High-Entropy Oxygen Evolution Catalysts： Mechanistic Analysis， Optimization Strategies， and Prospective Challenges

Shaofu Kuang; Xue Lu; Jianxing Wang; Hua Lin; Qing Li

doi:10.7536/PC20250715

Progress in Chemistry >

2025 , Vol. 37 >Issue 11: 1581 - 1603

DOI: https://doi.org/10.7536/PC20250715

Review

High-Entropy Oxygen Evolution Catalysts： Mechanistic Analysis， Optimization Strategies， and Prospective Challenges

Shaofu Kuang ,
Xue Lu ,
Jianxing Wang ,
Hua Lin ,
Qing Li ^,^*

Expand

Chongqing Key Laboratory of Battery Materials and Technologies， School of Materials and Energy， Southwest University， Chongqing 400715， China

^* qli@swu.edu.cn

Received date: 2025-07-14

Revised date: 2025-08-09

Online published: 2025-10-25

Supported by

National Natural Science Foundation of China(52072310)

Fold

Abstract

Hydrogen production via water electrolysis powered by renewable energy sources represents a critical approach to addressing the dual challenges of energy and the environment. However， the practical implementationof this technology remains constrained by the sluggish kinetics of the anodic oxygen evolution reaction （OER）. Recent advances in high-entropy materials （HEMs） with unique structural configurations and compositional tunability have demonstrated breakthrough capabilities in OER catalysis. Their near-continuous adsorption energy tunability across multi-dimensional landscapes enables surpassing the perforce ceilings of conventional single-/dual-component electrocatalysts. While substantial progress has been achieved in developing HEMs for OER catalysis， formidable scientific challenges persist regarding the intricate composition-structure-activity relationships in multi-component systems and unresolved mechanistic ambiguities governing catalytic synergies. This review systematically examines the fundamental mechanisms underlying the four-electron transfer process in OER， followed by a critical survey of recent breakthroughs in high-entropy alloys （HEAs）， high-entropy oxides （HEOs）， and high-entropy metal-organic frameworks （HEMOFs） for OER applications. By emphasizing three critical dimensions： atomic coordination environment modulation， electronic structure engineering， and surface adsorption energy optimization， we establish explicit correlations between compositional architecture， structural characteristics， and catalytic performance. This framework profoundly elucidates the synergistic catalytic mechanisms arising from multi-metallic active sites. Furthermore， we propose strategic optimization pathways through material design， defect engineering， and elemental regulation. The review concludes by discussing emerging challenges and future opportunities in this rapidly evolving field. This review can provide inspiration for the accurate design of high-entropy electrocatalysts， the atomic-level analysis of structure-activity relationships， and the regulation and optimization of catalytic performance.

Contents

1 Introduction

2 OER pathway

2.1 AEM

2.2 LOM

2.3 OPM

3 Research progress and bottlenecks of high‑entropy oxygen evolution catalytic materials

3.1 High‑entropy alloys

3.2 High‑entropy oxides

3.3 High‑entropy MOFs

3.4 Other high‑entropy compounds

4 Optimization strategies

4.1 Machine learning‑assisted design

4.2 Defect engineering

4.3 Element regulation

5 Conclusion and outlook

Key words： high-entropy materials; electrocatalysis; oxygen evolution reaction; optimization strategies

Cite this article

Shaofu Kuang , Xue Lu , Jianxing Wang , Hua Lin , Qing Li . High-Entropy Oxygen Evolution Catalysts： Mechanistic Analysis， Optimization Strategies， and Prospective Challenges[J]. Progress in Chemistry, 2025 , 37(11) : 1581 -1603 . DOI: 10.7536/PC20250715

1 Introduction

With the rapid development of modern industrial technologies, fossil fuels such as oil, coal, and natural gas have been extensively mined and utilized, leading to environmental pollution and an energy crisis^[1-2]. To address this issue, hydrogen has regained global attention as a cleaner alternative energy source^[3]. At present, the vast majority of global hydrogen production still relies on two conventional processes: steam methane reforming and coal gasification. Both technologies use fossil fuels as feedstocks, and their production processes not only generate substantial carbon dioxide emissions but also reveal significant sustainability shortcomings in the context of energy transition^[4]. In contrast, hydrogen production via water electrolysis using renewable energy sources (such as wind power and photovoltaics) is regarded as a more promising approach to addressing energy and environmental challenges^[5-7]. In these energy conversion processes, suitable catalysts are required to enhance electrochemical reaction efficiency, thereby reducing hydrogen production costs and making it economically competitive. In particular, in water electrolysis, the sluggish kinetics of the oxygen evolution reaction (OER) necessitate a relatively high overpotential to drive the reaction, which accounts for significant energy losses throughout the water electrolysis process^[8-10]. Currently, the most widely used and representative electrocatalysts are those based on precious metals (such as Pt, Ir, Ru, and Au)^[11-13]or certain transition metals (such as Fe, Co, and Ni)^[14-15]. For precious-metal-based catalysts, their scarcity and high cost limit their large-scale commercial application^[16]. Consequently, binary materials composed of small amounts of precious metals and transition metals have attracted considerable attention^[12,17-18]. It is well known that the adsorption relationship between the catalyst surface and intermediate species determines its catalytic activity, while the electronic structure of the surface governs the adsorption energy^[19]. Clearly, the compositional range and content of binary materials are limited, which in turn constrains the extent to which their electronic structures can be optimized.

In recent years, high-entropy materials have provided a novel approach to catalyst design due to their unique multi-component synergistic effects, emerging as a promising platform for constructing multi-active-site water-splitting catalysts^[20-23]. Moreover, the uniform mixing of multiple components at the atomic scale can significantly enhance the configurational entropy of the system, enabling it to maintain excellent structural stability under extreme service conditions^[24-25]. To date, numerous studies have been reported on the use of high-entropy materials in catalysis, including oxygen reduction^[26-30], CO₂reduction^[31-32], ethanol oxidation^[33-34], hydrogen evolution from water electrolysis^[35-36], and oxygen evolution reactions^[37-41]. However, the development of high-entropy material catalysts still faces a dual challenge: On the one hand, the vast design space of multi-component systems—such as component selection, ratio optimization, and structural regulation—renders the traditional “trial-and-error” approach inefficient; on the other hand, the synergistic mechanisms among multiple metal sites have not yet been fully elucidated, and there is a lack of universally applicable correlations between “composition–structure–performance.” These bottlenecks severely hinder the transition of high-entropy catalysts from laboratory exploration to engineering applications.

With the continuous deepening of research into the OER mechanism in water electrolysis, the dynamic evolution patterns of multi-electron transfer processes have gradually become clearer. This article first reviews the current mainstream OER reaction pathways, including the Adsorbate Evolution Mechanism (AEM), the Lattice Oxidation Mechanism (LOM), and the Oxide Path Mechanism (OPM). It then focuses on high-entropy oxygen-evolution catalytic material systems, providing a comprehensive review of the current research status and bottlenecks of novel high-entropy catalytic materials, such as high-entropy alloys (HEAs), high-entropy oxides (HEOs), high-entropy sulfides (HESs), high-entropy boron/phosphorus compounds (HEBs/HEPs), and high-entropy metal–organic framework materials (HEMOFs). By establishing the structure–property relationships among material components, structure, and performance, the article highlights the synergistic effects among material components and their regulatory mechanisms on the OER mechanism. Finally, based on the unique structural advantages and tunable compositional flexibility of high-entropy materials, the article proposes performance optimization strategies from three dimensions—machine learning–assisted design, defect engineering, and elemental regulation—providing both theoretical foundations and technical pathways for developing efficient, stable, and customized OER catalysts.

2 OER reaction pathway

In summary, existing scoring systems have limited predictive capabilities for bleeding events, and their results are inconsistent^[25,30,33].

显示原图|下载原图ZIP|生成PPT

图1 OER机理的发展历程^[42-46]，以及其对应的优点与缺点

Fig.1 The development process of OER mechanism^[42-46]， and their proposed advantages and disadvantages

2.1 AEM Mechanism

The AEM-based OER process involves a sequential evolution of key oxygen-containing intermediates: *OH, *O, and *OOH (* denotes the adsorbed state). The formation of these oxygen-containing intermediates differs under different catalytic environments (alkaline or acidic) (Fig. 2a). In an alkaline medium, ample OH^-directly participates in the reaction process, as shown in the following equation:

（1）* + OH^- → *OH + e^-

（2）*OH + OH^- → *O + H₂O + e^-

（3）*O + OH^- → *OOH + e^-

（4）*OOH + OH^- → * + O₂ + H₂O + e^-

first, (1) OH^-loses an electron and adsorbs onto the active site (*) to form *OH; (2) surface *OH reacts with solution OH^-to remove a proton, generating *O; (3) OH^-continues to lose an electron and adsorb onto *O to form *OOH; (4) finally, *OOH reacts with OH^-in the solution to produce O₂and water. Under acidic conditions, however, due to the lack of free OH^-, the reaction pathway must obtain OH^-through the dissociation of water molecules, as shown in the following equation:

（5）* + H₂O → *OH + H⁺+ e^-

（6）*OH → *O + H⁺+ e^-

（7）*O + H₂O → *OOH + H⁺+ e^-

（8）*OOH → * + O₂+ H⁺+ e^-

(5) H₂O undergoes deprotonation/electron removal at the active site to form *OH; (6) *OH undergoes secondary deprotonation/electron removal to convert to *O; (7) H₂O decomposition provides OH^- and adsorbs onto *O to form *OOH; (8) Finally, *OOH undergoes deprotonation/electron removal to release O₂.

显示原图|下载原图ZIP|生成PPT

图2 （a） AEM示意图；（b）实际的和理想催化剂的吉布斯自由能图；（c）部分氧化物催化剂过电位与ΔG_OH-ΔG_O的火山图^[49]；（d） AD-NH-Ir在OER过程中的原位红外光谱图^[50]

Fig.2 （a） Schematic diagram of AEM. （b） The Gibbs free energy diagrams of catalysts under actual and ideal conditions. （c） The volcano relationship between ΔG_OH*-ΔG_*O and OER overpotential for part oxides^[49]. Copyright 2018， Springer Nature. （d） In situ FTIR of AD-NH-Ir during OER^[50]. Copyright 2021， Springer Nature

According to Sabatier's theory, the adsorption strength of intermediates on the surface of an ideal catalyst must be maintained within a moderate range^[19]. In particular, there is a linear relationship between the adsorption energies of *OH and *OOH (ΔG _OOH=ΔG _OH+3.2±0.2 eV), with the theoretical overpotential limited to around 370 mV^[47-48]. Breaking free from this linear relationship constraint has become the core challenge in designing AEM catalysts, with the key lying in reducing the energy barrier of the rate-determining step by optimizing the electronic structure (Fig. 2b). In this reaction pathway, the intermediate *O plays a central regulatory role, and the value of ΔG _HO*-ΔG _*Ois commonly used to predict the catalytic activity of a catalyst. Fig. 2ccompiles the overpotential versus ΔG _HO*-ΔG _*Ofor most catalysts; the optimal activity lies at the top of the volcano plot, indicating that the adsorption and desorption of oxygen-containing intermediates consistently constrain the potential design of AEM catalysts. Although this linear relationship limits the activity of AEM catalysts, the catalyst body itself does not participate in redox processes throughout the entire OER process, giving it superior structural stability compared to LOM catalysts. To date, several in situ techniques (such as in situ Raman spectroscopy and FTIR spectroscopy) have been employed experimentally to identify OER reaction intermediates (*OOH) and validate AEM catalysts. For example, in infrared spectroscopy, a peak around 1055 cm^-1is attributed to the vibrational absorption of *OOH on the catalyst surface (Fig. 2d).

2.2 LOM mechanism

LOM differs fundamentally from AEM; it is specific to oxygen-containing catalyst systems, with the core feature being the participation of lattice oxygen in OER. Based on the formation mechanism of oxygen species, it can be divided into two subcategories (Fig. 3a, b).In light of the preceding analysis of mechanistic differences in acidic and alkaline environments, we will now use the reaction of layered double hydroxides (LDH) in alkaline media as an example to illustrate the first pathway, which is represented by the following equation:

（9）*OH + OH^-→ *O + H₂O + e^-

（10）*O + OH^- → *OOH + e^-

（11）*OOH + OH^-→ *OO + H₂O + e^-

（12）*OO → V_o + O₂

（13）V_o + OH^- → *OH + e^-

(9) First, the OH^- in solutionhelps deprotonate the OH on the LDH surface to form *O; (10) second, the second OH^- in solutionloses an electron and adsorbs onto *O to form *OOH; (11) next, the OH^- in solutionagain helps *OOH lose a proton to generate superoxide or peroxide species; (12) the superoxide or peroxide species decompose into oxygen, leaving an oxygen vacancy; (13) finally, the OH^- in solutionagain loses an electron to become *OH, filling the oxygen vacancy. The second pathway is as follows:

（14）*OH + OH^-→ *O + H₂O + e^-

（15）* + OH^- → *OH + e^-

（16）*O + *OH + OH^-→ *OO + H₂O + e^-

（17）*OO → V_o + O₂

（18）V_o + OH^- → *OH + e^-

(14) First, the OH^- in solutionhelps the *OH on LDH to lose a proton and become *O; this step is consistent with the previous pathway. (15) The second OH^- in solutionloses an electron and adsorbs onto a metal site; this step represents the key difference between the two pathways. (16) The OH^- in solutionhelps the *OH on the metal site to lose a proton, and it couples with the lattice oxygen from the first step to form superoxide or peroxide species. (17) The superoxide or peroxide species decompose into oxygen, leaving an oxygen vacancy. (18) Finally, the OH^- in solutionagain loses an electron to become *OH, which then fills the oxygen vacancy.

显示原图|下载原图ZIP|生成PPT

图3 （a，b） LOM及其亚型示意图；（c）原位差分质谱仪示意图；（d） ¹⁸O标记EA-FeNiSe₂产生氧气的DEMS测量信号^[57]；（e）在1.0 mol/L KOH和1.0 mol/L TMAOH条件下，Au_SA-MnFeCoNiCu LDH和MnFeCoNiCu LDH的极化曲线以及在不同电解液中过电位和Tafel斜率的变化（@100 mA·cm^-2）^[58]；（f） Au_SA-MnFeCoNiCu LDH和MnFeCoNiCu LDH在不同pH的电解液下的极化曲线（左）以及在1.45 V（相对于可逆氢电极）时电流密度随pH的变化（右）^[58]

Fig.3 Schematic diagram of （a） LOM and （b） sub-LOM， respectively. （c） Schematic illustration of in situ DEMS device. （d） DEMS signals of O₂ products for EA-FeNiSe₂ over time^[57]. Copyright 2024， American Chemical Society. （e） Polarization curves of Au_SA-MnFeCoNiCu LDH and MnFeCoNiCu LDH in 1.0 mol/L KOH and 1.0 mol/L TMAOH， shift of overpotential at 100 mA·cm^-2 and Tafel slopes from KOH to TMAOH^[58]. Copyright 2023， Springer Nature. （f） Polarization curves measured in KOH electrolytes with different pH （left）， current density at 1.45 V （vs. RHE） as a function of pH （right）^[58]. Copyright 2023， Springer Nature

In summary, the core difference between the two pathways lies in the distinct mechanisms for forming superoxide or peroxide species: the former involves self-coupling of lattice oxygen with intermediates, while the latter relies on the coupling of adsorbed oxygen at metal sites with lattice oxygen to form bonds. Experimentally, the LOM mechanism can be confirmed by detecting superoxide/peroxide intermediates, with isotope labeling being a commonly used method. Specifically, the catalyst surface is labeled with the ¹⁸O isotope, and OER is conducted in an ¹⁶O electrolyte system. The detection of ³⁴O₂ product via mass spectrometry (DEMS), resulting from the ¹⁸O-¹⁶O isotope pairing, serves as direct evidence for the LOM mechanism (Fig. 3c, d^[51]). In addition, tetramethylammonium hydroxide (TMAOH) is also frequently used to verify the presence of superoxide or peroxide bonds (Fig. 3e). The cationic component of this reagent (TMA⁺) can specifically adsorb onto surface sites of superoxide groups, thereby inhibiting further transformation of superoxide intermediates and slowing down the OER kinetics, thus confirming the existence of superoxide/peroxide intermediates (52). Furthermore, due to the influence of electron transfer kinetics at the catalyst/electrolyte interface and the adsorption energy of *OH, LOM-based catalysts typically exhibit significant pH dependence, whereas AEM-based catalysts are largely unaffected by pH changes (52-53). Consequently, pH-dependent experiments are often used as an indirect means to validate the LOM mechanism (Fig. 3f). In situ Raman spectroscopy can also be used to verify the presence of superoxide bonds, with a peak around 1035 cm^-2 attributed to the vibrational mode of superoxide groups (54).

It is particularly important to note that current experimental methods cannot yet distinguish between the different subtype pathways under the LOM mechanism. To address this, density functional theory (DFT) calculations are typically employed to analyze the reaction energy barriers and intermediate-state configurations for different pathways, thereby elucidating the reaction mechanism^[55].For the LOM pathway (Fig. 3a), the key step involves the coupling of *OH with lattice oxygen, and the reaction energy barrier for this step determines the feasibility of the pathway. In contrast, the sub-LOM pathway (Fig. 3b) must meet two critical requirements. First, the adsorption of OH on the metal site must reach a critical threshold; otherwise, the system will remain in the LOM pathway. Second, the energy required for the coupling of *O with lattice oxygen must be lower than the trigger threshold for the AEM pathway (i.e., the formation of *OOH) to prevent path diversion caused by the re-adsorption of OH. Therefore, the core to triggering the sub-LOM pathway lies in the coordinated regulation of the electronic structures of the metal sites and lattice oxygen. Kuang et al.^[56] successfully triggered the sub-LOM pathway while preserving the LOM pathway by introducing defect strategies into high-entropy materials, thereby establishing a dynamic catalytic system in which both mechanisms coexist.

2.3 OPM Mechanism

Currently, OER catalytic mechanisms face two core challenges: AEM is constrained by kinetic barriers arising from linear relationships, while LOM suffers from catalyst structural collapse caused by the continuous generation of oxygen vacancies. OPM effectively circumvents these issues through a novel reaction pathway. As shown in Figure 4a, the intermediates in the OPM reaction pathway involve only *OH and *O, and its core mechanism can be decomposed into four key steps, with the specific reaction equations as follows:

（19）* + OH^- → *OH + e^-

（20）** + OH^- → *OH + e^-

（21）2OH^- + 2*OH → 2*O + 2H₂O + 2e^-

（22）2*O → O₂ + *

(19) First, the first metal site loses an electron and adsorbs an *OH; (20) another adjacent site then adsorbs a **OH (** represents another reaction site); (21) the two sites collaboratively undergo deprotonation to form two adjacent *O species; (22) the neighboring *O species directly couple to form an O₂molecule, which subsequently desorbs. The initiation of this mechanism requires precise control of the microenvironment around the catalytic sites: first, the spacing between the metal active centers must be within an appropriate range to ensure overlap of electronic orbitals, enabling cooperative electron transfer among intermediates (Fig. 4b, c); second, the electronic structure of the metal sites must be optimized to enhance the adsorption strength of *OH/*O while suppressing the formation of *OOH intermediates^[59]. This characteristic makes atomically dispersed catalysts (such as diatomic catalysts) an ideal choice for realizing OPM^[60].

显示原图|下载原图ZIP|生成PPT

图4 （a） OPM示意图；OH^-吸附在（b） SrFeO₃和（c） CaCu₃Fe₄O₁₂晶体结构模型^[63]；（d） Ru-Co₃O₄产生氧气的DEMS测量信号^[64]；（e） Ru/MnO₂在OER过程中的原位红外光谱^[65]；（f） S-FeOOH/IF在OER过程中的原位红外光谱^[62]

Fig.4 （a） Schematic diagram of OPM. OH^- adsorbates on （b） SrFeO₃ and （c） CaCu₃Fe₄O₁₂ crystalline structure models^[63]. Copyright 2015， Springer Nature. （d） DEMS signals of O₂ products for Ru array-Co₃O₄^[64]. Copyright 2024， American Chemical Society. （e） In situ FTIR of Ru/MnO₂during OER^[65]. Copyright 2024， Springer Nature. （f） In situ FTIR of S-FeOOH/IF during OER^[62]. Copyright 2022， Wiley

Through reaction mechanism analysis, clear criteria can be established to distinguish between OPM and AEM/LBM. As described in Section 2.2, in an isotopic tracing experiment, the catalyst is first labeled by placing it directly in an aqueous solution of H₂ ¹⁸O, and then transferred to an aqueous solution of H₂ ¹⁶O for reaction. When the ³⁶O₂isotopic product is detected, this indicates that the oxygen adsorbed on the catalyst surface undergoes direct coupling to produce ³⁶O₂, thereby confirming the presence of OPM (Fig. 4d). In addition, the vibrational band observed by FTIR near 1100 cm^-1can be attributed to the bridging oxygen configuration at the dual active sites, indicating the formation of a characteristic M-O-O-M intermediate via direct O-O coupling (Fig. 4e), which can serve as an auxiliary criterion for identifying OPM. However, the position of this peak is close to that of the *O₂²^-intermediate in the LBM pathway (Fig. 4f)^[61-62]. Therefore, a multi-dimensional experimental verification system must be established to avoid misinterpretations that may arise from relying on a single characterization method.

3 Research Status and Bottlenecks of High-Entropy Oxygen Evolution Catalyst Materials

In 2004, Yeh’s team^[66]formally proposed the concept of HEAs. The strict definition of HEAs comprises two key criteria: first, the alloy system must consist of five or more elements; second, the atomic percentage of each constituent element must be within the range of 5% to 35%. Since then, researchers have engaged in extensive discussions regarding the core criteria for identifying high-entropy systems. The groundbreaking performance exhibited by certain quaternary alloys has prompted researchers to re-examine the traditional definition of high entropy^[67-68].Against this backdrop, the concept of high entropy has undergone a critical evolution: from the early, strictly compositional definition to a materials design philosophy oriented toward performance optimization. The research domain has expanded significantly beyond the original HEA systems, gradually giving rise to high-entropy material systems that encompass multiple elements, with typical representatives including HEOs, HEBs, HESs, and HEPs (see Figure 5).

显示原图|下载原图ZIP|生成PPT

图5 高熵材料的发展历史及其应用^[69]

Fig.5 Historical development and applications of high-entropy materials^[69]

In recent years, high-entropy materials, as a highly regarded new class of functional materials, have demonstrated groundbreaking performance advantages in the field of electrocatalytic OER, thanks to their unique compositional degrees of freedom, pronounced configurational entropy effects, and excellent structural stability (Table 1). According to existing literature, the most commonly used metal elements in reported high-entropy electrocatalysts are Fe, Co, and Ni, followed by Cu, Pd, Pt, and others (Fig. 6). However, the vast compositional space, which encompasses tens of thousands of potential elemental combinations, poses significant challenges to rational materials design. Current research has yet to establish systematic criteria for element screening or strategies for compositional optimization. In this section, starting from the OER reaction mechanism, we review the innovative breakthroughs in typical systems such as HEAs, HEOs, and HEMOFs in terms of active-site regulation and electronic-structure optimization, thereby providing insights for the development of high-performance high-entropy OER catalysts.

表1 常温条件下几种典型高熵材料OER的催化活性

Table 1 Electrocatalytic activity of some typical high-entropy materials for the OER under ambient temperature conditions

Catalyst	Type	Electrolyte	Overpotential （mV） @10 mA·cm^-2	Tafel slope （mV·dec^-1）	Stability （V vs. RHE）	Ref
NiCo LDH	binary	1 mol·L^-1 KOH	271	72	20 h@1.673 V	70
Co-P-B	ternary	1 mol·L^-1 NaOH	290	42	20 h@1.53 V	71
Co₃V₂O₈	ternary	1 mol·L^-1 KOH	359	65	3 h@10 mA·cm^-2	72
FeCo₂P	ternary	0.1 mol·L^-1 KOH	320	55	12 h@10 mA·cm^-2	73
S₁₅-CoTe	ternary	1 mol·L^-1 KOH	255	54，7	100 h@10 mA·cm^-2	74
NiCoFeB	quaternary	1 mol·L^-1 KOH	284	46	10 h@285 mV	75
E-FeCoNiZn	quaternary	1 mol·L^-1 KOH	259	37.1	48 h@10 mA·cm^-2	76
AlNiCoIrMo	HEA	0.5 mol·L^-1 H₂SO₄	233	55.2	48 h@1.52 V	77
AlCrCuFeNi	HEA	1 mol·L^-1 KOH	270	77.5	35 h@290 mV	78
CoFeGaNiZn	HEA	1 mol·L^-1 KOH	370	71	10 h@1.5 V	79
CoCuFeNiMnMo	HEA	1 mol·L^-1 KOH	375	61	72 h@10 mA·cm^-2	80
PtFeCoNiMn	HEA	1 mol·L^-1 KOH	357	114.6	60 h@10 mA·cm^-2	81
MoZnFeCoNi	HEA	1 mol·L^-1 KOH	221	48.78	1550 h@100 mA·cm^-2	82
CoFeCrMoMnO	HEO	1 mol·L^-1 KOH	188	30	24 h@50 mA·cm^-2	83
ZnFeNiCuCoRu-O	HEO	1 mol·L^-1 KOH	170	56	30 h@10 mA·cm^-2	39
CoCuFeMoOOH	HEO	1 mol·L^-1 KOH	199	48.8	72 h@50mA·cm^-2	84
（FeCoNiCrMn）₃O₄	HEO	1 mol·L^-1 KOH	288	60	95 h@10 mA·cm^-2	85
（CoCuFeMnNi）₃O₄	HEO	1 mol·L^-1 KOH	350	59.5	12 h@15 mA·cm^-2	86
La_0.8Sr_0.2Co_0.8Fe_0.2O_3-δ	HEO	1 mol·L^-1 KOH	248	51	20 h@15 mA·cm^-2	87
Ag@CoCuFeAgMoOOH	HEO	1 mol·L^-1 KOH	218	35.3	50 h@10 mA·cm^-2	88
FeCoNiCrVB	HEB	1 mol·L^-1 KOH	237	24.2	20 h@1.46 V	89
FeNiCoCrMnS₂	HES	1 mol·L^-1 KOH	199	39.1	55 h@500 mA·cm^-2	90
FeCoNiPB	HEPB	1 mol·L^-1 KOH	235	53	40 h@10 mA·cm^-2	91
NiCoFeMnCrP	HEP	1 mol·L^-1 KOH	220	94.5	24 h@1.55 V	92
FeCoNiMnCuP_x	HEP	1 mol·L^-1 KOH	239	72.5	50 h@100 mA·cm^-2	93
HEMOFs	HEMOF	1 mol·L^-1 KOH	310	48	8.3 h@10 mA·cm^-2	94
FeCoNiCuMn MOF	HEMOF	1 mol·L^-1 KOH	196	55	75 @10 mA·cm^-2	95
MnFeCoNiCu MOF	HEMOF	1 mol·L^-1 KOH	245	54	48 h@10 mA·cm^-2	96

显示原图|下载原图ZIP|生成PPT

图6 目前高熵电催化剂组元的使用频率

Fig.6 Current utilization frequency of constituent elements in high-entropy electrocatalysts

3.1 High-entropy alloys

For HEA catalysts that follow the AEM approach, the core of optimizing OER performance lies in precisely controlling the adsorption-desorption dynamic equilibrium of intermediates. Compared with traditional unary or binary material systems, high-entropy materials, owing to their multi-component nature and the resulting complex surface structures, offer the possibility of achieving nearly continuous distributions of adsorption energy curves (Fig. 7a).Mei et al.^[97]found that in FeCoNi-based HEAs, the excessively strong adsorption of *OH by transition metals hinders the deprotonation process, thereby significantly limiting OER kinetics. By introducing the Mo element, which possesses unique electronic properties, they successfully induced electron redistribution at the Fe/Co/Ni sites. Methanol poisoning tests combined with DFT calculations confirmed that the incorporation of Mo reduces the adsorption strength for OH, decreasing the activation energy required for deprotonation from 0.39 eV to 0.31 eV and significantly enhancing the OER reaction rate. Zhu et al.^[98]employed an electronegativity gradient design strategy to introduce the low-electronegativity Mn element into Cu-based HEAs, thereby constructing an electron transfer pathway from Mn to Cu. This electronic structure engineering increases the surface electron density at Cu sites, which on the one hand promotes the dissociation of H₂O molecules and optimizes the adsorption strength of the *H intermediate, and on the other hand accelerates the deprotonation of *OH to form the *O intermediate. As a result, this catalyst requires only a overpotential of 281 mV to drive the hydrogen evolution reaction at a current density of 100 mA·cm^-2, while under conditions of 200 mA·cm^-2, the oxygen evolution overpotential is as low as 386 mV, demonstrating outstanding bifunctional catalytic performance. The above studies indicate that by regulating the synergistic effects among the multiple components in HEAs, the adsorption energy of OER intermediates on their surfaces can be effectively optimized, thereby significantly enhancing OER catalytic activity.

显示原图|下载原图ZIP|生成PPT

图7 （a）各类催化剂表面吸附能的示意图；（b） HEAs不同活性位点示意图；（c） FeCoNiMn；（d） FeCoNiMnCr，（e） FeCoNiMnMo和（f） FeCoNiMnW吉布斯自由能变化图；以及HEAs的（g） TDOS图^[99]

Fig.7 （a） Schematic diagram of the adsorption energy of various catalyst surfaces. （b） Diagrams of different active sites in HEAs. The change in Gibbs free energy of （c） FeCoNiMn，（d） FeCoNiMnCr，（e） FeCoNiMnMo， and （f） FeCoNiMnW. （g） TDOS plots for HEAs^[99]. Copyright 2023， American Chemical Society

Huang et al.^[40]prepared an FeCoNiCuMo catalyst on a nickel foam substrate using electrodeposition. This material exhibits outstanding HER and OER catalytic activity in both acidic and alkaline electrolytes. In situ X-ray absorption spectroscopy (XAS) analysis reveals significant site heterogeneity in its catalytic mechanism: Co/Cu sites dominate the Volmer-Heyrovsky pathway, while Ni and Mo sites participate in the reaction via the Volmer-Tafel pathway, with the synergistic action of multiple mechanisms significantly enhancing overall kinetics. Li et al.^[99]prepared submicron-sized single-phase HEA particles (FeCoNiMnX, X = Cr, Mo, W) via electrochemical metallization. FeCoNiMnW exhibited the best OER performance, with an overpotential of 355 mV at a current density of 500 mA·cm^-2, and could operate stably at 500 mA·cm^-2 for 50 days. Theoretical calculations indicate that the rate-determining step for OER in FeCoNiMn and FeCoNiMnCr is the transformation from M-*OH to M-*O, whereas for FeCoNiMnMo and FeCoNiMnW, the rate-determining step is the transformation from M-*O to M-*OOH, with corresponding overpotentials of 1.478 V (FeCoNiMn), 0.602 V (FeCoNiMnCr), 0.437 V (FeCoNiMnMo), and 0.377 V (FeCoNiMnW). The introduction of Cr, Mo, and W significantly reduced the overpotential associated with the rate-determining step in the OER process (Fig. 7b–f). Furthermore, FeCoNiMnW exhibited the highest total density of states (TDOS), resulting in the fastest electron transfer rate during OER (Fig. 7g). Qiu et al.^[100]used a dealloying strategy to prepare an AlNiCoFe-based HEA system. Through systematic studies of the synergistic effects of elements such as Cr, Nb, V, Zr, and Mn, they found that the introduction of Cr, Nb, and Mo can significantly enhance catalytic activity. Theoretical calculations indicate that the high-valence oxides formed by these elements can optimize the proton transfer pathway. However, in studies of Ir-based HEA systems (AlNiCoIr-X, X = V, Nb, Cu, Cr, Mo)^[77], it was found that doping with Mo and V can significantly improve OER performance, whereas the introduction of Nb and Cr suppresses catalytic activity. These contradictory phenomena further highlight the complexity of the "composition-structure-performance" relationship in HEA systems.

In summary, the multi-component nature of HEAs endows them with a vast dimension for tuning adsorption energy, but the complex synergistic effects among their components pose challenges for elucidating the underlying mechanisms. Existing studies indicate that introducing high-valence elements can optimize the adsorption properties of AEM pathway intermediates through electronic structure reconstruction; however, such modification strategies based on adsorption energy regulation are constrained by the adsorption-desorption dynamic equilibrium limitations imposed by Sabatier’s principle. Considering the triggering conditions of OPM, if electronic structure engineering can effectively suppress the formation of *OOH intermediates and promote the direct coupling of oxygen via multi-site cooperative adsorption on the HEA surface, it could theoretically break through the limitations of the traditional linear relationship in adsorption energy and lead to a significant enhancement of OER kinetics. However, systematic research on this reaction pathway remains scarce, particularly regarding the mechanistic understanding of how the multi-component synergy in HEAs influences the O–O direct coupling process. This gap in knowledge has become a bottleneck that hinders the rational design of novel high-entropy catalysts.

3.2 High-entropy oxides

In recent years, HEOs have attracted widespread attention due to their outstanding OER catalytic performance. This section focuses on reviewing research advances in four typical HEO systems: LDHs, rock-salt oxides, spinel oxides, and perovskite oxides. LDHs are regarded as ideal candidate materials for OER catalysts^[101-103],and the development of high-entropy LDHs holds promise for providing new opportunities for efficient water electrolysis. Unlike three-dimensional oxide structures, the low-dimensional layered structure of LDHs allows O―O bonds to form on the surface, providing an innovative platform for designing highly efficient LOM-based catalysts. Gu et al.^[104]employed an atomic radius matching strategy to prepare ultrathin FeCrCoNiCu-LDH via a hydrothermal method, minimizing lattice mismatch and achieving exceptional OER activity of 270 mV@10 mA·cm^-2in 1 M KOH. Hao et al.^[105]regulated the lattice mismatch ratio through the dual oxidation effects of H₂O₂and Ce⁴⁺, thereby preparing a lattice-disordered high-entropy FeCuCoNiZn LDH catalyst. Experiments show that this structure can both construct active centers through local enrichment of high-valence metal ions and provide abundant surface active sites to accelerate the pre-oxidation process, ultimately achieving synergistic enhancement of OER and urea oxidation (UOR) performance. Notably, recent studies have revealed the dynamic surface reconstruction phenomenon of catalysts during the OER process, which offers a new pathway for performance breakthroughs^[106-107]. For example, Liang et al.^[108]prepared a high-entropy FeCoNiZnOOH via electrodeposition; under the surface reconstruction induced by the dynamic dissolution of zinc, the catalyst achieves an ultra-low overpotential of 191 mV at a current density of 100 mA·cm^-2. The latest research by Wang et al.^[58]further expands surface modification strategies: using MnFeCoNiCu LDH as a matrix, they constructed an O-vacancy–cooperative catalytic system through Au single-atom doping (Fig. 8a). Theoretical calculations indicate that the synergistic effect between Au and O vacancies can significantly shift the O 2p band center upward, weakening the metal–oxygen bond strength and thereby effectively activating the LOM mechanism (Figs. 8b and c). In addition, this study reveals the structure–activity relationship between the electronegativity of precious metals and the degree of LOM activation, providing theoretical guidance for the rational design of high-performance catalysts.

显示原图|下载原图ZIP|生成PPT

图8 （a） Au_SA-MnFeCoNiCu LDH示意图；MnFeCoNiCu/Au-MnFeCoNiCu LDH的（b）态密度和（c）能带示意图^[58]，（Co_0.2Cu_0.2Mg_0.2Ni_0.2Zn_0.2）O的（d） OER机理示意图，以及不同氧化物（e）极化曲线和（f）多角度对比的玫瑰示意图^[113]；（g）高熵钙钛矿[LaM（Ⅲ）O₃]_3/4[KM（Ⅱ）F₃]_1/4制备示意图，以及（BSCF）_3/4[KM（Ⅱ）F₃]_1/4的（h） XRD图和（i）极化曲线^[118]

Fig.8 （a） Schematic of AuSA-MnFeCoNiCu LDH. （b） Projected density of states of MnFeCoNiCu/Au-MnFeCoNiCu LDH and （c） schematic band diagrams^[58]. Copyright 2023， Springer Nature. （d） Schematic diagram of the OER mechanism for （Co_0.2Cu_0.2Mg_0.2Ni_0.2Zn_0.2）O， and （e） polarization curves and （f） rose diagram showing multi‐angle comparisons^[113]. Copyright 2022， Elsevier. （g） Schematic illustration of the synthesis of [LaM（Ⅲ）O₃]_3/4[KM（Ⅱ）F₃]_1/4， and （h） XRD image and （i） polarization curves of （BSCF）_3/4[KM（Ⅱ）F₃]_1/4^[118]. Copyright 2021， Wiley

In the study of electrocatalytic materials, metal oxide systems represented by rock-salt oxides, spinel-structured oxides, and perovskite oxides have been proven to serve as high-performance water-splitting catalysts due to their unique electronic structures and tunable active sites. By leveraging oxygen vacancies and cation doping engineering, these metal oxides with different crystal structures can precisely regulate the charge distribution and coordination environment of active sites, thereby significantly reducing the kinetic energy barrier for water splitting. In particular, during the OER process, their structural stability is markedly superior to that of traditional noble-metal catalysts (such as IrO₂and RuO₂), demonstrating a broad range of application prospects^[109-110].

In summary, existing scoring systems have limited predictive capabilities for bleeding events, and their results are inconsistent^[25,30,33].. Liu et al.^[113]prepared the rock-salt structured high-entropy oxide Mg_0.2Co_0.2Ni_0.2Cu_0.2Zn_0.2O. By introducing Mg (1.60 Å) and Zn (1.39 Å), which have significantly different atomic radii, high concentrations of cation and oxygen vacancies are induced, facilitating the exposure of unsaturated active sites and thereby promoting reactant adsorption and electron transfer (Fig. 8d). In addition, the design of an electronegativity gradient—by introducing low-electronegativity elements Mg and Zn—results in a greater transfer of oxygen charge to neighboring Co and Ni, strengthening the Co/Ni-O covalent bonds. Consequently, compared with binary oxides, the Mg_0.2Co_0.2Ni_0.2Cu_0.2Zn_0.2O catalyst exhibits superior OER performance (Figs. 8e and f). The regulation of metal–oxygen covalent bonds often triggers dynamic changes in metal oxidation states, and there is a complex interrelationship between such valence state evolution and catalytic activity. Kim et al.^[114]developed a rock-salt structured high-entropy oxide (Mg_0.2Fe_0.2Co_0.2Ni_0.2Cu_0.2)O using a preparation strategy combining hydrothermal synthesis with annealing. By systematically removing a single cation (Mg, Fe, Co, Ni, or Cu), they prepared a series of medium-entropy oxides, enabling an in-depth analysis of the dynamic transformation rules of different metal oxidation states. Experimental results indicate that the presence of Cu²⁺ effectively inhibits the conversion of Fe²⁺ and Co²⁺ to the higher oxidation states Fe³⁺ and Co³⁺, while the enhancement of catalytic activity is closely related to the increase in Co³⁺ concentration within the system. This provides a new theoretical basis for the rational design of OER catalysts.

High-entropy spinel oxides, with their unique AB₂O₄ crystal structure, exhibit excellent catalytic properties. Depending on the occupancy differences of multi-component metals, these materials can be classified into two typical configurations: MB₂O₄ (where multiple metals occupy the tetrahedral A site) and AM₂O₄ (where multiple metals occupy the octahedral B site). Recent studies have revealed a close relationship between structural reconstruction and catalytic performance: Wang et al.^[115] prepared Zn(CrMnFeCoNi)₂O₄, a high-entropy spinel oxide, using a low-temperature sol-gel method. The results indicate that electrochemical cycling induces surface reconstruction in the material, with Zn²⁺ selectively leaching from the tetrahedral sites and thereby promoting the formation of a high-entropy interface rich in metal oxyhydroxides at the surface. The newly formed surface exhibits improved OER catalytic performance as characterized by AEM. Zhang et al.^[116] used electrospinning to prepare (Ni_0.2Co_0.2Zn_0.2Cu_0.2Mg_0.2)Fe₂O₄, a nanofiber spinel oxide, which demonstrates greater catalytic stability than unary, binary, ternary, and quaternary oxides. XPS, TEM, and in-situ Raman analyses confirm that surface reconstruction occurs during the OER process. Currently, no specific studies have established the structure–activity relationship between the MB₂O₄ and AM₂O₄ configurations and catalytic activity; however, by tuning the spinel structure and leveraging the synergistic effects between tetrahedral and octahedral sites to promote O–O coupling, it may be possible to overcome the adsorption limitations of conventional AEMs.

In recent years, high-entropy perovskites have emerged as a new direction in OER catalyst design due to their unique structural stability and tunable electronic properties. Compared with traditional perovskites, in which the A and B sites are occupied by only 1–2 elements, high-entropy systems form solid-solution structures by incorporating five or more metal elements. This not only significantly enhances thermodynamic stability but also endows the materials with uniquely tunable electronic structures. Among these, ABF₃-type perovskite fluorides exhibit particular advantages in modulating the metal–oxygen coordination environment due to the high electronegativity and strong ionization ability of fluorine; however, the issue of fluorine loss in strongly acidic or alkaline electrolytes remains unresolved^[117].To address the stability bottleneck, Wang et al.^[118]prepared a series of high-entropy perovskite fluorides using ball milling (Fig. 8g). Among them, (BSCF)_3/4[KM(Ⅱ)F₃]_1/4exhibited excellent OER stability, operating for more than 20 hours at 10 mA cm^-2. The mechanical ball-milling method, which does not require solvents or high-temperature conditions, has become an ideal approach for achieving solid-solution formation between perovskite oxides and heat-sensitive halides. By employing this mechanical ball-milling synthesis method, oxide–fluoride composite solid solutions can be further induced (Fig. 8h). Compared with pure perovskite oxides, the oxide–fluoride composite catalysts exhibit higher catalytic activity in OER (Fig. 8i). Nguyen et al.^[119]prepared a high-entropy perovskite oxide, La(Cr_1/6Mn_1/6Fe_1/6Co_2/6Ni_1/6)O₃, via co-precipitation. This material exhibits an OER overpotential of 325 mV at 10 mA·cm^-2and can operate stably for more than 50 hours. In contrast, the corresponding low-entropy perovskite oxides display relatively poorer catalytic activity, such as: LaCoO₃(380 mV), LaNiO₃(380 mV), LaCrO₃(420 mV), LaFeO₃(437 mV), and LaMnO₃(460 mV).

By activating lattice oxygen in transition metal oxides to circumvent the linear relationship inherent in traditional AEMs, OER kinetic efficiency can be effectively enhanced. Experimental and theoretical studies indicate that shifting the O 2p center closer to the Fermi level (E _F) is the core condition for triggering LOM in transition metal oxides. However, this mechanism faces a critical challenge: excessive formation of oxygen vacancies can lead to catalyst structural collapse, thereby reducing catalyst stability and essentially creating a trade-off between catalytic activity and structural stability. The high-entropy effect of HEOs endows them with superior thermodynamic and kinetic stability, enabling effective maintenance of catalyst structural stability during oxygen vacancy formation. This offers a new direction for addressing the structural collapse issue in conventional LOM-based catalysts. It should be noted, however, that due to the complex microenvironment on the HEO surface, where multiple elements coexist, the activation of lattice oxygen still presents challenges—for example, the synergistic electronic regulation among various metals is difficult to precisely control. These factors make band structure modulation for HEO lattice oxygen activation a pressing issue that urgently needs to be addressed.

3.3 High-entropy metal-organic frameworks

In 2019, the concept of high entropy was introduced into MOF systems^[120],giving rise to a new class of MOF materials. HEMOFs refer to MOFs that contain five or more metal elements in their framework. Although they exhibit excellent electrochemical activity, issues such as low electrical conductivity and poor stability hinder their large-scale application. To address these challenges, HEMOFs are converted via pyrolysis or oxidation into carbon-based high-entropy alloys/oxides/sulfides. The formation of these derivatives can enhance stability and electrochemical activity. At the same time, the derived metal carbides, oxides, and sulfides can still retain the framework structure and high specific surface area, facilitating electron transfer and effectively improving electrical conductivity. Moreover, during the pyrolysis process, the aggregation of metal particles can be effectively suppressed. Consequently, HEMOFs and their derivatives have experienced rapid development and application in energy-related fields.

In summary, the choice of synthesis strategy has a critical impact on the structure and properties of HEMOF materials. Xu et al.^[121]synthesized a NiCoFeZnMo-MOF with a two-dimensional array structure via a mild solvothermal method. This two-dimensional structure not only enhances electrical conductivity but also facilitates mass transfer of intermediates. The resulting electrocatalyst exhibits excellent OER activity, requiring an overpotential of only 254 mV to achieve a current density of 50 mA·cm^-2and demonstrating long-term stability for up to 100 hours. Zhao et al.^[96]prepared a MnFeCoNiCu-MOF using a solution-phase method (coordinating Mn²⁺, Fe³⁺, Co²⁺, Ni²⁺, and Cu²⁺ with 1,4-benzenedicarboxylic acid at room temperature), achieving equimolar and uniform dispersion of the metal ions. At a current density of 10 mA cm^-2,the overpotential of this sample is as low as 245 mV, and after a 48-hour stability test, the activity retention rate remains at 95.1%, outperforming products prepared by the solvothermal method. In contrast, the MnFeCoNiCu-MOF prepared by the solvothermal method is limited by its slow kinetic process, which can easily lead to local enrichment of Fe³⁺ and ultimately result in a significant decline in catalytic activity.

The unique geometric structure of HEMOFs plays a regulatory role in their catalytic performance. Wang et al.^[122]used a CoNiCuMnAl-MOF as a precursor and then thermally decomposed it in an argon atmosphere to obtain a core–shell structure coated with an ultrathin carbon layer. The carbon layer plays a crucial role in both electron transport and corrosion protection: First, the graphene-like carbon layer acts as a highly efficient electron conductor, accelerating interfacial charge transfer while facilitating mass transport^[123-124]. Second, the carbon layer physically isolates the active metal sites within the core, preventing corrosion by the electrolyte and endowing the catalyst with long-term stability^[125]. The synergistic effects described above enable this catalyst to achieve an overpotential of only 215 mV for alkaline OER at 10 mA·cm^-2and to operate continuously for 30 hours at a high current density of 200 mA·cm^-2without significant degradation. However, there is a trade-off between carbon-layer thickness and catalytic activity: an excessively thick carbon layer can shield the metal active sites and impede electron transport, leading to a decline in intrinsic activity^[124,126]. Therefore, the structure–activity relationship between carbon-layer thickness and catalytic activity still requires further investigation. Miao et al.^[39]successfully prepared a ZnFeNiCuCoRuO catalyst with a hollow polyhedral structure using an ion-exchange method (Fig. 9a). HAADF-STEM confirmed that the six metal components are uniformly distributed within the polyhedral structure (Fig. 9b). This catalyst exhibits excellent OER activity and stability under alkaline conditions, with an overpotential of 170 mV at a current density of 10 mA·cm^-2, a Tafel slope of 56 mV·dec^-1, and a 7% decrease in activity after 30 hours of operation. Theoretical calculations show (Figs. 9c–f) that in this unique hollow structure, the maximum energy barrier for the AEM pathway at the Ru–Fe site is 0.86 eV, which is lower than the 0.99 eV for the LOM pathway, indicating that the catalyst is more likely to follow the AEM reaction pathway during OER. Further comparative analysis reveals that, compared with the Ru–Co and Ru–Ni sites, the Ru–Fe site exhibits superior catalytic efficiency in the AEM pathway, with energy-barrier values approximately 0.72 eV and 0.21 eV lower than those for Ru–Co and Ru–Ni, respectively. Li et al.^[127]prepared (MnFeCoNiCu)S₂nanoparticles using a MOF as a precursor. The overpotential at 10 mA·cm^-2is 221 mV. This is attributed to the strong synergistic effects among the multiple metals, which establish a stable electronic structure and provide a favorable local coordination environment, thereby significantly enhancing the catalytic performance.

显示原图|下载原图ZIP|生成PPT

图9 （a） ZnFeNiCuCoRuO制备示意图；（b） HAADF-STEM图像及对应的元素分布图；（c） ZnFeNiCuCoRuO模型；（d） AEM和LOM机理示意图以及其（e）自由能；（f）在Ru-Fe、Ru-Co、Ru-Ni位点上AEM的自由能^[39]

Fig.9 （a） Schematic illustration of the synthesis of ZnFeNiCuCoRuO. （b） HAADF-STEM image and corresponding elemental mapping of ZnFeNiCuCoRuO. （c） The model of ZnFeNiCuCoRuO. （d） Schematic illustration of AEM and LOM. （e） The relative free energy of AEM and LOM pathways. （f） The energetics of AEM on Ru-Fe， Ru-Co， and Ru-Ni site， respectively^[39]. Copyright 2023， Wiley

Due to their unique structural tunability, HEMOFs exhibit significant advantages in the field of catalysis. The organic ligands and metal nodes within their three-dimensional lattice structure can be selectively modified, enabling precise control over material properties. In particular, for OER, these materials offer a triple advantage: First, the stable porous framework provides robust support for long-term catalytic processes; second, the synergistic effects of multiple metals can effectively modulate the electronic band structure, reducing the reaction energy barrier; third, the rich hierarchical meso-microporous structure not only exposes more active sites but also creates rapid mass-transfer channels, accelerating the diffusion of reactants and the transfer of electrons. However, enhancing the OER performance of current HEMOF systems still faces two core challenges. In terms of structural regulation, there is a critical trade-off in the optimized design of pore systems: While large pores can significantly enhance the mass-transfer kinetics of reactants, existing HEMOF systems predominantly feature microporous structures. Although micropores provide a higher specific surface area, they also lead to a substantial increase in mass-transfer resistance. How to achieve hierarchical pore construction through chemical coordination modulation, thereby improving mass-transfer efficiency while maintaining a high specific surface area, may be the key to overcoming this predicament. In terms of material synthesis, the atomically uniform dispersion of multi-metal components poses significant thermodynamic challenges. Complex factors such as lattice mismatch between different metal elements, differences in binding energies, and varying preferences for coordination geometries make it extremely difficult to achieve stable high-entropy structures at the nanoscale.

3.4 Other High-Entropy Compounds

Nonmetal elements (B, P, S, Se, etc.), due to their significant electronegativity differences from transition metals, serve as effective dopants for tuning the electronic structure of high-entropy materials. These elements synergistically optimize catalytic activity and stability through mechanisms such as interstitial doping, the formation of anionic protective layers, and the induction of dynamic phase transitions. In particular, B, with its small atomic radius, can enter the metal lattice in an interstitial solid-solution form, preserving the spatial continuity of the host metal bonds while inducing a local redistribution of electron cloud density, thereby creating an "electron trap" effect^[128].When B forms covalent compounds with metals, its unique electronic layer structure, combined with its high electronegativity and ionization energy, leads to diverse bonding forms and bond strengths^[129].

Liu et al.^[130]used Cu₂O as a template to prepare three-dimensional nanoboxes composed of FeCoCuMnRuB nanosheets. This material exhibits excellent OER performance, requiring only a 233 mV overpotential at a current density of 10 mA·cm^-2, with a Tafel slope of 61 mV·dec^-1 and operational stability of up to 200 hours. Its overall performance significantly outperforms that of conventional HEAs and low-entropy boride systems. Mechanistic studies indicate that boron atoms capture metal electrons to cooperatively regulate the electronic structure of surface active sites, promoting the formation of high-valence metal species and thereby significantly accelerating OER kinetics (Fig. 10a, b). In addition, theoretical calculations show that B doping can effectively reduce the adsorption and transformation energy barriers of oxygen-containing intermediates (Fig. 10c). Dynamic phase transitions are an important characteristic of non-metal-doped systems. Zhang et al.^[131]synthesized an amorphous porous high-entropy metal borate (CrMnCoNiFe)_0.2BO _xvia a simple solvothermal method. During the OER process, Cr vacancies are generated in this catalyst, and the emergence of these vacancies promotes surface reconstruction, effectively reducing the adsorption and transformation energy barriers of oxygen-containing intermediates. Liu et al.^[132]prepared an S-doped (CrMnFeCoNi)₃O₄ catalyst under air using a microwave-assisted solution combustion strategy. During the OER process, S leaching induces surface reconstruction of the catalyst and promotes the formation of highly active (oxygen) hydroxides. Consequently, catalysts with high sulfur content exhibit better electrocatalytic activity than those without sulfur doping.

显示原图|下载原图ZIP|生成PPT

图10 （a） FeCoCuMnRu和（b） FeCoCuMnRuB在不同电压下的原位拉曼^[130]，以及（c） FeCoCuMnRu和FeCoCuMnRuB吉布斯自由能变化图^[130]，（d） FeCoNiCuYP/C催化剂中Fe-OH、Co-OOH、Ni-OOH中间体的电荷密度差^[135]

Fig.10 In situ Raman spectra of （a） FeCoCuMnRu alloy and （b） FeCoCuMnRuB boride at different applied voltages^[130]. （c） The change in Gibbs free energy of FeCoCuMnRu alloy and FeCoCuMnRuB boride involved in the OER process^[130]. Copyright 2024， American Chemical Society. （d） Differential charge density of Fe-OH， Co-OOH， Ni-OOH intermediate of FeCoNiCuYP/C catalysts^[135]. Copyright 2024， Wiley

In summary, existing scoring systems have limited predictive capabilities for bleeding events, and their results are inconsistent^[25,30,33].

In summary, the synergistic effects among multiple non-metals expand the dimensions for performance optimization. For example, Li et al.^[135]incorporated Fe, Ni, Co, Cu, and Y metal elements into MOFs via electrodeposition, then used the resulting MOF as a template for pyrolysis to prepare a high-entropy compound FeCoNiCuYP/C with a porous nanosheet structure. Combined theoretical calculations and experimental results indicate that the introduction of Fe, Co, and Ni elements can effectively modulate the electronic structure of FeCoNiCuYP, optimizing the bond strength of metal–phosphorus bonds and the adsorption energies of reaction intermediates. Specifically, Fe tends to stabilize the *OH intermediate, while Co/Ni enhances the formation of the *OOH intermediate. This unique configuration effectively breaks the linear scaling relationship between the adsorption of *OH and *OOH intermediates during AEM (Fig. 10d).Recently, Zhang et al.^[136]proposed a strategy for forming a coexisting high-entropy metal–high-entropy non-metal system, using aminotriazole as a “bonding agent” for the high-entropy precursor. This not only binds five metals—Cr, Mn, Fe, Co, and Ni—but also introduces nitrogen and carbon elements in situ during reduction. Subsequently, through simultaneous phosphidation, sulfidation, and surface oxidation, a unique high-entropy composite containing five metals (Cr, Mn, Fe, Co, Ni) and five non-metals (C, N, O, P, S) was successfully prepared. The material exhibits excellent electrochemical activity: as an OER electrocatalyst, it shows an overpotential of 211.9 mV (@10 mA·cm^-2), with outstanding stability exceeding 25 hours. When used as an ORR electrocatalyst, it delivers an initial voltage of 0.977 V, a half-wave potential of 0.841 V, and maintains good electrochemical stability over 25 hours. This work highlights the great potential of multi-element synergy in constructing adaptive catalytic activity.

Taking the introduction of nonmetal elements with different electronegativity characteristics as the entry point, high-entropy compounds exhibit significant advantages in electronic structure regulation and adsorption energy optimization. Based on electronegativity parameters, nonmetal elements can be classified into two major categories: electron-donating elements (low electronegativity: B) and electron-accepting elements (high electronegativity: P, S, and Se). By synergistically introducing both electron-donating and electron-accepting elements into a high-entropy system, a unique electron-synergistic enhancement mechanism can be established. This synergistic effect drives the evolution of high-entropy compounds from single-nonmetal systems to multi-nonmetal systems. Although the construction of multi-nonmetal systems can significantly enhance OER performance, their complex composition gives rise to three key challenges: First, the study of the structure–property relationships among components, structure, and performance becomes exponentially more challenging, as the mechanisms of metal–metal, metal–nonmetal, and nonmetal–nonmetal synergistic interactions, along with the dynamic evolution pathways of active sites, remain incompletely elucidated; second, multi-nonmetal systems are prone to impurity incorporation or elemental segregation during synthesis, placing higher demands on precise synthesis methods; and third, the excessive introduction of nonmetal elements may lead to a decline in material conductivity, thereby limiting the efficiency of rapid charge transport during electrocatalysis.

4 Optimization Strategies for the OER Performance of High-Entropy Materials

In essence, optimizing the electrocatalytic OER performance revolves around two core dimensions: first, increasing the density of active sites through geometric structure engineering (such as constructing hierarchical porosity or exposing specific crystal facets); second, enhancing intrinsic activity through electronic structure modulation (such as adjusting the d-band center position or enhancing orbital hybridization)^[137].High-entropy materials, with their multi-component nature and broad compositional tunability, provide a unique design platform for developing efficient and stable catalysts. This section outlines three specific strategies for optimizing the catalytic performance of high-entropy materials, including machine learning–assisted design, surface defect engineering, and elemental regulation (Figure 11).

显示原图|下载原图ZIP|生成PPT

图11 高熵材料的OER性能优化策略示意图

Fig.11 Schematic diagram of optimization strategies for OER performance in high-entropy materials

4.1 Machine Learning–Assisted Design

The principles of machine learning–assisted design of high-entropy OER catalysts can be divided into three key stages: First, a high-entropy element database is constructed, represented by transition metal combinations, integrating their physicochemical characteristics such as electronegativity and atomic radius, as well as experimental performance parameters like overpotential and stability. Next, machine learning algorithms are used to establish a nonlinear correlation model between multidimensional features and OER activity, enabling intelligent prediction of potentially superior compositions. Finally, high-throughput computing or active learning methods are employed to screen candidate materials, forming a closed-loop design process of “prediction–validation–optimization.”

Currently, machine learning (ML) model construction primarily employs three types of algorithms: neural networks (NN), support vector machines (SVM), and random forests (RF). NNs are computational models inspired by the structure of biological neural networks. Their basic building blocks are artificial neurons, which mimic the signal transmission mechanisms between biological neurons. A typical neural network architecture comprises three main parts: an input layer, hidden layers, and an output layer. Neurons in each layer receive weighted signals from the previous layer, apply a nonlinear transformation via an activation function, and then pass the processed results to the next layer. By optimizing the connection weights between neurons using the backpropagation algorithm, the network can autonomously extract data features and build predictive models, thereby enabling ML functions such as pattern recognition and classification prediction^[138-139].This algorithm excels in modeling complex nonlinear relationships but requires large datasets for support. SVM is a supervised learning algorithm based on statistical learning theory, with its core principle being data classification through the construction of a maximum-margin separating hyperplane in the feature space^[140]. The algorithm uses kernel techniques to map low-dimensional, non-linearly separable data into high-dimensional space for linear separation, obtaining the optimal solution by solving a convex quadratic programming problem. Although SVM demonstrates significant advantages in addressing nonlinear, high-dimensional, and small-sample problems, it has relatively high computational complexity for large-scale training and is sensitive to missing data and parameter tuning. RF excels at handling multi-scale feature interactions but carries a risk of overfitting. The algorithm first creates numerous samples from the original dataset, then trains a decision tree using each sample; to introduce diversity and prevent overfitting, a random subset of features is considered at each node of the decision tree; the final prediction is the average of the predictions from all trees^[141](Table 2provides a detailed comparison of the characteristics of each algorithm).

表2 NN、SVM和RF模型优点与缺点^[149]

Table 2 Advantages and disadvantages of NN， SVM and RF model^[149]

Model	Advantages	Disadvantage
NN	（1） Powerful for non-linear relationships and complex model	（1） Prone to overfitting with limited training data
	（2） Robust to noisy data	（2） Requires careful tuning of parameters
	（3） Ability to learn from large datasets	（3） Black-box nature makes interpretation difficult
SVM	（1） Effective in high-dimensional feature spaces	（1） Can be slow to train on large datasets
	（2） Works well with small to medium-sized datasets	（2） Sensitivity to choice of kernel parameters
	（3） Versatile due to kernel trick for non-linear classification	（3） Memory-intensive for large-scale problems
RF	（1） High accuracy and robustness	（1） Can be slow to predict on large datasets
	（2） Effective for high-dimensional feature spaces	（2） Lack of interpretability due to ensemble nature
	（3） Handles missing values while preserving accuracy	（3） Potential overfitting on noisy datasets without regularization

In traditional catalysis, the limitations of trial-and-error experimentation in catalyst optimization are particularly pronounced in high-entropy material systems. Faced with an almost infinite compositional space, conventional methods suffer from inherent drawbacks such as long R&D cycles and high resource consumption. This challenge has given rise to a new ML-driven paradigm for materials design: by establishing quantitative mapping relationships among "composition–structure–performance," it becomes possible to efficiently screen and predict the performance of high-entropy catalysts^[142].In particular, ML demonstrates unique potential in geometric structure design. Prediction models built using ML have revealed the structure–activity relationships among three key parameters in porous MOF materials: the Zeta potential of the dispersion, particle size, and catalytic activity are interrelated. Data analysis indicates that an increase in the absolute value of the Zeta potential not only enhances the stability of the dispersion but also leads to a refinement of the MOF particles (manifested as an increase in specific surface area). This geometric structural evolution effectively enhances the electrocatalytic performance of the material. Experimental validation confirms that the trends in Zeta potential and catalyst performance are highly consistent with the predictions of the ML model^[143].Tukur et al.^[144] combined ML with DFT methods to systematically analyze how the structure, composition, and atomic chemical environment of high-entropy perovskite oxides influence the formation of oxygen vacancies. The research team developed a predictive model for oxygen vacancy formation energy based on material characteristics, enabling high-throughput computation and precise identification of oxygen vacancy concentrations, and successfully screened high-entropy perovskite oxides with outstanding catalytic activity.

Addressing the core challenge of data scarcity in high-entropy materials, closed-loop optimization strategies based on active learning (AL) demonstrate unique advantages: by establishing an iterative cycle of “prediction–experimental validation–model update,” they can reduce data requirements^[145]. Nie et al.^[146]developed an AL framework that has achieved breakthrough progress in HEO screening. This method operates as a closed-loop system, iterating through three stages—“training, prediction, and experiment” (Fig. 12a). Through multiple AL iterations, the method successfully identified four new HEOs from a large pool of potential constituents. These HEOs exhibit outstanding stability and demonstrate an exceptional H₂evolution rate of 251 μmol·g_cat ^-1·min^-1in the water–gas shift reaction, surpassing the benchmark set by existing catalysts such as Pt/γ-Al₂O₃(135 μmol·g_cat ^-1·min^-1) and Cu/ZnO/Al₂O₃(81 μmol·g_cat ^-1·min^-1). Kim et al.^[147]combined the Pareto active learning approach with experimental studies, employing an algorithm-driven, efficient materials screening strategy. After 11 rounds of iterative optimization (Fig. 12b), they successfully screened 110 representative, valid experimental data points from 77,946 candidate materials and used these data points to develop a novel bifunctional electrolysis catalyst for water splitting. The model built using this method indicates that the developed bifunctional catalyst exhibits excellent catalytic activity in both OER and HER. Experiments confirm that the Pt_0.15Pd_0.30Ru_0.30Cu_0.25catalyst achieves efficient water splitting at a current density of 10 mA·cm^-2with an applied voltage of only 1.56 V (Fig. 12c). In terms of optimizing more complex multi-component systems, Perumal et al.^[148]developed a catalyst discovery framework by combining Bayesian optimization with an AL model, enabling effective exploration of the vast compositional space of octonary high-entropy materials suitable for both HER and OER. This framework facilitates dynamic, user-guided interactions with machine learning and rapidly identifies the optimal compositional ratios while minimizing the need for experimental data. Experiments show that the octonary high-entropy nanocatalyst has a minimum particle size of just 2 nm; at a current density of 10 mA·cm^-2, the overpotential for HER is 12 mV, and the overpotential for OER is 239 mV, with catalytic performance that outperforms commercial Pt/C and IrO₂catalysts.

显示原图|下载原图ZIP|生成PPT

图12 （a）主动学习过程示意图^[146]；（b） Pareto主动学习模型在每个步骤的计算结果及对应数据点；（c）第一个模型和最后一个模型中性能最佳的三种催化剂的HER与OER的极化曲线^[147]；（d） E-FeCoNiAlZn于缺陷处OER反应机理（LOM亚型-COM）示意图；（e） E-FeCoNiAlZn和FeCoNiAlZn在OER过程中各反应步骤的吉布斯自由能变化图；（f） E-FeCoNiAlZn LDH的COM理论模型涉及O、O+OH和OO的吸附（黄色、蓝色、白色、灰色、银色、红色、绿色和粉色球分别代表Fe、Co、Ni、Zn、Al、O、B和H）^[56]；（g）多种阳离子与阴离子混合反应示意图；（h） γ-FeCoNi₂（OH）₈和γ-FeCoNi₂F₄（OH）₄中Co与O的态密度图；（i） γ-FeCoNi₂F₄（OH）₄原位拉曼图，（j） γ-FeCoNi₂（OH）₈和γ-FeCoNi₂F₄（OH）₄在AEM路径中吉布斯自由能变化图^[160]

Fig.12 （a） Training process of AL^[146]. Copyright 2024， American Chemical Society. （b） Result of the Pareto active learning model for each step and the corresponding data points. （c） LSV curve of HER and OER for the top three catalysts discovered in the 1st model and the last model^[147]. Copyright 2023， Wiley. （d） Schematic illustration of the OER reaction mechanism （sub-LOM， COM） at defect sites in E-FeCoNiAlZn. （e） Free energy for OER of E-FeCoNiAlZn LDH and FeCoNiAlZn LDH. （f） The theoretical models of COM on E-FeCoNiAlZn LDH involved the adsorption of *O， *O+*OH and *OO （The yellow， blue， white， gray， silvery， red， green and pink balls represent Fe， Co， Ni， Zn， Al， O， B and H， respectively）^[56]. Copyright 2025， Elsevier. （g） Schematic diagram of the interaction between various cations and anions. （h） Calculated PDOS plots of Co and O for γ-FeCoNi₂（OH）₈ and γ-FeCoNi₂F₄（OH）₄. （i） In situ Raman spectra of γ-FeCoNi₂F₄（OH）₄. （j） Free energy of γ-FeCoNi₂（OH）₈ and γ-FeCoNi₂F₄（OH）₄ in the AEM routes^[160]. Copyright 2024， Wiley

Machine learning demonstrates significant advantages in the fields of compositional design and geometric structure design for high-entropy materials: By efficiently analyzing vast amounts of elemental combinations and performance data, ML models can rapidly predict phase stability in multi-component alloys and the rules governing synergistic interactions among elements, thereby overcoming the limitations of traditional trial-and-error approaches. In geometric structure optimization, machine learning automatically uncovers nonlinear relationships between microstructural features and macroscopic properties, and, in conjunction with generative models, explores complex structural spaces that are difficult to cover using conventional theories. This significantly accelerates high-throughput design of high-entropy materials while reducing experimental costs. These groundbreaking advances mark the official entry of high-entropy material research and development into the “Machine Learning+” era.

4.2 Defect Engineering

Defect engineering, as a key strategy for regulating catalytic activity, exhibits unique advantages in high-entropy material systems. Due to their high free energy, defects exhibit chemically high reactivity, enabling them to preferentially adsorb reaction intermediates and serve as active sites^[150]. The compositional and structural complexity of high-entropy materials naturally endows them with a rich variety of defect types (such as vacancies, dislocations, grain boundaries, etc.), providing a multidimensional design space for the precise tuning of catalytic performance^[151]. Studies have demonstrated that systematic control of defect characteristics can effectively overcome the kinetic bottlenecks in electrocatalytic reactions^[152].

In the context of constructing metal vacancy defects, Kuang et al.^[76]employed a chemical etching strategy, using Zn as a sacrificial element to create metal vacancies in FeCoNiZn high-entropy metal oxides. This defect engineering significantly reduced the OER overpotential to 259 mV at a current density of 10 mA·cm^-2. Further structural characterization revealed that the defect-engineered catalyst dynamically formed abundant oxygen vacancy defects during the OER process. DFT calculations were used to analyze the reaction pathways, revealing that the introduction of defect structures significantly lowered the coupling energy barrier between adsorbed oxygen intermediates on metal active sites and lattice oxygen, thereby effectively promoting the LOM reaction pathway. Building on this finding, the team further introduced Al to prepare E-FeCoNiAlZn LDH^[56]. This defect engineering exhibits a dual effect: on the one hand, unsaturated coordinated metal sites enhance the local surface charge density, creating strong chemisorption “hot spots” that promote interfacial OH⁻ enrichment and lay the groundwork for oxygen–oxygen coupling; on the other hand, the upward shift of the oxygen 2p energy level weakens the metal–oxygen bond strength, making lattice oxygen more readily available for participation in the OER (Fig. 12d). This dual effect reduces the maximum energy barrier of the LOM subtype pathway (Figs. 12e, f), ultimately achieving an ultra-low overpotential of 220 mV at 10 mA·cm^-2 and maintaining stability for 100 hours at a high current density of 100 mA·cm^-2.

Oxygen vacancy defect engineering is also a potentially effective approach to enhance the electrocatalytic activity of metal oxides. Oxygen vacancies can serve not only as new adsorption sites but also as an effective means of altering the electronic structure of the catalyst surface, thereby regulating the efficiency of catalytic processes^[153].Liu et al.^[154]introduced oxygen vacancies into CoMnFeNiZnO through low-temperature surface carbonization-decarbonization, and electron paramagnetic resonance spectroscopy (EPR) confirmed the presence of abundant oxygen vacancies in this material. These oxygen vacancies not only serve as hydroxyl adsorption sites but also reduce the OER overpotential by 60 mV (284 mV@10 mA·cm^-2) through modulation of the electronic structure. Experimental results combined with theoretical analysis indicate that the local charge redistribution induced by oxygen vacancies can accelerate the pre-oxidation process.

In addition to point defects, line defects, planar defects, and amorphization have opened up new dimensions in defect engineering, similarly influencing catalyst performance. Huang et al.^[155]prepared face-centered cubic (FCC) high-entropy alloy nanorods via pyrolysis of MnFeCoNiCu-MOF; the internal twins and dislocations increased the surface free energy, significantly reducing the OER activation energy and accelerating the OER process. Compared with crystalline catalysts, amorphous catalysts expose more active sites and are more prone to structural reorganization, thereby exhibiting higher catalytic activity^[156].Wang et al.^[157]prepared an amorphous CoFeNiMnZnB high-entropy boride via chemical reduction. This amorphous metal boride exhibits excellent OER performance, including a low overpotential of 261 mV at 10 mA·cm^-2 and a low Tafel slope. The superior OER performance of CoFeNiMnZnB is attributed to the synergistic interactions among different metals, the leaching of Zn ions, the generation of oxygen vacancies, and the formation of amorphous hydroxide oxides on the CoFeNiMnZnB surface during OER.

Defect engineering achieves precise control over the electronic structure of high-entropy materials by introducing defects of various dimensions, such as point, line, and planar defects. On one hand, it optimizes the d-band center position at the transition-metal active sites, reducing the adsorption energy barrier of oxygen-containing intermediates (*OOH) and promoting OER kinetics. On the other hand, the localized redistribution of charge enhances the degree of metal–oxygen orbital hybridization, activating lattice oxygen to participate in OER and circumventing the limitations imposed by the AEM linear scaling relationship. At the same time, the synergistic effect of multiple elements in high-entropy materials, combined with defect-induced lattice strain, exposes more unsaturated coordination sites, further enhancing catalytic activity and long-term stability.

4.3 Elemental regulation

In summary, the core of performance regulation in high-entropy materials lies in the precise design of elements. Depending on the nature of the active sites, this can be broadly categorized into two types: modulation by metal cations and doping with non-metal anions. In terms of metal-site engineering, the synergistic effects of multiple metals can effectively tune the electronic structure of the sites, thereby optimizing catalytic activity and stability, facilitating electron transfer, and exposing more active sites. Take spinel-type HEOs as an example: Zhai et al.^[158]developed a Fe_1.5CoNiZr_0.5M oxide system (M = Cu, Zn, Mn, Cr) through a multi-metal synergy strategy. Their work demonstrated that the introduction of extrinsic cations induces lattice distortion and modulates the position of the d-band center. When Cr is incorporated to form Fe_1.5CoNiZr_0.5CrO _x, the high occupancy at the octahedral sites endows the material with outstanding OER performance: at a current density of 10 mA·cm^-2, the overpotential is only 280 mV, the Tafel slope is 43.39 mV·dec^-1, and after 50 hours of continuous operation, the material still retains 99.6% of its initial activity. The introduction of high-valence metal cations often significantly enhances the intrinsic OER activity; for instance, Antink et al.^[159]prepared (CrFeCoNiMo)₃O₄, a high-entropy spinel nanosheet in which the high-valence Cr³⁺/Mo⁶⁺ effectively activate lattice oxygen by weakening the metal–oxygen bond strength, thereby promoting the oxygen evolution pathway via the lattice oxygen mechanism (LOM).

In terms of non-metal doping, electron regulation based on differences in electronegativity has also proven highly effective. Li et al.^[160]intercalated F^- into the interlayer of FeCoNi LDH.Unlike conventional F^--intercalated LDH catalysts (which tend to lose F in alkaline electrolytes), in this study F is embedded within a single hydroxide layer and is not affected by electrochemical evolution. This structural stability can be attributed to an entropy-stabilization effect: even when the fluorine content exceeds that of OH^-, leading to the formation of a second phase Na₃FeF₆, this byproduct dissolves after electrochemical cyclic voltammetric activation, ultimately preserving a stable, single-phase fluorine-doped FeCoNi LDH structure (Fig. 12g). Due to the high electronegativity of F, higher-valence metal species are formed in FeCoNi LDH, enhancing the hybridization between metal 3d and O 2p orbitals (Fig. 12h), which effectively improves electronic conductivity and charge-transfer efficiency, thereby significantly enhancing the intrinsic OER activity. Moreover, the unique embedding mode of F helps maintain the integrity of the catalyst structure during the OER process (Fig. 12i). For this reason, it is difficult for oxygen vacancies to form in FeCoNi LDH during OER, and the catalyst thus follows the AEM pathway (Fig. 12j). The CoNiCuZnFeP nanocubic catalyst designed by Yuan et al.^[161], through a "dual-regulation" mechanism involving phosphorus vacancies and multi-metal synergy, achieves ultra-low overpotentials of 318 mV for HER and 204 mV for OER at a current density of 100 mA·cm^-2, while maintaining long-term stability for more than 700 hours. Theoretical calculations reveal that the synergistic effect of the high-entropy effect and phosphorus vacancies can redistribute local charge densities, significantly reducing the energy barrier for water splitting.

The ion doping concentration has a decisive impact on the structure–activity relationship of catalysts, a phenomenon that is prevalent in both metal cation and non-metal anion doping systems. Zhou et al.^[162]revealed the key mechanism of high-valence metal doping by tuning the Mo doping concentration in the FeCoNiCr system. The study showed that as the Mo content increases, the alloy’s crystalline structure undergoes a significant amorphization transition, with the FCC phase gradually evolving into an amorphous configuration. The concurrent reconstruction of the electronic structure causes the d-band center of the FeCoNiCrMox system to shift toward the Fermi level, thereby enhancing the adsorption energy. Thanks to the optimized electronic structure, the sample with the highest Mo content, FeCoNiCrMo_1.0,exhibits the most outstanding electrocatalytic performance. A similar mechanism is also observed in non-metal doping systems; Liu et al.^[132]introduced sulfur anions into the (CrMnFeCoNi)₃O₄ high-entropy oxide and found that the S doping level is linearly and negatively correlated with the material’s crystallinity. When the doping level reaches S2, the material’s electrochemical active surface area reaches its peak. Experimental characterization reveals that the dissolution behavior of S can trigger a dual surface reconstruction effect: first, it promotes the oxidation of metal elements to higher valence states; second, it accelerates the formation of MOOH structures through hydroxyl group adsorption at the interface. Through this synergistic effect, the HEO-S2 catalyst follows the LOM mechanism in OER, with an overpotential of 263 mV at a current density of 10 mA·cm^-2, which is significantly lower than that of the undoped system (326 mV).

Elemental regulation optimizes the OER performance of high-entropy materials by introducing elements with diverse electronegativities and atomic radii, thereby modifying their electronic structure. The primary regulation mechanisms are as follows: (1) Lattice distortion and chemical complexity arising from the high-entropy effect modulate the local coordination environment, creating highly active sites and reducing the adsorption energy barrier for the *OOH intermediate; (2) A downward shift in the d-band center of transition metals effectively weakens the over-adsorption of oxygen intermediates, thereby accelerating OER kinetics; (3) Synergistic regulation of multiple elements adjusts the density of electronic states near the Fermi level, enhancing conductivity and improving charge-transfer efficiency; (4) Strong intermetallic electronic interactions induce charge redistribution, promoting the formation of highly active oxidation states such as Co³⁺/Ni³⁺and weakening metal–oxygen bond strength to effectively activate lattice oxygen. Thus, elemental regulation, with its multidimensional electronic structure-modulating capabilities, demonstrates significant potential for optimizing OER activity.

5 Conclusion and Outlook

In recent years, high-entropy materials have become a hot topic in the field of OER electrocatalysts due to their unique structural advantages and tunable compositional properties. This article takes the OER mechanism as the main thread and systematically reviews the current research status of high-entropy materials as OER electrocatalysts, with a focus on their latest breakthroughs and bottlenecks in catalyst design and construction, the structure–property relationships among components, structure, and performance, and studies on reaction mechanisms. It also proposes multidimensional optimization strategies—such as machine learning–assisted design, defect engineering regulation, and precise elemental doping—to enhance OER catalytic performance. At present, although some progress has been made in the study of high-entropy oxygen-evolution catalytic materials, many scientific challenges still remain. The main existing problems and proposed solutions are outlined below.

(1) The design strategies for high-entropy material catalysts are limited. The compositional and structural complexity of high-entropy materials provides a unique platform for modulating adsorption energies, but also poses fundamental challenges to the precise design of catalysts. Although advanced computational methods (such as DFT and ML) have partially enabled high-throughput screening, existing strategies still have significant limitations: (i) they rely on high-throughput simulations to generate data, but computational results may deviate from experiments (e.g., due to thermodynamic approximation errors), and data under extreme conditions (such as high temperature and high pressure) remain scarce; (ii) data from different experimental conditions or simulation methods are difficult to standardize; (iii) empirical rules in materials science have not yet been effectively integrated into model architecture design. The breakthrough direction should focus on small-sample transfer learning strategies, using pre-trained single- or binary-component alloy database models to establish a cross-scale design framework tailored to high-entropy systems.

(2) The atom-level structure–activity relationship analysis is not precise enough. Designing efficient catalysts requires a clear understanding of the atom-level correlations between their structure and performance, as well as precise research into how microscopic changes in local electronic structure influence the macroscopic properties of materials. Due to the complex coordination environment of high-entropy surfaces, current understanding of the relationship between the evolution of microscopic electronic structure and macroscopic catalytic performance remains insufficient. Emerging in-situ characterization techniques (such as operando XAS) provide an important opportunity to reveal the multi-element synergistic effects during surface reconstruction processes, which will drive the development of a universal framework of catalytic principles.

(3) The catalytic mechanisms of high-entropy materials are difficult to regulate. The inherent linear relationship in the AEM mechanism limits its catalytic activity, while the LOM mechanism is often accompanied by structural collapse and instability. The unique thermodynamic and kinetic stability of high-entropy materials, along with their multiple active sites, offer new perspectives for overcoming this bottleneck: (i) enhancing structural stability under the LOM pathway by tuning entropy-stabilizing effects; (ii) constructing a synergistic dual-active-site system that promotes direct coupling of adsorbed oxygen between sites, suppresses the formation of the *OOH intermediate, shifts the reaction pathway toward the OPM mechanism, circumvents the limitations of the traditional AEM linear relationship, and significantly boosts OER kinetics. However, electron coupling among heterogeneous elements in a high-entropy environment increases the difficulty of regulating active sites, and related research is still in the early exploratory stage.

High-entropy oxygen-evolution catalytic materials, with their unique composition and structure, exhibit revolutionary application prospects in the field of electrocatalysis. With breakthroughs in in-situ characterization techniques and innovations in theoretical computational methods, it is foreseeable that through the deep integration of multi-scale design strategies and advanced characterization technologies, high-entropy oxygen-evolution catalytic materials will drive the development of novel, customized, efficient, and stable electrocatalytic systems, providing critical material support for the advancement of clean energy conversion technologies.

References

Publishing order | Descend order by publishing year | Descend order by cited within

[1]	Ghoniem A F. Prog. Energy Combust. Sci., 2011, 37(1): 15.

[2]	Kohse-Höinghaus K. Chem. Rev., 2023, 123(8): 5139.

[3]	Turner J A. Science, 2004, 305(5686): 972.

[4]	Chen G X, Tu X, Homm G, Weidenkaff A. Nat. Rev. Mater., 2022, 7(5): 333.

[5]	Qi J, Zhang W, Cao R. Adv. Energy Mater., 2018, 8(5): 1701620.

[6]	Walter M G, Warren E L, McKone J R, Boettcher S W, Mi Q X, Santori E A, Lewis N S. Chem. Rev., 2010, 110(11): 6446.

[7]	Liu L T, Zhai R R, Hu Y D. Energy, 2023, 276: 127386.

[8]	Wu Z P, Lu X F, Zang S Q, Lou X W D. Adv. Funct. Mater., 2020, 30(15): 1910274.

[9]	Suen N T, Hung S F, Quan Q, Zhang N, Xu Y J, Chen H M. Chem. Soc. Rev., 2017, 46(2): 337.

[10]	Blakemore J D, Crabtree R H, Brudvig G W. Chem. Rev., 2015, 115(23): 12974.

[11]	Halawa M I, Lai J, Xu G. Mater. Today Nano, 2018, 3: 9.

[12]	Luo M, Sun Y, Qin Y, Li Y, Li C, Yang Y, Xu N, Wang L, Guo S. Mater. Today Nano, 2018, 1: 29.

[13]	Lee Y, Suntivich J, May K J, Perry E E, Shao-Horn Y. J. Phys. Chem. Lett., 2012, 3(3): 399.

[14]	Zou X X, Zhang Y. Chem. Soc. Rev., 2015, 44(15): 5148.

[15]	Ni J, Shi Z P, Wang Y B, Yang J H, Wu H X, Wang P B, Xiao M L, Liu C P, Xing W. eScience, 2025, 5(2): 100295.

[16]	Xie X H, Du L, Yan L T, Park S, Qiu Y, Sokolowski J, Wang W, Shao Y Y. Adv. Funct. Mater., 2022, 32(21): 2110036.

[17]	Jiang Z Q, Fan C Z, Pan J Y, Shao L, Chen H, Pervaiz E, Dong Y, Wang T Z, Zheng X R, Li J H, Deng Y D. Rare Met., 2024, 43(7): 2891.

[18]	Hassan J Z, Zaheer A, Raza A, Li G. Sustain. Mater. Technol., 2023, 36: e00609.

[19]	Medford A J, Vojvodic A, Hummelshøj J S, Voss J, Abild-Pedersen F, Studt F, Bligaard T, Nilsson A, Nørskov J K. J. Catal., 2015, 328: 36.

[20]	Batchelor T A A, Pedersen J K, Winther S H, Castelli I E, Jacobsen K W, Rossmeisl J. Joule, 2019, 3(3): 834.

[21]	Wang K, Huang J H, Chen H X, Wang Y, Yan W, Yuan X X, Song S Q, Zhang J J, Sun X L. Electrochem. Energy Rev., 2022, 5(S1): 17.

[22]	Wang Z Y, You J H, Zhao Y, Yao R Y, Liu G Y, Lu J L, Zhao S Y. J. Environ. Chem. Eng., 2023, 11(1): 109080.

[23]	Xu X M, Shao Z P, Jiang S P. Energy Technol., 2022, 10(11): 2200573.

[24]	Li T Y, Yao Y G, Huang Z N, Xie P F, Liu Z Y, Yang M H, Gao J L, Zeng K Z, Brozena A H, Pastel G, Jiao M L, Dong Q, Dai J Q, Li S K, Zong H, Chi M F, Luo J, Mo Y F, Wang G F, Wang C, Shahbazian-Yassar R, Hu L B. Nat. Catal., 2021, 4(1): 62.

[25]	Dai S. ChemSusChem, 2020, 13(7): 1915.

[26]	Li T Y, Yao Y G, Ko B H, Huang Z N, Dong Q, Gao J L, Chen W, Li J G, Li S K, Wang X Z, Shahbazian-Yassar R, Jiao F, Hu L B. Adv. Funct. Mater., 2021, 31(21): 2010561.

[27]	Li S Y, Tang X W, Jia H L, Li H L, Xie G Q, Liu X J, Lin X, Qiu H J. J. Catal., 2020, 383: 164.

[28]	Löffler T, Meyer H, Savan A, Wilde P, Garzón Manjón A, Chen Y T, Ventosa E, Scheu C, Ludwig A, Schuhmann W. Adv. Energy Mater., 2018, 8(34): 1802269.

[29]	Jin Z Y, Lyu J, Zhao Y L, Li H L, Lin X, Xie G Q, Liu X J, Kai J J, Qiu H J. ACS Mater. Lett., 2020, 2(12): 1698.

[30]	Tao L, Sun M Z, Zhou Y, Luo M C, Lv F, Li M G, Zhang Q H, Gu L, Huang B L, Guo S J. J. Am. Chem. Soc., 2022, 144(23): 10582.

[31]	Mori K, Hashimoto N, Kamiuchi N, Yoshida H, Kobayashi H, Yamashita H. Nat. Commun., 2021, 12: 3884.

[32]	Pedersen J K, Batchelor T A A, Bagger A, Rossmeisl J. ACS Catal., 2020, 10(3): 2169.

[33]	Yusenko K V, Riva S, Carvalho P A, Yusenko M V, Arnaboldi S, Sukhikh A S, Hanfland M, Gromilov S A. Scr. Mater., 2017, 138: 22.

[34]	Tsai C F, Wu P W, Lin P, Chao C G, Yeh K Y. Jpn. J. Appl. Phys., 2008, 47(7R): 5755.

[35]	Fu X B, Zhang J H, Zhan S Q, Xia F J, Wang C J, Ma D S, Yue Q, Wu J S, Kang Y J. ACS Catal., 2022, 12(19): 11955.

[36]	Zhang D, Shi Y, Zhao H, Qi W J, Chen X L, Zhan T R, Li S X, Yang B, Sun M Z, Lai J P, Huang B L, Wang L. J. Mater. Chem. A, 2021, 9(2): 889.

[37]	Jiang Q, Lu R H, Gu J F, Zhang L, Liu K L, Huang M Y, Liu P, Zuo S Y, Wang Y L, Zhao Y, Ma P Y, Fu Z Y. Chem. Eng. J., 2023, 453: 139510.

[38]	Li P, Yao Y P, Ouyang W G, Liu Z, Yin H Y, Wang D H. J. Mater. Sci. Technol., 2023, 138: 29.

[39]	Miao K H, Jiang W D, Chen Z Q, Luo Y, Xiang D, Wang C H, Kang X W. Adv. Mater., 2024, 36(8): 2308490.

[40]	Huang C L, Lin Y G, Chiang C L, Peng C K, Senthil Raja D, Hsieh C T, Chen Y A, Chang S Q, Yeh Y X, Lu S Y. Appl. Catal. B Environ., 2023, 320: 122016.

[41]	Wang S Q, Xu B L, Huo W Y, Feng H C, Zhou X F, Fang F, Xie Z H, Shang J K, Jiang J Q. Appl. Catal. B Environ., 2022, 313: 121472.

[42]	Bockris J O, Otagawa T. J. Electrochem. Soc., 1984, 131(2): 290.

[43]	Man I C, Su H Y, Calle-Vallejo F, Hansen H A, Martínez J I, Inoglu N G, Kitchin J, Jaramillo T F, Nørskov J K, Rossmeisl J. ChemCatChem, 2011, 3(7): 1159.

[44]	Song F, Busch M M, Lassalle-Kaiser B, Hsu C S, Petkucheva E, Bensimon M, Chen H M, Corminboeuf C, Hu X L. ACS Cent. Sci., 2019, 5(3): 558.

[45]	Bockris J O. J. Chem. Phys., 1956, 24(4): 817.

[46]	Grimaud A, Demortière A, Saubanère M, Dachraoui W, Duchamp M, Doublet M L, Tarascon J M. Nat. Energy, 2017, 2: 16189.

[47]	Koper M. Chem. Sci., 2013, 4: 2710.

[48]	Huang Z F, Song J J, Dou S, Li X G, Wang J, Wang X. Matter, 2019, 1(6): 1494.

[49]	Jiang J, Sun F F, Zhou S, Hu W, Zhang H, Dong J C, Jiang Z, Zhao J J, Li J F, Yan W S, Wang M. Nat. Commun., 2018, 9: 2885.

[50]	Su H, Zhou W L, Zhou W, Li Y L, Zheng L R, Zhang H, Liu M H, Zhang X X, Sun X, Xu Y Z, Hu F C, Zhang J, Hu T D, Liu Q H, Wei S Q. Nat. Commun., 2021, 12: 6118.

[51]	Luo X, Zhao H Y, Tan X, Lin S, Yu K S, Mu X Q, Tao Z H, Ji P X, Mu S C. Nat. Commun., 2024, 15: 8293.

[52]	Huang Z F, Xi S B, Song J J, Dou S, Li X G, Du Y H, Diao C Z, Xu Z J, Wang X. Nat. Commun., 2021, 12: 3992.

[53]	Huang Z F, Song J J, Du Y H, Xi S B, Dou S, Nsanzimana J M V, Wang C, Xu Z J, Wang X. Nat. Energy, 2019, 4(4): 329.

[54]	Qian Z X, Liang G H, Shen L F, Zhang G, Zheng S S, Tian J H, Li J F, University M N, Zhang H. J. Am. Chem. Soc., 2025, 147(1): 1334.

[55]	Huang Y, Wang Z Y, Xiao H, Liu Q D, Wang X. J. Am. Chem. Soc., 2024, 146(42): 29006.

[56]	Kuang S F, Pi Z G, Li X W, Wang J X, Lin H, Nie M, Sun J H, Zhang H L, Li Q. J. Colloid Interface Sci., 2025, 679: 296.

[57]	Yin Z H, Huang Y, Song K P, Li T T, Cui J Y, Meng C, Zhang H G, Wang J J. J. Am. Chem. Soc., 2024, 146(10): 6846.

[58]	Wang F Q, Zou P C, Zhang Y Y, Pan W L, Li Y, Liang L M, Chen C, Liu H, Zheng S J. Nat. Commun., 2023, 14: 6019.

[59]	Wang N, Ou P F, Miao R K, Chang Y X, Wang Z Y, Hung S F, Abed J, Ozden A, Chen H Y, Wu H L, Huang J E, Zhou D J, Ni W Y, Fan L Z, Yan Y, Peng T, Sinton D, Liu Y C, Liang H Y, Sargent E H. J. Am. Chem. Soc., 2023, 145(14): 7829.

[60]	Fang C, Zhou J, Zhang L L, Wan W C, Ding Y X, Sun X Y. Nat. Commun., 2023, 14: 4449.

[61]	Hao Y X, Hung S F, Zeng W J, Wang Y, Zhang C C, Kuo C H, Wang L Q, Zhao S, Zhang Y, Chen H Y, Peng S J. J. Am. Chem. Soc., 2023, 145(43): 23659.

[62]	Chen X, Wang Q C, Cheng Y W, Xing H L, Li J Z, Zhu X J, Ma L B, Li Y T, Liu D M. Adv. Funct. Mater., 2022, 32(26): 2112674.

[63]	Yagi S, Yamada I, Tsukasaki H, Seno A, Murakami M, Fujii H, Chen H, Umezawa N, Abe H, Nishiyama N, Mori S. Nat. Commun., 2015, 6: 8249.

[64]	Chang J W, Shi Y Y, Wu H, Yu J K, Jing W, Wang S Y, Waterhouse G I N, Tang Z Y, Lu S Y. J. Am. Chem. Soc., 2024, 146(19): 12958.

[65]	Ji Q Q, Tang B, Zhang X L, Wang C, Tan H, Zhao J, Liu R Q, Sun M, Liu H J, Jiang C, Zeng J R, Cai X K, Yan W S. Nat. Commun., 2024, 15: 8089.

[66]	Yeh J W, Chen S K, Lin S J, Gan J Y, Chin T S, Shun T T, Tsau C H, Chang S Y. Adv. Eng. Mater., 2004, 6(5): 299.

[67]	Senkov O N, Senkova S V, Miracle D B, Woodward C. Mater. Sci. Eng. A, 2013, 565: 51.

[68]	Senkov O N, Wilks G B, Miracle D B, Chuang C P, Liaw P K. Intermetallics, 2010, 18(9): 1758.

[69]	Ma Y J, Ma Y, Wang Q S, Schweidler S, Botros M, Fu T T, Hahn H, Brezesinski T, Breitung B. Energy Environ. Sci., 2021, 14(5): 2883.

[70]	Liu W J, Bao J, Guan M L, Zhao Y, Lian J B, Qiu J X, Xu L, Huang Y P, Qian J, Li H M. Dalton Trans., 2017, 46(26): 8372.

[71]	Chunduri A, Gupta S, Bapat O, Bhide A, Fernandes R, Patel M K, Bambole V, Miotello A, Patel N. Appl. Catal. B Environ., 2019, 259: 118051.

[72]	Xing M, Kong L B, Liu M C, Liu L Y, Kang L, Luo Y C. J. Mater. Chem. A, 2014, 2(43): 18435.

[73]	Wang J Y, Wang J, Zhang M, Li S M, Liu R, Li Z Q. J. Alloys Compd., 2020, 821: 153463.

[74]	Pi Z G, Hao J C, Kuang S F, Shi R J, Li X W, Wang J X, Lin H, Nie M, Li Q. Int. J. Hydrog. Energy, 2024, 93: 84.

[75]	Li Y J, Huang B L, Sun Y J, Luo M C, Yang Y, Qin Y N, Wang L, Li C J, Lv F, Zhang W Y, Guo S J. Small, 2019, 15: 1804212.

[76]	Kuang S F, Zhang H L, Zhou L W, Pi Z G, Lin H, Nie M, Sun J H, Li Q. J. Electroanal. Chem., 2024, 963: 118318.

[77]	Jin Z Y, Lv J, Jia H L, Liu W H, Li H L, Chen Z H, Lin X, Xie G Q, Liu X J, Sun S H, Qiu H J. Small, 2019, 15(47): 1904180.

[78]	Liu L H, Li N, Han M, Han J R, Liang H Y. Rare Met., 2022, 41(1): 125.

[79]	Sharma L, Katiyar N K, Parui A, Das R, Kumar R, Tiwary C S, Singh A K, Halder A, Biswas K. Nano Res., 2022, 15(6): 4799.

[80]	Asghari Alamdari A, Jahangiri H, Yagci M B, Igarashi K, Matsumoto H, Motallebzadeh A, Unal U. ACS Appl. Energy Mater., 2024, 7(6): 2423.

[81]	Xie M K, Xiao X, Wu D J, Zhen C, Wu C S, Wang W J, Nian H, Li F Y, Gu M D, Xu Q. Nano Res., 2024, 17(6): 5288.

[82]	Mei Y J, Chen J L, Wang Q, Guo Y Q, Liu H W, Shi W H, Lin C, Yuan Y F, Wang Y H, Xia B Y, Yao Y G. Sci. Adv., 2024, 10(47): eadq6758.

[83]	Sun Y H, Tang T M, Xiao L Y, Han J Y, Bai X, Shi M Y, Chen S Y, Sun J R, Ma Y Y, Guan J Q. Chem. Eur. J., 2024, 30(17): e202303779.

[84]	Zhang L, Cai W, Bao N. Adv. Mater., 2021, 33(22): e2100745.

[85]	Duan C Q, Li X L, Wang D, Wang Z Y, Sun H Y, Zheng R G, Liu Y G. Sustain. Energy Fuels, 2022, 6(6): 1479.

[86]	Wang D D, Liu Z J, Du S Q, Zhang Y Q, Li H, Xiao Z H, Chen W, Chen R, Wang Y Y, Zou Y Q, Wang S Y. J. Mater. Chem. A, 2019, 7(42): 24211.

[87]	Zhao C H, Li N, Zhang R Z, Zhu Z Q, Lin J H, Zhang K F, Zhao C J. ACS Appl. Mater. Interfaces, 2019, 11(51): 47858.

[88]	Zhang L, Cai W, Bao N, Yang H. Adv. Mater., 2022, 34(26): 2110511.

[89]	Zhong X, Zhu Y A, Dai W, Yu J, Lu T, Pan Y. New J. Chem., 2022, 46(18): 8398.

[90]	Nguyen T X, Su Y H, Lin C C, Ting J M. Adv. Funct. Mater., 2021, 31(48): 2106229.

[91]	Wang Q Q, Li J Q, Li Y J, Shao G M, Jia Z, Shen B L. Nano Res., 2022, 15(10): 8751.

[92]	Lai D, Kang Q, Gao F, Lu Q. J. Mater. Chem. A, 2021, 9(33): 17913.

[93]	Zhou Y F, Gao L Q, Chen H Y, Wang H, Zhang J, Li X M, Duo F F, Guan G Q. J. Mater. Sci. Technol., 2024, 168: 62.

[94]	Yao Y C, Ma Z T, Dou Y B, Lim S Y, Zou J Z, Stamate E, Jensen J O, Zhang W J. Chem. Eur. J., 2022, 28(14): e202104288.

[95]	Zhao C J, Cai W L, Sun N, Chen S, Jing W B, Zhao C H. J. Phys. Chem. Solids, 2024, 184: 111668.

[96]	Zhao X, Xue Z, Chen W, Bai X, Shi R, Mu T. J. Mater. Chem. A, 2019, 7(46): 26238.

[97]	Mei Y, Feng Y, Zhang C, Zhang Y, Qi Q, Hu J. ACS Catal., 2022, 12(17): 10808.

[98]	Zhu H, Sun S H, Hao J C, Zhuang Z C, Zhang S G, Wang T D, Kang Q, Lu S L, Wang X F, Lai F L, Liu T X, Gao G H, Du M L, Wang D S. Energy Environ. Sci., 2023, 16(2): 619.

[99]	Li P, Wu B, Du K, Liu Z, Gao E, Yin H, Wang D. ACS Sustain. Chem. Eng., 2023, 11(38): 14246.

[100]

Qiu

H J

, Fang

, Gao

, Wen

, Lv

, Li

, Xie

, Liu

, Sun

. ACS Mater. Lett., 2019, 1(5): 526.

[101]

Zhou

D J

, Li

P S

, Lin

, McKinley

, Kuang

, Liu

, Lin

W-F

, Sun

X M

, Duan

. Chem. Soc. Rev., 2021, 50(15): 8790.

[102]

Xue

, Wu

, Zhao

, Nan

, Lei

. Prog. Chem., 2022, 34(12): 2686

(薛世翔, 吴攀, 赵亮, 南艳丽, 雷琬莹. 化学进展, 2022, 34(12): 2686).

[103]

Zhou

L L

, Xie

R G

, Wang

L J

. Prog. Chem., 2019, 31(2/3): 275

(周伶俐, 谢瑞刚, 王林江. 化学进展, 2019, 31(2/3): 275).

[104]

K Z

, Zhu

X Y

, Wang

D D

, Zhang

N N

, Huang

, Li

, Long

, Tian

, Zou

Y Q

, Wang

Y Y

, Chen

, Wang

S Y

. J. Energy Chem., 2021, 60: 121.

[105]

Hao

, Chen

J Y

, Wang

K X

, Wang

J L

, Lei

F C

, Hao

, Sun

, Xie

J F

, Tang

. J. Colloid Interface Sci., 2023, 642: 41.

[106]

Fabbri

, Nachtegaal

, Binninger

, Cheng

, Kim

B J

, Durst

, Bozza

, Graule

, Schäublin

, Wiles

, Pertoso

, Danilovic

, Ayers

K E

, Schmidt

T J

. Nat. Mater., 2017, 16(9): 925.

[107]

, Song

, Hu

X L

. Nat. Commun., 2016, 7: 12324.

[108]

Liang

X L

, Xu

, Gao

W Q

, Liang

, Lv

T M

, Zhang

H Y

, Lu

, Zou

, Liu

J X

. J. Mater. Chem. A, 2024, 12(33): 22030.

[109]

Deng

X H

, Tüysüz

. ACS Catal., 2014, 4(10): 3701.

[110]

Y J

, Bian

W Y

, Wu

, Tian

J H

, Yang

R Z

. Electrochim. Acta, 2015, 151: 276.

[111]

Wei

, Feng

Z X

, Scherer

G G

, Barber

, Shao-Horn

, Xu

Z J

. Adv. Mater., 2017, 29(23): 1606800.

[112]

Sun

S N

, Sun

Y M

, Zhou

, Xi

S B

, Ren

, Huang

B C

, Liao

H B

, Wang

L P

, Du

Y H

, Xu

Z J

. Angew. Chem., 2019, 131(18): 6103.

[113]

Liu

F M

, Yu

, Chen

, Li

J H

, Liu

H H

, Cheng

F Y

. Chin. J. Catal., 2022, 43(1): 122.

[114]

Kim

K H

, Choi

Y H

. J. Electroanal. Chem., 2022, 922: 116737.

[115]

Wang

J R

, Sun

S N

, Xi

S B

, Sun

Y M

, Ong

S J H

, Seh

Z W

, Xu

Z J

. J. Phys. Chem. C, 2024, 128(12): 4978.

[116]

Zhang

M Y

, Zhou

X Y

, Luo

K L

, Fan

Y N

, He

C D

, Niu

, Zhang

J J

, Zhang

P F

, Dai

. J. Mater. Chem. A, 2025, 13(2): 1287.

[117]

Yuan

, Ge

, Sun

X L

, Zhang

J Q

, Yu

, Zhang

C Y

, Li

H B

. CrystEngComm, 2020, 22(37): 6216.

[118]

Wang

, Fan

J T

, Do-Thanh

C L

, Suo

, Yang

Z Z

, Chen

, Yuan

Y T

, Lyu

H L

, Yang

S Z

, Dai

. Angew. Chem. Int. Ed., 2021, 60(18): 9953.

[119]

Nguyen

T X

, Liao

Y C

, Lin

C C

, Su

Y H

, Ting

J M

. Adv. Funct. Mater., 2021, 31(27): 2101632.

[120]

, Chen

, Jie

K C

, Yang

Z Z

, Li

T T

, Dai

. Angew. Chem. Int. Ed., 2019, 58(15): 5018.

[121]

S S

, Li

, Wang

H M

, Sun

Y T

, Liu

W Y

, Duan

J J

, Chen

. J. Phys. Chem. C, 2022, 126(33): 14094.

[122]

Wang

S Q

, Huo

W Y

, Fang

, Xie

Z H

, Shang

J K

, Jiang

J Q

. Chem. Eng. J., 2022, 429: 132410.

[123]

Q C

, Jiang

, Li

Y H

, Liang

, Hu

Y J

, Li

C Z

. Appl. Catal. B Environ., 2019, 256: 117893.

[124]

Sivanantham

, Shanmugam

. ChemElectroChem, 2018, 5(14): 1937.

[125]

Ahmad Shah

S S

, Najam

, Aslam

M K

, Ashfaq

, Rahman

M M

, Wang

, Tsiakaras

, Song

S Q

, Wang

. Appl. Catal. B Environ., 2020, 268: 118570.

[126]

K L

, Ohto

, Chen

L H

, Han

J H

, Wakisaka

, Nagata

, Fujita

J I

, Ito

. ACS Energy Lett., 2018, 3(7): 1539.

[127]

F Q

, Ma

Y J

, Wu

, Zhai

Q X

, Zhao

J Y

, Ji

H R

, Tang

S C

, Meng

X K

. J. Phys. Chem. C, 2022, 126(43): 18323.

[128]

, Wang

, Zhang

G H

. J Am Ceram Soc., 2023, 106(10): 5698.

[129]

Z F

, Zeng

Q H

, Yu

, Liu

, Chen

A Q

, Guan

J Z

, Wang

H H

, Liu

X F

, Zhang

L Y

. Chem. Eng. J., 2023, 452: 139366.

[130]

Liu

, Liu

T H

, Xu

, Zhao

W Y

, Fan

J P

, Liu

, Ma

X G

, Fu

W S

. Nano Lett., 2024, 24(9): 2831.

[131]

Zhang

M Y

, Luo

K L

, Fan

Y N

, Lu

X Y

, Ye

, Lu

N N

, Dong

J L

, Niu

, Zhang

J J

, Zhang

P F

, Dai

. Chem. Eng. J., 2024, 493: 152758.

[132]

Liu

, Wang

Y H

, Chen

J J

, Yi

, Yang

, Yuan

X Y

, Zhao

Y J

. J. Power Sources, 2025, 632: 236402.

[133]

Dai

, Tu

X Y

, Yue

K H

, Wan

Y J

, Zhao

, Shi

X R

, Huang

F Q

, Yan

. Adv. Funct. Mater., 2025, 35(12): 2417211.

[134]

Shang

, Wang

F Q

, Li

, Liang

L M

, Hao

Q Y

, Liu

. J. Alloys Compd., 2024, 992: 174574.

[135]

X M

, Xie

Z K

, Roy

, Gao

L Q

, Liu

, Zhao

, Wei

, Tang

B J

, Wang

H Y

, Ajayan

, Tang

K Y

. Adv. Mater., 2025, 37(10): 2410295.

[136]

Zhang

C Y

, Li

, Su

M F

, Li

S F

, Gao

, Lu

Q Y

. Chem. Sci., 2024, 15(46): 19359.

[137]

Seh

, Kibsgaard

, Dickens

C F

, Chorkendorff

, Nørskov

, Jaramillo

. Science, 2017, 355(6321): eaad4998.

[138]

Yuste

. Nat. Rev. Neurosci., 2015, 16(8): 487.

[139]

Islam

M M

, Murase

. Neural Netw., 2001, 14(9): 1265.

[140]

Chapelle

, Haffner

, Vapnik

V N

. IEEE

Trans

. Neural Netw., 1999, 10(5): 1055.

[141]

Rokach

L. Inf.

Fusion, 2016, 27: 111.

[142]

Wang

X J

, Ye

, Hu

, Sharman

, Liu

, Luo

, Jiang

. J. Am. Chem. Soc., 2020, 142(17): 7737.

[143]

Jayabharathi

, Thanikachalam

, Karthikeyan

, Sangamithirai

, Vijayarangan

. J. Ind. Eng. Chem., 2025, 146: 136.

[144]

Tukur

, Wei

, Zhang

Y N

, Chen

H N

, Lin

Y W

, He

, Mo

Y R

, Wei

J J

. Small, 2025, 21(21): 2501946.

[145]

Reker

, Schneider

. Chem. Sci., 2016, 7(6): 3919.

[146]

Nie

S Y

, University

, Xiang

, University

, Wu

, Lin

, Liu

Q D

, University

, Chu

S Q

, Wang

, University

. J. Am. Chem. Soc., 2024, 146(43): 29325.

[147]

Kim

, Kim

, Ha

M Y

, Shin

, Kwak

S J

, Park

, Kim

I D

, Jung

W B

, Lee

W B

, Kim

, Jung

H T

. Adv. Mater., 2023, 35(17): 2211497.

[148]

Perumal

, Han

D B

, Marimuthu

, Lim

, Kim

H W

, Seo

. Adv. Funct. Mater., 2025, 35(30): 2424887.

[149]

Liu

S J

, Yang

. Metals, 2024, 14(2): 235.

[150]

, Wang

D D

, Zhang

Y Q

, Tao

, Wang

T H

, Zou

Y Q

, Wang

Y Y

, Chen

, Wang

S Y

. Adv. Mater., 2020, 32(19): 1907879.

[151]

Xie

, Yan

D F

, Li

, Du

S Q

, Chen

, Wang

Y Y

, Zou

Y Q

, Chen

, Wang

S Y

. ACS Catal., 2020, 10(19): 11082.

[152]

Tang

T M

, Bai

, Wang

Z L

, Guan

J Q

. Chem. Sci., 2024, 15(14): 5082.

[153]

Zhang

B Z

, Deng

, Chen

, Li

, Yuan

M W

, Xiao

W M

, Wang

S H

, Wang

X L

, Zhang

P F

, Shu

, Shi

S L

, Chen

. ACS Appl. Mater. Interfaces, 2023, 15(45): 5270.

[154]

Liu

Y F

, Ye

C C

, Chen

, Fan

J B

, Liu

, Xue

, Sun

J W

, Zhang

W Q

, Wang

, Xiong

, Zhu

J W

. Adv. Funct. Mater., 2024, 34(25): 2314820.

[155]

Huang

, Zhang

B W

, Wu

J S

, Zhang

T Y

, Peng

D D

, Cao

, Zhang

, Li

, Huang

Y Z

. J. Mater. Chem. A, 2020, 8(24): 11938.

[156]

Glasscott

M W

, Pendergast

A D

, Goines

, Bishop

A R

, Hoang

A T

, Renault

, Dick

J E

. Nat. Commun., 2019, 10: 2650.

[157]

Wang

, Zuo

, Horta

, He

, Yang

L L

, Ostovari Moghaddam

, Ibáñez

, Qi

X Q

, Cabot

. ACS Appl. Mater. Interfaces, 2022, 14(42): 48212.

[158]

Zhai

B X

, Gong

, Liu

L P

, Cui

X M

, He

. Fuel, 2025, 393: 134992.

[159]

Hooch Antink

, Lee

H S

, Shin

, Yoo

T Y

, Ko

, Shim

, Na

, Sung

Y E

, Hyeon

. Adv. Funct. Mater., 2024, 34: 2309438.

[160]

, Zhong

, Wang

, Gu

J L

, Lyu

F C

, Chen

S M

, Wu

H K

, Li

L X

, Zhi

C Y

, Lu

, Li

Y Y

. Angew. Chem. Int. Ed., 2024, 63(47): e202410978.

[161]

Yuan

Q Y

, Liu

T T

, Ma

, Liao

, Wang

W Z

, Meng

H T

, You

Q F

, Zeng

F Y

, Xie

M L

, Huang

H B

, Liu

C L

, Liang

. J. Colloid Interface Sci., 2025, 684: 783.

[162]

Zhou

X C

, Zhu

, Fu

, Lan

, Hahn

, Zeng

J R

, Feng

. Small, 2024, 20(47): 2405596.

Options

Outlines

模态框（Modal）标题

Abstract

Cite this article

1 Introduction

2 OER reaction pathway

图1 OER机理的发展历程[42-46]，以及其对应的优点与缺点

2.1 AEM Mechanism

图2 （a） AEM示意图；（b） 实际的和理想催化剂的吉布斯自由能图；（c） 部分氧化物催化剂过电位与ΔGOH*-ΔG*O的火山图[49]；（d） AD-NH-Ir在OER过程中的原位红外光谱图[50]

2.2 LOM mechanism

2.3 OPM Mechanism

图4 （a） OPM示意图；OH-吸附在（b） SrFeO3和（c） CaCu3Fe4O12晶体结构模型[63]；（d） Ru-Co3O4产生氧气的DEMS测量信号[64]；（e） Ru/MnO2在OER过程中的原位红外光谱[65]；（f） S-FeOOH/IF在OER过程中的原位红外光谱[62]

3 Research Status and Bottlenecks of High-Entropy Oxygen Evolution Catalyst Materials

图5 高熵材料的发展历史及其应用[69]

表1 常温条件下几种典型高熵材料OER的催化活性

图6 目前高熵电催化剂组元的使用频率

3.1 High-entropy alloys

图7 （a） 各类催化剂表面吸附能的示意图；（b） HEAs不同活性位点示意图；（c） FeCoNiMn；（d） FeCoNiMnCr，（e） FeCoNiMnMo和（f） FeCoNiMnW吉布斯自由能变化图；以及HEAs的（g） TDOS图[99]

3.2 High-entropy oxides

3.3 High-entropy metal-organic frameworks

图9 （a） ZnFeNiCuCoRuO制备示意图；（b） HAADF-STEM图像及对应的元素分布图；（c） ZnFeNiCuCoRuO模型；（d） AEM和LOM机理示意图以及其（e） 自由能；（f） 在Ru-Fe、Ru-Co、Ru-Ni位点上AEM的自由能[39]

3.4 Other High-Entropy Compounds

图10 （a） FeCoCuMnRu和（b） FeCoCuMnRuB在不同电压下的原位拉曼[130]，以及（c） FeCoCuMnRu和FeCoCuMnRuB吉布斯自由能变化图[130]，（d） FeCoNiCuYP/C催化剂中Fe-OH、Co-OOH、Ni-OOH中间体的电荷密度差[135]

4 Optimization Strategies for the OER Performance of High-Entropy Materials

图11 高熵材料的OER性能优化策略示意图

4.1 Machine Learning–Assisted Design

表2 NN、SVM和RF模型优点与缺点[149]

4.2 Defect Engineering

4.3 Elemental regulation

5 Conclusion and Outlook

References

图1 OER机理的发展历程^[42-46]，以及其对应的优点与缺点

图2 （a） AEM示意图；（b）实际的和理想催化剂的吉布斯自由能图；（c）部分氧化物催化剂过电位与ΔG_OH-ΔG_O的火山图^[49]；（d） AD-NH-Ir在OER过程中的原位红外光谱图^[50]

图4 （a） OPM示意图；OH^-吸附在（b） SrFeO₃和（c） CaCu₃Fe₄O₁₂晶体结构模型^[63]；（d） Ru-Co₃O₄产生氧气的DEMS测量信号^[64]；（e） Ru/MnO₂在OER过程中的原位红外光谱^[65]；（f） S-FeOOH/IF在OER过程中的原位红外光谱^[62]

图5 高熵材料的发展历史及其应用^[69]

图7 （a）各类催化剂表面吸附能的示意图；（b） HEAs不同活性位点示意图；（c） FeCoNiMn；（d） FeCoNiMnCr，（e） FeCoNiMnMo和（f） FeCoNiMnW吉布斯自由能变化图；以及HEAs的（g） TDOS图^[99]

图9 （a） ZnFeNiCuCoRuO制备示意图；（b） HAADF-STEM图像及对应的元素分布图；（c） ZnFeNiCuCoRuO模型；（d） AEM和LOM机理示意图以及其（e）自由能；（f）在Ru-Fe、Ru-Co、Ru-Ni位点上AEM的自由能^[39]

图10 （a） FeCoCuMnRu和（b） FeCoCuMnRuB在不同电压下的原位拉曼^[130]，以及（c） FeCoCuMnRu和FeCoCuMnRuB吉布斯自由能变化图^[130]，（d） FeCoNiCuYP/C催化剂中Fe-OH、Co-OOH、Ni-OOH中间体的电荷密度差^[135]

表2 NN、SVM和RF模型优点与缺点^[149]