Research advances in earth-abundant-element-based electrocatalysts for oxygen evolution reaction and oxygen reduction reaction

Metal-free C-based electrocatalysts

The expenditure of OER electrocatalysts is the first consideration for practical applications. C-based nanomaterials are generally easily available and can be used as OER electrocatalysts at a low cost and earth abundance, thereby significantly decreasing economic consumption. Furthermore, C-based nanomaterials possess high electrical conductivity, flexibility, and durability^[49]. However, pristine C materials, such as graphite and activated C (AC), are not promising candidates for OER due to their relatively low catalytic activity^[50,51]. Nevertheless, recent studies have demonstrated that the OER activity of C-based nanomaterials can be enhanced by modifying them with functional groups or heteroatoms, which can improve the adsorption of OER intermediates^[52,53]. Zhang et al. developed a straightforward method for producing carboxyl-rich multiwalled C nanotubes (COOH-MWCNTs) with high stability and OER electrocatalytic activity in acidic solutions^[54]. These COOH-MWCNTs exhibited an overpotential of 265 mV at a current density of 10 mA·cm^-2 and a Tafel slope of 82 mV·dec^-1, representing the best OER performance of metal-free electrocatalysts and a performance comparable to that of RuO₂ and IrO₂. The catalytic mechanism involves the loss of an electron from the COOH-MWCNT electrocatalyst at the beginning of the OER process, which triggers the oxidation of H₂O molecules via OH formation. An electron is then recaptured from the H₂O molecule to form oxidized H₂O and restore the initial state of the COOH-MWCNT electrocatalyst. This new OER mechanism provides insights into the critical roles of organic functional groups in electrocatalytic processes. Additionally, studies have demonstrated that doping C substrates with N, P, S, O, or B could optimize their electronic structure and further facilitate electrocatalytically active sites with excellent OER performance^[52,55-57]. For example, Hu et al. prepared ultrafine pentlandite NPs fixed by a porous N- and S-doped C network (P-NSC/Ni₄Fe₅S₈-1000)^[58]. Owing to the synergy between pyridinic N (p-N), S-doped C, and highly active Fe sites, P-NSC/Ni₄Fe₅S₈-1000 has emerged as an efficient and stable OER electrocatalyst under acidic conditions [Figure 4]. Kang et al. prepared ultrathin B-N codoped porous nanocarbon (BN-PCN) sheets using a carbonization-etching method^[59]. Owing to its unique structural characteristics, BN-PCN exhibits high electrocatalytic OER activity. In particular, Ma et al. developed a descriptor based on the p_z orbital of the active site to explain the catalytic performance of bi and tridoped graphene electrocatalysts for ORR and OER^[52]. This descriptor is universal and can account for multiple doping and predict the contributions of defects, edges, and defect-edge combinations. Ali et al. synthesized MWCNTs for OER in alkaline media by growing graphene in situ on Ni-silica nanocomposites at 650 °C using chemical vapor deposition^[60]. After removing the silica, the resulting nanomaterial exhibited an OER performance comparable to that of commercial Ir-supported C. This is due to the defective nature of MWCNTs that contain surface discontinuities on the rolled graphene layers of C nanotubes (CNTs) and the MWCNT-graphene interface, which may contain dangling bonds that function as active sites for the OER. The MWCNT-graphene hybrid C nanostructure material gives an overpotential of 310 mV at a current density of 10 mA·cm^-2 in a 1 M KOH solution at a scan rate of 5 mV·s^-1, whereas the commercial Ir/C exhibits an overpotential of 305 mV under similar conditions. Although this presents a new concept for modifying C-based nanomaterials by transforming catalytically inert materials into active electrocatalysts, the performance of C-based nanomaterials lags behind that of state-of-the-art electrocatalysts^[61]. Additionally, low stability is another disadvantage of C-based OER electrocatalysts, as studies have shown that they can undergo surface oxidative corrosion at high potentials and generate C dioxide, significantly decreasing their catalytic activity^[61,62].

Figure 4. (A) Schematic illustration of the synthesis of P-NSC/Ni₄Fe₅S₈-1000; (B) TEM images of P-NSC/Ni₄Fe₅S₈-1000: (B-1) 500 nm and (B-2) 100 nm; (C) Steady-state polarization curves and (D) the corresponding Tafel plots of NSC/Ni₄Fe₅S₈-1000 (without porous structure) and P-NSC/Ni₄Fe₅S₈-1000 (with porous structure) and commercial RuO₂ (Alfa Aesar). Reproduced with permission. Copyright 2019, Royal Society of Chemistry^[58].

Transition-metal-compound-based electrocatalysts

Similar to C-based nanomaterials, earth-abundant first-row-transition-metal oxides/hydroxides/oxyhydroxides exhibit high electrical conductivity and can be effectively enhanced in OER electrocatalytic activity through various strategies, such as heterostructure construction, alloying, elemental doping, coating, defect engineering, exfoliation, and hybridization^[63,64].

Kang et al. synthesized a heterostructure composed of Ni(OH)₂ nanosheets and tungsten (W) NPs on a conductive C cloth (W@Ni(OH)₂/CC) to address the problem of low catalytic activity^[65]. This heterostructure reflects the advantages of the metallic properties of the catalyst and reduces the length of the charge transfer path. Co phosphide has also been shown to be an effective OER catalyst. For example, Liang et al. constructed a CoP/TiO_x heterostructure using metal-organic framework (MOF) nanocrystals as templates, which significantly improved the OER performance^[66]. CoP/TiO_x has a unique hollow structure, with CoP NPs evenly distributed on TiO_x. The strong interaction between CoP and TiO_x in the heterostructure and the conductivity of TiO_x due to Ti³⁺ sites endow CoP-TiO_x hybrid materials with high OER activity, comparable to that of IrO₂ or RuO₂ electrocatalysts. The CoP/TiO_x heterostructure provides a path that facilitates electron transport and optimizes the H₂O adsorption energy, thereby enhancing the OER performance. In addition to Co phosphide, multishell mixed-metal P oxides have been reported to enhance the electrocatalytic activity of the OER, as shown in Figure 5^[67]. The results showed that activated manganese (Mn)-CoP oxides have excellent OER properties, with a Tafel slope of 52 mV∙dec^-1, which is the lowest compared to Mn-Co oxide (60 mV∙dec^-1) and RuO₂ particles. To achieve a current density of 10 mA∙cm^-2, the overpotentials required for the activated Mn-CoP oxide, Mn-Co oxide, and RuO₂ particles are 370, 420, and 366 mV, respectively.

Figure 5. (A) Formation of multishell mixed-metal oxyphosphide particle; (B) Characterization of the multishell Mn-Co oxyphosphide particles: (B-1) FESEM and (B-2) TEM images; (C) LSV curves and (D) Tafel plots of activated Mn-Co oxyphosphide particles, Mn-Co oxyphosphide particles, and Mn-Co oxide particles; (E) Chronoamperometry curves of activated Mn-Co oxyphosphide particles and Mn-Co oxide particles at overpotentials of 320 mV and 420 mV, respectively. Reproduced with permission. Copyright 2017, Wiley Online Library^[67].

Alloying is a viable method to adjust the catalytic performance of single-metal electrocatalysts. Ni-Fe alloys have been extensively researched as electrocatalysts for the OER, and the significance of Fe doping has been demonstrated through proportional control^[68]. Currently, Ni-Fe-based alloys are the most common OER electrocatalysts in the industry^[69,70]. Zhou et al. reported that Fe leaching could induce the electrochemical activation of strongly coupled Ni-Fe alloys on N-doped C composites, resulting in an enhanced OER performance^[71]. The leaching of the unstable FeOOH mixed layer from the FeNi₃ alloy surface leads to surface reconstruction and exposure of additional active sites. N-doped C, FeNi₃ alloy, and concomitant Fe-doped Ni(OH)₂ synergistically promote OER activity after full activation. High-valency Ni, p-N, and metallic N species were determined as the exact active sites, and the fully activated Fe_0.2Ni_0.8/NC-600 electrocatalyst is one of the best non-noble-metal OER electrocatalysts to date. It exhibits excellent OER performance with a low overpotential (η₂₀ = 290 mV) and a small Tafel slope (76 mV·dec^-1). This study contributes to the understanding of the fundamentals of Ni-Fe alloy electrocatalysts and provides guidance for designing advanced electrocatalysts for H₂O oxidation via in situ electrochemical activation. Jing et al. adjusted the electronic structure of CN-FeO_x through hydroxyl modification to enhance the binding capacity of O-intermediates in the OER process^[72]. Researchers found that increasing the Fe-Fe length in the material can improve the structural stability and activity of OER. Performance evaluation shows that the material has a low overpotential at 10 mA·cm^-2, close to commercial RuO₂.

Elemental doping has been shown to effectively enhance the performance of transition-metal-based compounds by tuning their electronic structure^[73-75]. Therefore, many efforts have been devoted to optimizing the elemental doping species and concentrations and identifying the electrocatalytically active sites. For example, trace amounts of Fe-doped Ni hydroxide in Fe-doped Ni compounds could significantly enhance the OER electrocatalytic activity^[47]. Liu et al. reported an amorphous NiFe-layered double hydroxide (a-NiFe-LDH) with abundant active sites that could expose additional NiFe active sites, as shown in Figure 6^[75]. The OER electrocatalytic activity of a-NiFe-LDH is much higher than that of RuO₂ or pristine NiFe-LDH. Furthermore, a-NiFe-LDH has a low OER overpotential of 241 mV at 10 mV·cm^-2 and a small Tafel slope of 55 mV·dec^-1.

Figure 6. (A) The schematic illustration of the preparation process for P-NiFe-LDH and A-NiFe-LDH; (B) FESEM images of (B-1) A-NiFe-LDH, (B-2) P-NiFe-LDH, and (B-3) M-NiFe-LDH; (C) LSV plots of P-NiFe-LDH, A-NiFe-LDH, M-NiFe-LDH, and RuO₂; (D) The corresponding overpotential at a current density of 10 mA·cm^-2; (E) The corresponding Tafel plots. Reproduced with permission. Copyright 2020, Elsevier^[75].

In addition to elemental doping, several new and advantageous strategies have been developed over the past few years to enhance the catalytic activity of transition-metal-based electrocatalysts. Zhang et al. proposed that gelled FeCoW oxyhydroxide is a ternary metal oxyhydroxide with better OER performance than Ni-Fe oxyhydroxide^[76]. A simple, fast, and scalable microwave method was reported for the first time, and a series of ultrasmall transition-metal selenide NPs (~5 nm in diameter) were supported on the surface of CNTs^[77]. By benchmarking the OER activity, Ni_0.27Co_0.28Fe_0.30Se@CNT was determined as the electrocatalyst with an optimized ratio and composition, exhibiting excellent OER performance and achieving a current density of 100 mA·cm^-2 at an overpotential of 291 mV, which is 48 mV lower than that of commercial RuO₂. Furthermore, Ni_0.27Co_0.28Fe_0.30Se@CNT exhibits long-term stability of up to 50 h. Methods such as coating, defect engineering, exfoliation, and hybridization have also been explored for enhancing the catalytic activity of transition-metal-based compounds. Wu et al. utilized the MOF precursor approach and polymer coating/encapsulation strategy to incorporate FeNi alloys into N-doped C with customized structures^[78]. Furthermore, defect engineering has been introduced recently to enhance the catalytic performance of Co₃O₄ nanosheets by generating active sites and increasing electrical conductivity^[79]. Ultrathin Ni-Co layered double hydroxide (LDH) nanosheets exhibit high OER catalytic activity owing to the presence of numerous exposed active sites^[80]. These results suggest that complex engineered electrocatalysts based on transition-metal-based compounds can significantly overcome the limitations of existing electrocatalysts. However, the unstable OER potentials of transition metals under alkaline conditions should be considered, as they can lead to their conversion into hydroxides/oxyhydroxides. Therefore, the oxidized surface metals should be considered as the actual OER active sites of transition-metal electrocatalysts.

Metal-organic framework (MOF) electrocatalysts

MOFs are a promising porous material for electrocatalysis in the OER due to their high specific surface area. The evenly distributed metal nodes in the MOF framework can potentially act as active sites for electrocatalysis. However, the presence of organic ligands in MOFs can hinder the transport of electrons during the reaction process, leading to a loss of catalytic activity^[81].

MOF-derived OER electrocatalysts include N-donor MOFs, carboxylate MOFs, and Prussian blue analogs (PBAs). N-donor MOFs, such as zeolitic imidazolate frameworks (ZIFs), consist of tetrahedral M(II) nodes and imidazolate-based ligands. Ding et al. synthesized α-Co(OH)₂-ZIF-L/NF-y by in situ surface reconstitution treatment of CoM-ZIF-L precursors, which could elevate the valence of the Co cations and promote the formation of highly active amorphous metal hydroxides^[82]. The optimized α-Co(OH)₂-ZIF-L/NF-40 exhibited higher catalytic activity than commercial RuO₂ and many other OER catalysts. Surface reconstruction could also occur on bimetallic precursors. Lei et al. studied the morphology and crystalline structure reconstruction of M-ZIF-67 and produced the CoOOH-Vo catalyst, which was driven by an electric field^[83]. The obtained M-ZIF-67-EA delivered low overpotentials of 175 mV at 10 mA·cm^-2, which could be attributed to the leaching of partial ligands and the generation of CoOOH species with oxygen vacancies.

Carboxylate MOFs have versatile coordination modes and stable frameworks, making it easier to adjust their surface reconstitution process. Huang et al. produced CoNi-MOF nanosheet arrays (CoNi-MOFNA) via an in situ self-dissociation-assembly (SDA) synthetic strategy^[84]. The microstructure of such MOFs was 2D nanosheets with a thickness of 8.2 nm, and during the long-term OER process, the surface of the CoNi-MOFNA was partially reconstructed. Two coordinatively unsaturated sites of Co and Ni played a significant role in catalysis. Similar excellent performance can be obtained on other carboxylate MOFs through structural transformation. For instance, Ni_0.9Fe_0.1-MOF achieved a low overpotential of 198 mV at 10 mA·cm^-2[85].

PBAs are cyanide-based coordination framework materials. The introduction of CN vacancy into PBA through ionized N plasma by Yu et al. enhanced the OER catalytic activity^[86]. The corresponding reconstruction improved the OER performance of PBA catalysts, accompanied by the formation of oxyhydroxides or hydroxides and the increase of the valence of metal ions^[87,88]. Therefore, the reaction activity of MOF catalysts can be optimized through crystal structure reorganization and the optimization of organic ligands.

Noble-metal-based ORR electrocatalysts

The ORR is a critical process in advanced fuel cell technologies and plays a determining role in their overall yield and commercial applicability. However, the ORR exhibits sluggish kinetics due to complex processes such as O₂ adsorption on the catalyst surface, activation/cleavage of O-O bonds, and product removal^[91]. Therefore, high-efficiency electrocatalysts are required to meet the stringent requirements of ORR applications, and developing such catalysts remains a significant challenge. During the ORR process, the transfer of O₂ and accumulation of liquid products severely hinder the active sites^[92]. To date, researchers have conducted precise design and elaborate synthesis of Pt-based electrocatalysts for ORR, which are identified as high-efficiency ORR electrocatalysts^[93]. Three research hotspots for Pt-based electrocatalysts have been identified, including (1) reducing the size of Pt NPs, increasing their dispersion and specific surface area^[94]; (2) exploring the preparation of Pt NPs with specifically oriented crystal planes; and (3) adding other metallic elements to Pt catalysts through various physical and chemical methods to alloy or disperse Pt into alloys with other transition metals or oxides, respectively, thereby forming mixtures or alloys at a specific scale, such as core-shell structures^[95]. Among these strategies, the precise control of nanoscale Pt structures and the tuning of morphology and composition are considered the most effective approaches for substantially improving the ORR performance and reducing the amount of expensive Pt in fuel cells^[94,96].

Huang et al. prepared a core (Pd)/shell (Pt₃Ni(Pt-skin)) electrocatalyst using Zn underpotential deposition (UPD) onto a Ni UPD-modified Pd/C catalyst, as depicted in Figure 7^[97]. The Pt₃Ni(Pt-skin) structure retains the advantages of the ultrathin layer structure and the synergy of the Ni sublayer. The synthesis process involved first electrodeposition of Ni_u onto the Pd surface of Pd20/C, followed by modification of Znu on Ni_u/Pd₂₀/C. Subsequently, Zn_u was replaced by Pt via the Galvanic process, resulting in a Pt atomic-layer cover on the Ni_u/Pd₂₀/C (PtNi/Pd₂₀/C). Repetitive Zn UPD-Galvanic deposition introduced the second and third Pt atomic layers onto PtNi/Pd₂₀/C, yielding the desired Pt₃Ni(Pt-skin)/Pd₂₀/C electrocatalyst. This electrocatalyst exhibits the best ORR performance, with a Pt specific activity of 16.7 mA·cm^-2 and a Pt mass activity of 14.2 A mg Pt^-1. These values are 90 and 156 times higher, respectively than those of commercial Pt/C electrocatalysts.

Figure 7. (A) The representative HRTEM image of a core (Pd)/shell (Pt₃Ni(Pt-skin)) nanoparticle; (B) The preparation process of Pt₃Ni(Pt-skin)/Pd₂₀/C; (C) ORR polarization curves; (D) Cyclic voltammograms. Reproduced with permission. Copyright 2018, Royal Society of Chemistry^[97].

Porous structures are efficient for electrocatalysis due to the presence of numerous exposed active sites on their surfaces and their smooth mass transfer capability. Various methods, including hard templating, dealloying, and surfactant-assisted synthesis, have been employed for synthesizing porous noble-metal electrocatalysts. Mesoporous silica has been extensively used as a hard template for preparing noble-metal mesoporous materials due to its low cost, highly ordered structure, and capability of facile silica removal via efficient etching processes. Chen et al. developed a new approach for synthesizing single-crystal-like MCM-48 nanospheres (NSs) for the templated preparation of mesoporous Pt NSs by molten salt impregnation, followed by H₂ reduction^[98]. The synthesized mesoporous Pt catalysts exhibit higher activity than commercial Pt/C electrocatalysts due to their open pore structure. Dealloying is another efficient technique for preparing porous Pt nanostructures, wherein non-Pt components are selectively dissolved from the alloy by chemical or electrochemical methods. Leaching of non-Pt atoms from the surface of Pt-based alloys results in large mesoporous structures, defects, and vacancies. For instance, Li et al. observed that Pt/NiO core-shell nanowires synthesized in the solution can be converted into Pt-Ni alloy nanowires through thermal annealing^[99]. These alloy nanowires can then be transformed into jagged Pt nanowires by electrochemical dealloying. The jagged nanowires have highly stressed, undercoordinated, and rhombus-rich surface configurations that improve ORR activity more effectively than relaxed surface configurations.

Soft templating is considered a flexible and efficient method for preparing porous nanomaterials. Attard et al. developed porous nanomaterials using the lyotropic liquid crystal (LLC) method to investigate Pt nanostructures^[100,101]. The LLC phase was formed using polyoctylethylene glycol monohexadecyl ether (CH₃(CH₂)₁₅(OCH₂CH₂)₈OH) as the nonionic surfactant and hexachloroplatinic acid (H₂PtCl₆) as the Pt source. They prepared mesoporous Pt NPs^[100] and films^[101] using direct chemical reduction and electrochemical methods, respectively, which exhibit excellent ORR performances. Ever since the development of the LLC method, researchers have attempted to develop more environmentally friendly methods by substantially decreasing the surfactant concentration. Li et al. reported a micelle assembly method, which requires a significantly lower surfactant concentration than the LLC method, for synthesizing mesoporous Pt nanomaterials^[102].

Although numerous preparation methods have emerged, they afford excellent ORR performance. However, Pt-based noble-metal electrocatalysts have low stability and undergo structural and compositional changes under typical proton exchange membrane fuel cell operating conditions, thereby significantly degrading the performance. The degradation is attributed to the following reasons: (1) electrocatalyst dissolution, such as Pt dissolution; leading to preferential leaching of active metal elements in the Pt-based alloy NPs; and (2) electrocatalyst agglomeration.

Metal-free (nonmetallic heteroatom-doped) C-based electrocatalysts

C nanomaterials generally include graphene, C nanofibers (CNFs), CNTs, ordered mesoporous carbons (OMCs), C black (CB), and graphene-related materials such as rGO^[103]. Compared with metals, C-based nanomaterials have more desirable properties, such as a large specific surface area, excellent mechanical stability, high electrical conductivity, high electrochemical potential, and low cost. Therefore, C-based nanomaterials are widely used as support for ORR electrocatalysts. Doping with heteroatoms such as nonmetallic N, B, P, and S and other light nonmetallic elements can significantly improve the ORR performance of C-based nanomaterials, which have long emerged as electrocatalysts. In addition, they are effective against corrosion and poisoning and have higher durability than transition-metal-based compounds. Through heteroatomic doping of C nanostructures, the charge and spin densities of C radicals can be altered, thereby accelerating the electron transfer of electrocatalysis.

In 2009, Gong et al. proposed the use of N-containing CNTs as ORR catalysts^[104]. The N-containing CNTs demonstrated ORR catalytic performance and long service life compared to those of Pt/C catalysts in an alkaline solution. Yan et al. prepared N-doped hollow C polyhedra (NHCPs) by direct pyrolysis of NaCl-templated zeolite imidazolate framework (ZIF)-8^[105]. The NHCPs exhibited high ORR electrocatalytic activity due to their hollow porous structure, large surface area, high degree of graphitization, and desired N bond type. Kim et al. proposed a novel N-doping approach by introducing a biomimetic synthesis of monodisperse, N-rich melanin-like polymers derived from marine cuttlefish sepia biopolymer^[106]. These polymers were subsequently used as highly N-doped synthetic C (MC) and biomass C (SC) spheres. The ultra-microporosity (<1 nm) of the N-doped MC and SC spheres can be fine-tuned by suitable CO₂ activation. N-doped MC and SC, with ultralarge surface areas of 2,677 and 2,506 m²·g^-1, respectively, exhibit high ORR activity with a good 4e^- reduction pathway, high durability, and excellent methanol tolerance, comparable to those of commercial Pt-based electrocatalysts.

Liao et al. utilized malic acid, tartaric acid, and citric acid as C sources to fabricate N-doped graphene for ORR^[107]. The authors found that N-doped graphene prepared from citric acid exhibits the highest electrocatalytic activity, comparable to that of commercial Pt/C electrocatalysts, and high durability due to the high p-N content in the composites, as shown in Figure 8A-D. Li et al. synthesized a metal-free N-doped γ-graphene catalyst and demonstrated that in situ p-N in N-doped γ-graphene acts as an active site for the ORR, accelerating the RDS of proton extraction, based on their theoretical calculations^[108]. Wang et al. presented a novel approach to uncover the active sites of N-doped graphene in the oxygen reduction ORR by selectively modifying the acetyl group at the p-N and ortho-C atoms of the pyridinic ring^[109]. Their research revealed that the ortho-C atom of the pyridinic ring is the active site for ORR, as evidenced by the complete inactivity towards ORR upon blocking this atom. Zhu et al. synthesized a highly active metal-free N-doped C catalyst (NPC-4) for ORR by KOH activation to increase the pore size and generate surface defects in ZIF-8-derived NPC nanomaterials [Figure 8E-G]^[110]. The authors attributed the high ORR activity of NPC-4 nanomaterials to the abundant and highly exposed active sites resulting from KOH activation. They found that the effect of numerous surface defects is more pronounced than that of the pyrrolic N and p-N functional groups. In contrast, graphitic N species in C-based materials can function as active sites for ORR in alkaline media^[111,112]. Lai et al. produced N-graphene by annealing N-containing polymer/rGO composites (polyaniline (PANI)/rGO or polypyrrole/rGO) and demonstrated that the electrocatalytic activity depends on the graphitic N content, which determines the limiting current density^[113]. Meanwhile, the p-N content increases the ORR onset potential (E_onset). Yasuda et al. attributed the 4e^- transfer process in the ORR in N-doped graphene catalysts to the existence of p-N, whereas graphitic N promotes a 2e^- transfer pathway. The authors concluded that p-N is a more effective active site for ORR than graphitic N^[114].

Figure 8. (A) SEM image of NG-C; (B) low-magnification TEM image and selected area electron diffraction (SAED) pattern (inset) of NG-C; (C) LSV curves of ORR on NG-M, NG-T, NG-C, and commercial Pt/C; (D) ORR forward peak maximum currents for NG-C and commercial 20 wt% Pt/C catalysts after different cycles. Reproduced with permission. Copyright 2015, ACS Publications^[107]; (E) Schematic diagram of the synthesis of the pore size-enlarged NPC nanomaterials derived from ZIF-8 precursors; (F) Linear sweep voltammograms of ZIF-8-derived NPC nanomaterial, KOH-activated ZIF-8-derived NPC nanomaterials and Pt/C (20%) in O₂-saturated 0.1 M KOH solution at a scan rate of 10 mV·s^-1 with 1,600 rpm; (G) LSV curves of NPC-4 with various rotation speeds, with the inset giving the K-L plots. Reproduced with permission. Copyright 2019, ACS Publications^[110].

The P doping of C-based electrocatalysts significantly influences their ORR activity compared to that of undoped nanomaterials^[115]. Wang et al. developed a technique for in situ P-doping of C dots (CDs) and simultaneous modification of graphene substrates via biomass derivatization^[116]. The resulting P-doped CD/graphene (P-CD/G) nanocomposite displayed an extremely high P-doping level of C nanomaterials and ORR activity comparable to commercial Pt/C electrocatalysts. Yang et al. proposed four types of P-doped graphene structures, namely, PC3G, OPC3G, PC4G, and OPC4G sheets, to study the effect of the chemical bonding state of the P atoms on stability, electronic structure, and ORR catalytic activity^[117]. Based on stability, electrical conductivity, Gibbs free energy barriers of the rate-determining step (ΔG_RDS), and the number of potential active sites of the P-doped graphene structure, OPC3G is the most efficient and stable P-doped graphene for ORR. Additionally, the C atom, which has a high negative charge, could be the most active site for ORR. The high electrocatalytic activity and durability of the electrocatalysts are mainly attributed to P doping in the C lattice, and the amount of P incorporated into the C mainly determines the ORR activity increase and modification of the ORR pathway^[118]. Wu et al. prepared P-doped C hollow spheres (P-CHS) through a hydrothermal method, using glucose as the C source and tetraphenylphosphonium bromide as the P source^[119]. The P-CHS possesses a microporous shell structure, exhibiting a large specific surface area exceeding 500 m²·g^-1. The electrocatalytic activity and long-term stability of the resulting P-CHS, containing 1.61 at% of P, were evaluated using rotating ring-disk electrode (RRDE) measurements in alkaline media. The microstructure and ORR activity are significantly influenced by the P content in the C. The P doping in the C lattice induces highly active catalytic sites, contributing to the high electrocatalytic activity and durability of P-CHS for ORR. Moreover, the hollow spherical structure of P-CHS facilitates the efficient mass transfer of reactants and products during the ORR. Bai et al. used a density functional theory (DFT) approach to investigate the ORR mechanism on P-doped divacancy graphene^[120]. Their study showed that P-doped graphene with intrinsic C defects has high electrocatalytic activity for ORR. Overall, P doping in C-based electrocatalysts significantly enhances their ORR activity, and the amount of P incorporated into the C lattice plays a crucial role in determining the ORR performance.

The ORR performance can be enhanced by doping S atoms in graphene, as C-S acts as a catalytically active site, improving the ORR kinetics^[121-123]. Chen et al. synthesized S-doped rGO (S-rGO) nanosheets using a one-pot hydrothermal method [Figure 9A]^[124]. The prepared S-rGO consisted of high-quality sulfide species (mainly C-S-C) and many exposed defects and edge sites on its surface. Figure 9B-D demonstrates that S-rGO has high conductivity and electrocatalytic activity for the ORR and significantly higher stability and better methanol tolerance than commercial Pt/C electrocatalysts. Wong et al. proposed that changes in the number of defects in a material affect its electrochemical performance and that the amount of S dopant is positively correlated with the decrease in the required overpotential^[125]. Although S-graphene has a lower onset potential than commercial Pt/C electrocatalysts, it has higher durability than N-graphene and commercial Pt/C electrocatalysts. To improve the catalytic efficiency of graphene quantum dots (GQDs) for the ORR, Banerjee et al. explored the role of S-dopants using a first-principles DFT-based approach^[126]. They found that the ORR efficiency varies with the dopant location and configuration on the S-doped GQDs at different active sites. The charge transfer from the dopant site to the active site and intermediates and the binding energy of the intermediates on the active site mainly determine the RDSs, onset potentials, and overpotentials required for the ORR. An increase in the S-dopant concentration or the presence of O₂ significantly increases the ORR efficiency of the GQDs. Similarly, B-doping in C materials can result in better catalytic performance than undoped C materials, although they exhibit moderate ORR activity^[127-129]. Zhang et al. used Na borohydride as a reducing agent to convert CO₂ into B-doped porous C (B-PC), which was further activated by heat treatment^[130]. The treated B-PC has the same ORR activity as the Pt-AC catalyst but shows better ORR selectivity. The improvement in electrocatalytic performance is attributed not to the change in C morphology but to the change in the bond between B and C atoms on the surface and the broadening of π states. Xiao et al. reported a method to synthesize porous graphene and B-doped porous graphene with an abundant porous structure and high B-doping level [Figure 9E]^[131]. The abundant pore structures and B-related active sites accelerate the ORR kinetics. Overall, doping of S and B in C-based electrocatalysts can significantly enhance their ORR activity, and the dopant concentration, location, and configuration play a crucial role in determining their electrocatalytic performance.

Figure 9. (A) Micrographs of S-RGO-180, (A-1) FE-SEM image of S-RGO-180 and (A-2) the corresponding C-, O-, and S-elemental mappings; (B) LSV curves of RGOs, S-RGOs, and Pt/C; (C) Rotating-disk voltammograms recorded for S-RGO-180 electrode in an O₂-saturated 0.1 M KOH solution at a different rotation rate. The inset in (C) shows the corresponding Koutecky-Levich plots (J^-1vs.ω^-0.5) at different potentials; (D) The magnified Nyquist plots of RGO-180 and S-RGOs. Reproduced with permission. Copyright 2014, Royal Society of Chemistry^[124]; (E) Schematic illustration of the synthesis process of B-hG and its charge/discharge processes as cathode for LOBs. Reproduced with permission. Copyright 2022, Elsevier^[131].

Single-heteroatom-doped C-based materials exhibit high electrocatalytic activity. However, bidoped nanocarbon exhibits better electrocatalytic performance^[132-135]. For example, Duan et al. synthesized N, S-bidoped graphene, which shows higher ORR electrocatalytic activity than S- or N-doped graphene^[136]. Zhang et al. demonstrated that N, S-codoped graphene/CNT composites (N, S-GCNT) exhibit superior electrochemical ORR activity compared to graphene-doped or N, S-codoped CNTs^[137]. Xue et al. reported that B, N-codoped graphene foams display significantly enhanced electrocatalytic ORR activity compared to foams doped with only B or N^[138]. In addition, Zhao et al. found that the ORR activity of CNTs doped sequentially with B and N is higher than that of CNTs doped simultaneously with B and N^[139]. Furthermore, the B, N-codoped graphene nanoribbons exhibit ORR activity comparable to that of the 20 wt% Pt/C electrocatalyst, owing to the synergy between N, B-codoping and active sites located at the edges of the nanoribbons^[140]. N, P-bidoped C electrocatalysts present higher ORR activity than N- or P-doped composites [Figure 10A-D]^[141]. Furthermore, N, P-bidoped porous C synthesized via high conversion of coconut shell residues exhibit high electrocatalytic ORR activity^[142]. Tridoping also leads to high ORR activity^[143-145]. For instance, N/S/O-doped OMCs (N/S/O-OMCs) exhibit high ORR activity [Figure 10E and F]^[146]. Nanocarbons with nanoporous morphology and heteroatom (N, P, and B) codoping exhibit higher ORR activity than single- or bidoped catalysts, as evidenced by their superior current density and onset potential^[147]. Li et al. doped F and S atoms into a graphene C matrix using N as the main dopant via a novel ternary-N-precursor-inspired strategy to promote the ORR^[148]. The use of secondary and tertiary N precursors enhances the configuration of active sites, resulting in increased ORR activity. Additionally, the ORR performance of N, F, S-tridoped porous graphene is significantly improved, largely due to the presence of additional active sites and the porous structure of the material. These results highlight the importance of bi and tridoping of electrocatalysts with different heteroatoms for increasing their ORR activity. The development of ternary or multi-doped C remains a significant challenge due to various factors. These challenges include difficulties in selecting appropriate precursors, a lack of established design principles, and problems in optimizing the beneficial composition of heteroatoms.

Figure 10. (A) Schematic illustration of the fabrication process and structure of the N,P-GCNS bifunctional oxygen electrocatalyst; (B) LSV curves of various catalysts at a rotation rate of 1,600 rpm; (C) K-L plots of N,P-GCNS at different potentials. (Inset) Electron transfer number (n) of various catalysts at different potentials; (D) Bifunctional catalytic activity of various catalysts toward both ORR and OER. Reproduced with permission. Copyright 2015, ACS Publications^[141]; (E) Schematic illustration of the synthetic process; (F) The dependences of onset and half-wave potentials, and percentages of different types of N sites on the carbonization temperatures for the N/S/O-OMCs. Reproduced with permission. Copyright 2021, Elsevier^[146].

Metal-N-C electrocatalysts

N-doped-C-containing transition-metal-based materials are currently considered the most promising non-Pt catalysts for the ORR, exhibiting high activity in both 0.1 M KOH and 0.5 M H₂SO₄ or 0.1 M HClO₄ solutions^[95]. The high ORR activity of macrocyclic compounds with metal-N (M-N₄) centers was first reported in 1964, leading to research on non-Pt-based ORR electrocatalysts. Since then, transition-metal macrocyclic compounds have been of interest to many researchers. Initially, N-containing macrocyclic compounds, such as phthalocyanine, tetraphenylporphyrin, tetramethoxyporphyrin, and tetrazene, were used as N source precursors to provide M-N₄ structures^[149]. More recently, a major breakthrough in the synthesis of M-N-C electrocatalysts has been achieved through the pyrolysis of simple and inexpensive C substrates, N precursors, and transition-metal salts in the absence of precursors with M-N₄ groups to obtain nanomaterials with M-N₄ structures.

Determining the correlation between the ORR catalytic activity and M-N₄ structure has proven difficult, leading to the use of various nonmacrocyclic N-containing compounds as N-source precursors. These include small inorganic molecules such as NH₃ and Na azide, small organic molecules such as acetonitrile, pyrrole, ethylenediamine, and phenanthroline, and N-containing organic polymers such as polyacrylonitrile (PAN), polypyrrole, polyethyleneimine, melamine resins, and PANI^[150-154]. N-containing polymers have advantages over small organic molecules as they can form ordered templates and more stable C-based active layers, preventing the collapse of the C skeleton and severe loss of active sites during catalyst pyrolysis. They are also cheaper and more accessible than macrocyclic polymers. The choice of N source precursor significantly influences the activity and stability of the catalyst, and the selection of C support, transition metal, and pyrolysis process affects the final properties. Current research is mainly focused on Fe- or Co-containing nanomaterials as they exhibit the highest ORR activity.

A one-pot method using dopamine as the N source and metal salts as precursors for the ORR was used to incorporate various transition metals into N-doped CNTs (M-N-C/CNT, M = Fe, Mn, Co, and Ni), as shown in Figure 11^[155]. Among the four as-synthesized catalysts, Mn-N-C/CNT-800 and Co-N-C/CNT-800 exhibit much higher electrocatalytic activities than Ni-N-C/CNT-800 and Fe-N-C/CNT-800. Mn-N-C/CNT-800, in particular, has an E_onset approximately 75 mV lower than that of Pt/C (-35 mV) and an E_1/2 approximately 20 mV lower than that of Pt/C (-150 mV), indicating its high ORR activity. The stability test results demonstrate excellent stability, with the curve obtained after 2,000 or 5,000 cycles highly consistent with that obtained initially by linear sweep voltammetry. Moreover, the limiting current density decreases by only 5.57% in approximately 11 h, indicating its high durability. The electrocatalytic performance of M-N-C/CNT catalysts is affected by the degree of metal graphitization and the metal state, which depend on the type of transition metal.

Figure 11. (A) The TEM images of M-N-C/CNT: (A-1) Co-N-C/CNT-800, (A-2) Mn-N-C/CNT-800, (A-3) Ni-N-C/CNT-800 and (A-4) Fe-N-C/CNT-800; (B) CV curves of M-N-C/CNT-800 s and N-C/CNT-800 in N₂-saturated (grey) and O₂-saturated (colored) 0.1 M KOH electrolytes; (C) LSV plots of M-N-C/CNT-800 s, N-C/CNT-800 and 20 wt% Pt/C. Reproduced with permission. Copyright 2018, Elsevier^[155].

Progress has been made in optimizing the C support and pyrolysis temperature of M-N-C electrocatalysts, depending on the preferred N source or transition-metal precursor. Single-atomic catalysts (SACs), in particular, have shown higher ORR catalytic activity and stability than commercial Pt/C electrocatalysts^[156,157]. Despite these advancements, improving the catalytic activity and stability of M-N-C electrocatalysts remains a significant challenge. Two key factors influence the catalytic activity and stability of M-N-C electrocatalysts:

(i) The catalytic activity and spatial distribution of ORR active sites in M-N-C catalysts are determined by complex interactions between the elemental compositions and different components. However, there is no consensus on the exact nature of the catalytically active sites in M-N-C catalysts^[158]. N plays a crucial role in M-N-C electrocatalysts, with its role in catalytic reactions divided into three categories. Firstly, N functional groups on the surface of C-based supports directly participate in the catalytic reaction and are responsible for the ORR activity^[159]. Secondly, the N functional group coordinates the environment, and the coordination bond formed by the lone electron pair provided by the transition-metal ion serves as the catalytically active center for the ORR^{[157,160,161]}. For example, in a Co-N-C catalyst prepared with PANI as the N source precursor, Co facilitates the entry of N into the structural matrix of C, and Fe can form an active center of Fe-N_x with N^[162,163]. Researchers have harnessed the bimetallic effect by simultaneously introducing Fe and Co to increase the activity and stability of CoFe-based electrocatalysts^[164-166]. Thirdly, a combination of the above two methods exists. In situ (isotope-labeled) Raman spectroscopy of a molecular Fe phthalocyanine model catalyst and a pyrolytic Fe-N-C catalyst has shown that besides the single-atom Fe-N_x sites, two C-N sites were identified in the pyrolyzed Fe-N-C catalyst and elucidated as separate active sites, showing distinct ORR intermediates and RDSs^[166,167]. In summary, the exact nature of the catalytically active sites in M-N-C catalysts remains unclear, but N plays a crucial role in catalytic reactions and is involved in three different categories of mechanisms. Researchers have used bimetallic effects and identified separate active sites to increase the activity and stability of M-N-C electrocatalysts.

(ii) The effective active site density and oxygen transportation of electrocatalysts are determined by their specific surface area and porous structure^[168,169]. Direct high-temperature heat treatment typically results in M-N-C electrocatalysts with a nonuniform active site distribution and difficulty forming a porous structure, which is not conducive to O₂ transport. The introduction of nanoporous structures could effectively resolve this challenge. A large specific surface area increases the number of catalytically active sites, while an ordered mesoporous structure increases the mass transfer rate. Xu et al. developed a facile dual molten salt (ZnCl₂ and NaCl)-mediated templating method for preparing Fe-N-C electrocatalysts with tailored porous frameworks^[170]. The resulting Fe-N-C material exhibits excellent ORR performance and high stability in both alkaline and acidic media. Furthermore, the electrocatalyst has a large surface area (1,605 m²·g^-1), which facilitates the mass transfer and exposure of the Fe-N₄ sites.

To obtain porous nanomaterials with large specific surface areas that can carry many active sites, new and efficient synthesis methods must be explored. Future research should focus on in-depth studies of the catalytic reaction mechanism and microstructure of M-N-C, which will be the key to the development of efficient electrocatalysts.

Transition-metal-oxide-, phosphide-, chalcogenide-, nitride-/oxynitride-, and carbide-based electrocatalysts

Nonprecious-metal-based ORR electrocatalysts, including TMOs, TMPs, transition-metal sulfides, TMNs/TMONs, and TMCs, have been extensively researched, along with nonmetallic and M-N-C electrocatalysts. Although the use of transition-metal-compound catalysts in acidic media poses a challenge, their performance in alkaline media is excellent^[171]. The ORR kinetics is significantly improved in alkaline media compared to acidic media. Rational structural design can endow transition-metal-compound electrocatalysts with high activity comparable to that of Pt-based electrocatalysts.

(i) Transition-metal oxides: TMOs have garnered significant interest as electrocatalysts for the electrochemical ORR in alkaline media due to their potential applications in metal-air batteries and fuel cells. They are often naturally abundant, cost-effective, and environmentally friendly, displaying a high catalytic activity for ORR. Among various TMOs, Mn oxides (MnO_x) have gained much attention due to their low cost, abundance, and high electrocatalytic activity for both ORR and OER. The relationship between the structure and composition of TMOs and their ORR catalytic activity has been a subject of research. In the 1990s, Bockris and Otagawa studied the impact of transition metal cation properties on oxygen electrocatalysis^[172]. Later publications extended the conclusion of the work of Bockris and Otagawa to the ORR and other types of TMOs. For instance, a study on TMOs with perovskite structures investigated the OER and reported the following order of ORR activity in a recent paper on the effects of crystal structure^[173]: α-MnO₂ > amorphous MnO_x > β-MnO₂ > δ-MnO₂ [Figure 12A-C]. The authors also studied the relationship between ORR activity and O₂ adsorption strength using a temperature-programmed desorption method. Liu et al. studied the effect of Co₃O₄ NP surface orientation and found that the ORR activity follows the order of (111) > (100) > (110)^[174]. Meanwhile, Tang et al. attributed the dependence of ORR activity on MnO_x structure and composition to the redox potential of surface Mn cations^[175]. In binary and mixed MnO_x, ORR activity was found to be linked to the surface Mn(IV)/Mn(III) redox transition potential (E_f). Mn₂O₃ demonstrated the highest ORR activity among all the studied MnO_x due to its highest E_f, which arises from the unique property of the crystal structure related to waffle Fe fluorite, making it suitable for larger Mn(III) cations but not for smaller Mn(IV) cations. Other studies have also investigated the effect of the redox properties of transition-metal cations on ORR activity. For instance, in the La_xCa_1-xMnO₃ perovskite^[176], the ORR activity was found to increase as the average Mn oxidation state decreased from 4.0 to 2.8. Qaseem et al. synthesized a hybrid nanomaterial consisting of MnCo₂O₄ and N-doped rGO (NGr) to serve as an ORR electrocatalyst. They found that the addition of spinel TMO NPs via nucleation and cation substitution on graphene flakes led to covalent coupling with NGr, resulting in higher activity than either NPs or NGr alone^[177]. The substitution of Mn also increased the ORR activity of the catalytic site of the hybrid nanomaterial compared to that of the pure CoxOy/NGr hybrid. The MnCo₂O₄/NGr hybrid electrocatalyst exhibited comparable current density to that of the Pt/C electrocatalyst at low overpotentials and higher current density at higher overpotentials. Additionally, the hybrid catalyst demonstrated excellent stability, retaining 96% of its initial ORR current density after continuous operation for over 5,000 s, while the Pt/C electrocatalyst retained only 78% of its initial current density. Recently, many studies have focused on the modification engineering of transition-metal oxide catalysts (e.g., regulation of crystallinity and oxygen vacancies)^[178], among which surface or interface engineering of transition-metal-oxide electrocatalysts is the most common approach^[179]. Nayak et al. developed a surface engineering strategy for Ni- and Co-based hybrids (NiO-Co₃O₄) to be used as efficient and robust ORR electrocatalysts^[180]. They achieved this by controlling the precursor amounts of Ni and Co during the wet chemical synthesis to easily tune the surface nanostructure of the NiO-Co₃O₄ nanohybrid. By optimizing the composition, they were able to create surface structures consisting of 1D nanorods and 2D nanosheets, which significantly increased the ORR electrocatalytic activity and stability of the NiO-Co₃O₄ nanohybrid. The excellent catalytic performance of the NiO-Co₃O₄ nanohybrids can be attributed to charge or mass transfer within the 1D and 2D open-surface nanostructures. Zha et al. investigated the interplay between nanostructure, chemical composition, and electrocatalytic ORR activity by constructing tunable anemone-like hierarchical Fe₃O₄/Co₃O₄ nanoheterojunctions with controllable porosity and structural parameters^[181]. They achieved this using a solvothermal quasi-inverse-emulsion method, as depicted in Figure 12D and E. The electrocatalysts developed by Zha et al. possess a large specific surface area ranging from 302 to 810 m²·g^-1, an optimized pore size distribution, and pore volume^[181]. The resulting nanoheterojunctions exhibit high electrocatalytic ORR activity, long-term stability, and excellent methanol tolerance in alkaline solutions. Furthermore, the correlation analysis revealed that the electrocatalytic ORR activity of the porous nanoheterojunction is proportional to its specific surface area and is governed by the surface chemistry of the electrocatalyst.

Figure 12. (A) SEM and images of materials: (A-1) SEM image of Ni/α-MnO₂-SF, (A-2 and A-3) TEM images of Ni/α-MnO₂-HT, (A-4) SEM image of α-MnO₂-SF, and (A-5 and A-6) TEM images of α-MnO₂-HT; (B) LSV curves studied on RDE electrodes in O₂-saturated 0.1 M KOH solution, at a rotation rate of 1,600 rpm with a scan rate of 5 mV·s^-1; (C) Nyquist plots obtained from EIS measurements in O₂-saturated 0.1 M KOH solution on the manganese oxide modified PG carbon electrodes at an anodic polarization potential of 0.65 V (1.66 Vvs. RHE). Reproduced with permission. Copyright 2014, ACS Publications^[173]; (D) Morphology and microstructure characteristics of Actiniae-25 and Actiniae-65: (D-1 and D-2) SEM images of Actiniae-65, (D-3) TEM image of Actiniae-65, (D-4) HRTEM image of Actiniae-65, (D-5) SEM image of Actiniae-25, and (D-6) TEM image of Actiniae-25; (E) Schematic illustration of the solvothermal quasi-reverse-emulsion induced fabrication of actiniae-like hierarchical Fe₃O₄/Co₃O₄ nanoheterojunctions. Reproduced with permission. Copyright 2020, Elsevier^[181].

(ii) Transition-metal phosphides: TMPs exist in many forms, among which phosphides with electrochemical catalytic properties mainly include Co phosphide^[182,183], Fe phosphide^[184,185], Ni phosphide^[186,187], and Fe/Co/Ni binary or ternary mixed phosphides^[188,189]. Hydrothermal treatment and templating are the main methods for preparing nanostructured TMP precursors, following which phosphating is performed at low temperatures using Na₂PO₂ to afford the phosphides^[190]. Some phosphides can be synthesized by oil-bath heating or high-temperature annealing. The nanostructure of phosphides can be significantly altered by changing the reaction temperature, reaction time, chemical composition, template type, or support (e.g., porous C, graphene, and C cloth).

While Co phosphide is typically used as an HER electrocatalyst, Liu et al. prefabricated CoP@N-C for the ORR and found that the electron transfer number for CoP@N-C ranged from 3.62 to 3.94, indicating that the reaction proceeds via a complete 4e^- transfer ORR process^[191]. The ORR catalytic activity of the CoP@N-C electrocatalyst was found to be comparable to that of the Pt/C electrocatalyst. Additionally, certain unique polymer structures were found to enhance the catalytic performance of phosphides. Wu et al. reported Co-porphyrin-based conjugated mesoporous polymer (CoP-CMP) as an ORR electrocatalyst with an onset potential of -0.12 V and a half-wave potential of -0.18 V, which is close to the values obtained for Pt/C electrocatalyst^[192]. The electron transfer number was calculated to be 3.83-3.86, indicating the dominance of the 4e^- transfer ORR process. This suggests that CoP-CMP800 has the highest ORR catalytic activity and outperforms many ORR electrocatalysts reported so far. Meanwhile, Fe phosphides are also commonly used as ORR electrocatalysts. Wang et al. developed a facile method for fabricating myricetus-like N, P-rich C NSs embedded in Fe₂P nanocrystals (Fe₂P/NPCs) by synthesizing Myrica-like PANI-NSs through microemulsion templating, incorporating Fe into the PANI-NS during hydrothermal treatment, and obtaining Fe_xP nanocrystals and N, P-rich carbonaceous supports through pyrolysis of hydrothermally treated PANI-NS [Figure 13]^[193]. The FexP/NPCs synthesized in this study have a uniform particle size of about 250 nm, a large specific surface area of 523 m²·g^-1, and a mesoporous structure with an average diameter of approximately 3.46 nm. The phase structure of FexP is determined by the Fe/P molar ratio used during hydrothermal treatment. The Fe₂P phase can be obtained when the Fe/P ratio is around 0.2. Fe₂P/NPCs exhibit high ORR electrocatalytic activity, with a half-wave potential of 0.820 Vvs. RHE and a diffusion-limited current density of 5.58 mA·cm^-2, comparable to those of commercial Pt/C electrocatalysts. Ran et al. also developed a honeycomb Ni₂P/Ni₁₂P₅ heterostructure (Ni₂P/Ni₁₂P₅@NF) with abundant P vacancies grown in situ on foamed Ni. The P vacancies can induce the delocalization of the bound electrons near the Ni-P bond, thereby adjusting the bandgap and facilitating conductivity and electrocatalytic activity enhancements^[187]. Furthermore, Fe/Co/Ni binary or ternary mixed phosphides have been found to exhibit high ORR activity. Vijayakumar et al. developed CoP-N-doped-C@NiFeP nanoflakes (CoP-NC@NFP) from MOFs rich in multiple active sites^[194]. Experimental results demonstrate that the multiple active catalytic sites of CoP-NC@NFP facilitate fast charge-transfer kinetics and excellent electrocatalytic performance for ORR.

Figure 13. (A) The synthesis process of Fe_xP/NPCs as efficient electrocatalysts for ORR; (B) LSV curves of different electrocatalysts obtained at a rotation rate of 1,600 rpm; (C) LSV curves obtained before and after 4,000 CV cycles. Reproduced with permission. Copyright 2021, Elsevier^[193].

(iii) Research on non-noble-metal chalcogenides as ORR electrocatalysts began in the 1970s. In 1974, Baresel et al. reported maximum current densities of 2,249 mA∙m^-2 and 1,000 mA∙g^-1 (at 600 mV) for Co-S and Co-Ni-S in acidic systems, respectively^[195]. This high catalytic activity was attributed to the minimal energy difference between the S 2p orbital and the highest occupied d orbital of the sulfide. In recent years, various transition-metal chalcogenides (M-X, where M=Co, Ni, Fe, Ru, Re, or Rh, and X=S, Se, or Te) have been utilized as ORR electrocatalysts, displaying excellent performance^{[171,196-198]}. For instance, Zhang et al. developed an efficient electrocatalyst for the electrosynthesis of H₂O₂ by designing a Ti-doped Zn Co sulfide hollow superstructure (Ti-ZnCoS HSS)^[199]. Ti-ZnCoS HSS was synthesized by sulfidation treatment of the MOF precursor followed by curing, forming a hollow-to-hollow superstructure (HSS) with small nanocages assembled around a large cake-like cavity [Figure 14]. The experimental and simulation results demonstrate that the multiple metallic components in Ti-ZnCoS HSS play a crucial role in altering the D-band center and their binding energy with O₂ species. The presence of abundant active sites in HSS promotes mass and electron transfer, resulting in excellent 2e^- transfer ORR performance at the atomic and nanometer levels. The synergy between D-band centers and superstructures in Ti-ZnCoS HSS leads to a highly selective ORR performance of 98%, with a high activity potential of 0.774 V vs. RHE at 1 mA·cm^-2 and a H₂O₂ production rate of 675 mmol h^-1 g_cat^-1. Moreover, Ti-ZnCoS HSS exhibits long-term stability under alkaline conditions. To tune ORR performance, crystal phase control is critical. For instance, co-encapsulating hexagonal (h-FeSe) and tetragonal FeSe (t-FeSe) ultrafine nanocrystals in an N-doped C matrix without agglomeration have been shown to be effective^[200]. The resulting material, which contains a large number of h-FeSe nanocrystals, exhibits excellent ORR performance with a positive onset potential of 0.97 V and a high limiting current density of 5.4 mA·cm^-2. In addition to its high catalytic activity, the material also exhibits remarkable methanol tolerance and stability. Theoretical studies have demonstrated that h-FeSe is superior to t-FeSe in terms of O₂ adsorption and *OOH intermediate dissociation of O-O bonds on the active Fe sites, making it more effective for alkaline ORR than t-FeSe. Therefore, h-FeSe is a promising alternative to t-FeSe for ORR applications. Moreover, Xiao et al. synthesized raisin-bread-shaped electrocatalysts composed of Fe sulfide (Fe_1-xS) and N, S-bidoped mesoporous graphite C spheres (N, S-MGCSs) via two-step pyrolysis and acid leaching^[201]. This catalyst demonstrates high electrocatalytic activity for ORR in both alkaline and acidic media. Its activity is directly correlated with the content and distribution of Fe sulfide species. Notably, the electrocatalytic activity of (Fe_1-xS/N, S-MGCS)_0.2 surpasses that of the commercial Pt/C electrocatalyst in alkaline media. This superior catalytic performance can be attributed to the introduction of Fe sulfide nanocrystals into high-conductivity C carriers, which significantly enhances the reactivity of the catalytic active sites. Additionally, the raisin-bread-like structure improves the electron- and ion-transfer characteristics and also inhibits the self-aggregation and leaching of Fe sulfide species.

Figure 14. (A-1) A scheme of the two-step synthesis of Ti-ZnCoS HSS; (A-2) Illustration of the 2e-ORR activity-selectivity of CoSx, ZnS, ZnCoS, and Ti-ZnCoS HSS; (B) Characterization of Ti-ZnCoS HSS: (B-1) SEM, (B-2) TEM, and (B-3) HADDF-STEM; (C) Free-energy diagram for oxygen reduction to H₂O₂; (D) Schematic illustration for the 2e-ORR process of Ti-ZnCoS HSS catalyst; (E) LSV polarization curves; (F) CV current density vs scan rate; the linear slope is equivalent to the double-layer capacitance (C_dl). Reproduced with permission. Copyright 2022, Wiley Online Library^[198].

(iv) Transition-metal nitrides and oxynitrides: The synthesis of noble-metal nitrides has demonstrated that all transition metals can form nitrides^[202]. These nitrides and oxynitrides possess high electrical conductivity and chemical resistance, making them attractive for various applications^[203-206]. Nitride formation is particularly advantageous for modifying catalyst electronic structure, as it contracts the D-band of TMN and increases electron density near the Fermi level, giving TMNs electronic properties similar to noble metals^[207]. This facilitates electron transfer to oxygen intermediates and accelerates ORR kinetics^[208]. Non-noble TMNs have been reported as highly effective ORR electrocatalysts in alkaline media^[208]. Air-exposed nitrides spontaneously form oxide shells several nanometers thick on conductive nitride cores, which serve as highly active catalytic structures. Among these, C-supported Co nitride (Co₃N/C) is the most active electrocatalyst, exhibiting a half-wave potential of 0.862 V and achieving a record peak power density of 700 mW·cm^-2 in an alkaline membrane electrode assembly. However, operando X-ray absorption spectroscopy studies indicate that Co₃N/C remains stable below 1.0 V but undergoes irreversible oxidation at higher potentials. Bimetallic nitrides have also been extensively studied in this context. Tang et al. synthesized Ti_0.95Co_0.05N, a binary nitride electrocatalyst with a 3D structure that exhibits high activity and stability in ORR^[209]. The 20%C-Ti_0.95Co_0.05N catalyst shows comparable ORR performance to 20% Pt/C in alkaline media, with the 4e^- pathway dominating the reaction. The 3D structure of the catalyst provides a larger surface area and more exposed active sites, while the incorporation of Co induces a change in the electronic structure, increasing the intrinsic ORR activity of Ti nitride. Huang et al. demonstrated that VO_xN_y-CNTs electrocatalyst has high stability and catalytic activity in a KOH medium^[210]. The porous structure facilitates full contact between the electrolyte, O₂, and electrocatalyst, while the formation of C-N bonds increases overall conductivity and decreases overpotential. The ORR mainly follows the ideal 4e^- pathway, as shown by the Koutecky-Levich equation analysis. Bimetallic oxynitrides exhibit better catalytic performance than single-metal oxynitrides due to the synergy occurring on them^[211]. Balamurugan et al. synthesized a new type of N-doped graphene (CuMo₂ON@NG) anchored by Cu molybdenum oxynitride using a pyrolysis method. The as-synthesized electrocatalyst exhibits trifunctional activity, the activity being higher than that of the benchmark Pt/C or IrO₂ electrocatalyst^[212]. The outstanding catalytic activity of CuMo₂ON@NG is attributed to the synergistic electron transfer among the active CuMo₂ON NPs, N-doped species, and graphitic C, as demonstrated by first-principles calculations. In the CuMo₂ON@NG nanohybrid, the RDS for achieving ORR activity is the creation of an O* intermediate on the CuMo₂ON lattice. Several experimental investigations have indicated that adjusting the oxidation states of metals in the O-N phase can further improve the ORR activity, which has the potential to match or exceed that of Pt/C electrocatalysts.

(v) Transition-metal carbides: TMCs have been identified as potential substitutes for Pt group metals due to their high electrical conductivity^[213,214]. However, their susceptibility to corrosion in acidic and oxidizing media can affect their catalytic stability. Among TMCs, TiC has received significant attention due to its high electrical conductivity and corrosion resistance^[215]. Researchers have synthesized TiC/C nanofibers as ORR electrocatalysts via electrospinning^[216], showing that the morphology, TiC content, and TiC-C interface of the nanofibers can be tuned by varying the preparation conditions. The TiC/C properties affect catalytic performance, with the Ti-CNFs/C samples exhibiting the best ORR performance, including an onset potential, peak potential, peak current density, and charge transfer number of 0.941 V vs. RHE, 0.812 V vs. RHE, 0.742 mA·cm^-2, and 3.29, respectively. These samples also showed high catalytic durability and excellent methanol tolerance, demonstrating the potential of TiC/C nanofibers as ORR electrocatalysts. Recent studies have shown that N doping can improve the ORR performance of TiC and that the introduction of metals such as Fe and Co into TiC powder enhances its ORR performance^[217]. One promising material for the ORR in alkaline environments is Fe-N-TiC. This material has been found to have the highest ORR activity among tested materials, with a reaction path of 4e^-. N plays a role in both the formation of C-N active sites and changes to the electronic structure of the TiC surface, contributing to enhanced catalytic activity. Fe doping of Fe-N-TiC leads to further increases in ORR catalytic activity by forming Fe-N active sites. Additionally, C-based materials with Fe₃C particles have also demonstrated excellent ORR properties [Figure 15A-D]. For example, Chen et al. developed highly efficient electrocatalysts for the ORR in alkaline solutions by preparing N-doped C-coated Fe₃C NPs (NS) through the carbonization of Fe precursors loaded with polyacrylamide microspheres [Figure 15E-H]^[218]. The N-doped C and Fe₃C work synergistically as catalytically active sites, and the C shell effectively prevents leaching from the Fe₃C, resulting in excellent electrocatalytic performance. This highlights the potential for doping TMCs as a research direction for ORR electrocatalysts. In recent years, Fe carbide encapsulated in N-doped graphene (NG/Fe₃C) has emerged as a promising ORR catalyst. Patniboon et al. investigated the stability and catalytic activity of N-doped graphene supported on metal-Fe carbide (NG/M-Fe₃C) for the ORR using DFT calculations^[219]. The NG/M-Fe₃C heterostructure was modeled by replacing Fe atoms near the NG/Fe₃C interface with metal alloy species “M” (M = Cr-Mn, Co-Zn, Nb-Mo, or Ta-W). The calculations showed that introducing M changes the work function of the N-doped graphene, which is correlated with the bond strength of the ORR intermediate. The introduction of Ni or Co atoms at the interface increases the ORR activity of NG/Fe₃C and stabilizes its heterostructure. Furthermore, increasing the concentration of Ni or Co atoms near the interface enhances the ORR activity, leading to the creation of stable heterostructures over a wide range of substitution concentrations. These findings provide approaches to improve the ORR activity of NG/Fe₃C electrocatalysts.

Figure 15. (A) STEM imaging and EELS spectra of FeN₄-TiC catalysts; (B) Kinetic currents extracted from linear sweep voltammograms of FeN₄-TiC and FeN₄-C catalysts at the beginning of life (BoL) and end of life (EoL); (C) DFT-calculated free energy diagram of FeN₄-TiC and FeN₄-C for ORR; (D) Cyclic voltammograms of the Fe-N-C and Fe-N-TiCcatalysts in their pristine state and after 1,000 cycles of the catalyst corrosion resistance stability protocol. Reproduced with permission. Copyright 2021, Chemistry Europe^[216]; (E) TEM and (F) high-resolution TEM (HRTEM) image of Fe₃C@N/C-1; (G) Linear sweep voltammetry (LSV) curves; (H) Current-time chronoamperometric response of Fe₃C@N/C-1 catalyst and Pt/C. Reproduced with permission. Copyright 2019, MDPI^[217].

Nonmetallic C-based electrocatalysts

In the field of electrocatalysts, C-based materials, such as CNTs, graphitic C nitride (g-C₃N₄), CB, graphene oxide (GO), and other metal-free C-based electrocatalysts, have shown great potential due to their high electrical conductivity and large specific surface areas. These materials promote electron transfer and mass diffusion during the electrocatalytic reaction^[221,222]. Efficient and low-cost bifunctional electrocatalysts have also been developed using heteroatom-doped C materials. The introduction of heteroatoms, such as N, S, O, B, and P, can effectively adjust the charge or spin redistribution in the sp2 conjugated C matrix, leading to an increase in the bifunctional electrocatalytic activity for OER and ORR^{[127,223,224]}. Based on research, Heteroatom-doped C materials mainly include three types: single-atom-doped, bidoped, and tridoped.

Single-atom doped: N-doped C materials have been extensively studied as electrocatalysts, with N-doped C nanomaterials being one of the earliest heteroatom-doped C materials investigated for this purpose^[225,226]. In 2013, few-layer N-doped graphene materials synthesized by pyrolyzing a composite of GO and PANI were found to have low OER activity, with a current density of less than 2.0 mA∙cm^-2 at an overpotential of 500 mV. Despite this, researchers have continued to explore this nanomaterial and have developed new methods for preparing N-doped C bifunctional electrocatalysts. For instance, Lu et al. used a two-step in situ growth method to create an air electrode using an N-doped CNT array coated with Co NPs on the surface of a C fiber cloth, resulting in a C-based electrode with high OER and ORR activities^[227]. Murugesan et al. developed a low-cost C-based electrocatalyst by introducing highly porous C (HPC) from waste leaves^[228]. To enhance its catalytic activity, HPC was subjected to treatment with urea (CO(NH₂)₂) and thiourea (CS(NH₂)₂) and doped with N and S. The OER and ORR activities of HPC, N-doped HPC, and S-doped HPC were investigated, with N-doped HPC displaying the highest bifunctional electrocatalytic activity. Moreover, Zhao et al. developed an electrochemically active interface to promote ORR/OER kinetics by coupling N-doped submicron C tubes (N-SMCTs) with N-doped rGO (N-rGO)^[229]. The resulting N-SMCTs@N-rGO catalyst exhibited high bifunctional OER/ORR activity and durability, with an E_1/2 of 0.87 V for ORR and η₁₀ of 351 mV for OER. DFT revealed a dual-site mechanism, with the C adjacent to the graphitic-N in N-SMCTs being the most active site for the OOH intermediate and the C neighboring the p-N in N-rGO being beneficial for the adsorption of O/OH* intermediates. The unique active interface with a dual-site mechanism demonstrates the potential of N-SMCTs@N-rGO to overcome the energy barrier bottleneck in traditional bifunctional electrocatalysts.

In addition to N, other types of heteroatom-doped C nanomaterials have been widely researched^[230-232]. For example, Jiang et al. conducted a first-principles study of graphene and N-doped, B-doped, and codoped graphenes as potential electrocatalysts^[233]. Among the studied samples, B-doped graphene has the lowest overpotential, which indicates that it is the best electrocatalyst for the OER and ORR. However, in the presence of Li atoms, N, B-codoping does not enhance the OER/ORR performance, suggesting that the synergy due to protons does not apply to this system. This behavior is attributed to the fact that the presence of Li atoms can change the most stable adsorption sites and the adsorption energies of the intermediates. To improve the catalytic performance of CNTs, Wang et al. synthesized B-doped CNTs (BC₃NTs) with different B content^[234]. Raman spectroscopy revealed that the number of defects increased with increasing B content. The BC₃NT-based hybrid Li-air battery (HLAB) exhibits excellent cycling performance and catalytic performance in air, with an overpotential of 0.3 V for 200 h at a current density of 0.05 mA·cm^-2. Furthermore, DFT-based theoretical calculations confirmed the effect of topological defects in BC₃NTs on the ORR and OER. BC₃NTs containing numerous defect rings were obtained owing to the doping of B into the CNTs. The defect generation is beneficial for battery reactions. Therefore, BC₃NTs are highly favorable cathode materials for the OER and ORR in HLABs, and the topological defect loops generated by them enhance the catalytic performances of the OER and ORR.

Moreover, P is a widely studied dopant^[235]. Liu et al. developed efficient electrocatalysts for OER and ORR applications by modifying CDs with P and amidogen^[236]. These modified CDs, synthesized from the same environmentally friendly C-based nanomaterial, exhibit similar electrocatalytic activities and higher stability compared to benchmark metal electrocatalysts. Amidogen-terminated CDs display slightly higher OER activity than IrO₂, while P-terminated CDs show ORR activity similar to that of Pt/C. Lei et al. synthesized large-scale 2D P-doped C nanosheets (2D-PPCNs) with tunable porosity, using P pentoxide and common carbohydrates in a stepwise multiple templating process for 2D construction, P doping, and regulated micro/mesoporous generation, as illustrated in Figure 16^[237]. The impact of various porous structures on the electrocatalytic activity of 2D-PPCN-based electrocatalysts was studied. The interconnected open-pore system and large specific surface area resulted in high catalytic efficiency for both OER and ORR. This reliable technology for producing 2D porous C has the potential to be applied in various fields related to energy storage and conversion.

Figure 16. (A) SEM and TEM images of 2D-PPCN; (B) Schematic illustration of the zinc-air battery configuration; (C) Galvanodynamic charge/discharge polarization curves of 2D-PPCN and Pt/C+Ir/C based air electrode; (D) ORR activity at 1,600 rpm; (E) ORR polarization curves of 2D-PPCN-2/6 at various rotating speeds; (F) OER activity at 1,600 rpm; (G) Potential differences between the E_j = 2 of ORR and E_j = 10 of OER for all the catalysts. Reproduced with permission. Copyright 2018, ACS Publications^[236].

Bidoped: Bidoping of C frameworks with two heteroatoms can produce a synergy, increasing the catalytic activity; hence, bidoping has become an increasingly popular method for optimizing the electrocatalytic activity. The electrocatalyst performance can be significantly improved via various types of doping, especially codoping of N with other heteroatoms such as B, P, and S^[52,237-240]. In 2016, Qu et al. introduced an efficient and environmentally friendly method for incorporating S into C frameworks, resulting in the synthesis of N, S-codoped mesoporous C nanosheets that exhibited excellent performance, fast kinetics, and high durability as bifunctional OER and ORR electrocatalysts^[238]. The study suggests that a combination of high concentrations of multiple dopants, abundant porous structures, and good electron transfer ability can significantly enhance the catalytic processes of OER and ORR.

Recently, Sun et al. found that the P-N bond is favorable for OER, and N dopants only affect ORR activity^[241]. By tuning the heterogeneous atomic weights of C nanomaterials, they obtained high bifunctional OER/ORR activity. They synthesized the P, N-codoped graphene framework (PNGF) using GO and N/P-containing organic precursors via a hydrothermal method, demonstrating its high OER/ORR bifunctional activity due to the presence of P-N-bond-rich oxygen electrocatalyst. The combined configuration of P and N was characterized by X-ray photoelectron spectroscopy, as shown in Figure 17A-C. The electrocatalyst was further optimized by adjusting the ammonium dihydrogen phosphate (ADP) concentration and adding an N source [PNGF(op)] during synthesis, resulting in the highest content of P-N bonds and p-N and exhibiting excellent electrocatalytic performance. PNGF(op) showed an ORR half-wave potential E_1/2 of 0.85 V, an OER overpotential E₁₀ of 1.55 V, and a bifunctional potential difference of 0.70 V, outperforming Pt/C and Ir/C hybrid electrocatalysts, as shown in Figure 17D. The activity remained unchanged after 5,000 cycles, with stability higher than that of the Pt/C and Ir/C mixed electrocatalysts. Zhang et al. synthesized N, S-codoped graphene nanosheets (NSGs) using a one-step pyrolysis method with different N-containing precursors, namely 5-aminouracil (ANA), ammonium persulfate, and 2,5-dithiobisurea (DBA)^[242]. The NSGs derived from ANA and DBA demonstrated exceptional bifunctional activity for OER and ORR, with a ΔE of 0.73 V. These NSGs were dominated by p-N and pyrrolic N dopants. The performance of NSG was validated in a rechargeable Zn-air battery (ZAB), yielding a peak power density of 146 mW·cm^-2 and a specific capacity of 796 mA·h·gZn^-1. Notably, these values were higher than those achieved by the state-of-the-art Pt/C and IrO₂ (1:1 wt%) electrocatalyst for the same loading. These excellent properties were achieved by optimizing the OER or ORR intermediate energies through the configuration of S and different N species in the graphene nanosheets, depending on the specific S and N precursors used.

Figure 17. XPS (A) P2p and (B) N1s spectra for PNGF_DAP and PNGF_DAP_800; (C) Relative ratio of XPS N1s binding configurations for PNGF_DAP, PNGF_ADP, PNGF_ADP(op) and PNGF(op); (D) Bifunctional ORR/OER activities. Reproduced with permission. Copyright 2017, Royal Society of Chemistry^[31].

Tridoped: Codoping has been shown to introduce additional active sites into the C lattice, and researchers have proposed that adding more types of heteroatoms could further improve catalytic activity, leading to the development of tridoped C materials as bifunctional electrocatalysts^[243-246]. For instance, N, S, P-tridoped graphene (NSP-Gra) exhibits a higher ORR onset potential (0.95 V vs. RHE) than binary N, S-codoped graphene and undoped graphene, approaching the performance of commercial Pt/C electrocatalysts^[247]. This improvement in electrocatalytic performance is mainly attributed to local changes in the charge distribution and the abundant active sites generated by the heteroatom dopants. Li et al. synthesized a metal-free C bifunctional electrocatalyst (NFS-CNF) through direct pyrolysis electrospinning of PAN/polymeric ionic liquid nanofibers [Figure 18A]^[248]. The electrocatalyst exhibits uniform multiheteroatom (N, F, and S) doping and a large specific surface area (1,450.7 m²·g^-1). As shown in Figure 18B-E, the NFS-CNF exhibits excellent catalytic performance and high durability, with the ORR and OER associated with a positive half-wave potential of 0.91 V and a low overpotential of 380 mV at a current density of 10 mA·cm^-2, respectively, in an alkaline medium. Zheng et al. synthesized a piperazine-containing covalent triazine framework (P-CTF) using an ultrasonic-initiated route, in addition to N, F, S-tridoped C materials^[249]. The polycondensation and assembly of monomers into a nanoflower-like morphology were initiated using ultrasonic energy without the use of a template. The subsequent carbonization of the P-CTF resulted in N, P, F-tridoped porous C (NPF@CNFs) with a well-retained nanoflower morphology [Figure 18F and G]. The resulting NPF@CNFs exhibited higher electrocatalytic activity and stability than commercial Pt/C and IrO₂ electrocatalysts for ORR and OER, respectively [Figure 18H-M]. In recent years, various heteroatom-doped C nanomaterials have been developed using specific tools and toxic precursors. For example, N, P, S-tridoped C NSs (NPSCteamSs) have been synthesized via polymer derivative groups, template etching, and secondary pyrolysis, endowing the NSs with hollow and mesoporous structures, numerous defects, large specific surface areas, and other advantageous properties, which make a great contribution to OER and ORR performance^[250].

Figure 18. (A) SEM images of the (A-1) as-spun PDADMAFSI-PAN-PMMA-0.5 and (A-2) NFS-CNF-0.5-1000 nanofibers; (B) LSV curves of ORR at an RDE (1,600 rmp) in O₂-saturated 0.1 M KOH electrolyte; (C) LSV curves of OER without IR compensation; (D) LSV curves on a RDE (1,600 rpm) in 0.1 M KOH; (E) Potential gap (ΔE) between E_j=10 and E_1/2 for all the catalysts. Reproduced with permission. Copyright 2021, Elsevier^[247]; (F) Procedures of Ultrasound-Triggered Assembly of P-CTF Nanoflowers for Synthesizing NPF@CNFs; (G) Characterization of P-CTF nanoflowers: (G-1) SEM image of P-CTF nanoflowers, (G-2) SEM, (G-3) TEM; (H) LSV curves in O₂-saturated 0.1 M KOH; (I) Chronoamperometric responses in O₂-saturated 0.1 M KOH; (J) OER polarization curves in 1 M KOH at 5 mV·s^-1; (K) Chronoamperometric responses in 1 M KOH; (L) LSV curves between the ORR and OER; (M) Potential differences between the OER current density at 10 mA·cm^-2 and the E_1/2 of ORR. Reproduced with permission. Copyright 2021, ACS Publications^[248].

Transition-metal-C-based electrocatalysts

Although heteroatom-doped nonmetallic C catalytic nanomaterials have been proven as effective ORR/OER bifunctional electrocatalysts, they are prone to electrochemical corrosion under high potential conditions^[251]. And their electrocatalytic activity is strictly limited. Therefore, researchers have found that hybrid electrocatalysts prepared by strongly coupling heteroatom-doped C and metal oxides can moderately solve the problem of low stability of C-based nanomaterials. Moreover, C-based nanomaterials can increase the conductivity and dispersibility of metal oxides^[252,253].

(i) Transition-metal oxide/carbon composites: TMO/C composite can facilitate strong interfacial chemical interactions or electronic coupling between the two components, leading to increased OER and ORR activities^[253-255]. In 2011, Liang et al. demonstrated that the excellent OER/ORR performance of Co₃O₄/N-graphene electrocatalyst was due to the formation of Co-N-C and Co-O-C bonds, proposing the concept of an interfacial chemical interaction between Co₃O₄ and N-doped graphene^[256]. Since then, numerous efforts have focused on designing and developing metal oxide/C nanomaterials with different structures and compositions as bifunctional electrocatalysts^[257-261]. For example, Li et al. prepared a unique bifunctional electrocatalyst using a one-dimensional bimetallic (Co, Mn) ZIF pyrolysis method. The resulting product is a fibrous material with surface-attached particles, referred to as Co₃O₄/Mn₃O₄/CN_x@CNFs, as shown in Figure 19A^[262]. Strong coupling exists between the nanostructure and N-C bonds of each component in the material, resulting in a highly efficient catalytic performance for the ORR and OER [Figure 19B]. In particular, the ZAB assembled with Co₃O₄/Mn₃O₄/CN_x@CNFs demonstrated a high open-circuit voltage of 1.518 V, maximum power density of 265 mW·cm^-2, high cycling stability, and excellent mechanical flexibility. These properties surpass those of previously reported MOF-derived bifunctional electrocatalysts [Figure 19C]. Li et al. prepared a 3D flower-like composite electrocatalyst by hydrothermal synthesis, which contained NiCo₂O₄ and N-doped C derived from g-C₃N₄^[263]. The resulting NiCo₂O₄-CN electrocatalyst had a hierarchical structure composed of interconnected nanosheets and exhibited multifunctional bifunctional electroactivity for OER and ORR. The half-wave potential for the ORR was 0.81 V, and the overpotential for the OER at a current density of 10 mA·cm^-2 was 383 mV. The ZAB with a NiCo₂O₄-CN cathode demonstrated high performance, including a peak power density of 149.6 mW·cm^-2 and a capacity of up to 738.7 mA·h·g^-1, along with exceptional durability. These results highlight the potential of NiCo₂O₄-CN for use in electrochemical energy devices.

Figure 19. (A) Schematic illustration of the preparation process of Co₃O₄, Mn₃O₄ and CN_x supported on 1D carbon nanofibers (Co₃O₄/Mn₃O₄/CN_x@CNFs); (B) LSV curves of Co₃O₄, CNFs, Co₃O₄/CN_x@CNFs, Co₃O₄/Mn₃O₄/CN_x@CNFs and Pt/C for ORR; (C) LSV curves of RuO₂, CNFs, Co₃O₄, Co₃O₄/CN_x@CNFs and Co₃O₄/Mn₃O₄/CN_x@CNFs for OER in 1 M KOH at 1,600 rpm. Reproduced with permission. Copyright 2020, Elsevier^[261].

(ii) Transition metal (or alloy)/C composites: Although TMOs are promising, they have not been developed for practical application because of their low electrical conductivity^[264,265]. In comparison, transition metals and their alloys are expected to provide more advantages as bifunctional electrocatalysts due to their high electrical conductivities, in contrast to other materials^[266,267]. Therefore, the combination of metals or their alloys with heteroatom-doped C affords another type of C composite, which is of great interest for bifunctional electrocatalysts^[268]. An important type of electrocatalytic structure is the monometallic M-N/C, exemplified by Fe-N/C and Co-N/C^[269-273]. Quantum chemical calculations indicate that the electronegativity of N (3.04) is considerably higher than that of C (2.55). Consequently, the introduction of doped N atoms can create active electrocatalytic sites due to their strong electron affinity. This, in turn, enhances the positive charge density of the adjacent C atoms^[274-276].

For instance, Zhong et al. developed a single-atom Fe-N-C electrocatalyst by in situ carbonizing wood porous C after a simple Lewis acid pretreatment^[275]. The pretreatment of wood cell walls with FeCl₃ not only created numerous microchannels but also introduced atomically dispersed Fe-N reactive species into the hierarchical structures. The resulting electrocatalyst, consisting of uniformly dispersed SACs on a layered structure, exhibited enhanced performance and durability for OER and ORR. Recently, Wang et al. developed a NaCl-template-assisted strategy for 3D honeycomb N-doped C substrates to integrate nanoscale Co species, resulting in 3D Co/N-C bifunctional, highly active OER and ORR electrocatalysts, as shown in Figure 20A-F^[272]. The results demonstrated that 3D Co/N-C exhibits a positive half-wave potential of 0.84 V (vs. RHE) for ORR and a low overpotential of 1.56 V (vs. RHE) at a current density of 10 mA·cm^-2 for OER. Tang et al. utilized the inherent structural defects of nanocarbons to fabricate Co/N/O ternary doped graphene meshes with single-atom dispersed Co-N_x-C active sites^[277]. The electrocatalyst Co-NGM-Co possesses desirable bifunctional activities for OER and ORR in alkaline media, owing to the presence of Co-N_x-C sites, N dopants, oxygen functional groups, and topological defects. Upon assembly into a flexible solid-state ZAB, it demonstrates a high open-circuit voltage of 1.44 V, a small charge-discharge potential difference of 0.7 V at 1.0 mA∙cm^-2, and a high energy efficiency of 63% at the same current density [Figure 20G-M]. These results suggest that defect engineering is an effective and innovative strategy for developing high-performance M-N-C nanomaterials.

Figure 20. Morphology characterization of 3D Co/N-C. (A) SEM; (B) TEM; (C) Graphical abstract;(D) Schematic diagram of an aqueous zinc-air battery; (E) Discharge polarization lines of 3D Co/N-C and 20% Pt/C based air electrodes for the rechargeable ZABs and the correspondent power density curves of ZABs; (F) Galvanostatic discharging plots at a current density of 10 mA·cm^-2 of 3D Co/N-C and 20% Pt/C + IrO_2. Reproduced with permission. Copyright 2022, Elsevier^[271]; Morphology and composition characterization of NGM-Co catalysts. (G) SEM; (H) Dark-field high-resolution TEM image; (I) Charge and discharge polarization curves of the NGM-Co and Pt/C + Ir/C catalysts; (J) Galvanostatic discharge-charge cycling curves at 5.0 mA·cm^-2 of the rechargeable Zn-air batteries with the NGM-Co or Pt/C + Ir/C catalysts; (K) Discharge curves of Zn-air batteries assembled from the NGM-Co and Pt/C + Ir/C catalysts at 20.0 mA·cm^-2 discharging rate; (L) Photograph of a toy car powered by one Zn-air battery based on NGM-Co catalyst; (M) Long time cycling test at 2.0 mA·cm^-2. Reproduced with permission. Copyright 2017, Wiley Online Library^[276].

In addition, the catalytic activity of an electrocatalyst depends on its electrical conductivity; the higher the conductivity, the higher the ORR/OER activity. The bonding between two transition metals of dissimilar types can create intrinsic polarity, resulting in a unique reaction pathway that exhibits higher activity than their homometallic counterparts^[278,279]. Many researchers have explored NiCo NPs anchored on C-based materials. For example, Yu et al. demonstrated the excellent bifunctional ORR/OER performance of a B, N-codoped C nanosheet network (BNC) encapsulating Ni-Co alloy NPs (NiCo@BNC)^[280]. The potential difference between E_ORR1/2 and E_OER10 is remarkably small, measuring only 0.61 V. This value exceeds those reported for most C nanosheets decorated with either monometallic or bimetallic alloys. The exceptional performance of the electrocatalyst can be attributed to the unique role played by NiCo, which not only protects the nanosheets from destruction but also facilitates in situ B, N-codoping, thereby exposing numerous active sites. Moreover, the synthesis of many-layer BNC-shell-encapsulated NiCo alloys with high electrocatalytic activity and electrical conductivity further enhances the catalytic process.

Similar findings were reported for other synergistic electrocatalysts, such as FeCo/C^[281-284] and NiFe/C^[78,285]. For example, Liu et al. employed a micro-mesopore confinement technique to produce ultrasmall FeCo bimetallic alloy NPs with a size of less than 4 nm^[281]. These NPs were anchored on biomass-treated porous N-C, which was synthesized using peanut shells as precursors. The synergy between the FeCo alloy NPs and N-C resulted in the formation of FeCo-NC, which exhibited high bifunctional activity for OER and ORR in alkaline media. When utilized in ZABs, the hybrid electrocatalyst showed remarkable stability even after 1,400 cycles. At a current density of 10 mA·cm^-2, the ZABs delivered an ultrahigh energy density of 922 W·h·kg^-1.Jing et al. utilized waste biomass (pomelo peel, PP) as a raw material to synthesize NiFe-PBAs on its surface for in-situ growth^[284]. The resulting material, NiFe@NC/PPC (carbonized PP), was obtained by carbonization and used as a 3D binder-free and free-standing cathode in Li-O₂ batteries. NiFe@NC/PPC possesses superior features compared to PPC and NiFe@NC/CP (NiFe@NC coated on C paper), including effective diffusion rate at the O₂/electrolyte interface and high catalytic activity for the OER/ORR. A Li-O₂ battery with NiFe@NC/PPC as the cathode exhibits a high specific capacity of 13.79 mA·h·cm^-2 and a long cycle life of 290 cycles at a current density of 0.1 mA·cm^-2. Zhong et al. combined FeM (M=Ni, Co) particles with NPC to create FeM/NPC^[285]. The resulting material demonstrated high bifunctional activity and stability toward ORR (E_1/2 of FeCo and FeNi is 0.78 V and 0.73 V, respectively) and OER (η₁₀ of FeCo and FeNi is 0.46 V and 0.5 V, respectively). This was attributed to the synergy between FeM and NPC, which provided a high electrical conductivity, N doping, porous structure, and close contact between the two components. In addition to bimetallic alloys, trimetallic alloy electrocatalysts exhibit high bifunctional activity^[286]. Liu et al. synthesized FeCoNi nanoalloys embedded in a hierarchical N-C matrix (FeCoNi@HNC), which functioned as efficient and durable electrocatalysts for the OER and ORR in alkaline media^[287]. The core-shell structure of FeCoNi@HNC was found to be highly beneficial for regulating the adsorption/desorption of oxygen intermediates on the N-rich C shell through electron penetration effects, leading to higher ORR activity than Pt/C and higher OER activity than IrO₂, as demonstrated by both experimental studies and DFT calculations. Additionally, the N-rich C shell acted as a barrier, preventing excessive oxidation and dissolution of the FeCoNi alloy NPs, thereby providing high stability in alkaline electrolytes. Prototype ZABs constructed using FeCoNi@HNCs exhibited a maximum power density of 109 mW∙cm^-2 without a significant increase in the voltage gap, outperforming batteries fabricated using Pt/C and IrO₂/C electrocatalysts in terms of battery performance.

Recently, synergistic combinations of transition-metal sulfides, phosphides, and C-based materials have been proposed to improve the bifunctional electrocatalytic performance^[288-290]. Gong et al. prepared Co₂P/CoP hybrid NPs (Co₂P/CoP@NPGC) embedded on the surface of core-shell MOFs by direct pyrolysis of P-containing MOFs precursor^[291]. Liu et al. developed a hierarchical nanostructure, CoP-HNTs@NCL, which is composed of 1D hollow CoP nanotubes surrounded by a N-doped C layer^[292]. The thickness of the NCL coating on the outer surface of the hollow CoP-HNTs is 5-8 nm. The superior OER/ORR activity and stability of CoP-HNTs@NCL-0.4 can be attributed to the synergistic effect between CoP and NCL. This enhances mass transfer through the 1D hollow structures and increases the exposure of active sites on both the inner and outer surfaces of the nanotube walls. Moreover, Tiwari et al. synthesized an advanced bifunctional electrocatalyst by combining W disulfide (WS₂) with CNTs using a W carbide (WC) catalyst^[293]. The resulting electrocatalyst utilizes the catalytically active sites provided by the WS₂ flakes on the CNT surface, while the CNTs act as conduction channels to provide a large surface area for oxygen transport. Additionally, the newly formed WC crystal structure facilitates spin-coupled electron transfer, increasing both the stability and electrocatalytic activity of the material. Liu et al. prepared S-Co_9-xFe_xS₈@rGO, a hybrid of Co_9-xFe_xS₈/Co, Fe-N-C, by semivulcanization and calcination of GO-wrapped bimetallic CoFe-ZIF precursors^[294]. As shown in Figure 21A-E, owing to the synergy between the CoFeS and Co components, high dispersity, and high conductivity, the obtained S-Co_9-xFe_xS₈@rGO-10 electrocatalyst shows an ultrasmall overpotential of ~0.29 V at a current density of 10 mA·cm^-2 for OER and a half-wave potential of 0.84 V for ORR, exhibiting high oxygen electrode activity of ~0.68 V in 0.1 M KOH. Bimetallic phosphides have also been studied. Chen et al. prepared a novel structured bimetallic phosphide FeNiP-NP-embedded composite electrocatalyst in an N, P-bidoped C matrix^[189]. It is a trifunctional electrocatalyst that can efficiently catalyze the HER, OER, and ORR. The introduction of oleylamine during the preparation process can tune the catalytic sites, resulting in the desired catalytic performance. The electrocatalyst obtained in this study demonstrates superior ORR catalytic performance in comparison to commercial Pt/C electrocatalysts, as depicted in Figure 21F-J. Additionally, its OER performance is comparable to that of RuO₂. Furthermore, the electrocatalyst exhibits exceptional HER performance. In a 0.1 M KOH solution, the ORR half-wave potential measures 0.879 V (vs. RHE). Conversely, the OER overpotential at a current density of 10 mA·cm^-2 is 280 mV. The potential difference between the ORR and OER is only 0.700 V. This study establishes the groundwork for amalgamating composite electrocatalysts with bimetallic phosphides and hybrid C substrates, enabling the advancement of high-performance multifunctional electrocatalysts and their utilization in energy devices.

Figure 21. (A) Characterization of CoFe-ZIF@GO-10 and S-Co_9-xFe_xS₈@rGO-10: (A-1) SEM image of CoFe-ZIF, (A-2) TEM image of CoFe-ZIF@GO-10, and (A-3) TEM image of S-Co_9-xFe_xS₈@rGO-10; LSV curves of (B) OER, (C) ORR, (D) the overall polarization in ORR and OER region, and (E) the ΔE of S-Co_9-xFe_xS₈@rGO-10 (I), M-Co_9-xFe_xS₈@rGO-10 (II), Co,Fe-N-C@rGO-10 (III), Co_9-xFe_xS₈@rGO-10 (IV), Pt/C (V), and IrO₂ (VI) in 0.1 m KOH. Reproduced with permission. Copyright 2018, Wiley Online Library^[293]; (F) Characterization of FeNiP@C-200: (F-1) SEM, (F-2) TEM and (F-3) HRTEM images; (G) LSV curves of various catalysts at 1,600 rpm; (H) OER polarization curves of various catalysts at 1,600 rpm tested in 1 M KOH solution with a scan rate of 10 mV·s^-1; (I) OER-ORR polarization curves tested in 0.1 M KOH solution; (J) ΔE (OER_j=10 - ORR_E=1/2) in 0.1 M KOH solution. Reproduced with permission. Copyright 2021, Royal Society of Chemistry^[189].