Electronic structure modulation of atomically dispersed metal electrocatalysts for the electrocatalytic oxygen reduction reaction

Abstract

High-efficiency, robust and low-cost electrocatalysts for the oxygen reduction reaction (ORR) are at the heart of new energy conversion and storage devices. Recently, atomically dispersed metal electrocatalysts (metal–nitrogen–carbon, M–N–C) for the ORR have received great attention. Herein, this review presents recent advances in the noble metal-free atomically dispersed metal electrocatalysts toward the ORR. Specifically, we first introduce the different mechanisms of 2e⁻ and 4e⁻ ORR on the catalyst. Then, the classification and corresponding recent advances in M–N–C electrocatalysts are reviewed, including metal coordination configuration (like the structure and coordination of N in M–N₄, heteroatom substitution, heteroatom doping in carbon skeleton and axial coordination), modulation of the second atom in diatomic catalysts, and the effect of metal nanoparticles/clusters in M–N–C catalysts. In parallel, the synthesis strategy, structure, electrochemical properties and reaction mechanism are highlighted. Finally, an outlook on the current advances and challenges and the potential of the M–N–C-based electrocatalysts towards 2e⁻ and 4e⁻ ORR are discussed.

---

1. Introduction

Oxygen conversion plays an essential role in sustainable energy techniques. The direct electrocatalytic four-electron oxygen reduction reaction (4e⁻ ORR) to produce water (H₂O) is an important reaction in fuel cells and metal–air batteries.¹ Producing hydrogen peroxide (H₂O₂) electrochemically through a two-electron oxygen reduction reaction (2e⁻ ORR) is a desirable technology that may replace the traditional anthraquinone process.^2,3 However, the intrinsically sluggish kinetics of the oxygen reduction reaction (ORR) limit the commercial application of these devices. Furthermore, the competition process between the 2e⁻ and 4e⁻ pathways complicates the design of highly active and selective ORR catalysts.^4,5 From this perspective, designing efficient and selective ORR electrocatalysts is important to promote the ORR kinetics and selectivity for energy conversion and storage devices in the industry.

Among numerous heterogeneous electrocatalysts, metal-based nanomaterials lay the foundation for nanoscience research and industrial applications. The size reduction of metal nanoparticles can effectively improve the number of exposed surface atoms, resulting in the evolution of the atomic, electronic, and defect structures of the surface, which is conducive to the catalytic performance. Besides, downsizing of nanoparticles, especially noble-metal nanoparticles, can increase metal atom utilization, thereby reducing the cost. Therefore, the fabrication of catalysts with atomically dispersed metals has attracted attention in recent years. Atomically dispersed metal catalysts, initially known as single-atom catalysts (SACs), were proposed by Zhang and co-workers in 2011.⁶ A single Pt atom dispersed on FeO_x was developed for the oxidation of carbon monoxide. Since the concept was introduced to catalysis more than a decade ago, the number of related studies reported has been growing exponentially year by year (Fig. 1a).

Fig. 1 (a) Statistics on the number of publications related to single-atom catalysts since 2011 (data obtained from Web of Science, collected in January 2024). Structures of SACs in different substrates, such as (b) surface alloys, reproduced from ref. 7 with permission from Springer Nature, copyright 2018; (c) oxide-supported metal atoms, reproduced from ref. 8 with permission from the American Association for the Advancement of Science, copyright 2016; (d) coordinated metal atoms; and (e) phthalocyanine-based molecular catalysts, reproduced from ref. 3 with permission from the American Chemical Society, copyright 2023.

Originally, SACs were defined as single metal atoms embedded in a solid substrate (e.g., metals, metal oxides or nitrides, and carbon substrates. The structures are shown in Fig. 1b–d). However, catalysts loaded on metals and metal oxides/nitrides face problems in electrocatalysis, such as poor corrosion resistance, poor conductivity, and difficulty in synthesis. Comparatively, carbon-based substrates are ideal candidates for anchoring metal atoms. Notably, it is rather challenging to spatially confine single metal atoms into ordered, defect-free carbon (e.g., graphene) because the migration of metal atoms can easily result in agglomeration to form clusters or nanoparticles.⁷ Therefore, many researchers have devoted their efforts to construct desirable defects in carbon-based substrates to anchor the metal atoms.

Nitrogen (N), which is more electronegative than carbon (C), can form both strong interactions with metals and neighboring C atoms and be integrated into the lattice structure. Thus, N is a suitable ligand atom for stabilizing metal atoms loaded on carbon substrates.^8–10 Besides, N doping can also modulate the electronic structure of the bonded metal and C atoms to redistribute the charge and regulate the adsorption energy of reactants in ORR; thus, they can affect the catalytic activity, selectivity and stability of catalysts. As early as 1964, catalysts featuring M–N coordination structures (Fig. 1e) were reported.¹¹ It was found that phthalocyanine compounds with defined M–N₄ structures exhibited oxygen reduction activity in alkaline electrolytes. Nevertheless, these types of molecular catalysts have the disadvantages of poor electrical conductivity and easy agglomeration, resulting in poor activity and low catalyst stability. In 1983, Scherson et al. found that cobalt porphyrin molecules could maintain the M–N coordination structure even after exposure to high temperatures, and could greatly improve the ORR activity and stability.¹² Since then, researchers realized that pyrolysis is an essential step in the preparation of atomically dispersed metal–nitrogen–carbon (M–N–C) catalysts. In 1989, Yeager et al. discovered that catalysts obtained by pyrolysis of mixtures of metal precursors, N-containing macromolecular polymers, and carbon materials showed comparable ORR performance to catalysts that were obtained from pyrolytic metal porphyrin-based molecules.¹³ Since then, the method of synthesis of M–N–C catalysts by pyrolysis of ligand-anchored metal atoms has been widely studied.

The distinct advantages of the application of M–N–C in electrocatalytic ORR are as follows: (1) maximum atomic utilization, which enables activity with ultralow metal loading. (2) Tunable coordination environments, in which saturated/unsaturated M-site configurations can be used to optimize interactions with reactants and intermediates. (3) Uniform active sites, which ensure high product selectivity. (4) Unique structures, which serve as ideal model systems for probing atomic-scale ORR mechanisms and structure–activity relationships. However, despite these attractive advantages of M–N–C, precisely controlling the active configurations of the metal center and maintaining its stability remains a serious challenge. Due to the extremely high surface energy of the isolated metal atoms, they can easily migrate and aggregate to form clusters/nanoparticles under harsh synthesis or catalysis conditions, leading to a decrease in activity and stability.^14,15 Therefore, the development of effective synthesis strategies is crucial for the controllable preparation of M–N–C. Moreover, there still exists a non-negligible activity gap between M–N–C and commercial Pt/C, especially in acidic electrolytes, suggesting that there is still vast room to improve the catalytic activity of M–N–C. Two synergistic strategies demonstrate significant potential: (1) enhancing intrinsic activity through electronic structure modulation, and (2) improving active site accessibility via increased density and exposure. So far, efforts have focused on the rational design of M–N–C with precise active configurations and the in-depth exploration of the origin of its activity. Therefore, it is necessary to summarize the current research results to provide new insights for further studies of M–N–C-based catalysts toward practical ORR.

In this review, we summarize and evaluate the recent advances in M–N–C-based catalysts in the field of electrocatalytic oxygen reduction reactions. Firstly, the commonly used pyrolysis synthesis strategy and the mechanism of ORR are briefly introduced. Then, the strategies for the electronic structure modulation of metal centers in M–N–C catalysts are described in detail, including heteroatom doping around the M site, constructing a second metal site, metal nanoparticle modulation, ligand engineering, and carbon substrate modulation. Finally, the challenges and future prospects for the M–N–C catalysts are presented.

2. Mechanism of the 2e⁻/4e⁻ oxygen reduction reaction

The 4e⁻ pathway is very meaningful for energy conversion devices (such as fuel cells and metal–air batteries),^15–17 while the 2e⁻ pathway is the optimal alternative to replace the conventional anthraquinone process for the production of hydrogen peroxide.^18,19 To design and synthesize ideal ORR electrocatalysts, understanding the ORR reaction mechanism at the atomic level is essential. The pathways exhibit strong pH dependence, demonstrating fundamentally distinct reaction kinetics and thermodynamic characteristics in acidic versus alkaline electrolytes. The overall reaction equations are as follows:²⁰

In acidic electrolytes:

4e⁻ O₂ + 4H⁺ + 4e⁻ → 2H₂O 1.229 V vs. RHE

2e⁻ O₂ + 2H⁺ + 2e⁻ → H₂O₂ 0.670 V vs. RHE

In alkaline electrolytes:

4e⁻  O₂ + 2H₂O + 4e⁻ → 4OH⁻ 0.401 V vs. RHE

2e⁻ O₂ + H₂O + 2e⁻ → HO₂⁻ + OH⁻ −0.065 V vs. RHE

2.1 The 4e⁻ ORR pathway

The ORR proceeds through a sophisticated sequence of steps involving oxygen adsorption, proton and electron transfer, bond cleavage/formation, and product release.^21,22 The pathway for complete conversion of oxygen to water depends on the state of the O₂ molecule (whether dissociation or association) before reduction (blue arrows in Fig. 2). The association mechanism proceeds through three key adsorbed intermediates (OOH*, O*, and OH*), initiating with molecular oxygen adsorption followed by sequential proton–electron transfers. The reaction begins with O₂ adsorption on catalytic sites, followed by initial proton-coupled electron transfer to form the OOH* intermediate (step 2). This intermediate then undergoes bifurcation into two distinct pathways: (i) OOH* protonation yielding OH* (step 2-2) or (ii) O–O bond cleavage producing O* and H₂O (step 2-1). The adsorbed species then sequentially undergo further protonation and reduction steps to ultimately form water molecules.^21,23 In contrast, the dissociation mechanism operates through direct O–O bond cleavage to O* before reduction, which subsequently combine with protons to form OH* (step 1); the OH* subsequently reacts in the same way as step 2-2. The intricate interplay between intermediates fundamentally governs the reaction kinetics, inevitably leading to the intrinsic sluggishness of ORR processes. The kinetic limitation stems from the complex energy landscape involving multiple bond-breaking/forming events and the delicate balance required for optimal intermediate adsorption/desorption thermodynamics.

Fig. 2 ORR process via 2e⁻ and 4e⁻ pathways in an acidic electrolyte (red and white balls represent O and H, respectively).

Gibbs free energy analysis serves as a fundamental probe for elucidating ORR reaction mechanisms, where the step exhibiting the maximum free energy difference determines the rate-determining step (RDS). Through density functional theory (DFT)-derived free energy diagrams, critical performance metrics, including reaction energy barriers and thermodynamic overpotentials, can be quantitatively evaluated. Contemporary computational approaches typically employ the thermodynamic framework established by Nørskov et al.,²⁴ which correlates intermediate adsorption energies with catalytic descriptors. Notably, the interaction between the active site and the reaction intermediates with different adsorption/desorption strengths is crucial for the selectivity and activity of catalysts. Based on Sabatier's principle, an ideal catalyst should have a moderate binding energy with the intermediates.²⁴

For M–N–C, modulation of the electronic structure of the M-sites enables precise tuning of intermediate binding energies, thereby enhancing both ORR efficiency and pathway control. Beyond thermodynamic parameters, DFT simulations provide atomic-scale insights into catalyst characteristics, including adsorption geometries, electron density redistribution, bond evolution trajectories, and Bader charge-transfer patterns. These microscopic descriptors synergistically reveal structure–activity relationships and dynamic reaction processes. Such a comprehensive mechanistic understanding forms the foundation for rationally engineering M–N–C catalysts with tailored electronic configurations, ultimately bridging theoretical predictions with experimental catalyst design.

2.2 The 2e⁻ ORR pathway

Compared with the 4e⁻ ORR, the generation of H₂O₂via the 2e⁻ ORR pathway is distinguished by the fact that the O–O bond of the intermediate (OOH*) is preserved or broken. As shown in Fig. 2 (red arrow), O₂ is adsorbed on the active site, followed by proton-coupled electron transfer to form the OOH* intermediate, which is the same as the first step of the 4e⁻ ORR. The subsequent pathway bifurcation is governed by the adsorption strength of OOH*: weak interfacial interaction facilitates sequential protonation and H₂O₂ desorption, while strong adsorption promotes O–O bond scission toward H₂O formation.^3,25,26 This delicate balance makes stabilization of the OOH* intermediate by optimizing the adsorption energy a key determinant of reaction selectivity.

The preservation of OOH* requires precise electronic control to maintain the metastable O–O bond while enabling timely H₂O₂ release before bond dissociation. Current challenges in 2e⁻ ORR catalysis stem from the inherent competition between requiring weak intermediate binding and maintaining sufficient reaction kinetics (needing moderate activation barriers). Advanced catalyst design must address this activity–selectivity paradox through atomic-level modulation of active site electronic environments, particularly on the M–N–C catalyst, where orbital hybridization characteristics directly influence intermediate stabilization.

3. Synthesis methods of M–N–C electrocatalysts

Currently, a large number of methods for the controlled synthesis of M–N–C have been reported, including pyrolysis, wet-chemical approaches, physical and chemical deposition, and mechanochemical processing. Among them, pyrolysis is the most common and well-established protocol for the preparation of M–N–C. Since the migration of metal atoms induced by high temperature is uncontrollable, it remains a great challenge to ensure the uniform distribution of metal atoms on the substrate. Addressing these issues requires complex engineering of precursors and pyrolysis conditions to simultaneously achieve objectives such as preventing the migration and agglomeration of metal atoms during pyrolysis, optimizing mass transfer and charge transfer capabilities, and maximizing the density of accessible active sites. Conventional pyrolysis involves controlled thermal decomposition of metal-containing precursors under an inert/reactive atmosphere such as nitrogen (N₂), hydrogen (H₂), ammonia (NH₃), or argon (Ar).¹⁴ The selection of precursor architecture fundamentally determines the final catalyst's coordination environment and atomic distribution. This section focuses on the pyrolysis strategies for achieving M–N–C catalysts, with particular emphasis on precursor design and spatial confinement approaches that collectively enable precise control over active site formation.

3.1 Pyrolysis of metal complexes

Electron-rich heteroatoms (N, O, S, P) with lone pairs of electrons can act as molecular anchors, forming robust coordination complexes with metal ions that resist thermal degradation. This coordination-driven immobilization mechanism effectively prevents metal atom migration/agglomeration during pyrolysis, ensuring atomic dispersion in the carbon substrates. Generally, some nitrogen-containing substances (e.g., phenanthroline,^12,27 4-dimethylaminopyridine,¹⁸ 2,2-bipyridine,²⁸ and graphitic carbon nitride²⁹), oxygen-containing functional groups (–COOH),³⁰ thiocyanide,³¹ and phytanic acid,³²etc., have been used to anchor the metal due to their high affinity for metal ions. The choice of chelating agent fundamentally determines the metal coordination environment, subsequently affecting intermediate adsorption energies and overall ORR activity.

Polymers. The pyrolysis of metal–polymer complexes has emerged as a versatile synthetic platform for M–N–C catalysts; the method involves two types of approaches: template-free and template-assisted. This methodology utilizes the inherent advantages of polymeric frameworks: (i) polymer matrices containing heteroatoms (O, N, S, and P) can adsorb and anchor the metal ions during pyrolysis, and (ii) the catalyst morphology and porosity can be engineered by modulating the polymer architecture, pyrolysis parameters, and template integration. For example, Wu et al.³¹ demonstrated this approach using a dicyandiamide/polyaniline hybrid precursor, achieving dual Fe–N₄ and Co–N₄ coordination sites through controlled pyrolysis. Similarly, Mu et al.³² obtained an Fe,Mn/N–C catalyst by carbonizing polymerized dicyandiamide, phytic acid, and metal salts, utilizing the phytic acid's phosphate groups for enhanced metal stabilization (Fig. 3a).

Fig. 3 Schematic of the synthesis of (a) Fe,Mn/N–C, reproduced from ref. 34 with permission from Springer Nature, copyright 2021; (b) (CM + PANI)-Fe-C, reproduced from ref. 38 with permission from Wiley-VCH GmbH, copyright 2017; (c) ISAS-Co/HNCS, reproduced from ref. 36 with permission from the American Chemical Society, copyright 2017; and (d) SA-Fe/CN, reproduced from ref. 37 with permission from the American Chemical Society, copyright 2017.

Nowadays, introducing a sacrificial template (e.g. SBA-15, SiO₂ spheres, metal hydroxides (FeOOH) nanorods, or inorganic salts (NaCl/KCl)) to the polymers has enabled hierarchical pore architectures that enhance the active site accessibility and promote mass transport. Moreover, template integration provides a certain range of spatial blocking, which prevents the migration and agglomeration of metal atoms during the pyrolysis process. For instance, Liu et al.³³ used dopamine to polymerize iron (Fe) ions on the surface of the template SBA-15; after pyrolysis and acid-etching, the FeSA/NSC catalyst with a configuration of FeN₄S₂ was formed (Fig. 3b). Similarly, Li et al.³⁴ used SiO₂ spheres as templates to synthesize Co–N–C (ISAS-Co/HNCS) catalysts (Fig. 3c). Briefly, the template and precursor were mixed hydrothermally and then pyrolyzed at high temperature under an atmosphere of Ar/H₂ to obtain the M–N–C catalyst with a SiO₂ template. After treatment with a strong base, the template was removed, which not only removes the template but also the excess metal particles agglomerated by pyrolysis, exposing more monatomic metal active sites. Meanwhile, Li et al.³⁵ used nanorod-shaped FeOOH as a template, which was coated with dopamine hydrochloride. The SA-Fe/CN catalyst was obtained by high-temperature pyrolysis and hydrochloric acid-etching (Fig. 3d). As mentioned above, the introduction of templates provides an effective way to endow the catalyst with a hierarchical porous structure. However, the template-assisted method has drawbacks, such as the complexity of the process and the easy introduction of impurities, and the removal of the template requires aggressive etching reagents that are harmful to the environment. Therefore, the use of eco-friendly templates (e.g. ice-templates) and self-templating polymers to mitigate the above issues while maintaining structural integrity has considerable potential for development.

Mixtures of complexes and carbon materials. The incorporation of highly conductive carbon materials (e.g. graphene, carbon black, and carbon nanotubes) into complexes for pyrolysis can further improve the conductivity of the catalysts while preserving the M–N–C structure. Inspired by this, Chen et al.¹⁸ obtained Co–N–C catalysts with different nitrogen coordination structures by pyrolysis of a mixture of ligands with varied nitrogen coordination geometries (4-dimethylaminopyridine, 2-methylimidazole, and dicyandiamide), cobalt salts, and carbon black (Fig. 4a), which demonstrated >90% selectivity for hydrogen peroxide under acidic conditions. In contrast, Xie et al.³⁰ functionalized the surface of CNTs with bipyridine, then introduced the Fe and thiocyanate ion, and subsequently pyrolyzed the assembly to obtain atomically dispersed Fe–N_x catalysts (denoted as S,N–Fe/N/C-CNT, Fig. 4b). Furthermore, sacrificial template-assisted methods can effectively inhibit metal atom agglomeration. For example, Sa et al.³⁶ adsorbed Fe porphyrin onto carbon nanotubes (CNTs), and then coated them with a SiO₂ layer, followed by high-temperature pyrolysis, which gave the porphyrin thin-layer-loaded carbon tube (CNT) catalyst with the template removed (CNT/PC) (Fig. 4c). This synthesis method enables atomic dispersion and isolation of the precursor and suppresses the migration and agglomeration of the atoms. As mentioned above, the method of using cheap and readily available complexes as precursors is favorable for large-scale preparations. Therefore, this method has become a mainstream method for the synthesis of M–N–C catalysts.

Fig. 4 (a) Schematic of the synthetic route for the three types of Co–N–C catalysts. Reproduced from ref. 20 with permission from the American Chemical Society, copyright 2022. (b) Synthetic process for S,N–Fe/N/C-CNT. Reproduced from ref. 32 with permission from Wiley-VCH GmbH, copyright 2017. (c) Synthetic scheme for the preparation of the CNT/PC catalysts. Reproduced from ref. 39 with permission from the American Chemical Society, copyright 2016.

3.2 Pyrolysis of MOFs

Compared with the complex precursors mentioned above, MOF materials have become ideal precursors due to their controllable metal coordination, periodic framework, and specific cavity structure. Taking advantage of the poor thermal stability of MOF, researchers prepared M–N–C catalysts by pyrolysis of MOF materials at high temperatures, which paved the way for new research on MOF materials. Using MOF materials as pyrolysis precursors has the following advantages: (1) the morphology, porous structure, and specific surface area of the precursor MOF can be maintained to a certain extent by controlling the pyrolysis conditions; (2) the spatial structure of MOF can block the metal, preventing metal atoms from migration and aggregation during pyrolysis. Given this, the properties of M–N–C catalysts derived from MOFs, such as high specific surface area, tunable composition, and unique structure, enable them to be directly used as catalysts or carbon substrates. Currently, many types of MOFs have been reported, such as zeolitic imidazolate frameworks (ZIFs), metal-containing covalent frameworks (M-COF), and others, such as MIL-101(Fe,Cr). Among them, ZIF-8 has been widely investigated for two main reasons: (1) ZIF-8 has a unique cavity structure that can act as a host for suitably sized metal precursor molecules, (2) zinc (Zn) metal, which has a boiling point of 907 °C, evaporates when it reaches the corresponding pyrolysis temperature, and the vacancies left behind can facilitate the formation of M–N–C catalysts by M and N coordination. To obtain M–N–C catalysts with different metals, ZIF-8 can be modulated using techniques such as doping and impregnation.

Doping method. This method refers to the partial replacement of Zn ions in ZIF-8 with transition-metal ions, resulting in the doping of the target metal into ZIF-8. The original Zn atom increases the distance between the metal atoms, which can effectively prevent them from aggregating. In addition, the evaporation of zinc promotes the formation of anchoring sites and stabilizes the M atoms. Based on this, Wu et al.³⁷ synthesized bimetallic Zn₁Co₁-BMOF with a Zn/Co molar ratio of 1 : 1 via a hydrothermal method. Then, pyrolysis (condition: 800 °C, Ar atmosphere) was performed to obtain Co-based single-atom catalysts (Co SAs/N–C) with precise N coordination (Fig. 5a). Based on this principle, the target MOF precursors can be obtained via a precisely controlled molar ratio of zinc/metal (M: Fe, Co, Ni, Cu, Ru, etc.), and the M–N–C catalysts can be obtained via a pyrolysis process. For example, Zelenay et al.³⁸ synthesized Fe-ZIF-8 precursors by precisely tuning the amount of Fe doping (0.05–9 at%) and then pyrolyzed them at high temperatures to obtain Fe–N–C catalysts. The density of the Fe–N active sites in the catalyst can be regulated by controlling the doping amount of Fe. In fact, if the pyrolysis temperature is too high, the density of the M–N active sites will be lower, but if the pyrolysis temperature is too low, the electrical conductivity of the catalyst will be affected. Therefore, it has been found that the electrical conductivity of the catalyst can be improved by adding conductive agents, e.g., carbon nanotubes (CNTs) or carbon black (CB), to the ZIF-8. Our group³⁹ obtained Fe-based catalysts with a better conductivity by adding the conductive CNT agent to the Fe-doped ZIF-8. Moreover, if two doped metal species are used and the doping amount is precisely regulated, a bimetallic atom catalyst can be obtained by pyrolysis. For example, He et al.⁴⁰ introduced both Fe and Co into ZIF-8, and the bimetallic atom catalyst FeCo-NC was obtained after pyrolysis (Fig. 5b). In addition, the regulation of M–N–C can be achieved by modulating the size, pore size, and morphology of ZIF-8 by controlling the amount of metal and organic ligands, as well as the type of solvent. For instance, Wu et al.⁴¹ synthesized Fe-doped ZIF-8 with different sizes by regulating the amount of metal and organic ligands, with subsequent pyrolysis to obtain the Fe–N–C catalysts.

Fig. 5 (a) Formation of Co SAs/N–C. Reproduced from ref. 40 with permission from Wiley-VCH GmbH, copyright 2016. (b) Synthesis scheme of the atomically dispersed dual-metal site FeCo-NC via pyrolysis of the chemically doped ZIF-8 precursors. Reproduced from ref. 43 with permission from the American Chemical Society, copyright 2022. (c) Schematic of model systems established by adsorbing Fe onto N-doped carbon. Reproduced from ref. 48 with permission from Wiley-VCH GmbH, copyright 2019. (d) Schematic of the formation of Fe-ISAs/CN. Reproduced from ref. 38 with permission from Wiley-VCH GmbH, copyright 2017.

Impregnation and adsorption method. In this method, metal salt ions are first impregnated and adsorbed, and then pyrolyzed to form the M–N–C catalysts. Generally, there are two routes of impregnation and adsorption: the first is to carbonize ZIF or ZIF derivatives to obtain a nitrogen-doped carbon substrate (NC), and then the metal salt ions are impregnated and adsorbed on the NC, and then pyrolyzed once again. The second route is to select metal source molecules with suitable sizes (e.g., acetylacetonate^42–44 or ferrocene⁴⁵) to be impregnated and adsorbed in the cavities of ZIF-8, then the assembly is pyrolyzed to obtain the M–N–C catalyst. Li and Zhang et al.⁴⁶ adsorbed Fe³⁺ ions on a host of N-doped carbon (ZIF-NC) derived from ZIF-8 and then carbonized the ZIF-NC-Fe³⁺ precursor to obtain the Fe–N–C catalyst (Fig. 5c). Generally, the derived ZIF-NC has a hierarchical porous structure, which is an ideal host for metal salt ions; moreover, this method is universal for other metal ions.^47,48

In addition, the cavity structure of ZIF-8 can be used to trap suitably sized molecules, which can avoid the migration and agglomeration of metal atoms during the pyrolysis process. For instance, Wang and Li et al.⁴⁴ adsorbed Fe(acac)₃ (diameter, 9.7 Å) into the cavity (diameter, 11.6 Å) of ZIF-8, which was subsequently pyrolyzed to obtain Fe–N–C (Fig. 5d), in which the loading of Fe can be as high as 2.16 wt%. Ferrocene (6.4 Å) and Ir(acac)₃ (9.8 Å) molecules, etc., were also trapped in the ZIF-8 cavities to prepare M–N–C catalysts for the ORR. Recently, our group⁴⁹ adsorbed ferric nitrate (Fe(NO₃)₃) on ZIF-8 and obtained diatomic Fe–N–C catalysts with different N-coordination structures by tuning the adsorption amount of Fe³⁺.

As described above, ZIF is an ideal precursor for the preparation of single-atom M–N–C catalysts. However, the synthesis process of ZIF precursors is complicated and time-consuming, which limits the large-scale preparation of M–N–C catalysts. On the other hand, to avoid agglomeration during pyrolysis, the amount of loading of M–N–C catalysts is typically very low, which hinders the further improvement of catalytic activity. Currently, various methods are available for the preparation of atomically dispersed M–N–C catalysts, but increasing the metal loading while avoiding the agglomeration of metal atoms is still the direction of future development, and methods for large-scale preparation are also needed to fulfill the practical requirements.

4. Strategies for modulating the electronic structure of metal sites in M–N–C catalysts

4.1 Effect of coordination configuration

Heteroatom doping serves as a foundational strategy for electronic structure modulation, aiming to perturb the intrinsic electron distribution of M–N–C catalysts by introducing heteroatoms (e.g., P, S, B, and O) with distinct electronegativities into the carbon matrix or near M–N_x active sites.^1,50,51 This perturbation regulates the charge density and d-band center of metal centers, thereby optimizing the adsorption/desorption behavior of oxygen intermediates during ORR.^52,53 A key advantage of this strategy is its compatibility with large-scale synthesis processes and its ability to synergize with other modulation approaches to enhance catalytic activity, selectivity, and stability simultaneously.

4.1.1 Structural modulation of coordination N in M–N₄. Most of the M–N–C electrocatalysts feature a central metal atom coordinated with four N atoms, forming an M–N₄ coordination structure. The N coordination geometry structure is typically classified as pyridine-type (Py-N) or pyrrole-type (Pr-N), which critically influences the catalytic selectivity of M–N–C.⁵³ For instance, combining experimental and theoretical investigations, Cao et al.⁵⁴ found that the 4e⁻ ORR activity of Fe@Py-N₄ was higher than that of Fe@Pr-N₄. It has been shown that the pyridine-N coordination is more favorable for the 2e⁻ ORR than pyrrole-N configurations. Wu et al.⁵⁵ converted pyrrole-type into pyridine-type coordination via NH₃ treatment of a Fe–N₄ catalyst, obtaining high-purity HP-Fe–N₄. This modified catalyst possessed a higher onset potential (E_onset) and a more positive half-wave potential (E_1/2) than the pristine pyridine-type Fe–N₄ (Fig. 6a). DFT calculations revealed that pyrrole-type Fe–N₄ exhibited stronger electron depletion around the Fe atoms than pyridine-type coordination, leading to superior adsorption energy of O₂ on the Fe sites, which promoted efficient 4e⁻ ORR (Fig. 6b and c). Furthermore, the selectivity of the ORR primarily depended on whether the O–O bond is preserved in OOH*. Gibbs free energy analysis indicates that the ΔG_OOH* on pyridinic-N is lower than that on pyrrolic-N.⁵⁵ This suggests that the O–O bond is more readily retained on the pyrrolic-N structure, which favors higher H₂O₂ selectivity.

Fig. 6 (a) ORR polarization curves of the HP-FeN₄, FeN₄ and NC catalysts in O₂-saturated 0.5 M H₂SO₄. (b) Calculated charge density difference of pyrrole-type FeN₄ and pyridine-type FeN₄, (c) free energy diagram of the oxygen reduction reaction on pyrrole-type FeN₄ and pyridine-type FeN₄. Reproduced from ref. 57 with permission from the Royal Society of Chemistry, copyright 2020. (d) Structure of Co–N–C and the corresponding H₂O₂ selectivity in 0.1 M HClO₄. (e) 3d electron configuration of pyridine-type and pyrrole-type CoN₄ with the adsorption of OOH*. Reproduced from ref. 20 with permission from the American Chemical Society, copyright 2022.

Building on the strategy mentioned above, Chen et al.¹⁸ engineered Co–N–C catalysts with controlled N-coordination (pyrrole-N and pyridine-N) to adjust the ORR selectivity (Fig. 6d). Experimental and computational analyses demonstrated that the pyrrole-type Co–N₄ structure primarily drives H₂O₂ production, while the pyridine-type mainly facilitates H₂O production. This significant difference was induced by the interaction of the electrons in the Co center with the transition state of OOH* and the difference spin states of the Co center between the different coordination modes (Fig. 6e).

4.1.2 Heteroatom substitutions. The substitution of N atoms in the M–N₄ structure by other atoms (e.g., C, S, P, and O) disrupts the electronic symmetry of the M center, thereby modifying its electronic structure and ultimately influencing the ORR activity. For example, Yuan et al.⁵⁶ leveraged phytic acid, pyrrole, and ferric chloride pyrolysis to obtain a P-substituted Fe–N_x catalyst (Fe–N₃P), which greatly improved the ORR reaction activity under alkaline conditions compared to Fe–N₄ (Fig. 7a). DFT calculations showed that the charge distribution and electronegativity of the Fe sites were optimized due to the asymmetric N/P coupling. As a result, the P-doped Fe sites are more favorable for the desorption of OH*, which accelerated the reaction kinetics and improved the ORR reaction activity (Fig. 7b). This strategy of localized substitution of N is applicable to other M–N–C catalysts as well. For example, Co–N₃C has the optimal ORR catalytic activity among Co–N₃C₁, Co–N₂C₂ and Co–N₄ (Fig. 7c).²⁹ This is attributed to the fact that C atom doping rearranges the d-electrons in the Co center, which shifts the d-band center of Co upwards, close to the Fermi energy level, and the d-orbitals are enhanced by hybridization with the p-orbitals of O₂ (Fig. 7d). In addition, the Mn–N_xO_y active site with N, O coordination exhibits higher ORR activity than Mn–N₄, and the Mn–N₃O₁ conformation displayed the optimal intrinsic ORR activity among all reported Mn–N_xO_y sites.⁵⁷

Fig. 7 (a) Schematic of N,P-coordinated single Fe site (Fe-N₃P). (b) ORR polarization curves of the Fe–N/P–C catalysts. Reproduced from ref. 58 with permission from the American Chemical Society, copyright 2020. (c) PDOS of the Co center and oxygen in Co–N₃C₁. (d) LSV curves of Co–N₄, Co–N₃C₁, and Co–N₂C₂ in O₂-saturated 0.1 M KOH. Reproduced from ref. 31 with permission from the American Chemical Society, copyright 2020. (e) Free energy diagram of Fe-ISA/SNC and Fe-ISA/NC during the ORR at U = 1.23 V. Reproduced from ref. 61 with permission from Wiley-VCH GmbH, copyright 2018. (f) Correlation between the intermediate adsorption energy and d-band center. (g) ORR polarization curves of FeN₄, FeN₄-PN, FeN₄-GN, and Pt/C. (h) Simulated local electric field on FeN₄-PN. (i) DOS diagram of O₂ activation on FeN₄-PN. Reproduced from ref. 63 with permission from the American Chemical Society, copyright 2021.

4.1.3 Heteroatom doping in the carbon skeleton around M–N₄. Beyond direct coordination of N atoms with metal atoms, doping heteroatoms (such as N, S, and P) in the carbon skeleton surrounding the M–N_x site can induce localized strain effects. As early as 2015, Sun et al.⁵⁸ demonstrated the enhancement of 4e⁻ ORR kinetics in S-doped Fe–N–C systems. Subsequently, Li et al.⁵⁹ synthesized S and N co-doped Fe–N–C catalyst (Fe-ISA/SNC) with controllable amounts of S and N via a copolymer pyrolysis strategy, achieving superior alkaline ORR activity to Fe–N–C without S doping. The introduction of S in the carbon skeleton around Fe–N induces negative charge aggregation on N atoms, which repels the adsorption of the intermediate OH* and promotes its reductive release (Fig. 7e). Lee et al.⁶⁰ further investigated the role of different S-containing functional groups (such as thiophene-S and oxidized-S) on the ORR activity. It was found that oxidized-S has an electron-withdrawing effect, which can shift the d-band of Fe downwards, thus lowering the adsorption energy of the transition state and increasing the ORR activity at the Fe–N₄ site (Fig. 7f). In contrast, the thiophene-S, with an electron-donating effect, reduced the intrinsic ORR activity of Fe–N₄ sites. Lin et al.⁶¹ constructed a pyrrole-nitrogen structure (PN) in the carbon plane near the Fe–N₄ site, distorting the local electric field, which achieved a charge redistribution at the Fe–N₄ site. The local electric field induced by the PN structure enhanced the charge cycling around the Fe–N₄ sites, which benefited the ORR activity (Fig. 7g and h). Further theoretical calculations showed that more electrons in FeN₄-PN were transferred from the d_z² orbitals of Fe to the p orbitals of O₂, which sufficiently activate the O₂ molecule (Fig. 7i).

In addition, adjusting the doping structure of the carbon skeleton around the M center can change the catalytic pathway of the ORR. Recently, Qiao et al.⁶² redirected the ORR pathway of Co-based single-atom catalysts from 4e⁻ ORR to 4e⁻ ORR by constructing a C–O–C group (second coordination spheres) around the metallic Co–N/O site. Theoretical and in situ experiments have demonstrated that this unique change in selectivity resulted from the migration of the active site from Co atoms to C atoms adjacent to the O-ligands. Moreover, this local structure optimizes the electronic structure of Co and the adsorption behavior of the transition state OOH* on the active site, resulting in >95% selectivity for acidic H₂O₂ electrosynthesis (Fig. 8a and b). These studies establish that the local coordination structures fundamentally govern ORR mechanisms and kinetics.

Fig. 8 (a) LSV curves of the ORR obtained in O₂-saturated 0.1 M HClO₄, showing the disk current density (j_disk) and ring current (i_ring). (b) Calculated H₂O₂ selectivity as a function of applied potential. Reproduced from ref. 64 with permission from the American Chemical Society, copyright 2021. (c) Electron localization functions of FeAB-O and FePc/AB. (d) Projected density of states of adsorption on FeAB-O. Reproduced from ref. 66 with permission from Springer Nature, copyright 2020.

4.1.4 Effect of axial coordination. Crucially, the partially filled d-orbitals perpendicular to the M–N plane in the centered metal engage small ligands (e.g., small molecules or anions) to form axial coordination structures. This axial bonding modulates the electronic configuration of M–N sites, thereby regulating ORR activity. For example, Cho and Liu et al.⁶³ obtained five-coordinated (Fe–N₅) catalyst (FePc-Py-CNT) by immobilizing Fe phthalocyanine (FePc) molecules on pyridine-functionalized carbon nanotubes (CNTs), which enhanced alkaline ORR activity. Meanwhile, Liu's group⁶⁴ used O₂ plasma radiation to graft O-containing groups onto acetylene black (AB) substrate to make an axial coordination (Fe–O) between the Fe center and the FePc. DFT calculations showed that the axial O coordination (O–FeN₄) greatly disrupted the symmetry of the electron density at the Fe site and changed the spin state of the Fe, resulting in the confinement of the electrons at the oxygen site (Fig. 8c). The electron confinement effect enhanced the interaction between the 3d orbitals of Fe and the p orbitals of adsorbed O₂, which was conducive to the adsorption and activation of O₂ (Fig. 8d). Some anions that have strong coordination with metals can also be selected as axial ligands. Recently, Li's group⁶⁵ constructed synergistic interactions between the axial chloride ligands at the near-range Fe sites and long-range S doping. This modulation optimized the Fe electronic structure, which greatly improved the alkaline ORR activity of the Fe–N–C catalysts.

4.2 Modulation of the second atom in dual-atomic catalysts

Building upon the structural constraints of conventional M–N–C catalysts in multi-molecular reactions, incorporating a secondary metal to form diatomic catalysts (DACs) represents a key electronic structure modulation strategy.^66,67 Beyond cleaving critical bonds (e.g., O–O and C=O) and serving as both electronic modulators and active sites, the secondary metal enables precise adjustment of the electronic configuration at adjacent active centers.^68–70 This electronic synergy, together with mutual metal stabilization and strengthened metal–support interactions, not only mitigates the structural limitations of M–N–C systems but also promotes concerted optimization of charge distribution, intermediate adsorption/desorption, and active-site stability. Thus, DACs constitute a targeted electronic engineering approach that surpasses the performance limits of traditional single-atom M–N–C catalysts, establishing a basis for exploring various configurations (e.g., bonded, X-bridged, and non-bonded) to tailor ORR reactivity further.

Nowadays, DACs are categorized by atomic connectivity: bonded configuration (direct metal–metal bonding) and X-bridged configuration (metal atoms linked via bridging atoms). Each type can be further classified as homonuclear (same metal) or heteronuclear (different metals, Fig. 9).²⁰

Fig. 9 Schematic models of different DACs (the balls in orange/blue, white, and dark gray represent metal atoms, bridging atoms, and substrate atoms, respectively). Reproduced from ref. 22 with permission from Elsevier Inc., copyright 2023.

4.2.1 Bonded diatomic catalysts. Compared with single-metal sites, dual-site binding can reduce O–O bond cleavage energy and adsorption/desorption energies of the transition states, which promote the selectivity of the 4e⁻ ORR. Wang et al.⁷¹ synthesized bonded heteronuclear Fe–Co dual-atom catalysts that outperformed Fe or Co single-atom catalysts in acidic 4e⁻ ORR (Fig. 10a and b). DFT calculations showed that Fe and Co provided two anchor sites for O₂ molecules, which led to the adsorption of O₂ at the Fe and Co sites in a bridge adsorption mode. The bridged O₂ adsorption mode alters O–O bond lengths, transition-state adsorption/desorption energies, and reaction pathways, collectively promoting efficient 4e⁻ ORR kinetics. Similar catalytic mechanisms have been demonstrated in dual-atom catalysts such as Zn–Co,⁷² Fe–Ni,⁶⁹ and Fe–Mn.⁷³

Fig. 10 (a) Proposed architectures of Fe–Co dual sites. Reproduced from ref. 73 with permission from the American Chemical Society, copyright 2017. (b) LSV curves of Pt/C, Co SAs/N–C, Fe SAs/NC, and (Fe,Co)/N–C in O₂-saturated 0.1 M HClO₄; the inset is E_onset. (c) O₂ adsorption configurations, O₂ adsorption energies, and O–O bond lengths of Fe₂–N–C, Fe₃–N–C. (d) LSV curves for Fe₁–N–C, Fe₂–N–C, Fe₃–N–C, NC and Pt/C. (e) Stability test for Fe₂–N–C. (f) LSV curves for FeN₄, Fe₂N₆, Fe–N NPs and Pt/C. Reproduced from ref. 69 with permission from Elsevier Inc., copyright 2019. (g) Proposed ORR reaction pathways on the planar-like Fe₂N₆ structure. (h) H₂O₂ yield and electron transfer number of isolated FeN₄, planar-like Fe₂N₆ structure, Fe–N nanoparticles, and Pt/C during the ORR process. Reproduced from ref. 68 with permission from Elsevier Inc., copyright 2020.

Ye et al.⁶⁷ synthesized bonded homonuclear Fe₂ DACs that showed superior activity and selectivity to Fe–N–C and bonded Fe₃ catalysts. It was found that O₂ preferentially formed bridge adsorption on the Fe₂/Fe₃ sites, which promoted the activation through O–O bond elongation (Fig. 10c). Compared with Fe₃ samples, Fe₂ can provide more active sites, a more suitable N coordination structure, and a higher graphitized carbon substrate, collectively enabling improved activity and stability (Fig. 10d and e). Meanwhile, Zhang et al.⁶⁶ developed a planar Fe₂N₆ catalyst with exceptional 4e⁻ ORR activity and selectivity in acidic media (Fig. 10f). Furthermore, in situ characterization explored the Fe–Fe bond and valence changes during the electrocatalytic process, thereby proposing the related 4e⁻ ORR reaction mechanism (Fig. 10g and h). Our work further demonstrated that planar N₆-coordination in Fe₂ DACs facilitated O–O bond cleavage as well as the suppression of H₂O₂ production.³ Due to the high inhibition of 4e⁻ ORR, the DACs usually exhibited superior 4e⁻ ORR activity and stability.

4.2.2 X-bridged diatomic catalysts. Most X-bridged diatom catalysts (DACs) feature heteronuclear configurations.⁷⁴ Yao et al.⁷⁵ designed a Co–Pt diatomic catalyst that exhibited unprecedented catalytic activity for the ORR. Theoretical calculations revealed that the high 4e⁻ ORR activity was due to the charge transfer from Pt to Co atoms and the downshift of the Co d-band center. Subsequently, Hunter et al.⁷⁶ computationally screened metal pairs in two systems (Fig. 11a–c). They found that CoPt@N8V4 exhibited optimal ORR properties due to Co-site distortion induced by Pt doping, which modulates reduction potentials. However, the metal pairs in M1M2N6V4 all displayed poor oxygen reduction activity owing to excessive OH* adsorption energy.

Fig. 11 (a) and (b) Metal pairs in two different systems. (c) ORR volcano plot for metal pairs in two different systems. Reproduced from ref. 78 with permission from the American Chemical Society, copyright 2019. (d) Optimized structure of Fe,Mn/N–C. (e) Magnetic susceptibility of Fe,Mn/N–C. (f) Chronoamperometric response of Fe,Mn/N–C, Fe/N–C, Mn/N–C and Pt/C in an O₂-saturated 0.1 M HClO₄ solution at a potential of 0.6 V. Reproduced from ref. 34 with permission from Springer Nature, copyright 2021.

Mu's group³² reported an Fe,Mn–N–C diatomic catalyst (Fig. 11d), which revealed that 3d orbital spin states in the transition metals critically govern the ORR activity. Magnetic measurements and theoretical calculations both proved that Mn–N groups dispersed close to each other can effectively activate the Fe^III site, leading to the ideal single-electron-filled intermediate spin state t_2g⁴e_g¹ of Fe^III in the FeN₄/C system (Fig. 11e). This electron configuration facilitated electron penetration into the oxygen antibonding π*-orbitals, enhancing O₂ activation through spin-state transitions. In addition, the introduction of Mn promotes the graphitization of carbon and the generation of MnO₂, which improves the stability of carbon nanoparticles and reduces the corrosion of the carbon substrate. Consequently, the Fe,Mn/N–C delivers exceptional ORR activity in both alkaline/acidic media with superior stability (Fig. 11f).

Based on the above analyses, bonded DACs mainly modified the O₂ adsorption mode in the ORR and promoted O–O bond cleavage, thereby accelerating the 4e⁻ ORR kinetics. In contrast, X-bridged DACs mainly modulated the d-electronic structure of the active site by secondary metal incorporation, which induced electron redistribution at the active center. This electronic modulation altered both O₂ adsorption behavior and the subsequent ORR reaction pathway.

To intuitively compare the aforementioned control strategies for the structures of single- and double-atom catalysts and their mechanism on the activity and selectivity of the oxygen reduction reaction, we summarized the data in Table 1.

Table 1 Comparison of the effects of the modulation strategies on the structure of heteroatom doping catalysts and DACs

Category	Strategies		Mechanism	References
Heteroatom doping	N coordination geometry structure of M–N4		a. Pyrrole-N coordination is more favorable for 2e^− ORR	55
			b. Pyridine-N coordination is more favorable for 2e^− ORR	18
	Heteroatom substitutions (MNaXb, X = P, O, and C)		a. Disrupted the symmetry of the electron density at the M site	29, 58, 59
			b. Rearranged the d-electrons
			c. Shifted the d-band center of M close to the Fermi energy level
	Heteroatom doping in the carbon skeleton around M–N4		a. Utilized the electron-donating effect of heteroatoms to modulate the d-band center	60, 61
			b. Distorted the local electric field to alter the charge distribution
	Axial coordination		a. Disrupted the symmetry of the electron density at the M site	64, 77
			b. Facilitated the adsorption and activation of O2
Dual-atomic catalysts (DACs)	Bonded DACs	Hetero-nuclear	a. Provided two anchor sites for O2	72, 73, 75
			b. Changed the adsorption mode of O2 and facilitates the activation of O2
		Homo-nuclear	a. Reduced the activation energy of O2	49, 66, 67
			b. Afforded two identical active sites
			c. Enhanced the selectivity of products
	X-bridged DACs		a. Promoted the charge transfer from M1 to M2, helping to modulate the electronic structure of each other	32, 78, 79
			b. Changed the spin-state of M, promoting the activation of O2

---

4.2.3 Others. Recently, a distinct class of heteronuclear diatomic catalysts (DACs) features non-bonded metal centers. The non-bonded DACs involve fixing dual-atom units separately on a support, with each atom forming an individual catalytic site. This structure facilitates fine-tuning of individual sites through interactions between the carrier and the metal atoms, contributing to a stable and balanced catalytic environment. For instance, Du et al.⁸⁰ anchored Ce and Se on a nitrogen-doped carbon substrate, achieving long-distance charge transfer, optimizing the charge distribution state between the dual atomic sites, and thereby enhancing their intrinsic activity. Luo et al.⁸¹ used DFT calculations to validate experiments, and proposed a DFT-guided strategy for designing non-bond Fe–M DACs for electrochemical ORR. They found that the counter-metal plays an important role in lowering the ORR overpotential of Fe sites, which can deliver charges through the intervening ligands to the Fe site to modify its spin state and ORR activity. Similar electronic modulation has been achieved by introducing Co–N₄,⁸² Ni–N₄,⁷⁸ and Mn–N₄ (ref. 79) sites near other M–N centers.

Utilizing noble metals to modulate the electronic structure of non-noble metals in non-bonded DACs represents a prevalent trend, given the superior performance of noble metals in electrocatalytic ORR. For example, Han et al.⁴³ introduced Pt–N₄ sites into Fe–N–C catalysts to form non-bonded DACs (Fe–N₄/PtN₄@NC), the ORR activities of which were superior to those of Fe–N₄ and Pt–N₄ catalysts. Crucially, Pt atoms modulated the d-electronic structure of adjacent Fe sites, which assisted the activation of O₂ molecules on the Fe sites and the rehybridization of the Fe 3d orbitals with the O 2p, thereby promoting the 4e⁻ ORR pathway. It is worth noting that the modulation of the distance between catalytic sites significantly influences the adsorption behavior of reactants, their interactions with catalytic sites, and the electronic transfer processes during the reaction. Recent studies indicate that the optimal distance between the two metal atoms in DACs design should be less than 1.2–1.5 nm.^81,83

This type of DAC offers three advantages: (1) adjacent metal sites cooperatively tune adsorbate binding states and energies; (2) the synergistic effect of adjacent dual active centers lowers the O–O bond cleavage barriers. (3) The adjacent M–N_x centers may change the reaction routes of the intermediates, which leads to changes in the activity and selectivity of the reaction.^20,84 Therefore, the rational design of M–N_x centers adjacent to each other on the atomic scale can improve the intrinsic catalytic activity of the catalysts. Despite these advantages, the random anchoring of sites on the carrier surface introduces uncertainties in catalytic behavior, complicating the regulation and prediction of catalytic pathways. Moreover, precisely anchoring two co-planar and separated sites within a support is a challenging engineering task that requires high synthetic precision.

4.3 Effect of metal nanoparticles/clusters on the M–N–C

As previously noted, pyrolysis-derived M–N–C catalysts typically co-produce metallic particles/clusters, carbides, and nitrides alongside atomically dispersed M–N_x sites without specific migration-inhibiting strategies. However, the effect of inorganic particles/clusters on the activity of M–N–C is still controversial. Generally, these metal particles/clusters are considered to be electrochemically inert. Recently emerging evidence indicates that proximal metal particles/clusters can modulate the electronic structure of the M–N_x site to enhance the ORR activity and stability. These catalysts leverage both electronic regulation of metal centers and intrinsic M–N–C electrocatalytic properties.⁸⁵ For example, Wan et al.⁸⁶ reported a composite catalyst of Fe clusters and Fe–N–C (Fig. 12a), which showed good ORR activity and stability in an acidic electrolyte (Fig. 12b). The adsorption of oxygen reduction intermediates on FeN₄ was modulated by the electronic interactions between the Fe clusters due to the short distance between the clusters and the neighboring FeN₄ (Fig. 12c). It was shown that the catalyst exhibited about 60% improvement in stability and resistance to demetallization. This work displayed the great potential of the electronic interactions between multiphase metals for improving M–N–C catalysts.

Fig. 12 (a) Model structure of Fe–N₄/Fe₄–C₆. (b) LSV curves of Fe_SA-2DNPC, Fe_SA/Fe_AC-2DNPC, and Fe_SA/Fe_NP-2DNPC in O₂-saturated 0.5 M H₂SO₄ (0.1 M HClO₄ for Pt/C). (c) ORR free energy diagram at 1.23 V for Fe–N₄ in Fe–N₄/Fe₄–C₆. Reproduced from ref. 88 with permission from Springer Nature, copyright 2022. (d) LSV curves of FeSAs-N-C, Co-SAs-N-C, FeCo-SAs-N-C, M/FeCo-SAs-N-C, and Pt/C. (e) Charge density differences of Fe₄/FeN₄. (f) O₂ dissociation barriers on FeN₄, Fe₄/FeN₄, and Fe₂Co₂/FeN₄ at high-spin states, reproduced from ref. 89 with permission from Wiley-VCH GmbH, copyright 2020. (g) Top/side views of Fe-N₄-S-1-Fe_cluster. (h) PDOS of the Fe d-band in Fe-N₄, Fe-N₄-S-1, and Fe-N₄-S-1-Fe_cluster. (i) Free energy diagrams of Fe-N₄, Fe-N₄-S-1, and Fe-N₄-S-1-Fe_cluster at 0 and 1.23 V. Reproduced from ref. 91 with permission from Wiley-VCH GmbH, copyright 2022.

Recently, Sun and his colleagues developed an interfacial engineering strategy between Fe₄ clusters and FeN₄ sites, and employed it in acidic 4e⁻ ORR (Fig. 12d).⁸⁷ DFT calculations illustrated that the Fe₄ clusters increased the local charge density of FeN₄, which elongated the O–O bond and thereby inhibited the formation of reactive oxygen species (ROS, Fig. 12e and f). The suppression of ROS formation increased the stability of the Fe sites while promoting the 4e⁻ pathway. Jiang et al.⁸⁸ demonstrated experimentally that the high ORR activity of this electrocatalyst could be attributed to the enhancement of Fe–N_x activity by the Fe/Fe₃C nanoparticles. It was found that the Fe/Fe₃C nanoparticles amplified Fe–N_x activity via electron transfer to the coordinated N, and the introduction of heteroatoms (S and O) in the vicinity of the outer M–N_x site disrupted the electronic configuration of the M–N_x sites, which enhanced the ORR activity. As reported by Zhai et al.,⁸⁹ Fe clusters and S doping can synergistically modulate the electronic structure of the Fe–N₄ site, and the synthesized FeSA-FeNC@NSC had an ORR activity superior to that of commercial Pt/C (Fig. 12g). The S atoms and Fe nanoclusters can disrupt the symmetric electronic structure of Fe–N₄, upshifting the Fe d-band center (Fig. 12h), and enhancing the (d–σ)* antibonding occupancy to weaken OH* adsorption on the Fe site (Fig. 12i).

The studies discussed above suggest that metal nanoparticles/clusters in M–N–C catalysts accelerate the interfacial charge transfer between multiphases, remodel the electronic structure of M–N_x, facilitate O₂ activation, stabilize the M–N_x structure, and further improve the stability.^{87,88,90–92}

5. Summary and challenges

Recent research progress in M–N–C-based catalysts has been systematically reviewed herein. We have highlighted pyrolysis-based synthetic strategies, with particular emphasis on suppressing metal migration/aggregation across diverse carbon substrates. Leveraging the structural tunability of M–N–C systems, we further analyzed how three key modulation strategies (including heteroatom doping proximal to M–N sites, construction of secondary metal sites, and introduction of metal nanoparticles/clusters) enhanced ORR activity, stability, and selectivity. Furthermore, advanced characterization techniques coupled with DFT calculations have elucidated active-site configurations and reaction mechanisms, and thereby established critical synthesis–structure–property relationships.

The regulatory mechanisms of these strategies can be summarized as three aspects: electronic structure dominance, adsorption/desorption energy mediation, and active site geometry effects. 1) The electronic structure of the M–N site, with respect to characteristics such as valence, energy band, orbital occupancy, and the spin state, governed the catalytic activity,³² while the charge density and d-band center position dictated the electron transfer efficiency in the catalytic reaction.⁵⁰ 2) The different electronic structures of the M–N sites lead to different adsorption strengths of O₂ and oxygen-containing transition states (OOH*, OH*, O*, etc.) on the active site, which results in different activities and stability. 3) Different structures of the M–N sites altered the O₂/transition-state adsorption modes and thereby tuned the activity and selectivity.

Despite their promise as noble-metal alternatives, M–N–C-based catalysts still face persistent challenges to achieve large-scale application.

(1) Most of the M–N–C-based catalysts still exhibit excellent ORR performance exclusively in alkaline electrolytes, while the performance in acid is suboptimal, and there is a gap between them and noble metal catalysts. Therefore, future efforts should prioritize selecting acid-stable metal centers and supports to bridge this gap, such as single-atom catalysts formed by acid-resistant noble metals (such as iridium and rhodium) and non-noble metals to enhance the acid resistance of the catalysts.

(2) Current characterizations are predominantly measured under static and non-operando conditions. Given the dynamic process of ORR, more in situ/operando measurement techniques need to be performed in the future to monitor and track the dynamic morphology and structure of the active sites. For example, to address the in situ problem, in situ XAFS may be combined with in situ Raman spectroscopy to monitor, in real-time, changes in metal center coordination (e.g., M–N_x bond length and coordination number) and surface intermediates during the ORR.

(3) Most of the reported catalyst metrics are derived from laboratory-scale evaluations. Critical data on the durability and activity under practical operating conditions remain scarce. Therefore, the collection of performance data under practical application conditions is essential.

(4) Most of the reported M–N–C catalysts are suitable for the 4e⁻ ORR, electrocatalytic H₂O₂ synthesis over 2e⁻ ORR is relatively rare. Therefore, the catalytic structure–property mechanism of the 2e⁻ ORR can also be investigated, which offers the possibility of electrocatalytic H₂O₂ production instead of the traditional anthraquinone process.

Conflicts of interest

The authors declare no conflict of interest.

Data availability

No new experimental data were generated in this review.

Acknowledgements

This study was supported by the National Natural Science Foundation of China (22172195 and 22308387) and the Guangdong Basic and Applied Basic Research Foundation (No. 2021A1515110907 and 2023A1515011935).

References

1. C. X. Zhao, B. Q. Li, J. N. Liu and Q. Zhang, Angew. Chem., Int. Ed., 2021, 60, 4448–4463.
2. S. Chen, X. Li, C. Kao, T. Luo, K. Chen, J. Fu, C. Ma, H. Li, M. Li, T. S. Chan and M. Liu, Angew. Chem., Int. Ed., 2022, 61, e202206233.
3. Y. Xu, S. Chen, L. Chen, X. Zhu, X. Qiu and M. Liu, J. Phys. Chem. Lett., 2023, 14, 5850–5859.
4. X. Yang, Y. Zeng, W. Alnoush, Y. Hou, D. Higgins and G. Wu, Adv. Mater., 2022, 34, 2107954.
5. L. Zhang, Q. Meng, R. Zheng, L. Wang, W. Xing, W. Cai and M. Xiao, Nano Res., 2023, 16, 4468–4487.
6. B. Qiao, A. Wang, X. Yang, L. F. Allard, Z. Jiang, Y. Cui, J. Liu, J. Li and T. Zhang, Nat. Chem., 2011, 3, 634–641.
7. Y. Wang, H. Su, Y. He, L. Li, S. Zhu, H. Shen, P. Xie, X. Fu, G. Zhou, C. Feng, D. Zhao, F. Xiao, X. Zhu, Y. Zeng, M. Shao, S. Chen, G. Wu, J. Zeng and C. Wang, Chem. Rev., 2020, 120, 12217–12314.
8. F. Dong, M. Wu, Z. Chen, X. Liu, G. Zhang, J. Qiao and S. Sun, Nano-Micro Lett., 2022, 14, 36.
9. B. Lu, Q. Liu and S. Chen, ACS Catal., 2020, 10, 7584–7618.
10. B. Singh, V. Sharma, R. Gaikwad, P. Fornasiero, R. Zbořil and M. Gawande, Small, 2021, 17, 2006473.
11. T. K. H. M. Cota, M. Chin and F. J. Schoenweis, Nature, 1964, 203, 1281.
12. D. Scherson, S. Gupta, C. Fierro and E. Yeager, Electrochim. Acta, 1983, 28, 1205–1209.
13. S. Gupta, D. Tryk, I. Bae, W. Aldred and E. Yeager, J. Appl. Electrochem., 1989, 19, 19–27.
14. H. Xu, D. Wang, P. Yang, A. Liu, R. Li, Y. Li, L. Xiao, X. Ren, J. Zhang and M. An, J. Mater. Chem. A, 2020, 8, 23187–23201.
15. D. Wang, P. Yang, L. Liu, W. Wang and Z. Chen, Mater. Today Energy, 2022, 26, 101017.
16. Y. Chen, H. Zhang, Y. Li, W.-W. Li, G.-P. Sheng and Y. Wang, ACS Nano, 2025, 19, 14187–14199.
17. M. Zhang, X. Lu, Z. Wu, S. Wang and D. Yan, Chem. Commun., 2025, 61, 15716–15734.
18. S. Chen, T. Luo, X. Li, K. Chen, J. Fu, K. Liu, C. Cai, Q. Wang, H. Li, Y. Chen, C. Ma, L. Zhu, Y. R. Lu, T. Chan, M. Zhu, E. Cortes and M. Liu, J. Am. Chem. Soc., 2022, 144, 14505–14516.
19. S. Chen, T. Luo, K. Chen, Y. Lin, J. Fu, K. Liu, C. Cai, Q. Wang, H. Li, X. Li, J. Hu, H. Li, M. Zhu and M. Liu, Angew. Chem., Int. Ed., 2021, 60, 16607–16614.
20. L. Yan, P. Li, Q. Zhu, A. Kumar, K. Sun, S. Tian and X. Sun, Chem, 2023, 9, 280–342.
21. M. Chen, Y. He, J. Spendelow and G. Wu, ACS Energy Lett., 2019, 4, 1619–1633.
22. X. Jin, M. Chang, H. Sun, C.-W. Chang, M. G. Sendeku, Y. Li, M. Wang, J. Fang, Y. Li, Q. Zhu, B. Li, J. Yu, Y. Liu, Z. Chang, G. Zhang, Z. Zhuang, L. Bai, Q. Ma, Z. Feng, W. Liu, J. Li and X. Sun, J. Am. Chem. Soc., 2025, 147, 2689–2698.
23. A. Kulkarni, S. Siahrostami, A. Patel and K. Norskov, Chem. Rev., 2018, 118, 2302–2312.
24. J. Nørskov, J. Rossmeisl, A. Logadottir and L. Lindqvist, J. Phys. Chem. B, 2004, 108, 17886–17892.
25. E. Jung, H. Shin, B. Lee, V. Efremov, S. Lee, H. Lee, J. Kim, W. Antink, S. Park, K. Lee, S. Cho, J. Yoo, Y. Sung and T. Hyeon, Nat. Mater., 2020, 19, 436–442.
26. S. Siahrostami, A. Verdaguer-Casadevall, M. Karamad, D. Deiana, P. Malacrida, B. Wickman, M. Escudero-Escribano, E. A. Paoli, R. Frydendal, T. W. Hansen, I. Chorkendorff, I. E. Stephens and J. Rossmeisl, Nat. Mater., 2013, 12, 1137–1143.
27. R. Jasinski, Nature, 1964, 201, 1212–1213.
28. X. Xie, L. Peng, H. Yang, G. I. N. Waterhouse, L. Shang and T. Zhang, Adv. Mater., 2021, 33, e2101038.
29. X. Hai, X. Zhao, C. Na Guo, C. Yao, W. Chen, Y. Liu, H. Du, C. Yan, S. J. Su, C. Pennycook, S. Xi Zhang and J. Lu, ACS Catal., 2020, 10, 5862–5870.
30. P. Chen, T. Zhou, L. Xing, K. Xu, Y. Tong, H. Xie, L. Zhang, W. Yan, W. Chu, C. Wu and Y. Xie, Angew. Chem., Int. Ed., 2017, 56, 610–614.
31. G. Wu, K. More, C. Johnston and P. Zelenay, Science, 2011, 332, 443–447.
32. G. Yang, J. Zhu, P. Yuan, Y. Hu, G. Qu, B. A. Lu, X. Xue, H. Yin, W. Cheng, J. Cheng, W. Xu, J. Li, J. Hu, S. Mu and J. N. Zhang, Nat. Commun., 2021, 12, 1734.
33. X. Liu, X. Zhai, W. Sheng, J. Tu, Z. Zhao, Y. Shi, C. Xu, G. Ge and X. Jia, J. Mater. Chem. A, 2021, 9, 10110–10119.
34. Y. Han, Y. G. Wang, W. Chen, R. Xu, L. Zheng, J. Zhang, J. Luo, R. A. Shen, Y. Zhu, W. C. Cheong, C. Chen, Q. Peng, D. Wang and Y. Li, J. Am. Chem. Soc., 2017, 139, 17269–17272.
35. M. Zhang, Y. Wang, W. Chen, J. Dong, L. Zheng, J. Luo, J. Wan, S. Tian, W. Cheong, D. Wang and Y. Li, J. Am. Chem. Soc., 2017, 139, 10976–10979.
36. Y. Sa, D. Seo, J. Woo, J. Lim, J. Cheon, S. Yang, J. M. Lee, D. Kang, T. J. Shin, H. S. Shin, H. Jeong, C. S. Kim, M. G. Kim, T. Kim and S. Joo, J. Am. Chem. Soc., 2016, 138, 15046–15056.
37. P. Yin, T. Yao, Y. Wu, L. Zheng, Y. Lin, W. Liu, H. Ju, J. Zhu, X. Hong, Z. Deng, G. Zhou, S. Wei and Y. Li, Angew. Chem., Int. Ed., 2016, 55, 10800–10805.
38. H. Zhang, H. Chung, D. Cullen, S. Wagner, U. Kramm, K. More, P. Zelenay and G. Wu, Energy Environ. Sci., 2019, 12, 2548–2558.
39. L. Ran, Y. Xu, X. Zhu, S. Chen and X. Qiu, ACS Nano, 2024, 18, 750–760.
40. Y. He, X. Yang, Y. Li, L. Liu, S. Guo, C. Shu, F. Liu, Y. Liu, Q. Tan and G. Wu, ACS Catal., 2022, 12, 1216–1227.
41. H. Zhang, S. Hwang, M. Wang, Z. Feng, S. Karakalos, L. Luo, Z. Qiao, X. Xie, C. Wang, D. Su, Y. Shao and G. Wu, J. Am. Chem. Soc., 2017, 139, 14143–14149.
42. M. Xiao, J. Zhu, G. Li, N. Li, S. Li, Z. P. Cano, L. Ma, P. Cui, P. Xu, G. Jiang, H. Jin, S. Wang, T. Wu, D. Su and Z. Chen, Angew. Chem., Int. Ed., 2019, 58, 9640–9645.
43. A. Han, X. Wang, K. Tang, Z. Zhang, C. Ye, K. Kong, H. Hu, L. Zheng, P. Jiang, C. Zhao, Q. Zhang, D. Wang and Y. Li, Angew. Chem., Int. Ed., 2021, 60, 19262–19271.
44. Y. Chen, S. Ji, Y. Wang, J. Dong, W. Chen, Z. Li, R. Shen, L. Zheng, Z. Zhuang, D. Wang and Y. Li, Angew. Chem., Int. Ed., 2017, 56, 6937–6941.
45. J. Wang, G. Han, L. Wang, L. Du, G. Chen, Y. Gao, Y. Ma, C. Du, X. Cheng, P. Zuo and G. Yin, Small, 2018, 14, e1704282.
46. J. Li, H. Zhang, W. Samarakoon, W. Shan, D. A. Cullen, M. Chen, D. Gu, K. L. More, G. Wang, Z. Feng, Z. Wang and G. Wu, Angew. Chem., Int. Ed., 2019, 58, 18971–18980.
47. D. Zhang, W. Chen, Z. Li, Y. Chen, L. Zheng, Y. Gong, Q. Li, R. Shen, Y. Han, W. Cheong, L. Gu and Y. Li, Chem. Commun., 2018, 54, 4274–4277.
48. J. Wang, W. Liu, G. Luo, Z. Li, C. Zhao, H. Zhang, M. Zhu, Q. Xu, X. Wang, C. Zhao, Y. Qu, Z. Yang, T. Yao, Y. Li, Y. Lin, Y. Wu and Y. Li, Energy Environ. Sci., 2018, 11, 3375–3379.
49. Y. Xu, W. Li, L. Chen, W. Li, W. Feng and X. Qiu, Inorg. Chem., 2023, 62, 5253–5261.
50. Y. Shi, Z. Ma, Y. Xiao, Y. C. Yin, W. Huang, Z. Huang, Y. Zheng, F. Y. Mu, R. Huang, G. Y. Shi, Y. Y. Sun, X. H. Xia and W. Chen, Nat. Commun., 2021, 12, 3021.
51. X. Zhu, Y. Xu, L. Ran, S. Chen and X. Qiu, Inorg. Chem., 2024, 63, 3893–3900.
52. L. Ran, Y. Zhang, W. Tong, L. Chen, M. Wang, H. Zhou, P. Farràs, S. Chen and X. Qiu, Angew. Chem., 2025, 137, e202514542.
53. Y. Gao, B. Liu and D. Wang, Adv. Mater., 2023, 35, 2209654.
54. L. Yang, D. Cheng, H. Xu, X. Zeng, X. Wan, J. Shui, Z. Xiang and D. Cao, Proc. Natl. Acad. Sci. U. S. A., 2018, 115, 6626–6631.
55. N. Zhang, T. Zhou, M. Chen, H. Feng, R. Yuan, C. a. Zhong, W. Yan, Y. Tian, X. Wu, W. Chu, C. Wu and Y. Xie, Energy Environ. Sci., 2020, 13, 111–118.
56. K. Yuan, D. Lützenkirchenhecht, L. Li, L. Shuai, Y. Li, R. Cao, M. Qiu, X. Zhuang, M. Leung, Y. Chen and U. Scherf, J. Am. Chem. Soc., 2020, 142, 2404–2412.
57. Y. Yang, K. Mao, S. Gao, H. Huang, G. Xia, Z. Lin, P. Jiang, C. Wang, H. Wang and Q. Chen, Adv. Mater., 2018, 30, e1801732.
58. Y. Wang, Y. Lai, L. Song, Z. Zhou, J. Liu, Q. Wang, X. Yang, C. Chen, W. Shi, M. Rauf and S. Sun, Angew. Chem., Int. Ed., 2015, 127, 10045–10048.
59. Q. Li, W. Chen, H. Xiao, Y. Gong, Z. Li, L. Zheng, X. Zheng, W. Yan, W. C. Cheong, R. Shen, N. Fu, L. Gu, Z. Zhuang, C. Chen, D. Wang, Q. Peng, J. Li and Y. Li, Adv. Mater., 2018, 30, 1800588.
60. Y. Mun, S. Lee, K. Kim, S. Kim, S. Lee, J. Han and J. Lee, J. Am. Chem. Soc., 2019, 141, 6254–6262.
61. Y. Lin, K. Liu, K. Chen, Y. Xu, H. Li, J. Hu, Y.-R. Lu, T. Chan, X. Qiu, J. Fu and M. Liu, ACS Catal., 2021, 11, 6304–6315.
62. C. Tang, L. Chen, H. Li, L. Li, Y. Jiao, Y. Zheng, H. Xu, K. Davey and S. Qiao, J. Am. Chem. Soc., 2021, 143, 7819–7827.
63. R. Cao, R. Thapa, H. Kim, X. Xu, M. Kim, Q. Li, N. Park, M. Liu and J. Cho, Nat. Commun., 2013, 4, 2076.
64. K. Chen, K. Liu, P. An, H. Li, Y. Lin, J. Hu, C. Jia, J. Fu, H. Li, H. Liu, Z. Lin, W. Li, J. Li, Y. R. Lu, T. S. Chan, N. Zhang and M. Liu, Nat. Commun., 2020, 11, 4173.
65. Y. Han, Y. Wang, R. Xu, W. Chen, L. Zheng, A. Han, Y. Zhu, J. Zhang, H. Zhang, J. Luo, C. Chen, Q. Peng, D. Wang and Y. Li, Energy Environ. Sci., 2018, 11, 2348–2352.
66. N. Zhang, T. Zhou, J. Ge, Y. Lin, Z. Du, C. a. Zhong, W. Wang, Q. Jiao, R. Yuan, Y. Tian, W. Chu, C. Wu and Y. Xie, Matter, 2020, 3, 509–521.
67. W. Ye, S. Chen, Y. Lin, L. Yang, S. Chen, X. Zheng, Z. Qi, C. Wang, R. Long, M. Chen, J. Zhu, P. Gao, L. Song, J. Jiang and Y. Xiong, Chem, 2019, 5, 2865–2878.
68. Z. Zhang, X. Huang and H. Xu, ACS Appl. Mater. Interfaces, 2021, 13, 43632–43640.
69. M. Ma, A. Kumar, D. Wang, Y. Wang, Y. Jia, Y. Zhang, G. Zhang, Z. Yan and X. Sun, Appl. Catal., B, 2020, 274, 119091.
70. B. Zhang, J. Rui, Y. Zhang, L. Yang, S. Kubuki, Y. C. Yong and L. Zhang, Inorg. Chem. Front., 2025, 12, 4968–5002.
71. J. Wang, Z. Huang, W. Liu, C. Chang, H. Tang, Z. Li, W. Chen, C. Jia, T. Yao, S. Wei, Y. Wu and Y. Li, J. Am. Chem. Soc., 2017, 139, 17281–17284.
72. Z. Lu, B. Wang, Y. Hu, W. Liu, Y. Zhao, R. Yang, Z. Li, J. Luo, B. Chi, Z. Jiang, M. Li, S. Mu, S. Liao, J. Zhang and X. Sun, Angew. Chem., Int. Ed., 2019, 58, 2622–2626.
73. S. Sarkar, A. Biswas, T. Purkait, M. Das, N. Kamboj and R. S. Dey, Inorg. Chem., 2020, 59, 5194–5205.
74. L. Wang, Q. An, X. Sheng, Z. Mei, Q. Jing, X. Zhao, Q. Xu, L. Duan, X. Zou and H. Guo, Appl. Catal., B, 2024, 343, 123509.
75. L. Zhang, J. Fischer, Y. Jia, X. Yan, W. Xu, X. Wang, J. Chen, D. Yang, H. Liu, L. Zhuang, M. Hankel, D. J. Searles, K. Huang, S. Feng, C. L. Brown and X. Yao, J. Am. Chem. Soc., 2018, 140, 10757–10763.
76. M. Hunter, J. Fischer, Q. Yuan, M. Hankel and D. Searles, ACS Catal., 2019, 9, 7660–7667.
77. R. Cao, R. Thapa, H. Kim, X. Xu, M. Gyu Kim, Q. Li, N. Park, M. Liu and J. Cho, Nat. Commun., 2013, 4, 2076.
78. J. Chen, H. Li, C. Fan, Q. Meng, Y. Tang, X. Qiu, G. Fu and T. Ma, Adv. Mater., 2020, 32, e2003134.
79. S. Gong, C. Wang, P. Jiang, L. Hu, H. Lei and Q. Chen, J. Mater. Chem. A, 2018, 6, 13254.
80. L. Yin, S. Zhang, M. Sun, S. Wang, B. Huang and Y. Du, Nano Res., 2024, 17, 4753–4763.
81. M. Tamtaji, Q. Peng, T. Liu, X. Zhao, Z. Xu, P. R. Galligan, M. D. Hossain, Z. Liu, H. Wong, H. Liu, K. Amine, Y. Zhu, W. A. Goddard Iii, W. Wu and Z. Luo, Nano Energy, 2023, 108, 108218.
82. H. Li, Y. Wen, M. Jiang, Y. Yao, H. Zhou, Z. Huang, J. Li, S. Jiao, Y. Kuang and S. Luo, Adv. Funct. Mater., 2021, 31, 2011289.
83. Z. Jin, P. Li, Y. Meng, Z. Fang, D. Xiao and G. Yu, Nat. Catal., 2021, 4, 615–622.
84. Y. Ying, X. Luo, J. Qiao and H. Huang, Adv. Funct. Mater., 2020, 31, 2007423.
85. X. Zhu, Y. Xu, L. Ran, S. Chen, Z. Pei and X. Qiu, Small Struct., 2024, 6, 2400548.
86. X. Wan, Q. Liu, J. Liu, S. Liu, X. Liu, L. Zheng, J. Shang, R. Yu and J. Shui, Nat. Commun., 2022, 13, 2963.
87. S. Yin, J. Yang, Y. Han, G. Li, L. Wan, Y. Chen, C. Chen, X. Qu, Y. Jiang and S. Sun, Angew. Chem., Int. Ed., 2020, 59, 21976–21979.
88. W. Jiang, L. Gu, L. Li, Y. Zhang, X. Zhang, L. Zhang, J. Wang, J. Hu, Z. Wei and L. Wan, J. Am. Chem. Soc., 2016, 138, 3570–3578.
89. W. Zhai, S. Huang, C. Lu, X. Tang, L. Li, B. Huang, T. Hu, K. Yuan, X. Zhuang and Y. Chen, Small, 2022, 18, 2107225.
90. P. Yu, L. Wang, F. Sun, Y. Xie, X. Liu, J. Ma, X. Wang, C. Tian, J. Li and H. Fu, Adv. Mater., 2019, 31, e1901666.
91. J. C. Li, Y. Meng, L. Zhang, G. Li, Z. Shi, P. X. Hou, C. Liu, H. M. Cheng and M. Shao, Adv. Funct. Mater., 2021, 31, 2103360.
92. X. Cui, L. Gao, S. Lei, S. Liang, J. Zhang, C. D. Sewell, W. Xue, Q. Liu, Z. Lin and Y. Yang, Adv. Funct. Mater., 2020, 31, 2009197.

---