The interface between the metal center and the support will cause the re-arrangement of electrons originating from the support and anchored atoms [69,73]. The re-arranged electrons that have a significant impact on the catalytic performance will be confined in a space several atomic layers thick at the interface [74]. The magnitude and direction of the charge transfer are driven by differences in the Fermi level of the catalytic center and the support [50]. In addition, due to the special microenvironment at the interface, the interface sites will be in direct contact with the catalytic center, the carrier, and the reactants in order to promote the occurrence of synchronous reactions [37]. Additionally, the interface is conducive to the accumulation of excess charge during the charge transfer process, which will strongly promote the catalytic reactions at the interface [75].

SnO₂ and TiO₂-based compounds are commonly used as supports due to their characteristic of stability [76,77]. In order to improve their low charge-transfer rate, extensive studies have modified them to improve the conductivity. For SnO₂, Sb doping is often used to prepare Sb-SnO₂ (ATO), and the conductivity of the material will be improved due to the increased electron carrier density caused by donor doping [78]. Moreover, the specific surface area and pore volume of SnO₂ can be improved by destroying the long-range order of the original atomic arrangement so as to provide more anchor sites for IrO₂ nanoparticles [77]. The interaction between IrO₂ and the support, the cross-linking morphology of IrO₂, and the porous structure can improve the OER performance of the catalyst. Wang’s group designed a kind of Sb-SnO₂ nanowire carrier by an electrospinning method (Figure 3a) [37]. The conductivity can reach 0.83 S·cm⁻¹. Compared with pure IrO₂, the catalytic activity of supported IrO₂/Sb-SnO₂ exhibits significantly improved mass activity, benefiting from the porous structure and the high electronic conductivity of the Sb-SnO₂ support [37].

In addition to metal oxides, metal carbides have also emerged as promising OER carriers because of their high conductivity and stability. A TaC-supported IrO₂ catalyst sprayed by Polonsky et al. showed the lowest charge transfer resistance and the highest current density when the loading of IrO₂ reached 70 wt%, which was significantly improved compared with unsupported IrO₂ (Figure 3b) [79]. When TiC is employed as the support for Ir in PEMWE, Ir nanoparticles can be evenly distributed on the TiC surface, and the pore volume of Ir/TiC is twice that of pure Ir. All these advantages make the OER catalytic performance of Ir/TiC much better than that of pure Ir [80].

In general, the existence of a support has two major advantages. On one hand, catalyst particles can be well dispersed on the support surface and facilitate the construction of a three-phase interface consisting of the catalyst, the reactant water molecule, and the produced oxygen [81]. In fact, this effect has been widely used to explain the increased activity of supported catalysts. For example, Ir nanoparticles can be well dispersed onto the TiN carrier. The IrO₂@Ir/TiN catalyst prepared by Xing’s group showed an enhanced catalytic performance [50]. The overpotential was only 265 mV at a current density of 10 mA·cm⁻². Yang et al. synthesized iridium dioxide nanoparticle catalysts with α-MnO₂ nanorods as supports by a simple two-step hydrothermal method. They found that iridium dioxide nanoparticles were subjected to compressive strain due to the lattice mismatch between IrO₂ and α-MnO₂ (Figure 3c) [45].

Incorporation of heteroatoms or groups will destroy the periodicity of the lattice, resulting in the modification of the local coordination environment and the electronic structure of active sites [35,82]. This change can effectively regulate the adsorption energy of reaction intermediates and improve the intrinsic activity of electrocatalysts. One of the effective strategies is to incorporate easily soluble non-noble metals, which suffer in-situ dissolution to form an amorphous structure, and increase the degree of coordination unsaturation of the metal in the active center during the oxygen evolution reaction [67]. Zaman et al. chose Ni and Co as the dopants to substitute 50% of the precious metal Ir. The Ni-Co co-doped IrO₂ showed a low overpotential of 285 mV at a current density of 10 mA·cm⁻² (Figure 4a) [83]. Besides Ir-based catalysts, researchers have also done a lot of work on Ru-based catalysts [84]. For example, SrRuO₃ exhibited low OER activity in acid electrolytes due to rapid Sr leaching and Ru dissolution at high potential ranges in 0.1 M HClO₄ [85]. Surprisingly, incorporating a small amount of Na⁺ into the lattice of SrRuO₃ by substituting Sr²⁺ results in significantly enhanced OER performance, both in terms of activity and stability (Figure 4b) [86]. The doped sample exhibited 85% of activity retention after a stability test, which was assigned to the stabilization of Ru centers with a positive shift in dissolution potentials and less distorted RuO₆ octahedra.

Preparation of a perovskite ABO₃ structure (or A₂BB′O₆) and regulation of the valence band structure of B-site cations (usually Ir and Ru) by the atoms at A-site cations have commonly been used to improve the performance of the OER in recent years. Catalysts with this perovskite structure can greatly reduce the usage of noble metals. Pseudocubic SrIrO₃ was the first AIrO₃ single perovskite oxide reported for usage as an OER electrocatalyst in acid media [87]. It was found that the formation of IrO_x by Sr leaching on the surface of SrIrO₃ through surface reconstruction was responsible for the enhanced OER activity (Figure 4c). The active surface area increased significantly owing to the formation of IrO_x groups on the surface of catalysts during cycling. Moreover, the intrinsic activity of Ir in the SrCo_0.9Ir_0.1O_3-δ electrocatalyst was more than two orders of magnitude higher than that of IrO₂ and approximately 10-fold higher as compared with the unmodified benchmark SrIrO₃ [88]. The observed high activity was attributed to the surface reconstruction through Sr and Co leaching during the electrochemical cycling, which contains a large number of oxygen vacancies with corner-shared and under-coordinated IrO_x octahedrons in the oxide crystal lattice. Furthermore, surface reconstruction of AIrO₃ (A = Sr or Ba) perovskite oxides was visualized by atomic-resolution scanning transmission electron microscopy (Figure 4d) [89]. Leaching of Sr or Ba happened by applying an anodic current, resulting in surface roughening and a structure change by the continuous formation of the mosaic-shaped, Sr-deficient IrO_x on the surface. Moreover, the lattice strain induced by the substitution of smaller lanthanides or yttrium atoms can effectively decrease the adsorption energy toward oxygen-containing intermediates [51]. Although promising in terms of providing enhanced activity, such a surface reconstruction during the electrocatalysis process may be challenged by the uncertainty of the exact structure formed during reactions [87].

It should be mentioned that different kinds of crystal structures have a significant impact on the OER performance. It was proposed that monoclinic SrIrO₃ underwent less surface reconstruction than the pseudocubic SrIrO₃ owing to the better thermodynamic stability during OER tests in an acidic electrolyte [90]. Strong Ir–Ir metallic bonding and Ir–O covalent bonding in monoclinic SrIrO₃ together induced its high structural and compositional stability. Only about 1% of the leached Sr was detected after 30 h in a chronopotentiometry test, which was much less than that (24%) from the pseudocubic SrIrO₃. Additionally, Zou et al. put forward another way to decrease the cation leaching and surface reconstruction of pseudocubic SrIrO₃ in acid media. They prepared pseudocubic, low-Ir-containing SrIr_xTi_1−xO₃ perovskite oxides with 0 ≤ x ≤ 0.67. The inert Ti sites in pristine SrTiO₃ were activated by Ir-substitution and showed remarkable OER activity with a reserved crystal structure of Ir-SrIrO₃ during OER cycling [91].

For metal alloy/oxide-based electrocatalysts, the surface oxidation of metal atoms accompanied by de-alloying (surface dissolution of unstable metals) under acid OER conditions is considered to be a “surface engineering” strategy to design stable and efficient OER electrocatalysts. For example, Travis Jones and Peter Strasser found that electrochemical de-alloying and surface oxidation treatment of IrNi_3.2 nanoparticle precursors can lead to the dissolution of Ni, and the formed Ni nanoparticles showed better catalytic performance than core–shell-structured IrNi@IrO_x [57]. In order to investigate the relationship between the reconstructed structure and the enhanced OER performance, an operando X-ray absorption spectroscopy (XAS) analysis was performed to characterize the local electronic properties under the OER process. The results show that iridium titania and d-band holes appear in the IrNiO_x electrocatalyst when the potential increases from 0.4 to 1.5 V_RHE. More importantly, due to the higher oxidation state of iridium, the iridium oxygen bond length in IrNiO_x was significantly shortened. Based on this unique phenomenon, a structural model of the iridium oxygen ligand environment was proposed. The model shows that the hole-doped iridium ion sites around the electrophilic oxygen ligands form hole-doped IrO_x (caused by Ni leaching) during the OER. Therefore, more electrophilic oxygen ligands are susceptible to the nucleophilic attack of water molecules or hydroxyl ligands, resulting in the formation of oxygen bonds and the reduction of the kinetic energy barrier, which ultimately greatly improves the reaction activity.

Although great efforts have been made to improve the efficiency of OER electrocatalysts, the majority of active sites inside their bulk phases remain inaccessible [23]. In order to maximize the utilization of each active site (approaching 100%) as well as shed light on the effect of the structure of active centers and ligating atoms on the OER activity, catalysts have thus been continuously downsized to the single-atom (SA) level. Single-atom catalysts (SACs) have emerged as a hot new branch of heterogeneous catalysts due to their excellent catalytic performance and financial benefits [28]. Owing to the high requirements for the dispersion, activity, and stability of the atoms, an appropriate support must be selected to optimize the physiochemical properties of the metal atoms anchored [81]. The unique coordination configuration of the dispersed metal atoms and coordinating atoms together create active sites for OER catalysts; therefore, the coordination−activity relationship has a strong impact on the catalytic performance [92]. In fact, targeted synthesis of precious metal single-atom-based OER electrocatalysts remains a bottleneck and has been pursued in the search for better OER catalysis, especially under acid conditions [28]. Generally, carbon supports often suffer severe corrosion problems under acid conditions and oxide supports are not always conductive, which brings about huge trouble for the design of single-atom catalysts [93]. In order to solve these problems, a series of Pt-Cu alloys with an atomically dispersed Ru₁ decoration were studied by Yao et al. [94]. An ultralow overpotential of 170 mV at a current density of 10 mA·cm⁻² was reached by Ru₁-Pt₃Cu in acid media, together with a ten times longer lifetime than a commercial RuO₂ catalyst. Density functional theory calculations suggested that the electronic structure of single Ru sites at the corner or step sites of the Pt-rich shell was modulated by the compressive strain in the Pt skin shell, contributing to the optimized binding of oxygen species and improved resistance to over-oxidation and dissolution. Yin et al. demonstrated that surface-exposed Ir single atom couplings with oxygen vacancies anchored in ultrathin NiCo₂O₄ porous nanosheets exhibited remarkable OER activity and stability in acid media, with an overpotential of only 240 mV at a current density of 10 mA·cm⁻² and a long-term durability of 70 h [28]. Based on density functional theory calculations, high electronic exchange and transfer activities of the surface contributed by Ir atoms anchored at Co sites near oxygen vacancies were determined to be responsible for the prominent OER performance (Figure 5). The synergetic mutual activation between Ir and Co reached the desired H₂O activation level and stabilized *O to boost OER performance.

The interface between the catalyst and the electrolyte plays an important role in water electrocatalysis [95]. As a very important aspect of a surface structure, morphology has attracted much attention in recent years [23,96]. To date, many methods to control the morphology have been proposed [23,62].

The morphology of the catalyst can be modified by adopting a suitable preparation strategy, such as the template method, the solvothermal method, or the seed crystal method [97,98]. A nano-porous or ultra-thin structure can increase the number of exposed active sites. Luo et al. prepared a new kind of Ir nanowire with a diameter of 1.7 nm by a wet chemical method (Figure 6a) [62]. Due to the high aspect ratio and large specific surface area, the OER activity increased greatly. The overpotential at 10 mA·cm⁻² in 0.5 M HClO₄ is only 270 mV, which is significantly higher than that of Ir nanoparticles.

In fact, the electronic structure of catalytic centers can be modified by defects as well as the morphology. For example, Qiao et al. prepared a core–shell Ru@IrO_x icosahedral nanocrystal structure by the sequential polyol method, in which a highly distorted lattice can be observed (Figure 6b) [99]. Due to the interaction between the ruthenium core and the iridium shell, ruthenium is subjected to compressive strain, which was confirmed by the decrease in the distance between ruthenium and ruthenium atoms observed by EXAFS. This strain may lead to a shift in the d-band center, which can regulate the binding energy of oxygen intermediates and the activity of the OER. In addition to the lattice strain, morphology control can also lead to changes in the chemical composition, which directly affects the electronic structure. An example is the ultra-thin ruthenium oxide nanosheets prepared by Lotsch’s group (Figure 6c) [100]. Owing to the acid treatment in the stripping process, the actual composition of ruthenium oxide is H_xRuO₂, in which the valence of ruthenium is +3/+4. Therefore, ruthenium oxide nanoparticles exhibit enhanced OER activity and stability.