Carbon-based materials for electrocatalytic energy conversion: from understanding to designing

Synthesis

Pyrolysis strategy has always been the main preparation strategy of M-N-C catalyst. Early on, inexpensive precursors such as polyacrylonitrile were mixed with Co (II) or Fe (II) salts and Vulcan XC-72 CBs, then subsequently subjected to heat treatment, yielding catalysts with activity comparable to transition metal macrocycles and polypyrrole black-based catalysts^[85]. However, the direct synthesis of M-N_x/C catalysts using CBs as precursor often lacks precise control of morphology and structure. To regulate the morphology and structure of the catalyst, the precursors of the carbon and nitrogen sources were changed. Co particles were synthesized and coated with graphite shells by pyrolysis of sucrose and urea^[86]; alternatively, using 2D porous carbon nanosheets (PCNs) mixed with nitrogen sources, the metal ions were reduced to metal particles during the thermal treatment process, while facilitating the escape of nitrogen-containing gases released by cyanamide decomposition, thus achieving nitrogen doping in N-doped PCNs (NPCNs)^[87]; The metal particles can also be encapsulated by synthetic CNTs and then N-doped in an ammonia atmosphere^[25]. Although the carbon material encapsulation metal can control the material morphology and improve the catalytic activity, it is also difficult to conduct detailed analysis of its active sites because of the limited exposed active sites. At present, metal-organic framework (MOF) materials are the most widely used as precursors of oxygen reduction catalysts. Pyrolysis of MOF precursors can increase the specific surface area and pore distribution of materials, which is conducive to the exposure of catalytic active sites and enhances the electrocatalytic performance. In addition, due to the clear location and distribution of metal sites in MOF materials, good dispersion of metal sites is still maintained after pyrolysis^[88]. In the carbonization process, MOF pyrolysis converts organic ligands to porous graphitic carbon matrices and metal nodes to oxide NPs, metallic NPs, or single-atom dopants. Pyrolysis conditions (atmosphere/temperature/precursor) significantly affect catalyst structure and properties^[89,90]. The precursors of which are crucial. Recently, Huang et al. reported comprehensive chemical and morphological evolution and transformation of two commercial MOFs, ZIF-8 and ZIF-67, which share the same sodalite topology, using a combination of in situ and ex-situ techniques^[91]. The reason why ZIF-67 loses the favorable porous structure after pyrolysis is carbon graphitization, which is catalyzed by Co clusters formed during pyrolysis; most of the metal in ZIF-8 evaporates initially and does not aggregate to form metal particles. In addition to MOF precursors, metal sources are also crucial in the pyrolysis process. Gao et al. reported that the Fe(acac)₃ precursor results in exclusive formation of atomically dispersed Fe-N_x coordination, while the FeCl₃ and Fe(NO₃)₃ lead to the formation of more inactive Fe₃C NPs^[92]. MOFs with precise porous structures and adjustable compositions are also considered ideal platforms for stabilizing SACs, enabling detailed structure-activity relationship studies to guide the design of high-performance catalysts. Wang et al. obtained a CoN_x coordination catalyst with excellent electrochemical properties by pyrolysis co-doping MOF^[93]. Notably, single-atom catalysts easily aggregate into clusters and particles, the formation of particles. In order to avoid such a situation, particles can be added to the MOF to support its structure and prevent its shrinkage during pyrolysis. Wu et al. successfully synthesized a Co monatomic catalyst using KCl particles as the supporting template^[94].

Active sites and regions

Engineering active sites at the atomic/cluster scales
The central atom of the M-N-C catalyst is often considered to be the active center and directly affects the electron cloud arrangement. Choi et al. investigated the effect of different metal types on catalytic activity^[95]. Pyrolysis of dicyandiamide with metal chlorides at 900 °C, using dicyandiamide as both the carbon and nitrogen sources, revealed that the type of metal seed can change the electrochemical and physical characteristics of the resulting N-doped carbon. N-doped carbon derived from Co demonstrated the highest ORR activity in 1M HClO₄, exhibiting the most extensive sp²-carbon network compared to other metal types. However, due to the residual metal particles in the catalyst, it is not conducive to exploring the structure of the active site. Zitolo et al. synthesized Fe-N-C materials nearly devoid of crystallographic iron structures following pyrolysis in argon or ammonia atmospheres. Although Fe-N-C catalysts pyrolyzed in argon and ammonia have almost identical FeN₄C₁₂ structural units, their ORR activity is significantly different^[96]. This difference is due to the highly basic N-groups that formed during the pyrolysis of NH₃, which significantly enhanced the ORR activity of FeN₄C₁₂. According to the present studies, the order of ORR activities of transition metal-based M-N-C is (Fe > Co > C ≥ Mn > Ni) -N-C under alkaline conditions [Figure 2A]. Peng reported that the active N content, residual metal content, surface area and pore structure of M-N-C catalysts were influenced by the choice of central metals, and the improvement of ORR performance resulting from combined effect of these factors rather than any single factor [Figure 2B-D]^[97].

Figure 2. (A) Current density and half-wave potential of M-Pani /C-Mela (M = Co, Fe, Mn, Cu and Ni) and metal-free catalysts in 0.1M KOH; (B) N content (at%) of oxidized, Py-, Pyr-, and Gr-N; (C) BET surface areas of M-Pani /C-Mela; (D) Pore-size distribution of transition metal-codoped carbon catalysts. This figure is quoted with permission from Peng et al.^[97]; (E) Projected density of states of Fe-3d orbitals of FeN₄ and FeMoN₆OH. This figure is quoted with permission from Zhu et al.^[99]; (F) Atomic structures of electrocatalysts (blue: N, orange: Fe, purple: Co, and green: Ni). This figure is quoted with permission from Zhai et al.^[101]. BET: Brunauer–Emmett–Teller.

The ORR performance of Fe-N-C catalyst is on par with that of the precious metal Pt, but the change of catalytic intensity of Fe-N-C catalyst intermediate makes it easy to produce Fenton effect and affect the stability of Fe-N-C catalyst. To mitigate this, introducing a second metal can alter the band center of Fe to optimize intermediate adsorption energy^[98]. The Fe-Mo diatomic center within the N-doped carbon substrate can effectively downshift the d-band center of Fe by adjusting the electron configuration, thus reducing the adsorption of ORR intermediates and significantly improving the ORR performance of the Fe-Mo active site [Figure 2E]^[99]. Unlike the site where the diatomic center is combined, the introduction of Ni atoms near the FeN₄ site can regulate the electronic structure and adsorption strength of ORR intermediates^[100]. Zhai et al. reported the synthesis of a trimetallic monatomic catalyst (Fe_3%Co_3%Ni_9%-NC) with excellent catalytic activity and stability^[101]. Compared with monometallic and bimetallic single-atom structures, the d-band center of the FeCoNi-NC shifts toward the Fermi level [Figure 2F]. The upshift of d-band center can reduce the electron filling in the antibonding orbital, which is beneficial to the absorption of the reaction intermediates on the FeCoNi-NC surface. In addition to introducing other metal atoms into the structure, the introduction of metal clusters into the structure also generates synergistic effects^[33]. Wan et al. reported an Fe-N-C catalyst featuring nitrogen-coordinated iron clusters and closely surrounding Fe-N₄ active sites^[102]. Due to unblocked electron transfer pathways and extremely short interaction distances, strong electronic interactions arise between the iron clusters and adjacent Fe-N₄ sites. The iron clusters optimize the adsorption strength of oxygen reduction intermediates on Fe-N₄ sites and shorten the bond amplitude of Fe-N₄ with incoherent vibrations.

Coordination structure
The bond between the coordination atom and the metal atom produces the coordination bond. The bond length and the bond energy greatly affect the stability of the coordination bond. This alters the reactant’s adsorption behavior on the metal atom, ultimately influencing the catalyst’s performance. Additionally, the number of coordinating nitrogen atoms plays a critical role in modifying the charge distribution at metal sites, thereby adjusting the adsorption strength of oxygen-containing species at the active sites. Taking Fe-N-C catalyst as an example, its active site is N-coordinated iron atom FeN_x (x = 1-5). FeN₄ synthesized by high temperature pyrolysis is the ideal active site. DFT calculations show that the ORR activity and formation energies of FeN_x species show an “inverted volcano” and “volcano” relationship with the Fe-N coordination number x, respectively^[103]. The ORR activity order is calculated to be FeN₄ > FeN₃ > FeN₂ > FeN₁ > FeN₅ [Figure 3A]. Among a series of FeN_x (x = 1-5) and the configuration of FeN₄ is the most stable. Xue et al. explained the high activity of FeN₄ through the electron spin states^[104]. The Fe active center of FeN₅ (Fe-N₅-LS) is in the low spin state (LS), and FeN₄ (Fe-N₄-HS) and FeN₃ (Fe-N₃-HS) are in the high spin state (HS) [Figure 3B-D]. The high spin Fe active center enables it to penetrate the antibonding π-orbital of oxygen more easily. The 3d-electronic structure of Fe-N₄-HS is t_2g³_eg², and this structure endows the catalyst with more suitable O₂ adsorption capacity.

Figure 3. (A) DFT-calculated FeN_x (x = 1-6) formation energies and FeN_x (x = 1-5) ORR overpotentials. This figure is quoted with permission from Li et al.^[103]; Temperature-dependent inverse susceptibility and electron configurations of (B) Fe-N₅-LS, (C) Fe-N₄-HS and (D) Fe-N₃-HS. This figure is quoted with permission from Xue et al.^[104]; (E) The Fe 3d orbital energy levels and the electron configurations of PFePc and PFePc-L were calculated; (F) Molecular orbitals of HO-PFePc-L. This figure is quoted with permission from Zhao et al.^[109]; (G) ORR active volcano curve through the 4e^- path (onset potential plotted as a function of ΔG_OH). This figure is quoted with permission from Sabhapathy et al.^[110]. (H) Schematic representation of the formation of Fe/Co/Ni-SAs/NS in different coordination environments. This figure is quoted with permission from Zhang et al.^[111]. DFT: Density functional theory; ORR: oxygen reduction reaction; LS: low spin state; HS: high spin state.

In addition, some special coordination structures have been proposed, which have excellent catalytic activity and stability due to their unique catalytic mechanism. Different from the planar FeN₄ structure, the planar Fe₂N₆ structure can greatly improve the catalytic activity and excellent stability. With optimized oxygen intermediate adsorption and synergistic effects from multiple active sites, the structure exhibits unique ORR redox behavior and strong O–O bond cleavage driving force, accelerating ORR kinetics and inhibiting side reactions^[105]. Inspired by heme enzymes, the central iron ion in porphyrin rings coordinates with four nitrogen atoms and an axial ligand. In the FeN₅ coordination structure, the axially oriented fifth nitrogen ligand is capable of exerting an electronic push-pull effect and/or a steric effect on the monoatomic site, thereby regulating the interaction energy between the active site and reaction intermediates^[106,107]. Gong et al. incorporated axial oxygen into Fe centers, forming Fe-O-Fe bridges that stabilized adjacent FeN₄ moieties^[108]. O ligands act as modulators of the electronic structure, weakening the bond between Fe and ORR intermediates and moving the ORR activity closer to the volcano-curve peak. Zhao et al. used strong-field and weak-field ligands as axial coordination groups to modulate the three-dimensional (3D) orbital configuration and electron-spin state of the Fe center in Fe-N₄ [Figure 3E]^[109]. The energy level of electron donor (d_z2) can be adjusted based on the field strength of axial ligands; when the I^- ligand with a weak field strength undergoes axial coordination with PFePc (PFePc-I), the Fe 3d_z2 in it has the lowest energy level, thus demonstrating higher ORR activity [Figure 3F]. The axial coordination can not only enhance the catalytic activity, but also reduce the toxic reaction. Fe-N-C catalysts are known accelerators of Fenton reaction, which convert H₂O₂ into reactive oxygen species (ROS). The ROS attack the carbon substrate and rapidly degrade the activity affecting the stability of the catalyst in the ORR process. The introduction of Cl into FeN₄ via axial coordination inhibits the formation of ROS, and Cl integration provides the best OH-free adsorption energy, the highest calculated starting potential, and higher H₂O₂ tolerance for ORR at FeN₄ sites compared to the original FeN₄ [Figure 3G]^[110]. When the heteroatom directly replaces the N atom in the M-N-C coordination structure and directly bonds with the M atom, the active center is greatly affected. Zhang et al. reported the preparation of single atoms (Fe, Co, Ni) on N, S co-doped porous carbon as a highly efficient catalyst for ORR [Figure 3H]^[111]. The catalytic activity of the active center of Fe-SAs/NSC, (FeN₄S₂) was superior to that of Co-SAs/NSC (CoN₃S₁) and Ni-SAs/NSC (NiN₃S₁), which originated from the significant effect of the bonding difference of the S-atoms on the metal’s localized structure and electronic configuration.

Carrier effect
In addition to the above exploration of M-N-C coordination as the active center, it is also considered that the C structure around the N structure often affects the charge distribution of the active center, which indirectly affects the catalytic activity. The proposed configurations are FeN₄-C₈ with pyridine-N, FeN₄-C₁₀ with pyridine-N, pyrrolic and FeN₄-C₁₂ with pyrrolic^[96,112]. Theoretical calculation proves that FeN₄-C₈ and FeN₄-C₁₂ exhibit better catalytic activity than FeN₄-C₁₀ in acidic media [Figure 4A]. In-situ Mössbauer spectroscopy results were conducted on electrodes subjected to 150 h of aging at varying potentials. The D1 signal is predominantly associated with either low-spin FeN₄C₈ or high-spin FeN₄C₁₂ configurations. Among these, FeN₄C₁₂ demonstrates superior activity and stability, while FeN₄-C₈’s instability is due to its experiencing more intense demetallation [Figure 4B]^[113]. However, other studies have suggested that the formation of FeN₄-C₁₀ sites from an intact graphene lattice is energetically favorable. In contrast, both FeN₄-C₁₂ and FeN₄-C₈ active sites require pre-existing suitable pore structures within the graphene layer to maintain their stability^[114].

Figure 4. (A) The free energy diagram of typical ORR pathway of various FeN₄ sites; (B) Fe-N-C in-situ Mössbauer spectra and the relative content of different types of FeN₄. This figure is quoted with permission from Xu et al.^[113]; (C) Optimized structure of a perfect FeN₄ site and Divacancy defect models 6, 6a, 6I, and 6II. This figure is quoted with permission from Liu et al.^[115]; (D) Molecular orbital distributions showing the HOMO and LUMO for c-ND-Fe and e-ND-Fe configurations. This figure is quoted with permission from Wang et al.^[116]; (E) Calculated free energy diagrams for the ORR on in-plane-type FeN₄P₂ and edge-type FeN₄P₂ at zero electrode potential. This figure is quoted with permission from Yin et al.^[117]; (F) The optimized structures of basal-plane Co-N₄/G and edge-hosted Co-N₄ configurations (brown: C; grey: N, blue: Co, and pink balls: H); (G) Computationally derived volcano plot for the 2e^- ORR, illustrating the relationship between limiting potential (UL) and Gibbs free energy change for OOH^* formation (ΔG_OOH*). This figure is quoted with permission from Tian et al.^[118]. ORR: Oxygen reduction reaction.

Because the ability of divacancy defects to modulate the electronic properties of carbon substrates and consequently alter the catalytic performance of Fe-N₄ centers. Liu et al. engineered diverse divacancy configurations adjacent to the Fe-N₄ site to investigate their influence on ORR activity. Various carbon defects are constructed near the FeN₄ site, for example, 5-8-5 defects, 555-777 defects, 5-7-7-5 defects. DOS analysis showed divacancy defects (6, 6a, 6I) significantly altered Fe 3d orbital distribution, particularly 3d_xz and 3d_z2 states, compared to pristine FeN₄ sites [Figure 4C]^[115]. The strong hybridization between the Fe 3d_z2 and O 2pz orbitals results in substantial splitting of the O₂ 2pz orbital state, transferring electrons from the Fe site to ^*O₂, reducing Δd and magnetic moments, and thus better activation of the O₂ molecule. In our past studies, we believe that CoN₄ and FeN₄ have similar structures and good catalytic activity. When Co atoms are dispersed in N-doped carbon fibers (10Co-N@CNF) and defective carbon fibers (10Co-N@DCNF), the catalytic activity of 10Co-N@DCNF is close to that of Pt/C. Carbon deficiency is the root cause of enhanced ORR activity.

To further understand the role of nitrogen-modified divacancies (ND) configurations in M-N-C electrocatalysts, it is essential to create controllable ND trapped metal atoms and analyze how their location differences impact intrinsic activity. In DFT calculations, edge-trapped Fe sites (e-ND-Fe) and center-trapped Fe sites (c-ND-Fe) were modeled. The results reveal that e-ND-Fe shows greater electron redistribution than c-ND-Fe, leading to stronger electron density localization near the Fe-N₄ active sites^[116]. These results suggest that e-ND-Fe has a stronger electron-donating ability than c-ND-Fe, leading to improved catalytic reactivity. Additionally, the smaller highest occupied molecular orbital (HOMO)-lowest unoccupied molecular orbital (LUMO) gap in e-ND-Fe (0.06 eV) compared to c-ND-Fe (0.18 eV) facilitates more efficient electron transfer in catalytic reactions [Figure 4D]. We regulated the pore configuration by adjusting the amount of FeCl₃ to regulate the exposure of e-ND-Fe species. Experimental results confirmed a strong link between ORR activity and the concentration of e-ND-Fe sites, aligning with theoretical predictions. Based on the edge effect, Yin et al. demonstrated that incorporation of long-range P at edge-type FeN₄ can optimize the adsorption energy of the intermediate and reduce the energy barrier of the reaction [Figure 4E]^[117]. The introduction of P atoms into the carbon-base substrate can induce an interfacial charge density redistribution, which enhances the internal electron conduction. This optimization facilitates the adsorption of key intermediates on the charge-density-enriched surface Fe atoms, thus lowering the total energy barrier of the reaction. Highly efficient and selective ORR electrocatalysts are crucial for developing advanced energy storage systems that combine high performance, cost-effectiveness, and sustainable operation Tian et al. demonstrated different ORR pathways mediated by varying Co-N₄ sites: edge-bearing sites (i.e., atomic Co-N₄ partially located at the edge of the defects and in pockets) preferentially catalyze the 2e^- ORR pathway, whereas the basal planar-bearing sites promote the 4e^- ORR process by theoretical calculations [Figure 4F and G]. They synthesized a series of edge-to-bulk ratios carbon-supported Co-SACs catalysts and found that the exposed edges could be readily oxidized by oxygen functional groups during the ORR process, which enhances the kinetics of the 2e^- ORR transfer pathway while consistently maintaining selectivity levels above 90%^[118].

This extensive investigation into M-N-C catalysts underscores both the complexity and the promising potential of these materials for applications in the ORR. M-N-C catalysts present numerous advantages, making them viable alternatives to precious metal catalysts in ORR applications. These advantages include their cost-effectiveness, the ability to fine-tune their properties by modifying the central metal atom, coordination structure, and carrier effects, and their high catalytic activity, which is comparable to that of platinum-based catalysts, particularly in designs featuring iron (Fe) and cobalt (Co). Furthermore, M-N-C catalysts exhibit versatility, allowing for customization to suit specific applications, such as selective 2e^- or 4e^- ORR processes. These attributes position M-N-C catalysts as promising candidates for reducing costs and enhancing the performance of fuel cells and other electrochemical devices.

Heteroatom-doped carbon catalysts

Synthesis
Pyrolysis: Thermal decomposition of organic compounds generally yields two primary products: volatile substances and carbon-enriched solid residues. Pyrolysis is a straightforward and commonly employed technique for producing diverse functional carbon materials. Heteroatom doping can be accomplished through thermal processing of combined carbon precursors and/or doping sources. The pyrolysis temperature, heating rate, and duration critically influence the composition, structure, and morphology of the resulting doped carbon materials. Carbon precursors are diverse, including chemical powders, polymers, covalent organic frameworks (COFs), covalent organic polymers (COPs), and MOFs. For example, furfuryl alcohol can be used to prepare N, O co-doped nanoporous carbon with a honeycomb structure, which has high uniformity and controllable pore size^[119]; using self-assembled urea-formaldehyde (UF) resin as a single precursor can synthesize N-doped nanoporous carbon spheres composed of nanoplates^[120]; The covalent organic framework (F-COF) can incorporate F and N into the precursor of COF in advance, and direct pyrolysis can prepare the dispersive F and N co-doped nanoporous carbon (F-NPC) catalyst^[121]; Heteroatom-enriched covalent triazine polymers serve as effective self-doping precursors for the synthesis of nitrogen-phosphorus-fluorine tri-doped carbon nanospheres, simultaneously incorporating carbon, nitrogen, phosphorus, and fluorine elements^[122]; direct carbonization of thin MIL-101-NH₂ flakes and simple acid etching can construct an interesting porous carbon structure similar to brain platygyra coral, realizing the morphology control and uniform N atom doping of carbon nanostructures^[123]. Currently, biomass, as an excellent substitute for fossil fuels, can be used to prepare carbon materials of different morphologies through pyrolysis, including porous carbon, layered porous carbon, carbon quantum dots, heteroatom-doped porous carbon, and carbon fibers. Using biomass as a precursor is also a method for synthesizing heteroatom-doped carbon materials that is environmentally sustainable and cost-effective. Garlic skin^[124], tobacco^[125], pine pollen^[126] and other biomasses, as carbon and heteroatom precursors, make porous carbon and heteroatoms self-doped, without using any external heteroatom precursors, thus making the synthesis green and environmentally friendly. There is also the use of plant roots, stems, leaves, flowers, fruits mixed with urea as external N precursors to synthesize nitrogen/oxygen doped carbon materials with layered porosity^[127].

Chemical vapor deposition (CVD): CVD differs from pyrolysis in that the process typically involves controlled exposure of the substrate to volatile precursor gases, followed by a reaction or decomposition process on the substrate surface. Kumar et al. annealed chitosan and graphene oxide simultaneously under the argon flow using CVD, they did not have any physical contact between them, chitosan pyrolysis generates volatile N-heterocycles that react with graphene oxide to synthesize NrGO^[128].

Hydrothermal carbonization (HTC): HTC can convert organic compounds into carbon materials, commonly used for the production of nanostructured carbon. The HTC reaction occurs in a water medium, usually at temperatures above 100 °C and under high pressure. Song and co-worker reported using ultra-thin tellurium nanowires (Te NWs) as a template, nitrogen-carbon hydrate compounds as carbon sources, and a template-guided HTC process to synthesize nitrogen-doped carbonaceous nanofibers (N-CNFs) aerogels on a macro scale with diameters spanning from tens to hundreds of nanometers^[129]. Biomass is also a promising precursor for the HTC method, such as lignin extracted from beech wood by alkaline hydrothermal treatment and subsequently functionalized by aromatic nitration, synthesizing microporous, mesoporous and macroporous N-doped carbon^[130]; or using biomolecular guanine and various carbohydrate compounds (glucose, fructose, and cellulose) as carbon precursors to synthesize layered nitrogen-doped porous nano carbon^[131].

Ball milling: Ball milling is a widely used large-scale mechanical technique for grinding and co-grinding, valued for its established methodology, solvent-free operation, and cost-effectiveness^[132,133]. Recently, ball milling has also been researched and applied to the structural design of carbon materials, including intrinsic and extrinsic defects (such as heteroatoms, functional groups). The mechanical energy imparted by ball milling facilitates C=C bond cleavage and particle fragmentation, simultaneously generating abundant reactive sites including lattice defects, unsaturated bonds, and carbon-free radicals. Continuous mechanical energy input enables carbon atoms at these newly formed active sites to reorganize into stable graphite lattices through cold welding mediated by C–C and C–O–C bond formation, thereby facilitating defect healing and pore structure elimination^[134]. In addition, some reaction sites will react with air (O₂, CO₂ or H₂O) or under N₂ and NH₃ atmosphere, forming oxygen-containing functional groups or N-doped carbon^[135]. Yang et al. transformed small organic molecules directly into heteroatom-doped MC (HMCs) through a simple mechanical ball milling process^[136]. A variety of heteroatoms (B, N, O, P, S, F, Cl, etc.) can be conveniently embedded in the carbon matrix and generate abundant mesoporous structure in situ.

Plasma treatment: Plasma-related technology offers an environmentally friendly and energy-efficient approach for both conventional etching procedures and advanced surface modification techniques in material engineering^[137]. Plasma, consisting of ions, free radicals, electrons, and neutral atoms, can instantaneously generate free radicals, which are efficiently used for the synthesis of doped carbon materials. Based on four initial precursors (aniline, pyrrole, benzonitrile, and nitrobenzene), amino-N, pyrrolic-N, nitrile-N, and oxide-N can be successfully prepared through a room temperature plasma synthesis process^[138]. Niu et al. synthesized nitrogen-doped graphene through solution plasma. Generated by discharge in the liquid, it has a single dielectric barrier that can reduce excessive current, thereby stabilizing the plasma and maintaining the overall temperature^[139]. This promotes the preservation of nitrogen atoms during the synthesis process without evaporation, forming a graphite carbon skeleton. Additionally, precise control over nitrogen incorporation content in plasma-treated carbon materials can be synthesized by optimizing plasma exposure time and intensity, providing good flexibility for adjusting material properties^[140].

Activity and selectivity
The arrangement of nitrogen doping is influenced by the chemical environment. N-doped graphitic structures exhibit multiple distinct configurations, primarily comprising pyridinic N, pyrrolic N, graphitic N, N oxides, and sp-hybridized N species [Figure 5A]. Different N configurations will affect the electronic properties of adjacent carbon atoms and regulate the adsorption energy on the surface of electrocatalysts, resulting in different catalytic performances. At present, research shows that the content of graphite nitrogen determines the limit current density, while the content of pyridine nitrogen increases the initial voltage of ORR^[141], and the promotion of different types of N configurations for ORR reactions is inconsistent. Unfortunately, the synthesis of N-doped carbon catalysts results in heterogeneous mixtures of multiple N configurations, making the targeted fabrication of N-doped carbon materials with a single configuration a major scientific challenge. Therefore, how to evaluate the contribution of nitrogen configuration to ORR has certain research significance. N-doped highly oriented pyrolytic graphite (HOPG) was used to prepare specific types of N configurations. Guo et al. used this method to prepare pyridine-N-based HOPG (pyri-HOPG) and graphite-N-based HOPG (graph-HOPG)^[142]. Under acidic conditions, thanks to the pairing of pyridine N with the lone pair of electrons on the carbon base plane, it can break the O–O bond, promoting ORR performance. Compared with graphite-N, the increase in pyridine N content can increase current density and initial potential [Figure 5B]. Interestingly, under alkaline conditions, graphite-N is considered to have a strong correlation with ORR performance. Wang et al. observed a strong correlation between the limiting current decay and the decrease in graphite N content, while identifying and quantitatively characterizing three different N-doping configurations. As the carbon atoms near the graphite nitrogen show smaller Gibbs free energy changes than those near the pyridine-N or pyrrole-N, graphite-N can be considered as a potential active site for the ORR reaction^[143]. Furthermore, Zhao et al. reported and synthesized sp-N-doped graphene, and believed that the incorporated sp-N atoms possess significantly higher electron density compared to pyridinic and graphitic nitrogen species, which induces substantial positive charge accumulation on adjacent carbon atoms, creating a favorable chemical environment for O₂ adsorption [Figure 5C]^[144]. Therefore, it has good application prospects in ORR.

Figure 5. (A) Schematic diagram of different types of N; (B) The ORR performance of the model catalysts was evaluated, and the corresponding quantification of the nitrogen content is shown in the inset. This figure is quoted with permission from Guo et al.^[142]; (C) Bader effective charges of N and adjacent C in N-doped graphdiyne (sp-N, pyri-N, grap-N). This figure is quoted with permission from Zhao et al.^[144]; (D) Volcano plot from Tafel and DFT-calculated ΔG_OOH* for doped graphene catalysts. This figure is quoted with permission from Jiao et al.^[145]; (E) SEM images of N, F-MCFs. This figure is quoted with permission from Gong et al.^[149]; (F) The impact of doping concentration on the electronic properties of a sulfur and pyridine nitrogen co-doped graphitic carbon system was studied for three compositions: (1,2) C:N:S = 112:8:2 (SN/C-900), (3,4) C:N:S = 120:4:1 (SN/C-1000), and (5,6) C:N:S = 91:23:3 (SN/C-400). The color coding represents the atomic species (dark brown: C, gray: N, yellow: S, pink: H). This figure is quoted with permission from Li et al.^[151]; (G) Computationally optimized structural configurations and corresponding adsorption energies of O₂ and ^*OOH intermediates on graphite N/C-P model systems. This figure is quoted with permission from Cheng et al.^[156]; (H) Selectivity of the 2e^- reaction of the coupling of pyrrolic N with carbonyl O. This figure is quoted with permission from Koh et al.^[158]. ORR: Oxygen reduction reaction; DFT: density functional theory; SEM: scanning electron microscopy.

At present, single heteroatom doping cannot show good performance. Based on this, co-doping carbon materials with nitrogen and other heteroatoms have been widely explored to modify electron distribution and structural properties, thereby improving electrocatalytic performance. Jiao et al. confirmed through experiments and DFT calculations that non-metal heteroatoms such as B, O, S, P, etc., in graphene can obtain higher valence orbit energies of active atoms and cause stronger adsorption of ^*OOH and ^*OH intermediates, whose performance is comparable to that of Pt-based catalysts [Figure 5D]^[145]. The incorporation of electronegative heteroatoms, particularly N, into carbon matrices induces localized positive charge density on neighboring carbon atoms, which facilitates enhanced oxygen adsorption and improved charge transfer kinetics, ultimately boosting ORR activity. Moreover, the opposite strategy has been reported again, where the introduction of elements with lower electronegativity (e.g., boron or phosphorus) can also be effective in facilitating oxygen molecule adsorption and the subsequent reduction process. Choi et al. increased the ORR activity of N-doped carbon by enhancing the ratio of pyridine-N with additional B/P doping; P doping enhanced the charge delocalization of carbon atoms, and constructed a morphology with many splits and wrinkles at the open edge^[146]. In addition, the S atom has a similar electronegativity to the C atom, and the individual S doping does not introduce additional unpaired electrons. However, the co-doped graphene model of S and graphite N has the highest carbon atom spin density. The ^*OOH intermediate is more inclined to adsorb carbon atoms with higher spin density in terms of energy^[147].

The electronegativity of halogen elements is relatively high; carbon material samples doped with halogens (i.e., F, Cl, Br, and I) also showed improved electrocatalytic activity; for example, both Br-, and I-doped reduced graphene oxide (RGO) had more effective ORR than undoped RGO^[148]. Notably, the high electronegativity of fluorine (χ = 4.0) makes it an efficient electron acceptor when forming ionic or semi-ionic bonds with carbon, which significantly enhances the charge transfer between the C–F bonds, thus dramatically increasing the electrical conductivity and altering the electronic properties of pristine carbon materials. The co-doping of nitrogen and fluorine atoms creates a synergistic effect that significantly boosts ORR catalytic performance. This enhancement stems from the combined influence of heteroatom doping and the distinctive nanofiber morphology featuring microporous walls, enabling N, F-MCFs to demonstrate both superior activity and remarkable stability in alkaline electrolytes [Figure 5E]^[149]. In addition, the dual doping of fluorine and sulfur on carbon fiber substrates produced synergistic effects via charge delocalization between molecules and spin state modulation, achieving catalytic performance rivaling conventional metal-based catalysts^[150]. In our previous study, we developed a facile self-sponsored doping strategy to synthesize sulfur and nitrogen co-doped graphene, demonstrating that beyond the intrinsic effects of dopant elements, the catalytic performance is significantly influenced by the doping concentration^[151]. At low doping concentrations, individual dopant sites exhibit pronounced electronic modifications, yet the overall catalytic impact remains limited due to insufficient active site density. Conversely, high doping levels increase the quantity of active centers but diminish the individual site effectiveness, ultimately compromising the overall catalytic performance [Figure 5F].

Although a large number of studies have demonstrated the reaction mechanism of N-doped carbon materials through experiments and theoretical tests, the selectivity of different configurations for 2e^- reaction or 4e^- reaction paths is still uncertain. Some studies have highlighted graphite N’s superiority over other N species in promoting the 2e^- pathway, and some studies have emphasized the excellent selectivity of pyrrole N and pyridinic N for 2e^- ORR, which may be due to the differences in the surrounding environment of the structure^[152-154]. The addition of other heteroatoms based on nitrogen doping is very important to the catalytic mechanism. The carbon between pyridine-N and thiophene-S is considered to be the active site because it exhibits more positive charge distribution, and the larger the charge delocalization of this site, the stronger the absorption of O-intermediates, resulting in the O–O bond breaking in favor of 4e^- ORR^[155]. Cheng et al. synthesized N, P co-doped hollow MC spheres (N, P-HMCS) with rich graphitic N sites and rich electron domain C–P bonds as electron donors^[156]. This configuration not only de-stabilizes the ^*OOH intermediate and desorbs H₂O₂ more easily [Figure 5G], but also induces significant distortions in the graphene sp² lattice, leading to a synergistic catalysis of 2e^- ORR. When reaction conditions are relatively mild, most defective carbon-based materials inevitably have surface oxygen groups that are one of the determinants of 2e^- ORR in N-doped carbon catalysts. Chen et al. selectively targeted C=O, C-OH, and COOH functional groups by chemical titration to distinguish the catalytic contributions of different oxygenated species, and found that the carbonyl group (C=O) is the most active site in the two-electron ORR^[157]. Moreover, it has been shown that the coupling effect of pyrrolic N and the O functional groups [including ether (O-C-O) and carbonyl (C=O)] tends to affect 2e^- ORR, and the coupling of pyrrole N and carbonyl groups favors the 2e^- ORR, with a H₂O₂ selectivity of 98.3% at 0.0 V (relative to reversible hydrogen electrode) in alkaline electrolytes [Figure 5H]^[158].

This section delves into the influence of nitrogen doping configurations and heteroatom co-doping on the activity and selectivity of carbon-based catalysts in the ORR. Various N configurations - such as pyridinic-N, pyrrolic-N, graphitic-N, oxide-N, and sp-hybridized N - affect catalyst performance in distinct manners. For example, graphitic-N is known to influence the limiting current density, whereas pyridinic-N contributes to an increased initial ORR voltage. Co-doping with additional heteroatoms, including boron (B), oxygen (O), sulfur (S), phosphorus (P), and halogens, can enhance electrocatalytic performance through synergistic interactions. The density of doping is a critical factor in determining catalyst activity and selectivity. Moreover, the selectivity for 2e^- or 4e^- ORR pathways is influenced by specific N configurations and their interactions with other functional groups.

However, during the evolution of N-doped carbon materials, researchers discovered that the observed enhancement in activity was not solely attributable to nitrogen doping. This realization has led to the exploration and development of carbon defect materials, which will be further discussed in the subsequent section.

Defective carbon catalysts

Synthesis and defect recognition
Essentially, carbon defects are a form of local disordered structure in the carbon lattice. The creation of carbon defects usually requires additional energy input to disrupt the long-range ordered carbon structure and undergo a structural reorganization process. In various carbon materials, there are many methods to create defects, such as heteroatom removal, thermal reduction of carbon, plasma etching, etc. Zhao et al. synthesized N-doped carbon and vacancy-deficient carbon by controlling different pyrolysis temperatures using N-enriched porous organic framework material (PAF-40)^[159]. Thermal treatment induces carbonization of PAF-40, accompanied by partial nitrogen atom evolution, leading to the generation of monatomic vacancy defects in the interior. There is a monotonically decreasing trend of nitrogen content with increasing calcination temperature. Samples with lower nitrogen content (corresponding to higher defect densities) exhibit superior ORR performance. After the N atom was removed by heat treatment, the reconfiguration of carbon atoms caused by multiple monatomic vacancies forms various topological defects (pentagons, heptagons, and octagons), such as 585, 75585, and 5775. Raman analysis can clearly show the D band intensity of graphene is lower than the G band indicating that the carbon structure is highly regular and has relatively few defects [Figure 6A]. The removal of nitrogen-doped atoms leads to an increase in defects, and the increase in the ratio of ID/IG indicates a wider range of defect domains. And then they proposed that Zn-MOF was carbonized at 950 °C to synthesize a porous carbon material containing only carbon and oxygen, which showed excellent ORR activity despite the absence of any element doping, mainly due to the effect of defects arising from the removal of Zn atoms during the calcination process^[160]. Tao et al. utilized the pristine surface of HOPG as the starting substrate and ideal model, successfully regulating defect levels via an efficient plasma etching strategy^[161]. Through surface charge and potential analysis, it was found that the surface charge of HOPG increases with the increase of defects. Following Ar plasma etching, defect sites emerge on the HOPG surface. The partial removal of carbon atoms leads to an abundance of defects on the surface, which also becomes increasingly rough as the plasma irradiation time is extended. When the irradiation time exceeds 5 min, obvious nano-cones are produced on the HOPG surface. Post-etching, the I(D)/I(D′) remains approximately 2.4, suggesting that the defects produced by plasma irradiation are predominantly vacancy-type defects. Additionally, Kelvin probe force microscopy [Figure 6B] reveals a domain-like structure with significant potential differences on the P-HOPG surface. The potential differences primarily arise from the presence of defective graphene (DG) layers that are well-linked and coupled along the c-axis direction, thereby forming conductive regions, as opposed to the less conductive perfect graphene layers that exhibit minimal coupling and retain a more two-dimensional character.

Figure 6. (A) Raman spectra of different types of catalysts. This figure is quoted with permission from Zhang et al.^[159]; (B) KPFM test of HOPG before and after plasma etching. This figure is quoted with permission from Tao et al.^[161]; (C) Synthetic scheme for the preparation of a D-HOPG sample; (D) The HAADF-STEM images derived from D-G [enlarged image of the dashed box in (D)]. This figure is quoted with permission from Jia et al.^[162]; (E) Synthesis method of N-doped type carbon materials; (F-H) C K-edge XANES of different types CNF and CM. This figure is quoted with permission from Wang et al.^[163]; (I) Schematic design and fabrication of HDPC; (J) EPR spectrum of LDPC and HDPC. This figure is quoted with permission from Wu et al.^[165]. KPFM: Kelvin probe force microscopy; HOPG: highly oriented pyrolytic graphite; HAADF-STEM: high-angle annular dark field - scanning transmission electron microscopy; XANES: X-ray absorption near-edge structure; CNF: carbon nanofiber; CM: carbon monolith; EPR: electron paramagnetic resonance; LDPC: low density porous carbon; HDPC: high density porous carbon.

However, how to selectively synthesize a single type of carbon defect and evaluate its impact on ORR activity is a meaningful direction of research. Jia et al. first etched the original HOPG with a perfect graphite-carbon structure by argon plasma and formed a uniform concave through a nickel mesh with a square window [Figure 6C]^[162]. After removing the residual nickel from the surface, it is then annealed in an ammonia stream at 700 °C for 3 h. Thus, a specific N-doping (Pr-N) is formed at the edge of the N-HOPG groove. Finally, N-HOPG was annealed at 1,150 °C N₂ for 2 h to remove the N atom and D-HOPG was obtained. Removing one Pr-N atom forms a pentagon on the edge of the graphene, and removing two adjacent Pr-N atoms confirms by calculation that the type of defect on the D-HOPG edge is most likely to be a reconstructed pentagon carbon defect. The defect areas of the materials can be visualized in practice through aberration-corrected high-resolution transmission electron microscopy (AC-HRTEM) [Figure 6D]. Structural defects primarily accumulate at pore boundaries, whereas the intrinsic hexagonal graphene configuration is predominantly preserved in the bulk regions of the material. To explore controlled synthesis of other configurations, edge engineering methods are available through one-to-one transformations of confirmative N configurations, such as graphite-N leads to C585, pyridine-N leads to S-C5, and pyrrole N leads to A-C5 [Figure 6E]^[163]. High-temperature pyrolysis converted zinc-free phenol-formaldehyde nanofiber (PFN) precursors into edge-deficient CNFs with desirable carbon network structures, while PFNS materials synthesized with a small amount of zinc ion doping yielded porous CNFs (P-CNFs) with abundant edge structures and hexagonal lattice configurations. CNF with graphite N (GN-CNF) and CNF with pyridine N (PDN-CNF) can be obtained by subsequent N doping process. The samples dominated by pyrrorole-N were prepared by the same N-doped phenol-formaldehyde monolith (PFM) containing a large number of Zn species prepared by the hypersaline polymerization method. For accurate detection of accurately detect the structural evolution of carbon materials, the characteristic peak at ~285.5 eV in the normalized C K-edge X-ray absorption near-edge structure (XANES) spectra (corresponding to the C 1s → π^* transition) can effectively reflect the structural integrity of the sp² hybridized carbon network [Figure 6F-H]. In GN-CNF, the dominant graphite-N is located in the center of the carbon structure, leading to severe destruction of graphite sp² bonds. In contrast, pyridine-N and pyrrole-N are always located at the edge, with little effect on the internal π^* conjugation. After removing N atoms in GN-CNF, the intensity of the π^* peak increases, but it is still less than the original CNF; for PND-CNF and PON-CM, by subtracting the edge N dopants (pyridine-N and pyridine-N), the intensity of the π^* peak decreases, indicating that edge reconstruction may occur, leading to the formation of additional non-hexagonal defects, which slightly disrupt the over π^* conjugation network.

Creating high-density defect sites in carbon materials is essential for further enhancing the activity of metal-free catalysts. A significant challenge remains in effectively introducing nitrogen heteroatoms to create numerous catalytically active pentagons following denitrification. It is therefore essential to investigate the key factors influencing the nitrogen doping process and to elucidate their correlation with the catalytic performance of carbon materials post-denitrification. Li et al. employed DFT calculations to reveal that native in-plane pentagonal defects in carbon materials alter the preferential nitrogen doping sites from edge hexagons to in-plane pentagons^[164]. As the number of pentagons increases, the formation energy for nitrogen heteroatoms decreases progressively. Furthermore, post-denitrification, the nitrogen-containing in-plane pentagons transform into thermodynamically stable pentagon-heptagon-pentagon secondary defects, thereby increasing the density of pentagonal active sites. Inspired by these findings, they demonstrated that carbon materials derived from treated C60 exhibited excellent performance in alkaline oxygen reduction and zinc-air batteries, attributed to the abundance of secondary pentagonal defects, thus validating the theoretical predictions.

In contrast to traditional thermal decomposition and plasma etching methods, Wu et al. propose an interfacial self-etching method strategy by etching carbon bases with ZnO quantum dots [Figure 6I]. In this process, CO₂ gas is generated inside the closed carbon cavity, which in turn for ultra-high density carbon defects^[165]. The electron paramagnetic resonance (EPR) spectrum has confirmed the results of the Raman spectroscopy [Figure 6J]. The hyperdense defects in porous carbon (HDPC) formed through the interfacial self-corrosive strategy have stronger peak signals at the magnetic field value of 2.003G and higher I_D/I_G values, indicating that HDPC has a higher defect density in the bulk phase.

Despite the rapid development of defect characterization techniques, they are still insufficient in support of more fundamental and systematic research on the defects of electrocatalysts in the future. Therefore, it is necessary to develop more advanced characterization methods to study defects more accurately on time and spatial scales. Moreover, under harsh conditions (strong acid/base, high current density, long-term operation), the defect structure may be unstable and may undergo reconstruction or even oxidation. Therefore, it is necessary to develop in-situ characterization techniques for monitoring the dynamic evolution of defect structures during reaction processes and the actual interaction of reaction intermediates. For example, in-situ Raman spectroscopy technique can analyze intermediate products during the reaction process based on chemical structure, providing valuable information for exploring reaction mechanism. In-situ X-ray absorption spectroscopy (XAS) can reflect the chemical valence state and electronic structure of the measured elements during the reaction process well. By fitting the data, we can get important information such as the real spatial distribution of atoms, bonding conditions and coordination environment; In-situ XPS technology can analyze the valence changes and coordination environment of specific elements in real time accurately. However, the application of in-situ characterization techniques is often filled with challenges in future applications due to harsh operating environments or limited information obtained.

Cognition and design of active sites
Indeed, different types of carbon defects can effectively enhance the catalytic properties of materials. However, there is currently less research on the synthesis of single defects, so in order to deeply understand the role of single defects in ORR activity, it is particularly necessary to verify their importance with the help of theoretical calculations. Currently, exploring the impact of topological carbon defects, usually composed of one or more non-hexagonal rings, on catalytic activity through theoretical calculations [e.g., pentagon (C5), pentagon-octagon-pentagon (C5-8-5), pentagon-heptagon-heptagon-pentagon (C5-7-7-5)] can help develop catalysts with higher activity. Theoretical calculations show that the addition of nitrogen hinders the reduction of chemisorbed oxygen (O^*), and that a large energy barrier of 1.03 eV is required for the rate-determining O^* → OH^* transition. The G585 defect facilitated the O₂ → OOH^* transition, reducing the Gibbs free energy change (ΔG) from 0.88 to 0.41 eV for pristine graphene, and all basic steps on G585 exhibited thermodynamic advantages approaching the performance of an ideal catalyst [Figure 7A]^[159]. It is found that different topological defect configurations show unique catalytic advantages in various electrochemical processes, e.g., the C5-1 defect shows excellent performance in ORR and OER reactions, while the C7-5-5-7 configuration performs better in HER [Figure 7B-E]^[166]. The HOMO and LUMO are mainly distributed on the edge atoms of graphene holes, and the introduction of C5-1, 585, and 7557 defects will produce HOMO/LUMO orbitals around the edge atoms of the defects, thus improving the catalytic activity.

Figure 7. (A) Schematic diagram of G585 defects in graphene; (B) Computationally derived free energy profiles for ORR at equilibrium potential, comparing pristine monolayer graphene (G), NG, G585-defected graphene (G585), and an ideal catalyst (Ideal). This figure is quoted with permission from Zhao et al.^[159]. Comparative energy maps of defect configurations of ORR, OER, and HER reaction pathways on DG under alkaline and acidic conditions: (C) edge pentagon, (D) 5-8-5 defect, (E) 7-55-7; This figure is quoted with permission from Jia et al.^[166]; (F) Charge densities and oxygen adsorption energies of C6 and C5. This figure is quoted with permission from Zhu et al.^[168]; (G) Theoretical models, and (H) the free energy diagram of typical ORR pathway of C4, C5 and C6. This figure is quoted with permission from Srinivas et al.^[172]; (I) The electron density corresponding to adsorbed ^*OOH on different sites. This figure is quoted with permission from Wu et al.^[165]. ORR: Oxygen reduction reaction; NC: nitrogen-doped graphene; OER: oxygen evolution reaction; HER: hydrogen evolution reaction; DG: defective graphene.

It is reported that carbon atoms with higher charge density are more likely to be active sites^[167]. The HOMO of C5 has greater electron donor potential and better reactivity. The reduced HOMO-LUMO energy gap observed in C5 configurations, compared to C6 structures, facilitates more efficient electron transfer during electrochemical processes^[168]. In addition, local electron redistribution results in a large difference in charge density between C6 and C5. When they are at the edge, the average charge density of C5 (-0.190) is still greater than that of C6 (-0.115) [Figure 7F]. Moreover, the adsorption energy of oxygen at the C6 and C5 sites shows that C6 has weaker adsorption capacity for oxygen molecules (0.40 eV), while C5 has much stronger adsorption capacity for oxygen (5.95 eV). Therefore, intrinsic pentagonal carbon has good chemical activity. When C5 and C7 defects are adjacently configured to form a curved C5+7 structure, the catalytic overpotential is significantly reduced. Oxygen species preferentially adsorb at the five-seven ring junction, where the inherent C5-C7 dipole effectively weakens the O–O bond in OOH intermediates through bond elongation, thereby facilitating oxygen reduction^[169]. The generation and desorption of H₂O₂ in C57 are both thermodynamically uphill processes, with a total free energy change of 0.32 eV^[170]. Conversely, the O–O cleavage pathway from ^*OOH to ^*O + OH^- is exothermic, thus highly favorable. Similarly, the formation and desorption of H₂O₂ on the C7557 active site are also unfavorable. Therefore, at the pentagon/heptagon defect sites, the four-electron reduction to H₂O is the preferred ORR pathway.

Besides pentagonal carbon defects, edge defects are also good ORR active sites. The different electronic configurations between the sawtooth-edge and armchair-edge structures induce different reaction patterns to the ORR intermediates, ultimately resulting in significantly different ORR performance characteristics. In sawtooth edge configurations, localized unpaired π electrons reside on individual edge carbon atoms, whereas armchair edge structures exhibit paired π electrons between adjacent carbon atoms, forming stable covalent bonding interactions. Jiang et al. found that the unpaired π electrons of the sawtooth edge are conducive to electron transfer to O₂ and easy to form ^*OOH, with a free energy change of -0.222 eV^[171]. In contrast, the armchair edge lacking unpaired π electrons must rearrange adjacent bonds to transfer an electron to O₂ to form ^*OOH, thereby increasing the free energy by 1.089 eV. Comparative analysis of the vacancy configurations shows that the armchair pentagonal defect (C5-site) exhibits a pathway of reduced free energy required for the ORR process compared to the zigzag hexagonal (C4-site) and armchair hexagonal (C6-site) vacancy structures [Figure 7G]^[172]. From a thermodynamic perspective, in addition to zigzag edge carbon, the C5 site is also an actual active site for the ORR [Figure 7H]. Xia et al. proposed that the synergistic interaction between carbon vacancy defects and neighboring pentagonal structures enhances molecular oxygen adsorption and boosts catalytic performance^[173]. Moreover, vacancies cause the redistribution of electrons in adjacent pentagons, thereby forming more conjugated π electrons. When conjugated π electrons are enriched near the Fermi surface, the ^*OH is easier to desorb due to the weakened binding strength of ^*OH. Wu et al. theoretically established single pentagon defects (SC5), two pentagon defects with three-hexagon (C-56665), two-hexagon (C-5665), and one-hexagon (C-565) intervals, representing reduced spatial distances (1.2-0.5 nm) between defects^[165]. The results showed that the excellent electrocatalytic performance comes from the “proximity effect” generated by high-density carbon defects between carbon defect sites at different spatial distances. The smaller the spacing between pentagonal defects, the higher the number of transferred charges, reflecting the stronger interactions between defects and thus increasing ORR performance [Figure 7I]. In-plane defect models impose limitations on the target atomic structure. In contrast to planar configurations, local 3D curved structures induce more strained bonds due to their pronounced geometric curvature. These distinctive strain effects can alter the electronic structure of active sites and the adsorption-desorption dynamics of reaction intermediates, potentially enhancing catalytic efficiency. Wang et al. identified a correlation between the curvature of carbon defects and their activity in the ORR, highlighting the advantages of highly curved topological defects in carbon^[174]. By augmenting the curvature of carbon materials, it is possible to modulate the electronic structure of edge pentagon defects, optimize the adsorption-desorption process of reaction intermediates, and lower the energy barrier for ORR.

Complex defects catalysts

A key advantage of carbon defect engineering is its extreme compatibility with a wide range of heteroatomic species, allowing the incorporation of non-metallic heteroatoms (N, P, S, and O) at the defect site. When the defect is combined with the dopant, the obtained catalyst usually exhibits the highest ORR activity, demonstrating a synergistic enhancement of electrocatalytic activity through combined defect-dopant interactions in nanocarbon materials. Zhang et al. proposed several possible combination models: pyridinic-N (PN6) + pentagon (EC5) carbon rings, PN6 + heptagon (EC7) carbon rings, pyrrolic-N (PN5) + EC5 and PN5 + EC7. The PN6 + EC5 composite exhibited the most favorable ^*OH adsorption characteristics, with a binding energy of 0.88 eV, suggesting significant potential for optimizing ORR performance [Figure 8A and B]^[175]. The improved catalytic performance in the synergistic region may stem from the high electronegativity of the PN6 dopant and induced redistribution of charge within the pentagonal carbon ring.

Figure 8. (A) Schematic representation of graphene at various potential active sites; (B) corresponding ^*OH adsorption energies for different reaction centers. This figure is quoted with permission from Zhang et al.^[175]; (C) Schematic N-doped sites in graphene. This figure is quoted with permission from Fernandez-Escamilla et al.^[176]; (D) Correlative analysis between experimentally determined half-wave potentials and computationally derived overpotentials for various samples under acidic conditions. This figure is quoted with permission from Yan et al.^[177]; (E) The optimized structures of the N-S-D-G model; the electron density corresponding to N-S-D-G; the blue and red regions indicate electron-donating and electron-withdrawing areas, respectively. This figure is quoted with permission from Li et al.^[178]; (F and G) The atomic structures of the examined O-groups on edge and pentagon defect sites; (H) The calculated activity volcano map This figure is quoted with permission from Wu et al.^[179].

Charge redistribution resulting from defect-dopant interactions enhances binding affinity toward key reaction intermediates, consequently increasing adsorption energetics. This enhancement boosts the performance of ORR and alters the mechanism of ORR. The introduction of pyridine nitrogen into the basal plane of graphitic carbon substrates produces associated vacancy defects. Fernandez-Escamilla et al. employed a combined DFT and experimental approach to examine single vacancy (SV) configurations with various pyridinic nitrogen doping arrangements [Figure 8C]^[176]. The results showed that the ORR on graphite N and pyridine (SV + 3N) CNTs, which lack topological defects or dangling bonds, proceeds via a two-electron pathway through a physical adsorption mechanism. Yan et al. used computational simulations to explore the synergistic role of a variety of heteroatoms and carbon defects (C5) in ORR catalysis [Figure 8D]^[177]. The band gap of C5 co-modified by N and S can be optimized to achieve the highest ORR activity. With an electronegativity of 3.04 eV, nitrogen atoms exhibit superior electron-withdrawing capability compared to S and C atoms, creating localized regions of high electron density [Figure 8E]^[178]. In addition, the doping of N atoms can change the degree of π-conjugation and magnetic properties of neighboring atoms. The doping of electronegative N in S-D-G induces a significant charge redistribution, leading to a decrease in the band gap and a shift of the valence band towards the Fermi level of the N-S-D-G model. Also based on the C5 defect, Wu et al. modified carbonyl pentagonal defects (C5=O) and showed excellent H₂O₂ production efficiency [Figure 8F-H]^[179]. The work underscores the significance of atomic-level engineering in catalyst development, illustrating how seemingly minor structural modifications can lead to substantial improvements in performance. Our research team has demonstrated the critical role of carbon defects in ORR catalysis. Through controlled synthesis and precise regulation of defect density, we have provided compelling evidence for a more profound understanding of the catalytic mechanisms associated with carbon defects.

The synergistic interaction between defects and dopants presents a promising avenue for further catalyst optimization [Table 1], potentially leading to materials that outperform traditional precious metal catalysts. When carbon defects are combined with heteroatom doping, two primary benefits arise. Firstly, carbon defects create additional active sites for the incorporation of heteroatoms, facilitating a more uniform distribution throughout the carbon matrix. Secondly, the presence of heteroatoms modulates the electronic structure and chemical properties of carbon defects, significantly enhancing their catalytic activity.