Advanced electrocatalysts with unusual active sites for electrochemical water splitting

Abstract

Electrochemical water splitting represents a promising technology for green hydrogen production. To design advanced electrocatalysts, it is crucial to identify their active sites and interpret the relationship between their structures and performance. Materials extensively studied as electrocatalysts include noble-metal-based (e.g., Ru, Ir, and Pt) and non-noble-metal-based (e.g., 3d transition metals) compounds. Recently, advancements in characterization techniques and theoretical calculations have revealed novel and unusual active sites. The present review highlights the latest achievements in the discovery and identification of various unconventional active sites for electrochemical water splitting, with a focus on state-of-the-art strategies for determining true active sites and establishing structure–activity relationships. Furthermore, we discuss the remaining challenges and future perspectives for the development of next-generation electrocatalysts with unusual active sites. By presenting a fresh perspective on the unconventional reaction sites involved in electrochemical water splitting, this review aims to provide valuable guidance for the future study of electrocatalysts in industrial applications.

INTRODUCTION

Electrochemical water splitting is an energy conversion technology that allows renewable energy (e.g., wind, tide, and solar energy) to be stored in the form of hydrogen.^1,2 Because water electrolysis involves an oxygen evolution reaction (OER) at the anode and a hydrogen evolution reaction (HER) at the cathode, advancements in OER and HER electrocatalysis are crucial for improving the efficiency and reducing the costs of water electrolyzers.^3–6 However, the scaling relationships associated with multi-reaction intermediates during the OER contribute to intrinsically high overpotentials in the designed electrocatalysts.^7–14 Several design principles have been developed to address this, including structural engineering techniques, such as nanostructuring and porous structuring to increase availability of surface-active sites, and surface engineering, compositional regulation, and structural modification techniques, such as defect creation, heteroatom doping, phase control, and heterostructure engineering, to improve the intrinsic catalysis of active species.^3,15–22 Ultimately, the rational design of advance electrocatalysts necessitates elaborate modulation and a fundamental understanding of catalytically active sites.

Conventional catalysis typically relies on well-established active sites such as metal ions or specific functional groups. Current state-of-the-art electrocatalysts for the OER are predominantly Ru/Ir oxide-based materials, whereas Pt-based materials are widely recognized as benchmark electrocatalysts for the HER.^23–25 However, the low abundance, high cost, and poor stability of precious metals limit their practical use in large-scale applications. To minimize the reliance on noble metals, it is crucial to identify earth-abundant alternatives that would considerably facilitate the development of renewable energy technologies. Moreover, in alkaline electrolytes, the OER activity of transition-metal-based materials is superior to that of noble-metal-based electrocatalysts, whereas the HER activity of sulfides and phosphides, such as MoS₂, is comparable to that of commercial Pt/C catalysts.^26,27 Consequently, various materials consisting of 3d transition metals, such as Mn, Fe, Co, and Ni, have been reported to exhibit excellent OER and HER electrocatalytic properties.^12,28–38 Conventionally, 3d transition metals and noble metals have been considered as true catalytic active sites, particularly high-valence metal sites in the pre-catalysts or those formed in situ under OER conditions via surface reconstruction.^3,12,39,40 However, advanced characterization techniques have revealed previously unrecognized active sites with unique structures and reactivity, deepening the understanding of reaction mechanisms and further guiding the design of electrocatalysts. Furthermore, these findings are based on rigorous methodologies and supported by reasonable calculations, as in situ/operando characterizations and theoretical calculations are some of the most powerful tools for identifying active sites during electrochemical reactions.^8,28,41–46 However, few reviews have been conducted on this body of research, indicating a possible gap in current knowledge or understanding.

The purpose of this review is to offer a comprehensive and up-to-date summary of unusual active sites for the OER and HER. First, we provide fundamental descriptions of both conventional and unusually active sites. To confirm the dynamic reconstruction of active sites and interpret the reaction mechanisms, we examine a combination of in situ/operando characterizations (experimental methods) and theoretical calculations (computational simulations). Next, we summarize recent advancements in the development of electrocatalysts with unusual active sites. Finally, we discuss forthcoming challenges that lie ahead and present the future scope of this field. We hope that this review will contribute to the identification of active sites and guide the rational design of advanced electrocatalysts for water electrolysis.

CONVENTIONAL AND UNUSUAL ACTIVE SITES Conventional active sites

Noble metals such as Ru, Rh, Pd, Ir, and Pt are employed as benchmark catalysts for a wide range of electrochemical reactions owing to their moderate chemical bond strength with electrocatalytic reaction intermediates.²¹ For example, when using hydrogen adsorption free energy as the descriptor for HER, Pt is situated near the summit of the hydrogen volcano, indicating its outstanding HER activity (Figure 1A).⁴⁷ In terms of experimental results, Pt requires negligible overpotentials to achieve high reaction rates (Figure 1B). Meanwhile, 3d transition metal-based materials such as Co-, Fe-, and Ni-based compounds exhibit relatively high electrochemical activity for water electrolysis owing to their distinctive electronic properties, including flexible valence states, appropriate electronic configurations, and versatile compositions.²⁸ In contrast to the HER, which has only one reaction intermediate, the OER, which involves four-electron transfer, has multiple intermediates with strongly correlated binding energies that cannot be easily disentangled owing to the presence of scaling relationships. This presents a considerable challenge in optimizing individual reaction intermediates for the design of advanced OER electrocatalysts. However, scaling relationships enable a rapid transition in electrocatalyst screening from a trial-and-error approach to a theory-guided approach. Additionally, the adsorption energy calculated by density functional theory (DFT) computations allows the establishment of a catalytic activity trend for a series of electrocatalysts in “volcano–plot activity relationships”. Transition-metal-based oxides exhibit outstanding OER activity (Figure 1C,D). Emerging catalytic design strategies such as the generation of surface defects, low-coordination sites, and disordered lattices have been documented extensively. Furthermore, active-site engineering in newly designed materials may enhance electrocatalytic activity, contributing to the optimization of adsorption energies and reaction barrier for intermediates and electrocatalytic reactions, respectively.⁴⁸

FIGURE 1. (A) HER volcano plot for metals and MoS2. (B) Turnover frequency with linear sweep voltammetry curves of various HER electrocatalysts. (C) OER volcano plot for metal oxides. (D) Chronological trend in overpotentials of various OER electrocatalysts in acidic and alkaline media. Reproduced with permission.47 Copyright 2017, American Association for the Advancement of Science.

Unusual active sites

The exploration and characterization of previously unknown or underappreciated sites that contribute to the electrocatalytic process may expand the knowledge based on water-splitting reaction mechanisms and open new possibilities for the development of electrocatalysts with enhanced electrochemical performance. Unconventional active sites can provide alternate reaction pathways with lower activation barriers, facilitating reactions that are typically restricted by high energy requirements. There active sites may be unique owning to the presence of new materials, novel arrangements of atoms, or defects in the electrocatalytic structure. For example, recent studies have shown that defects such as vacancies or doping in the electrocatalyst structure can substantially enhance electrocatalytic performance by facilitating charge transfer and improving the adsorption and activation of reactants.^49–51 Thus, research on unconventional active sites in water splitting has yielded valuable insights into the electrocatalytic mechanism, providing excellent prospects for the further development of electrocatalysts. The advanced experimental and computational techniques used in these studies have considerably expanded our understanding of electrocatalytic and may lead to further breakthroughs in this field.

IDENTIFICATION OF ACTIVE SITES

Active sites are identified using advanced experimental techniques such as in situ/operando spectroscopy and microscopy, and computational simulations are used to predict their behavior and interactions with reactants and intermediates. In situ spectroscopy techniques, such as x-ray absorption spectroscopy (XAS) and Fourier-transform infrared spectroscopy, provide insights into the electronic and structural changes that occur during catalytic processes.^52,53 In situ microscopy techniques, such as scanning tunneling microscopy and transmission electron microscopy, provide real-time observations of the electrocatalytic structure and activity under working conditions.⁵⁴ Computational simulations, including DFT and molecular dynamics simulations, are invaluable tools for identifying and interpreting unusual active sites. These simulations predict the behavior and interactions of reactants and intermediates with the electrocatalyst surface to interpret the reaction mechanism and kinetics.

In situ/operando characterizations

The discovery and identification of true active sites can help clarify reaction mechanisms and guide the design of efficient electrocatalysts. Conventional methods for active site identification include the preparation of single-crystalline materials of different sizes, qualitative and quantitative comparisons of catalytic performance, and ex situ characterization techniques. However, most electrochemical reactions involve transfers of multiple electrons. The complexity and uncertainty inherent to these processes necessitate highly advanced and effective characterization techniques to provide direct correlations between active sites and observed activities. Advanced in situ/operando characterization techniques enable the detection of the surface and solid/liquid interfaces of catalysts as well as dynamic surface reconstruction during electrochemical reactions. Moreover, combining such techniques facilitates the comprehensive study of the structures, morphologies, and electronic states of electrocatalysts, reaction mechanisms, and products under given reaction conditions.^55–57

Recently, NiFe-based materials have demonstrated great potential as OER electrocatalysts. A comprehensive understanding of factors influencing OER activity has been achieved through in situ/operando studies conducted on NiFe-based OER electrocatalysts.^45,58–62 In the present review, we focus on Cu-based materials, which are not as commonly employed for OER catalysis. Our objective is to underscore the pivotal importance of in situ/operando characterization in enhancing our understanding of the less-utilized Cu-based electrocatalysts. Cu-based materials have received considerable research interest as electrocatalysts for CO₂ reduction and photocatalytic water reduction operations.^63–65 In contrast to other oxides containing 3d transition metals (such as Fe, Co, and Ni), Cu-based oxides typically exhibit unsatisfactory OER activity. Considering the high abundance and low cost of Cu, the investigation of effective strategies to improve the OER kinetics of these materials and determinate the related active sites is a challenging but urgent task. Recently, Peng et al. proposed that Cu₂O oxide with unsaturated coordination via facile hydrogenation can exhibit enhanced catalytic activity and stability toward the OER.⁶⁶ Particularly, Cu₂O with a low valence state was not expected to demonstrate high OER activity owing to its limited reactivity.³ However, following rational regulation of the Cu center, two steps of surface reconstruction were observed based on multiple operando spectroscopic techniques. For example, based on operando soft XAS (Figure 2A) at the Cu L₃- and O K-edges, an initial Cu¹⁺ (3d¹⁰) state was identified, and a sharp peak at 931.3 eV, ascribed to the Cu²⁺ state from Cu(OH)₂, was observed at an applied voltage of 1.4 V. Furthermore, a high broad shoulder feature at 932.4 eV was observed under a higher applied voltage, resulting from the doped hole for the Cu³⁺ state with the 2p⁵3d¹⁰L final state. Notably, the Cu³⁺ state was not observed, and only Cu²⁺ remained when the applied voltage was switched off (Figure 2B). The reaction intermediate Cu³⁺ was also observed in the O-K XAS spectrum (Figure 2C). To capture the real-time surface changes of catalysts, an operando rapid XAS technique was applied (Figure 2D). This unusual observation is markedly different from that of the commonly proposed scheme, which cannot be interpreted entirely through ex situ experiments. This study also highlights the importance of regulating the charge and spin states of transition-metal oxides for high-performance OER electrocatalysts. Based on a series of operando X-ray spectroscopy results, the covalence between Cu 3d and O 2p increased from Cu¹⁺ to Cu²⁺, and further increased to the Cu³⁺ state, leading the O 2p character to gain weight about E_F and shift near E_F. Thus, Cu³⁺ contributes to the high performance of the designed electrocatalysts.

FIGURE 2. (A) Schematic of operando soft-XAS for three-electrode electrochemical reactions. (B) Cu L3-edge and (C) O K-edge of the active Cu center with unsaturated coordination. (D) Schematic of operando liquid electrochemical cell for in situ x-ray diffraction spectra and in situ XAS spectra. Reproduced with permission.66 Copyright 2023, Nature Publishing Group.

Theoretical calculations

Based on a comprehensive understanding of the electronic structures before and after electrochemical reactions, DFT calculations can provide further insights into the relationship between the electrochemical behavior and physical/chemical properties of catalysts.⁶⁷ The inherent surface atomic and electronic structures considerably affect the intrinsic activity of the active species. The adsorption strengths of intermediates reflect their electronic structural properties, enabling the direct determining of reaction kinetics. Therefore, the Gibbs free energies of reaction intermediates can be effectively used as parameters to demonstrate the catalytic behavior, whereby the structure–activity relationship is built successfully.⁶⁸ Moreover, the energy band structure and density of states intrinsically characterize electronic properties. Descriptors to correlate structures with electrochemical activity have also been widely established.⁸ For example, Shen et al. designed S-doped Co_0.85Se (Co_0.85Se_0.5S_0.5), which demonstrated an exceptional HER activity with an overpotential of 108 mV at 10 mA cm⁻².⁶⁹ DFT calculations were performed to demonstrate the role of S-doping in improving intrinsic catalytic activity, and the hydrogen adsorption Gibbs free energy (∆G_H) was used to evaluate the HER activity. The values of ∆G_H at S sites (0.037 eV) were substantially lower than those at the Co and Se sites before and after S-doping (Figure 3A,B), suggesting that the S sites serve as active species. Moreover, electrons transfer from H to S occurred owing to the higher electronegativity of S (Figure 3C,D). The reaction barriers of catalysts before and after S-doping in the HER process were calculated. The energy barrier for the HER of Co_0.85Se was 0.620 eV, whereas that of Co_0.85Se_0.5S_0.5 was only 0.220 eV (Figure 3E,F). These results strongly confirm that the active sites were converted from conventional Co sites to unusual S sites, and the enhanced HER activity can be ascribed to the optimization of the corresponding energy barrier. Therefore, this study offers an effective approach for optimizing intrinsic activity, and presents comprehensive evidence for the identification of unusual active sites.

FIGURE 3. ∆GH on different sites of (A) Co0.85Se and (B) Co0.85Se0.5S0.5. Charge transfer difference for (C) Co0.85Se and (D) Co0.85Se0.5S0.5. Reaction pathway of the HER for (E) Co0.85Se and (F) Co0.85Se0.5S0.5. Reproduced with permission.69 Copyright 2021, Wiley-VCH.

ENGINEERING OF ACTIVE SITES

After an element has been confirmed as an active site, design strategies such as morphological tuning, crystal facet engineering, single-atom dispersion, and composition regulation can be deployed to maximally expose the surface-active sites.^70–76 To ensure sufficient catalytic activity, the following requirements are defined for catalytically active species: (1) The active species must exhibit high intrinsic activity toward a specific electrochemical reaction. (2) The active species are predominantly exposed on the catalytic surface. (3) The active species must be easily accessible to the reactant molecules.^12,77–82

RECENT ADVANCES IN EXPLORATIONS OF UNUSUAL ACTIVE SITES Beyond noble metal sites

For Ru-based alloys, the introduction of other metals can further modulate the electronic structure of active sites by exerting ligand effects and chemical strain. Consequently, the optimal hydrogen bond strength for Ru has been obtained for H adsorption and H₂ desorption. For example, Mo-modified Ru sites exhibit enhanced HER activity.⁸³ As an electron donor to Ru sites, Au can also regulate the Ru-H binding energy.⁸⁴ In these cases, Ru typically serves as a highly active site. In contrast, Shen et al. reported that Si sites in LaRuSi may serve as HER active species, as confirmed through both theoretical and experimental analyses.⁸⁵ LaRuSi has exhibited superior HER activity in alkaline solutions with a low overpotential of 72 mV at 10 mA cm⁻² (Figure 4A,B). The adsorption free energy of atomic hydrogen (ΔG_H*) was calculated to evaluate HER activity. Values of ΔG_H* for the Ru and La sites were −0.574 and 0.224 eV, respectively, indicating that hydrogen exhibits a high affinity toward Ru sites and a low affinity toward La sites. The Si sites exhibited a low ΔG_H* value of 0.063 eV, reflecting their status as the true active sites (Figure 4C). The d-band center of Ru shifted upward owing to the incorporation of La based on RuSi, leading to an enhanced Ru-H affinity (Figure 4D). In situ Raman spectroscopy was conducted to further verify the active sites of Si for the HER and validate the robust adsorption of H by the Ru sites (Figure 4E). Thus, outstanding HER electrocatalysts were developed and a new research platform was demonstrated for the rational design of such electrocatalysts.

FIGURE 4. (A) HER polarization curves and (B) Tafel plots of the unannealed sample (LaRuSi-ua), LaRuSi, and Pt/C. (C) Adsorption free energy of hydrogen at various sites in LaRuSi. (D) D-projected density of states of Ru atoms in RuSi and LaRuSi. (E) In situ Raman spectra of LaRuSi under HER conditions at different potentials. Reproduced with permission.85 Copyright 2022, Wiley-VCH. (F) HER polarization curves and (G) mass activities at different overpotentials of IrMo0.59 nanoparticles, Ir nanoparticles, and Pt/C. (H) Calculated Gibbs free energy diagram of various HER intermediates on Ir and Ir3Mo1 (H2O). (I) Proposed alkaline HER mechanism of Ir3Mo1 (H2O). Reproduced with permission.86 Copyright 2020, American Chemical Society.

A similar finding has been reported for Ir-based materials. Fu et al. synthesized IrMo alloy nanoparticles, with IrMo_0.59 exhibiting the highest HER activity, approximately 10 times higher than that of commercial Pt/C (Figure 4F,G).⁸⁶ Subsequent DFT calculations demonstrated that the Mo sites of IrMo exhibit the highest stability for the adsorption of H₂O, suggesting that H₂O molecules tend to readily occupy Mo sites on the surface. Unfavorable HER kinetics at the Ir sites can be inferred from the endothermic nature of the first and second steps (Figure 4H). In contrast, Ir₃Mo₁ (H₂O) displayed a downward trend, indicating that the adsorption and dissociation of water molecules on IrMo (H₂O) are considerably more favorable than those on Ir. These results indicate that the improved HER performance of IrMo (H₂O) is primarily attributed to the synergistic effect of Mo (H₂O) and Ir, which effectively modulates the adsorption and dissociation strengths of water, thereby promoting the Volmer–Heyrovsky steps during the catalytic process (Figure 4I).

Beyond 3d transition metal sites

Early 3d transition metals (e.g., Ti and V) and 4d transition metals (e.g., Nb and Mo) generally exhibit lower OER activity than late 3d transition metals (e.g., Fe, Co, and Ni).¹² Moreover, the electronic structures of late 3d transition metals and noble metals can be effectively modulated using 4d/5d transition metal dopants to satisfy the requirements of electrochemical reactions.^87–90 In particular, 4d/5d metals have a large d-electronic wave function spatial extent, which can provide prominent electronic structures via the interaction between 4d/5d and 3d orbitals for enhanced electrochemical activity. For example, Zhang et al. designed gelled FeCoW oxyhydroxides with an atomically homogeneous metal distribution.⁸⁷ The remarkable OER activity was attributed to the synergistic interplay between W, Co, and Fe by the formation of a favorable electronic structure and local coordination environment, as evidenced by XAS and theoretical calculations.

Notably, 4d/5d transition-metal oxides can also function as effective OER/HER electrocatalysts. For example, Jin et al. reported that the OER and HER activity of porous MoO₂ nanosheets grown on Ni foam is superior to that of their compact MoO₂ counterparts (Figure 5A).⁹¹ The excellent HER activity was comparable to that of commercial Pt materials (Figure 5B,C). MoO₂ is a transition metal oxide and potential electrocatalyst for water splitting, featuring a porous structure and a wide surface with a considerable number of active sites owing to porous nanostructure engineering. Redox couples in metal oxides are also critical for electrochemical reactions. VO₂ nanoparticles were found to be highly efficient OER electrocatalysts in alkaline solutions.⁹² In particular, VO₂ nanoparticles/carbon fiber paper has exhibited satisfactory OER activity with a small overpotential of 350 mV at 10 mA cm⁻² in an alkaline medium. Moreover, the morphology of the material was well-preserved following long-term OER testing. Active sites were ascribed to the V⁴⁺ and V^4+/5+ redox couples in VO₂, and V⁴⁺ species were observed to be more OER-active than V⁵⁺ species in V-based oxides. However, owing to surface reconstruction during the OER, the concentration of V⁵⁺ increased on the surface of VO₂, resulting in activity degradation with prolonged reaction times. This study presents V⁴⁺ and its redox couples as novel active sites for the OER in metal-based oxide materials.

FIGURE 5. (A) Schematic for synthesis of porous MoO2 nanosheets on nickel foam. (B) Polarization curves of porous MoO2, compact MoO2, Pt/C, and nickel foam toward OER and HER. (C) Polarization curves of porous MoO2 toward overall water electrolyzer. Reproduced with permission.91 Copyright 2016, Wiley-VCH.

In contrast to active-site engineering, Kawashima et al. evaluated the structural transformation of V₈C₇ microparticles as a pre-catalyst for the OER in an alkaline electrolyte.⁹³ Extended CV testing indicated an anisotropic morphological transformation, which was attributed to the preferential self-oxidation and dissolution of the (110) and (111) surfaces of V₈C₇, leading to the exposure of the more stable (100), (010), and (001) surfaces. This study reveals a novel strategy that improves the understanding of transition-metal-based OER pre-catalysts, and enhances their OER performance and stability by manipulating their crystal growth orientation.

Beyond metal sites

Metals are widely employed as active sites owing to their ability to undergo redox reactions and facilitate electron transfer during electrochemical reactions. As previously mentioned, noble and transition metals are generally expected to be active sites for OER electrocatalysts. Perovskite oxides (ABO_3−δ) have emerged as promising OER electrocatalysts.^75,94,95 The active sites of perovskite oxides typically contain metal-oxygen bonds, which provide the necessary redox properties and facilitate the transfer of oxygen molecules during the OER.⁹⁶ Highly OER-active perovskite oxides typically contain metals with high OER catalytic activity. The electrocatalytic activity of these oxides can be optimized by further tuning the composition, structure, and morphology of the material and controlling the oxygen vacancy concentration.^75,97–105 Noble and transition metals are typically involved in the B site, which is crucial for determining the efficiency and selectivity of catalytic reactions. Oxygen-deficient BaTiO_3−δ has been reported as an efficient bifunctional electrocatalyst for the OER and oxygen reduction reaction (ORR) (Figure 6A).¹⁰⁶ Numerous oxygen vacancies generated near the surface can function as donors and acceptors for semiconductor behavior over a thin surface layer. These vacancies can provide an additional source of electrons that go beyond the surface screening charges and may be emitted during electrocatalytic reactions. The OER activity of BaTiO_3−δ is comparable to that of state-of-the-art IrO₂ nanoparticles (Figure 6B,C). Thus, this study presents a suitable perovskite oxide model to demonstrate the positive role of oxygen vacancies, which are direct active sites beyond conventional metal sites for the OER.

FIGURE 6. Crystal structures of (A) tetragonal and hexagonal BaTiO3−δ. (B) Polarization curves of IrO2 and BaTiO3−δ (BTO)-based materials toward OER and ORR. (C) Comparison of mass activity between BTO and IrO2. Reproduced with permission.106 Copyright 2015, Elsevier. (D) Relation between MOF structure and its properties. (E) Polarization curves of bare Pt and MOF-808-based materials toward HER. (F) Illustration of capping of open metal sites in MOF-808 using benzoic acid. Reproduced with permission.107 Copyright 2022, American Chemical Society.

The OER and HER are proton-coupled electron transfer (PCET) reactions, with the latter involving a proton reduction to form hydrogen gas. As an alternative to metal-site engineering, Basu et al. designed a four-sister metal–organic framework (MOF) with controlled concentrations of missing-linker defects.¹⁰⁷ Among them, MOF-808 C exhibited the highest proton conductivity (2.6 × 10⁻¹ S cm⁻¹) reported thus far for pure MOF-based proton conductors without additional structural modifications (Figure 6D). Such a super-protonic conductor can significantly improve HER activity (Figure 6E). The designed MOF structure contained only Zr, which is considered a less OER-active metal (Figure 6F). These functional materials have enriched the understanding of electrocatalytic design beyond the focus on metal site regulation.

Carbon-based materials are widely acknowledged for their outstanding electrical conductivity and chemical stability, which make them ideal catalysts for advanced applications requiring rapid electron transport and long-term operation. Moreover, the typically large specific surface areas of carbon-based materials provide numerous anchoring sites to facilitate the isolated dispersion of metal atoms without clustering. Owing to these properties, carbon-based materials are highly suitable supports for single-atom catalysis. As an alternative to 3d transition metal and noble metal single-atom materials, Li et al. synthesized C₆₀-supported V single atoms to catalyze water splitting.¹⁰⁸ Infrared multiple photon dissociation spectroscopy was utilized to probe the chemical composition and reactivity of the C₆₀V⁺(H₂O)₂ and C₆₀V⁺O₂ intermediates, and the distinct geometric and electronic roles of the C₆₀-based support in facilitating the efficient splitting of water into hydrogen and oxygen were examined. Li et al. also investigated hydrogen production through C₆₀V⁺ + H₂O → C₆₀VO⁺ + H₂.¹⁰⁹ The C₆₀ support was instrumental in decreasing the energy barrier of the reaction by more than 70 kJ mol⁻¹. This decrease is attributed to the strong orbital overlap between one carbon atom in the C₆₀ support and one of the two hydrogen atoms in the water molecule, resulting in a more efficient and effective catalytic process. These studies offer new insights into overall water-splitting reactions, specifically those catalyzed by single atoms. Moreover, these results demonstrate the previously underestimated electron-donating ability of carbon materials such as C₆₀-fullerene.

Beyond normal valence states

Ir/Ru-based oxides are conventionally regarded as state-of-the-art OER electrocatalysts, with IrO₂ being more stable in acidic and alkaline media. However, Ir-based catalysts encounter several problems during the electrochemical OER. One of the most critical problems is an unsatisfactory operational stability owing to the transformation of surface Ir⁴⁺ to water-soluble IrO₄²⁻ anions. Improving OER performance and the breaking scaling relationships may be possible through the challenging strategy of stabilizing in situ-formed metastable high-valence metal sites (e.g., Ir⁵⁺ and Ir⁶⁺ species). Li et al. conducted an OER study of the double perovskite oxide catalyst Sr₂CoIrO_6−δ, wherein two B-site cations contained a structurally ordered three-dimensional arrangement of corner-sharing CoO₆ and IrO₆ units in three dimensions.¹¹⁰ Operando x-ray absorption near-edge structure (XANES) and extended x-ray absorption fine structure spectroscopy were used to examine the electronic and crystal structures of Sr₂CoIrO_6−δ at the Co-K and Ir-L₃ edges across a range of applied voltages. Combined operando characterizations and DFT calculations demonstrated a voltage-dependent transformation of Ir⁴⁺ and Co³⁺ ions to Ir^5+/6+ and Co⁴⁺ states, respectively, during the OER. The improved OER performance of Sr₂CoIrO_6−δ was attributed to the unique corner-shared Co-O-Ir network, which yields a synergistic enhancement of catalytic activity compared with that of conventional perovskite oxides such as SrCoO₃ and SrIrO₃. The use of operando XANES resulted in the detection of the Ir⁶⁺ high-valence state during the OER for the first time. Considering transition-metal-based materials, Zhang et al. reported the anionic redox, which allowed the formation of O–O bonds and enabled a higher OER activity than that of the conventional metal sites.¹¹¹ During the OER, the use of multiple in situ/operando spectroscopies facilitated the observation of the Ni^III → Ni^IV transition, along with the removal of Li. Furthermore, a theoretical analysis indicated that the presence of Ni^IV ($3{\mathrm{d}}^8{\underset{\_}{L}}^2$) promotes direct O–O coupling between the lattice oxygen and *O intermediates, thereby increasing OER activity. The results of this study provide a novel strategy for designing OER catalysts by promoting lattice oxygen redox reactions with sufficient ligand holes generated during the OER, along with the formation of uncommonly high-valence Ni. Additionally, these results demonstrate the advantages of using in situ/operando characterization to track dynamic changes in various catalytically relevant properties, such as the crystalline phase, local coordination environment, oxidation state, and electronic structure.

Beyond single active sites

The previous discussion primarily focused on catalytic materials in which a single active site exerts a maximal contribution to OER/HER electrocatalysis. Recently, it has been found that some emerging types of catalysts and catalyst systems contain multiple active sites that may cooperatively enhance catalysis. Examples of this include complex catalysts that contain multiple active sites within a single-phase structure. For instance, Ba₄Sr₄(Co_0.8Fe_0.2)₄O₁₅, a complex oxide with a unique hexagonal structure comprising a honeycomb-like network, was uncovered to have both the tetrahedral Co cation and the octahedral O anion serving as active sites, as confirmed through an XAS analysis and DFT calculations, wherein the alkaline OER was collectively catalyzed using a dual-center catalytic pathway.¹¹² In another example, a single-phase SrTi_0.7Ru_0.3O_3−δ perovskite oxide was found to possess multiple active sites that could accurately catalyze different elementary steps of the alkaline HER, that is, a Ti site for water dissociation, Ru site for OH* desorption, and O site for H* adsorption and H₂ desorption eventually leading to excellent alkaline HER activity.¹¹³

Another example of this configuration is a catalyst system wherein surface-adsorbed species, such as hydrogen and oxygen intermediates during water electrolysis, dynamically migrate between surface active sites in a phenomenon known as spillover.^114,115 The spillover phenomenon is usually found in multicomponent, supported, and composite catalysts.^116–118 For example, Chen et al. constructed an oxygen-deficient tungsten oxide (WO_3−x)-supported Ru nanoparticulate system, where protons inserted into the WO_3−x could spill over to the Ru nanoparticles during the HER, which significantly increased hydrogen coverage on Ru and produced a 24-fold increase in HER activity compared with the commercial Ru/C (Figure 7A).¹¹⁶ The hydrogen spillover from WO_3−x to Ru was further explicitly revealed by coupling electrochemical measurements and in situ Raman spectroscopy with DFT simulations (Figure 7B,C). Dong et al. reported that when Cu and Fe₂O₃ nanoparticles support porous N-doped carbon catalysis, the active oxygen species (i.e., *O oxidized from OH⁻) spill from the Fe₂O₃ surface to the Cu surface and experience subsequent OER steps (Figure 7D).¹¹⁷ This was found to overcome the high energy barrier generally required to initiate the OER on the Cu surface, leading to a considerable reduction in overpotential compared with the counterparts where only Cu or Fe₂O₃ is supported on the porous N-doped carbon.

FIGURE 7. (A) HER activity of Ru-WO3−x/CP, Ru/C (5.0 wt%)/CP, and WO3−x/CP in 1.0 M PBS. (B) Calculated free energy diagram for HER on Hx-WO3−x and Ru-HxWO3−x. (C) Illustration depicting how the transfer of hydrogen atoms from WO3−x to Ru boosts HER performance in a neutral environment. Reproduced with permission.116 Copyright 2022, Nature Publishing Group. (D) Description of oxygen spillover pathway at the Fe2O3/Cu interface on carbon, and a diagram depicting the spillover effect in Cu-Fe2O3/PNC composites during OER. Reproduced with permission.117 Copyright 2022, American Chemical Society. (E) Schematic illustration of hydrogen spillover on two-type catalyst systems for HER in acidic media. (F) HER activity of La2Sr2PtO7+δ and Pt black in 0.5 M H2SO4 acidic solutions. Reproduced with permission.119 Copyright 2022, Nature Publishing Group.

More recently, spillover has also been revealed to occur in single-phase complex catalysts. For instance, Dai et al. reported an atomic-scale hydrogen spillover effect between the multifunctional catalytic sites of the single-phase complex oxide La₂Sr₂PtO_7+δ.¹¹⁹ Compared with hydrogen spillover in a composite catalyst system, that achieved in this single-phase La₂Sr₂PtO_7+δ was argued to yield advantages including a short reaction path and interface-free feature (Figure 7E). Through combined experimental studies and DFT calculations, an unusual atomic-scale hydrogen spillover was found to occur with the capture of protons by the O site, followed by the facile diffusion of H from the O site to the Pt site with the thermoneutral La-Pt bridge site acting as a mediator, and finally, the favorable release of the formed H₂ from the Pt site. Benefiting from this multi-active-site collaborative catalytic process, La₂Sr₂PtO_7+δ exhibited excellent HER performance in acidic solutions (with a small overpotential of 13 mV at 10 mA cm⁻² and a low Tafel slope of 22 mV dec⁻¹) (Figure 7F).

SUMMARY AND OUTLOOK

Electrocatalysts are crucial for facilitating the efficient conversion of electrical energy into chemical energy. Consequently, they enable the implementation of sustainable water, carbon, and nitrogen cycles using renewable energy. This review summarizes recent endeavors in the investigation of unusual active sites for advanced electrocatalyst designs in the context of electrochemical water splitting. Active-site engineering and the interpretation of reaction mechanisms are highly desirable tasks that require advanced characterization techniques and theoretical calculations. Despite considerable advances in electrocatalysis, including the discovery of new and unconventional active sites (Table 1), several challenges remain to be addressed in this field. Accordingly, we present several perspectives that can facilitate further progress.

TABLE 1 Recently reported HER/OER electrocatalysts with unusual active sites.

[Table omitted. See PDF]

Electrocatalyst design is a complex endeavor involving intricate interactions between catalyst composition, structure, electronic properties, and surface reactivity.^81,82,122 Conventional active sites may be guided by certain general principles, and the design of electrocatalysts with unusual active sites requires a more intricate understanding of both the underlying electrochemical processes and the materials themselves. We emphasize that no universal strategy or descriptor can be uniformly applied to create electrocatalysts with unconventional active sites. Instead, the creation of such electrocatalysts must rely on rational design and computational modeling. Computational methods can be used to design and predict the behavior of unconventional active sites based on their crystalline structures and electronic properties, which can be further corroborated by the aforementioned in situ/operando characterizations. Therefore, advancements in computational modeling will enable more accurate predictions of active site behavior, allowing for the targeted design of active sites for specific reactions and applications.

Although existing computational methods have been considerably successful in tracking unusual active sites and establishing their structure–activity relationships, they are highly time- and resource-intensive. Machine learning techniques have recently emerged as powerful tools that enable the reorganization of information structures and support multidimensional features in material informatics.^123–126 Using these techniques, a model can be developed to learn input material features during the training phase, and subsequently predict catalytic activity according to designated descriptors. Thus, computer-aided material design provides insight into optimized high-throughput computational and experimental tests for the discovery of advanced water-splitting catalysts with unusual active sites.

Although some electrocatalysts with unusual active sites have already demonstrated promising water-splitting activity, their durability requires further investigation. This research task has attracted increasing attention owing to the demand for real-world applications with long-term operability.^127–130 To achieve a better understanding, extended durability tests must be performed,¹³¹ which requires recommended practices to assess durability. A combination of in situ/operando approaches can be adopted to identify potential structural changes during water electrolysis stability tests. However, a challenge in ensuring that the testing conditions correspond to real-world conditions (e.g., elevated temperatures and concentrated electrolytes) is a major challenge that may arise for in situ/operando characterizations.² Post-mortem analyses of catalysts and electrolytes may therefore provide useful information for durability studies.^132–134

In addition to the categories of unusual active sites discussed in this review, there are emerging catalysts whose active sites are also considered unusual. We anticipate that the material design principles, characterization techniques, and calculation/computation methods discussed in this review will be applicable to discovering new electrocatalysts and active sites for the OER/HER, as well as other heterogeneous catalytic processes. Unusual active sites have also been reported for other electrochemical reactions such as CO₂, nitrogen, oxygen reduction, as well as organic transformations.^135,136 In situ/operando characterizations and theoretical calculations have also been extensively applied to examine these active sites and catalytic mechanisms, providing further insights for the rational design of efficient electrocatalysts. Therefore, the information and methodology discussed in this review may be extended to a broader range of electrocatalytic applications.

ACKNOWLEDGMENTS

This research was supported by National R&D Program through the National Research Foundation of Korea (NRF) funded by Ministry of Science and ICT, grant number NRF-2020M3H4A3105824 and 2022K1A4A8A01080242.

CONFLICT OF INTEREST STATEMENT

The authors declare no conflict of interest.