Electronic and geometric engineering. J. Am. Chem. Soc. 2019;141:17763–17770. doi: 10.1021/jacs.9b08362. [ DOI ] [ PubMed ] [ Google Scholar ] 61. Zhu J., Xiao M., Ren D., et al. Quasi-covalently coupled Ni–Cu atomic pair for synergistic electroreduction of CO2. J. Am. Chem. Soc. 2022;144:9661–9671. doi: 10.1021/jacs.2c00937. [ DOI ] [ PubMed ] [ Google Scholar ] 62. Huang K., Li R., Qi H., et al. Regulating adsorption of intermediates via the sulfur modulating dual-atomic sites for boosting CO2RR. ACS Catal. 2024;14:8889–8898. [ Google Scholar ] 63. Yuan Y., Zheng Z., Wang Y., et al. Asymmetric diatomic site catalysts for highly efficient industrial-level CO2 electroreduction. Appl. Catal. B Environ. Energy. 2025;365 [ Google Scholar ] 64. Ma J., Lee W., Kim J.H., et al. Leveraging the intermetal distance in dual-atom catalysts: Revealing optimized synergistic interactions for CO2 electroreduction. ACS Nano. 2025;19:18698–18707. doi: 10.1021/acsnano.5c03160. [ DOI ] [ PubMed ] [ Google Scholar ] 65. Chen Y., Zhao J., Pan X., et al. Tuning the inter-metal interaction between Ni and Fe atoms in dual-atom catalysts to boost CO2 electroreduction. Angew. Chem. Int. Ed. 2024;63 doi: 10.1002/anie.202411543. [ DOI ] [ PubMed ] [ Google Scholar ] 66. Liu M., Li Y., Yang L., et al. Defect-triggered orbital hybridization in FeMn dual-atom catalysts toward sabatier-optimized oxygen reduction. Angew. Chem. Int. Ed. 2025;64 doi: 10.1002/anie.202505268. [ DOI ] [ PubMed ] [ Google Scholar ] 67. Tang B., Zhou Y., Ji Q., et al. A Janus dual-atom catalyst for electrocatalytic oxygen reduction and evolution. Nat. Synth. 2024;3:878–890. [ Google Scholar ] 68. Wang K., Liu J., Tang Z., et al. Establishing structure/property relationships in atomically dispersed Co–Fe dual site M–Nx catalysts on microporous carbon for the oxygen reduction reaction. J. Mater. Chem. A. 2021;9:13044–13055. [ Google Scholar ] 69. Li H., Wang J., Qi R., et al. Enhanced Fe 3d delocalization and moderate spin polarization in FeNi atomic pairs for bifunctional ORR and OER electrocatalysis. Appl. Catal. B Environ. Energy. 2021;285 [ Google Scholar ] 70. Yang G., Zhu J., Yuan P., et al. Regulating Fe-spin state by atomically dispersed Mn-N in Fe-N-C catalysts with high oxygen reduction activity. Nat. Commun. 2021;12:1734. doi: 10.1038/s41467-021-21919-5. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 71. Zhu P., Xiong X., Wang X., et al. Regulating the FeN4 moiety by constructing Fe–Mo dual-metal atom sites for efficient electrochemical oxygen reduction. Nano Lett. 2022;22:9507–9515. doi: 10.1021/acs.nanolett.2c03623. [ DOI ] [ PubMed ] [ Google Scholar ] 72. Lin X., Li Q., Hu Y., et al. Revealing atomic configuration and synergistic interaction of single-atom-based Zn-Co-Fe trimetallic sites for enhancing oxygen reduction and evolution reactions. Small. 2023;19 doi: 10.1002/smll.202300612. [ DOI ] [ PubMed ] [ Google Scholar ] 73. Li R., Gu C., Rao P., et al. Ternary single atom catalysts for effective oxygen reduction and evolution reactions. Chem. Eng. J. 2023;468 [ Google Scholar ] 74. Guo Y., Zhang Z., Chen D., et al. Precise construction of asymmetrically coordinated PtCuZn trimetallic atom catalysts for efficient oxygen reduction. Angew. Chem. Int. Ed. 2025;64 doi: 10.1002/anie.202507395. [ DOI ] [ PubMed ] [ Google Scholar ] 75. Chen C., Sun M., Zhang F., et al. Adjacent Fe Site boosts electrocatalytic oxygen evolution at Co site in single-atom-catalyst through a dual-metal-site design. Energy Environ. Sci. 2023;16:1685–1696. [ Google Scholar ] 76. Pei Z., Lu X.F., Zhang H., et al. Highly efficient electrocatalytic oxygen evolution over atomically dispersed synergistic Ni/Co dual sites. Angew. Chem. Int. Ed. 2022;61 doi: 10.1002/anie.202207537. [ DOI ] [ PubMed ] [ Google Scholar ] 77. Xu N., Jin Y., Liu Q., et al. Rational design of diatomic active sites for elucidating oxygen evolution reaction performance trends. Angew. Chem. Int. Ed. 2025;64 doi: 10.1002/anie.202413749. [ DOI ] [ PubMed ] [ Google Scholar ] 78. Liu H., Huang J., Feng K., et al. Reconstructing the coordination environment of Fe/Co dual-atom sites towards efficient oxygen electrocatalysis for Zn–air batteries. Angew. Chem. Int. Ed. 2025;64 doi: 10.1002/anie.202419595. [ DOI ] [ PubMed ] [ Google Scholar ] 79. Chen Y., Mao J., Zhou H., et al. Coordination shell dependent activity of CuCo diatomic catalysts for oxygen reduction, oxygen evolution, and hydrogen evolution reaction. Adv. Funct. Mater. 2024;34 [ Google Scholar ] 80. Pan F., Jin T., Yang W., et al. Theory-guided design of atomic Fe–Ni dual sites in N,P-co-doped C for boosting oxygen evolution reaction. Chem Catal. 2021;1:734–745. [ Google Scholar ] 81. Sun Z., Li C., Wei Z., et al. Sulfur-bridged asymmetric CuNi bimetallic atom sites for CO2 reduction with high efficiency. Adv. Mater. 2024;36 doi: 10.1002/adma.202404665. [ DOI ] [ PubMed ] [ Google Scholar ] 82. Yang Y., Zhang W., Wu G., et al. Electronic structure tuning in Cu–Co dual single atom catalysts for enhanced COOH* spillover and electrocalytic CO2 reduction activity. Angew. Chem. Int. Ed. 2025;64 doi: 10.1002/anie.202504423. [ DOI ] [ PubMed ] [ Google Scholar ] 83. Pei J., Yang L., Lin J., et al. Integrating host design and tailored electronic effects of yolk–shell Zn−Mn diatomic sites for efficient CO2 electroreduction. Angew. Chem. Int. Ed. 2024;63 doi: 10.1002/anie.202316123. [ DOI ] [ PubMed ] [ Google Scholar ] 84. Jin Z., Yang M., Dong Y., et al. Atomic dispersed hetero-pairs for enhanced electrocatalytic CO2 reduction. Nano Micro Lett. 2023;16:4. doi: 10.1007/s40820-023-01214-2. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 85. Yi J.-d., Gao X., Zhou H., et al. Design of Co-Cu diatomic site catalysts for high-efficiency synergistic CO2 electroreduction at industrial-level current density. Angew. Chem. Int. Ed. 2022;61 doi: 10.1002/anie.202212329. [ DOI ] [ PubMed ] [ Google Scholar ] 86. Min Z., Shao C., Chang B., et al. Constructing triple-atom sites for H2O participated electrocatalytic CO2 reduction. Chem. Eng. J. 2024;502 [ Google Scholar ] 87. Wang Y., Yin H., Dong F., et al. N-coordinated Cu–Ni dual-single-atom catalyst for highly selective electrocatalytic reduction of nitrate to ammonia. Small. 2023;19 doi: 10.1002/smll.202207695. [ DOI ] [ PubMed ] [ Google Scholar ] 88. Liu K., Sun Z., Peng X., et al. Tailoring asymmetric RuCu dual-atom electrocatalyst toward ammonia synthesis from nitrate. Nat. Commun. 2025;16:2167. doi: 10.1038/s41467-025-57463-9. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 89. Hu Y., Lan H., He J., et al. Entropy-engineered middle-in synthesis of dual single-atom compounds for nitrate reduction reaction. ACS Nano. 2024;18:23168–23180. doi: 10.1021/acsnano.4c05568. [ DOI ] [ PubMed ] [ Google Scholar ] 90. Xiao H., Lv Y., Hua W., et al. Atomically dispersed dual frustrated Lewis pairs for efficient relay conversion of nitrate into ammonia. Appl. Catal. B Environ. Energy. 2025;377 [ Google Scholar ] 91. Meng X., Wang K., Zhao Z., et al. Preventing nitrite desorption via switching hydrogenation position: A dual-site approach for selective nitrate reduction to ammonia. Small. 2025;21 doi: 10.1002/smll.202407216. [ DOI ] [ PubMed ] [ Google Scholar ] 92. Li Y., Qiu P., Qi C., et al. Synergistic electroreduction of nitrate to ammonia via bimetallic Fe-Cu sites supported on hierarchical porous carbon. Trans. Mater. Res. 2025;1 [ Google Scholar ] 93. Pu T., Ding J., Zhang F., et al. Dual atom catalysts for energy and environmental applications. Angew. Chem. Int. Ed. 2023;62 doi: 10.1002/anie.202305964. [ DOI ] [ PubMed ] [ Google Scholar ] 94. Cao L., Luo Q., Liu W., et al. Identification of single-atom active sites in carbon-based cobalt catalysts during electrocatalytic hydrogen evolution. Nat. Catal. 2019;2:134–141. [ Google Scholar ] 95. Wang X., Zhang N., Shang H., et al. Precisely designing asymmetrical selenium-based dual-atom sites for efficient oxygen reduction. Nat. Commun. 2025;16:470. doi: 10.1038/s41467-025-55862-6. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 96. Zhao X., Cheng H., Chen X., et al. Multiple metal–nitrogen bonds synergistically boosting the activity and durability of high-entropy alloy electrocatalysts. J. Am. Chem. Soc. 2024;146:3010–3022. doi: 10.1021/jacs.3c08177. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 97. Zhang Z., Zhao H., Xi S., et al. Breaking linear scaling relationships in oxygen evolution via dynamic structural regulation of active sites. Nat. Commun. 2025;16:1301. doi: 10.1038/s41467-024-55150-9. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 98. Ma R., Zhang J., Gong J., et al. The cooperative effects of the Rh-M dual-metal atomic pairs in formic acid oxidation. Angew. Chem. Int. Ed. 2025;64 doi: 10.1002/anie.202503095. [ DOI ] [ PubMed ] [ Google Scholar ] 99. Sun B., Zhang S., Yang H., et al. Revealing the active sites in atomically dispersed multi-metal–nitrogen–carbon catalysts. Adv. Funct. Mater. 2024;34 [ Google Scholar ] 100. Pan Y., Li Y., Dong C.L., et al. Unveiling the synergistic effect of multi-valence Cu species to promote formaldehyde oxidation for anodic hydrogen production. Chem. 2023;9:963–977. [ Google Scholar ] 101. Jiang H., He J., Zhou S., et al. Electron donor–acceptor activated anti-Fenton property for the ultradurable oxygen reduction reaction. ACS Nano. 2025;19:12161–12169. doi: 10.1021/acsnano.5c00085. [ DOI ] [ PubMed ] [ Google Scholar ] 102. Chen R., Zhao J., Li Y., et al. Operando Mössbauer spectroscopic tracking the metastable state of atomically dispersed tin in copper oxide for selective CO2 electroreduction. J. Am. Chem. Soc. 2023;145:20683–20691. doi: 10.1021/jacs.3c06738. [ DOI ] [ PubMed ] [ Google Scholar ] 103. Yin S., Yi H., Liu M., et al. An in situ exploration of how Fe/N/C oxygen reduction catalysts evolve during synthesis under pyrolytic conditions. Nat. Commun. 2024;15:6229. doi: 10.1038/s41467-024-50629-x. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 104. Cao X., Ding Y., Chen D., et al. Cluster-level heterostructure of PMo12/Cu for efficient and selective electrocatalytic hydrogenation of high-concentration 5-hydroxymethylfurfural. J. Am. Chem. Soc. 2024;146:25125–25136. doi: 10.1021/jacs.4c08205. [ DOI ] [ PubMed ] [ Google Scholar ] 105. Bai Y., Wang Z., Qin N., et al. Two-step redox in polyimide: Witness by in situ electron paramagnetic resonance in lithium-ion batteries. Angew. Chem. Int. Ed. 2023;62 doi: 10.1002/anie.202303162. [ DOI ] [ PubMed ] [ Google Scholar ] 106. Li X., Cao C.S., Hung S.F., et al. Identification of the electronic and structural dynamics of catalytic centers in single-Fe-atom material. Chem. 2020;6:3440–3454. [ Google Scholar ] 107. Zhou S., Shi J., Liu S., et al. Visualizing interfacial collective reaction behaviour of Li–S batteries. Nature. 2023;621:75–81. doi: 10.1038/s41586-023-06326-8. [ DOI ] [ PubMed ] [ Google Scholar ] 108. Yuan W., Zhu B., Li X.Y., et al. Visualizing H2O molecules reacting at TiO2 active sites with transmission electron microscopy. Science. 2020;367:428–430. doi: 10.1126/science.aay2474. [ DOI ] [ PubMed ] [ Google Scholar ] 109. Tamtaji M., Peng Q., Liu T., et al. Non-bonding interaction of dual atom catalysts for enhanced oxygen reduction reaction. Nano Energy. 2023;108 [ Google Scholar ] 110. Li Y., Wei B., Zhu M., et al. Synergistic effect of atomically dispersed Ni–Zn pair sites for enhanced CO2 electroreduction. Adv. Mater. 2021;33 doi: 10.1002/adma.202102212. [ DOI ] [ PubMed ] [ Google Scholar ] 111. Liang X.M., Wang H.J., Zhang C., et al. Controlled synthesis of a Ni2 dual-atom catalyst for synergistic CO2 electroreduction. Appl. Catal. B Environ. Energy. 2023;322 [ Google Scholar ] 112. Wang Y., Ma F., Zhang G., et al. Precise synthesis of dual atom sites for electrocatalysis. Nano Res. 2024;17:9397–9427. [ Google Scholar ] 113. Zhang Z., Dai Y., Guo P., et al. Noncovalent assembly strategy for efficient synthesis of dual-atom catalysts. Adv. Funct. Mater. 2024;34 [ Google Scholar ] 114. Zhao Q.P., Shi W.X., Zhang J., et al. Photo-induced synthesis of heteronuclear dual-atom catalysts. Nat. Synth. 2024;3:497–506. [ Google Scholar ] 115. Wang B., Yang X., Xie C., et al. A general metal ion recognition strategy to mediate dual-atomic-site catalysts. J. Am. Chem. Soc. 2024;146:24945–24955. doi: 10.1021/jacs.4c06173. [ DOI ] [ PubMed ] [ Google Scholar ] 116. Yan H., Lin Y., Wu H., et al. Bottom-up precise synthesis of stable platinum dimers on graphene. Nat. Commun. 2017;8:1070. doi: 10.1038/s41467-017-01259-z. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 117. Zhang L., Si R., Liu H., et al. Atomic layer deposited Pt-Ru dual-metal dimers and identifying their active sites for hydrogen evolution reaction. Nat. Commun. 2019;10:4936. doi: 10.1038/s41467-019-12887-y. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 118. Tong M., Sun F., Xing G., et al. Potential dominates structural recombination of single atom Mn sites for promoting oxygen reduction reaction. Angew. Chem. Int. Ed. 2023;62 doi: 10.1002/anie.202314933. [ DOI ] [ PubMed ] [ Google Scholar ] 119. Liu F., Shi L., Lin X., et al. Fe/Co dual metal catalysts modulated by S-ligands for efficient acidic oxygen reduction in PEMFC. Sci. Adv. 2023;9:eadg0366. doi: 10.1126/sciadv.adg0366. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 120. Tang X., Zhang Y., Tang S., et al. Structural and chemical origin of dual-atom sites for enhanced oxygen electroreduction. ACS Catal. 2024;14:13065–13080. [ Google Scholar ] 121. Chen H., Zhang Y., He Q., et al. A facile route to fabricate double atom catalysts with controllable atomic spacing for the r-WGS reaction. J. Mater. Chem. A. 2020;8:2364–2368. [ Google Scholar ] 122. Zhu A., Cao Y., Zhao N., et al. Geminal synergy in Pt–Co dual-atom catalysts: From synthesis to photocatalytic hydrogen production. J. Am. Chem. Soc. 2024;146:33002–33011. doi: 10.1021/jacs.4c09611. [ DOI ] [ PubMed ] [ Google Scholar ] 123. Zhang Y.X., Zhang S., Huang H., et al. General synthesis of a diatomic catalyst library via a macrocyclic precursor-mediated approach. J. Am. Chem. Soc. 2023;145:4819–4827. doi: 10.1021/jacs.2c13886. [ DOI ] [ PubMed ] [ Google Scholar ] 124. Zhao L., Cai Q., Mao B., et al. A universal approach to dual-metal-atom catalytic sites confined in carbon dots for various target reactions. Proc. Natl. Acad. Sci. USA. 2023;120 doi: 10.1073/pnas.2308828120. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 125. Chen Y., Lin J., Pan Q., et al. Inter-metal interaction of dual-atom catalysts in heterogeneous catalysis. Angew. Chem. Int. Ed. 2023;62 doi: 10.1002/anie.202306469. [ DOI ] [ PubMed ] [ Google Scholar ] 126. Leng K., Zhang J., Wang Y., et al. Interfacial cladding engineering suppresses atomic thermal migration to fabricate well-defined dual-atom electrocatalysts. Adv. Funct. Mater. 2022;32 [ Google Scholar ] 127. Lee D., Lee S., Choi I., et al. Positional functionalizations of metal–organic frameworks through invasive ligand exchange and additory MOF-on-MOF strategies: A review. Smart Mol. 2024;2 doi: 10.1002/smo.20240002. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 128. Yan X., Liu D., Guo P., et al. Atomically dispersed Co2MnN8 triatomic sites anchored in N-doped carbon enabling efficient oxygen reduction reaction. Adv. Mater. 2023;35 doi: 10.1002/adma.202210975. [ DOI ] [ PubMed ] [ Google Scholar ] 129. Zhao Q., Wang Y., Lai W.H., et al. Approaching a high-rate and sustainable production of hydrogen peroxide: Oxygen reduction on Co–N–C single-atom electrocatalysts in simulated seawater. Energy Environ. Sci. 2021;14:5444–5456. [ Google Scholar ] 130. Gu J., Jian M., Huang L., et al. Synergizing metal–support interactions and spatial confinement boosts dynamics of atomic nickel for hydrogenations. Nat. Nanotechnol. 2021;16:1141–1149. doi: 10.1038/s41565-021-00951-y. [ DOI ] [ PubMed ] [ Google Scholar ] 131. Men Y., Wu D., Hu Y., et al. Understanding alkaline hydrogen oxidation reaction on PdNiRuIrRh high-entropy-alloy by machine learning potential. Angew. Chem. Int. Ed. 2023;62 doi: 10.1002/anie.202217976. [ DOI ] [ PubMed ] [ Google Scholar ] 132. Duan X., Li Y., Zhao J., et al. Machine learning accelerated discovery of entropy-stabilized oxide catalysts for catalytic oxidation. J. Am. Chem. Soc. 2025;147:651–661. doi: 10.1021/jacs.4c12838. [ DOI ] [ PubMed ] [ Google Scholar ] 133. Wang Y., Li C., Han X., et al. General negative pressure annealing approach for creating ultra-high-loading single atom catalyst libraries. Nat. Commun. 2024;15:5675. doi: 10.1038/s41467-024-50061-1. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 134. Lin X., Wang Y., Chang X., et al. High-throughput screening of electrocatalysts for nitrogen reduction reactions accelerated by interpretable intrinsic descriptor. Angew. Chem. Int. Ed. 2023;62 doi: 10.1002/anie.202300122. [ DOI ] [ PubMed ] [ Google Scholar ] 135. Han L., Xiang Z. Intelligent design and synthesis of energy catalytic materials. Fundam. Res. 2025;5:624–639. doi: 10.1016/j.fmre.2023.10.012. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 136. Sun M., Wong H.H., Wu T., et al. Stepping out of transition metals: Activating the dual atomic catalyst through main group elements. Adv. Energy Mater. 2021;11 [ Google Scholar ] 137. Ling C., Ouyang Y., Li Q., et al. A general two-step strategy–based high-throughput screening of single atom catalysts for nitrogen fixation. Small Methods. 2019;3 [ Google Scholar ] 138. Lv X., Wei W., Huang B., et al. High-throughput screening of synergistic transition metal dual-atom catalysts for efficient nitrogen fixation. Nano Lett. 2021;21:1871–1878. doi: 10.1021/acs.nanolett.0c05080. [ DOI ] [ PubMed ] [ Google Scholar ] 139. Zhu C., Wang M., Wen C., et al. Establishing the principal descriptor for electrochemical urea production via the dispersed dual-metals anchored on the N-decorated graphene. Adv. Sci. 2022;9 doi: 10.1002/advs.202105697. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 140. Guan X., Gao W., Jiang Q. Design of bimetallic atomic catalysts for CO2 reduction based on an effective descriptor. J. Mater. Chem. A. 2021;9:4770–4780. [ Google Scholar ] 141. Shu Z., Chen H., Liu X., et al. High-throughput screening of heterogeneous transition metal dual-atom catalysts by synergistic effect for nitrate reduction to ammonia. Adv. Funct. Mater. 2023;33 [ Google Scholar ] 142. Li D., Sun P., Xu H., et al. A revised high-throughput screening model on oxygen reduction reaction over dual atom catalysts based on the axial pre-adsorption and O2 adsorption. Adv. Energy Mater. 2025;15 [ Google Scholar ] 143. Mai H., Le T.C., Chen D., et al. Machine learning for electrocatalyst and photocatalyst design and discovery. Chem. Rev. 2022;122:13478–13515. doi: 10.1021/acs.chemrev.2c00061. [ DOI ] [ PubMed ] [ Google Scholar ] 144. Zahrt A.F., Henle J.J., Rose B.T., et al. Prediction of higher-selectivity catalysts by computer-driven workflow and machine learning. Science. 2019;363:eaau5631. doi: 10.1126/science.aau5631. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 145. Xu H., Cheng D., Cao D., et al. Revisiting the universal principle for the rational design of single-atom electrocatalysts. Nat. Catal. 2024;7:207–218. [ Google Scholar ] 146. Sun J., Tu R., Xu Y., et al. Machine learning aided design of single-atom alloy catalysts for methane cracking. Nat. Commun. 2024;15:6036. doi: 10.1038/s41467-024-50417-7. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 147. Lei Y.J., Lu X., Yoshikawa H., et al. Understanding the charge transfer effects of single atoms for boosting the performance of Na-S batteries. Nat. Commun. 2024;15:3325. doi: 10.1038/s41467-024-47628-3. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 148. Mok D.H., Li H., Zhang G., et al. Data-driven discovery of electrocatalysts for CO2 reduction using active motifs-based machine learning. Nat. Commun. 2023;14:7303. doi: 10.1038/s41467-023-43118-0. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 149. Chang Y., Benlolo I., Bai Y., et al. High-entropy alloy electrocatalysts screened using machine learning informed by quantum-inspired similarity analysis. Matter. 2024;7:4099–4113. [ Google Scholar ] 150. Suvarna M., Pérez-Ramírez J. Embracing data science in catalysis research. Nat. Catal. 2024;7:624–635. [ Google Scholar ] 151. Tran R., Lan J., Shuaibi M., et al. The open catalyst 2022 (OC22) dataset and challenges for oxide electrocatalysts. ACS Catal. 2023;13:3066–3084. [ Google Scholar ] 152. Lin X., Du X., Wu S., et al. Machine learning-assisted dual-atom sites design with interpretable descriptors unifying electrocatalytic reactions. Nat. Commun. 2024;15:8169. doi: 10.1038/s41467-024-52519-8. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 153. Sun M., Wu T., Dougherty A.W., et al. Self-validated machine learning study of graphdiyne-based dual atomic catalyst. Adv. Energy Mater. 2021;11 [ Google Scholar ] 154. Tan Z., Li X., Zhao Y., et al. Machine learning-driven selection of two-dimensional carbon-based supports for dual-atom catalysts in CO2 electroreduction. ChemCatChem. 2024;16 [ Google Scholar ] 155. Wang M., Xiang Y., Lin Y., et al. p–d Orbital coupling in silicon-based dual-atom catalysts for enhanced CO2 reduction: Insight into electron regulation of active center and coordination atoms. J. Mater. Chem. A. 2024;12:31902–31913. [ Google Scholar ] 156. Cournia Z., Allen B.K., Beuming T., et al. Rigorous free energy simulations in virtual screening. J. Chem. Inf. Model. 2020;60:4153–4169. doi: 10.1021/acs.jcim.0c00116. [ DOI ] [ PubMed ] [ Google Scholar ] 157. Chen C., Zuo Y., Ye W., et al. A critical review of machine learning of energy materials. Adv. Energy Mater. 2020;10 [ Google Scholar ] 158. Sun X., Araujo R.B., dos Santos E.C., et al. Advancing electrocatalytic reactions through mapping key intermediates to active sites via descriptors. Chem. Soc. Rev. 2024;53:7392–7425. doi: 10.1039/d3cs01130e. [ DOI ] [ PubMed ] [ Google Scholar ] 159. Hu Y., Chen J., Wei Z., et al. Recent advances and applications of machine learning in electrocatalysis. J. Mater. Inf. 2023;3:18. [ Google Scholar ] Articles from Fundamental Research are provided here courtesy of The Science Foundation of China Publication Department, The National Natural Science Foundation of China ACTIONS View on publisher site PDF (7.7 MB) Cite Collections Permalink PERMALINK Copy RESOURCES Similar articles Cited by other articles Links to NCBI Databases Cite Copy Download .nbib .nbib Format: AMA APA MLA NLM Add to Collections Create a new collection Add to an existing collection Name your collection * Choose a collection Unable to load your collection due to an error Please try again Add Cancel Follow NCBI NCBI on X (formerly known as Twitter) NCBI on Facebook NCBI on LinkedIn NCBI on GitHub NCBI RSS feed Connect with NLM NLM on X (formerly known as Twitter) NLM on Facebook NLM on YouTube National Library of Medicine 8600 Rockville Pike Bethesda, MD 20894 Web Policies FOIA HHS Vulnerability Disclosure Help Accessibility Careers NLM NIH HHS USA.gov Back to Top