The escalating global energy crisis, coupled with growing environmental concerns, has necessitated urgent advances in clean and efficient energy conversion technologies. Among the emerging approaches, electrocatalytic water splitting has garnered substantial interest as a carbon-neutral strategy for hydrogen production, positioning hydrogen as a potential replacement for non-renewable fossil fuels [
1]. This process primarily involves two coupled half-reactions: the hydrogen evolution reaction (HER) and the oxygen evolution reaction (OER). In particular, the OER at the anode is hindered by intrinsically slow kinetics due to multi-electron transfer steps, electron-proton coupling, and adsorption/desorption processes. As a result, highly efficient electrocatalysts are required to reduce the overpotential. In this context, deciphering the actual catalytic sites and concomitant evolution of their electronic structure during OER under
operando conditions have become a critical imperative. Such mechanistic insights establish structure-property correlations that underpin the rational engineering of high-performance electrocatalysts.
In recent decades, iron group transition metal chalcogenides (TMCs, TM = Fe, Co, Ni; chalcogen = S, Se) have emerged as efficient OER pre-catalysts, offering key advantages such as natural abundance, cost-effectiveness, environmental benignity, and tunable electronic structures. Notably, several of these catalysts exhibit superior performance compared to benchmark Ru/Ir precious metal oxides for the OER [
2–
7]. It has been documented that these TMCs function as pre-catalysts, undergoing surface reconstruction processes that result in the formation of oxidized hydroxides/oxyhydroxides with amorphous or highly disordered structures, which are responsible for the catalytically active sites in these pre-catalysts [
8–
10]. During OER, the surface reconstruction of TMC pre-catalysts refers to the dynamic and irreversible (or quasi-reversible) transformation of their initial surface structures, including atomic arrangement, chemical valence states, and phase composition, triggered by under reaction conditions such as high applied electrode potentials, the electrolytic environment, and interactions with reaction intermediates. This transformation results in a surface structure that differs significantly from pristine material, forming catalytically competent phases under operational conditions and enhancing electrocatalytic efficiency. This adaptive process is fundamental to catalyst performance and is particularly prevalent among transition metal-based catalysts involving Co, Ni, and Fe [
11–
16]. Therefore, elucidating the surface reconstruction mechanisms of TMCs during the OER can provide profound insights into the nature of the true surface-active species that dominate the electrocatalytic activity, especially at the nanoscale or even angstrom scale.
Thanks to rapid advancement in analytical instrumentation, real-time monitoring of catalytic processes has become increasingly practical through the synchronized integration of electrochemical workstations with
insitu/
operando spectroscopic techniques [
17,
18]. The term
operando denotes characterization performed under actual operating conditions with simultaneous monitoring of catalytic performance, enabling direct correlation between functional metrics and structural dynamics [
19]. In contrast,
insitu (meaning “on-site”) methods often involve mimicking operational environments using customized electrochemical cells, allowing for detailed spectroscopic analysis under pseudo-working conditions. This distinction emphasizes that
operando methodologies are essential for capturing authentic structure-function relationships in real devices, while
insitu approaches often prioritize compatibility with spectroscopic tools. A comparative summary is provided in Table 1.
Recently, integrated
operando electrochemistry-spectroscopy approaches have been implemented to interrogate the dynamic surface reconstruction behavior TMCs during the OER. These include
insitu/
operando Fourier-transform infrared (FTIR) spectroscopy, Raman spectroscopy, and X-ray absorption spectroscopy (XAS), as shown in Fig. 1, which provide atomic-level insights into coordination geometry alterations and oxidation state dynamics under anodic polarization [
20,
21].
XAS can be divided into two regions: X-ray absorption near edge structure (XANES) and extended X-ray absorption fine structure (EXAFS), which provide insights into the electronic structure and local coordination, environment respectively. Operando XAS is particularly powerful in capturing the dynamic structural evolution of TMCs under real OER operating conditions. This technique enables the identification of active centers and elucidation of reaction mechanisms in real time.
For example, a mechanistic study by Wu et al. [
22] revealed the structural reconstruction pathway of Ni–Fe selenide as-prepared during oxygen evolution electrocatalysis. Quasi-
in situ XAS analysis after the OER test indicated increased valence states for both Ni and Fe, accompanied by the disappearance of Ni–Se coordination. Simultaneously, new Ni–M and Ni–O bond configurations emerged. These observations confirm the transformation of Ni–Fe–Se from its initial selenide form into a Ni–Fe–Se–OOH structure. The enhanced OER performance is attributed to the synergistic interaction between the two metals, each with different oxophilicities, which leads to a modulated electronic structure favorable for catalysis.
In another case, the post-formed oxo-bridged Fe–Ni coordination ((Ni
III–O–Fe
III) structure at the NiSe
2/FeSe
2 heterointerface was identified as the actual catalytic site during OER [
23]. In this system, FeSe
2 acts as the adsorption site for OH
−, promoting the formation of O–O
− intermediates, while NiSe
2 catalyzes the transformation of O–O
− species, promoting O
2 generation.
Moreover,
operando XAS measurements have revealed dynamic surface reconstruction in CoS
x during anodic polarization, characterized by a self-limiting conversion to active cobalt oxyhydroxide (CoOOH) matrices [
24]. Notably, the strategic integration of FeS
x heterophases was found to induce interfacial charge transfer, accelerating the activation of CoS
x species. Fe
3+-mediated electronic perturbations optimized OH
- adsorption energy at the cobalt catalytic sites and lowered the activation barrier for structural transformation in metastable oxyhydroxide CoOOH-(Fe) superlattices.
However,
operando XAS by Zhao et al. [
25] provided direct evidence that high-valent Co(IV) species, rather than CoOOH, serve as the true intermediates initiating the OER in low-crystalline CoS under basic conditions, as shown in Fig. 2(a). The dynamic low-crystalline CoS reconstruction mechanism was elucidated through a potential-dependent phase evolution pathway:
(1) Initiation of anion-ligand exchange (E < 1.0 V): Sulfur ligands in the CoS lattice are replaced, forming CoOOH;
(2) Topotactic phase transition (E = 1.0 V), complete conversion from sulfhydryl to oxo coordination;
(3) Coordination geometry optimization (E = 1.0–1.2 V): potential-driven stabilization of CoOOH, forming a 6-fold Co–O structure;
(4) Active phase formation via deprotonation (E > 1.325 V), deprotonation of μ-OH bridges lead to formation of catalytically active CoO2 moieties;
(5) Post-catalysis reversibility: Upon removal of applied potential, a slight decrease in Co valence suggests redox reversibility, with CoOOH forming post-reaction.
To quantify the relative proportions of sulfide and (oxy)hydroxide phases, a Fe–Ni–S catalyst measured at 1.43 V was analyzed [
26]. Principal component analysis (PCA) and linear combination fitting (LCF) were applied to the XAFS spectra, with LCF fitting reference spectra to that of the sample to determine composition. The LCF method has been validated as a reliable approach for identifying and quantifying the dominant chemical species. Analytical results reveal that the top surface of Fe-Ni-S@1.43V consists of 59.3% Fe-Ni-S and 40.7% Fe-Ni-OOH, suggesting an intermediate state during the transition from sulfide to (oxy)hydroxide.
Operando attenuated total reflection infrared spectroscopy (ATR-IR) combined with differential electrochemical mass spectrometry (DEMS) employing
18O isotopic labeling has revealed that electrocatalysts with the same components can exhibit different reconstruction behaviors under working conditions. For example, due to differences in valence states of metal ions and atomic arrangements, copper sulfide (CuS) and cuprous sulfide (Cu
2S) show different reconstruction dynamics, as illustrated in Fig. 2(b) [
27]. Specifically, in CuS nanomaterials, S
2− is first oxidized to SO
42−, followed by the conversion of Cu
x+ to CuO via a Cu(OH)
2 intermediate. Enhanced spectral signatures of *OOH and *OH intermediates observed on CuS, compared to Cu
2S, indicate a more favorable formation of hydroxide/oxyhydroxide species through surface oxidation, attributing to the higher susceptibility of Cu(I) species in CuS to oxidative transformation in oxygen-rich environments. Moreover, the coexistence of tetrahedral and trigonal planar coordination geometries in CuS helps establish an optimal electronic structure that facilitates enhanced electronic conductivity, dynamic structural rearrangement, and superior OER activity. Sulfur atoms close to structural defects are particularly prone to dissociative processes, thus accelerating reconstruction via vacancy-mediated pathways.
To further investigate the role of residual chalcogens in reconstructed layers, Liu et al. [
28] used Cu
0.4Co
0.6S as a model system to explore how residual sulfur modulates OER activity and to identify its chemical forms post-reconstruction. They also proposed a dual-pathway mechanism for the OER process. As illustrated in Fig. 3, given that peroxo species (O
22−) originate from direct O–O coupling in the lattice oxygen mechanism (LOM), chemical probes such as tetramethylammonium cations (TMA
+) can be used to track O
22− intermediates generated during OER via the LOM pathway. The OER activity of Cu
0.4Co
0.6OSH, formed by anodic oxidation of Cu
0.4Co
0.6S, is significantly reduced after the addition of TMA hydroxide (TMAOH) to the electrolyte, indicating that crystal lattice oxygen participates actively in the reaction via the LOM pathway.
The operando isotopic labeling experiments using 18O and in situ Raman spectroscopy confirmed the positive role of residual sulfur in promoting lattice oxygen activation. Integrating density functional theory (DFT) calculations with 18O isotopic labeling experiments, the study revealed that residual sulfur in the reconstructed Cu0.4Co0.6S layer effectively balances the adsorbate evolution mechanism (AEM) and LOM. This balance is achieved by both activating and optimizing the adsorption/desorption dynamics at the metal active sites, without causing a complete transition from AEM to LOM.
Cao et al. [
29] investigated the surface reconstruction behavior of Co
9S
8 supported on single-walled carbon nanotubes (Co
9S
8/SWCNT) under different pH conditions using
operando XAFS analysis. At pH = 7, the structural self-optimization of Co
9S
8/SWCNT becomes evident with increasing applied electrode potential from 1.1 to 1.6 V versus RHE, eventually stabilizing into oxygenated CoS-SWCNT species. In contrast, at pH = 14, this self-optimization is observed under OCP, and Co
9S
8/SWCNT gradually transforms into S-CoOOH-SWCNT as the applied electrode potential reaches 1.2 versus RHE. Fourier Transform Analysis analysis of EXAFS spectra with k
3-weighting reveals the simultaneous emergence of Co–O and Co–Co coordination with the attenuation of Co–S bonds as the applied potential increases. These findings indicate that Co
9S
8/SWCNT acts as a self-optimizing pre-catalyst, forming an active phase under operational conditions (e.g., pH and potential), a hallmark of authentic electrocatalysts.
Raman spectroscopy has also emerged as a critical
operando characterization technique for probing solid-liquid interfacial dynamics in electrochemical systems, due to its inherent compatibility with aqueous media, sensitivity to low-frequency vibrations [
30], and minimal spectral interference from water due to its weak Raman scattering cross-section—unlike the strong O–H stretching observed in IR spectroscopy. Particularly, Raman provides access to the low-wavenumber fingerprint regime, where vibrations of critical metal- ligand bonds (e.g., M–Se, M–S, M–O) serve as direct indicators of structural evolution at catalytic centers.
The potential-dependent phase transition of Ni
3S
2 during OER has been resolved via
operando Raman spectroscopic [
31]. The analysis reveals: (1) At the open circuit potential, the pristine spectrum exhibits characteristic vibrations between 100–400 cm
−1, consistent with the Ni
3S
2 phase; (2) Upon anodic activation (
E ≥ 1.24 V versus RHE), new vibrational modes emerge at ≈ 475 and ≈ 556 cm
−1, corresponding to the
Eg bending and
A1g stretching vibrations of Ni–O in the γ–NiOOH phase [
32–
34]. This transformation involves bond-length contraction, electron redistribution, and symmetry breaking—associated with charge storage. The metastability and reversibility of the γ–NiOOH phase are confirmed via cathodic polarization. During the reverse scan (1.8 → 1.1 V versus RHE), the Ni–O vibrational modes diminish, reflecting a partial reversion to the original state [
35].
However, this metastable γ–NiOOH phase exhibits significant susceptibility to chloride-induced corrosion in saline electrolytes, which compromises its OER durability during seawater electrolysis [
36]. In contrast, surface-adsorbed chalcogenates (SO
42− and SeO
32−) demonstrate dual functionality in alkaline media: (1) enhancing OER kinetics by modulating of ligand-to-metal charge transfer, and (2) providing selective anion adsorption capacity.
In situ Raman spectra manifest that at ≥1.2 V versus RHE, M–OOH vibrational peaks appear, while M–S peaks disappear completely, and the intensities of the sulfate peaks decrease [
37]. This indicates transformation of FeNiCoCrMnS
2 into a metal oxyhydroxide phase, leaving behind residual SO
42−, which itself contributes to OER activity. Although selenite (SeO
32−) is isoelectronic with sulfate, its lower pKa and larger ionic radius confer superior electrostatic shielding, especially against chloride attack [
38–
40].
Taking NiSe
2 as an example (Fig. 4),
insitu Raman spectroscopy reveals sequential oxidation of Se–Se to selenite (SeO
32−) and eventually to selenates (SeO
42−). These severe leaching of Se and the strong signal from selenates suggest a surface enrichment of selenates species, which play significant roles in enhancing OER performance. Electrochemically adsorbed or generated selenites can dramatically enhance OER activity in these pre-catalysts [
38,
41].
Additionally, Cheng’s group clarified the role of Se atoms in selenide-based OER systems [
14]. Their analysis of the mechanism demonstrated that the adsorption strength of the *OOH intermediate on the Ni sites of
in situ generated NiOOH is optimized by the electron-modulating effect of the Se atoms at the Ni
3Se
4/NiOOH interface, and thus decreases the reaction barrier for OER.
The phenomenon of catalysts reconstruction during OER is usually evidenced by changes in morphology and phase structure, which can be directly monitored by in situ or quasi-in situ microscopy techniques. In particular, advanced in situ electron microscopy methods, such as quasi-in situ identical location transmission electron microscopy (TEM) and in situ liquid cell TEM, have been developed to systematically evaluate solid-liquid electrochemical reactions, including OER and ORR.
Notably, the two-step structural evolution pathway (CoS
x → Co(OH)
2 → CoOOH) has been captured using advanced
insitu Cs-corrected TEM with (spherical aberration corrected TEM) [
42]. In detail, high-resolution
insitu TEM imaging reveals that the initial material (CoS
x) exhibits an amorphous structure. After 2.5 min of anodization, nanocrystals begin to nucleate within the amorphous hollow shell of CoS
x as shown in Fig. 5(a). With prolonged electrolysis (up to 12.5 min), distinct CoOOH lattice fringes appear.
An intriguing observation is the dynamic thickness of the inner CoSx shell, which initially increases and subsequently decreases, as illustrated in Fig. 5(b). This behavior is attributed to the generation and accumulation of oxygen gas, which increase the local concentration of the OOH* intermediate. This intermediate gradually corrodes the inner shell of CoSx, thus facilitating the formation of Co(OH)2 and CoOOH nanocrystals. The accumulation of OER intermediates and oxygen leads to the formation of an oxygen-rich inner shell, which eventually cannot withstand the internal pressure of the encapsulated oxygen gas. As a result, the shell becomes pierced or even collapses entirely, resulting in the formation of nanosheet-shaped catalyst morphologies.
Additionally, in situ FTIR spectroscopy of CoSx conducted over the course of the OER (0−1000 s) under a constant current of 1 mA further confirms the formation of hydroxide species, as evidenced by the appearance of a distinct vibrational band at 892 cm−1, characteristic of hydroxide formation during the reaction.
In summary, the reconstruction mechanism of TMCs in alkaline environments mainly involves three synergistic steps: (1) oxidative dissolution of lattice metal cations (e.g., Ni2+, Co2+, Fe3+) under anodic polarization; (2) reconfiguration of the coordination environment via anion exchange, where chalcogenide ligands [S2−/Se2−] are replaced by OH−/O2− species; and (3) recrystallization into disordered structures enriched with oxygen vacancies and undercoordinated metal sites. The resulting amorphous or defect-rich phases, characterized by high-valence MOOH and/or MO2-type metastable motifs, demonstrate enhanced catalytic activity due to optimized adsorption energetics and dynamic adaptability to OER intermediates. Additionally, following oxidation and severe leaching, surface-adsorbed or electrolyte-present chalcogenates (e.g., SO42− and SeO32−) significantly contribute to OER promotion, imparting exceptional durability to the electrocatalyst during seawater electrolysis.
The fundamental question—“
How do factors such as applied potential and the limitations of in situ characterization techniques influence the reconstruction pathway of cobalt sulfides”? —remains to be fully addressed. The preceding discussion illustrates that surface reconstruction of TMC pre-catalysts occurs via electrochemical activation, which induces structural modifications in TMC pre-catalysts, such as surface oxidation, ion leaching, and phase transformation. This surface evolution is governed by electrochemical operating conditions, including cycle voltammetry, galvanostatic, and potentiostatic modes [
43].
For instance, under identical alkaline conditions (1 mol/L KOH), Zhao et al. [
25] attributed OER activity to active Co(IV) species formed via the transformation of CoOOH, identifying CoOOH as the post-catalysis surface product. In contrast, Fan et al. [
42] proposed a sequential reconstruction mechanism involving the transformation of CoS
x → Co(OH)
2 → CoOOH.
The following analysis addresses this discrepancy. First, the magnitude of the applied electrode potential dictates both the direction and rate of the reaction. A higher electrode potential can provide sufficient energy to directly oxidize low-crystalline CoS into high-valent Co(IV) species, as demonstrated via stepwise chronoamperometry. Conversely, a lower potential may favor a gradual oxidation process, in which CoSx is first oxidized to Co(OH)2, followed by further oxidation to CoOOH, as observed in a chronopotentiometric experiment. Furthermore, variations in the initial potential can also affect the efficiency of the S–O exchange reaction in cobalt sulfides.
Second, the characterization techniques used in different studies, such as XAS, Raman spectroscopy, and
in situ electrochemical microscopy, vary in their sensitivity to intermediate species.
In situ XAS, as used by Zhao et al. [
25], is highly sensitive to changes in electronic structure and oxidation states, as indicated by edge shift and fine structural features. This makes it particularly effective to detect transient, high-valent species such as Co(IV) (e.g., CoO
2 or Co
4+ in distorted coordination environments), even in low concentrations or for short lifetimes. However, XAS lacks high spatial resolution (typically nanometers) and cannot directly visualize morphological transitions, such as the formation of Co(OH)
2 or CoOOH, limiting its capacity to capture gradual structural evolution.
In contrast, the
in situ TEM employed by Fan et al. [
42] provides atomic-scale insights into morphology and crystallography, enabling direct observation of phase transformations, including CoS
x dissolution, Co(OH)
2 nucleation, and CoOOH growth. Despite these advantages, TEM is less sensitive to fine changes in oxidation state, especially for light elements such as oxygen or short-lived, high-valent species. Moreover, it often relies on specialized liquid cells or environmental chambers, which may introduce altered experimental conditions such as electrolyte confinement, potential gradients, or electron beam-induced effects, that can bias the system toward slower, gradual reaction pathways.
The OER catalyst research mainly centers on unresolved surface reconstruction mechanisms. A particularly critical and open question is how the kinetics of surface reconstruction in TMCs affect the long-term stability of industrial electrolyzers. Commercial application faces several practical challenges. For example, surface reconstruction is difficult to control under industrial operating conditions. Laboratory-scale evaluations are typically conducted in idealized environments, featuring low current densities, high-purity electrolytes, and small-scale cells with efficient mass transport; whereas industrial systems operate at significantly higher current densities (> 500 mA/cm2) and use cost-effective electrolytes that contain various impurities.
These harsher operating conditions accelerate surface reconstruction, as localized overpotentials and non-uniform mass transport trigger heterogeneous structural transformations, such as sulfide-to-oxide transitions and active site aggregation. This phenomenon, often referred to as “uncontrolled reconstruction,” causes a gradual decrease in catalytic performance over time. Although this degradation is rarely observed in short-term laboratory tests (≤ 1000 h), it poses a major challenge to achieving the long-term operational stability (≥ 10000 h) required in industrial applications.
Furthermore, current stability assessments often emphasize short-term performance indicators, such as overpotential at 10 mA/cm2 or cyclic stability over ≤ 10000 cycles, which fail to capture the slow, cumulative structural changes occurring over extended periods. For example, sulfur leaching, typically negligible in lab-scale experiments, can significantly degrade the structural integrity of the active sites in real-world applications. Additionally, impurities in industrial feedstocks (e.g., seawater) may block active sites or alter reconstruction pathways, further compromising catalyst performance.
In conclusion, bridging the gap between lab-scale TMC research and commercial application requires more than elucidating reconstruction mechanisms. It demands attention to engineering adaptability, scalable synthesis routes, and operational resilience under real-world conditions—factors that are often overlooked in current studies.
Although significant progress has been achieved in the development of in situ/operando characterization techniques, which enable real-time detection of metastable reaction intermediates, precise identification of active sites, and continuous monitoring of dynamic structural/compositional evolution, several key challenges remain. In particular, the development of efficient OER electrocatalysts for large-scale commercialization remains insufficient, and fundamental questions related to OER mechanisms remain unresolved. To address these current obstacles, the following potential perspectives and strategies are proposed.
1) Alkaline water electrolysis is currently used in small- and medium-scale industrial applications. In contrast, proton exchange membrane (PEM) acid electrolyzers have attracted considerable interest due to their advantages, which include higher electrolytic conductivity, lower operating costs, effective suppression of gas crossover, and the ability to deliver high output currents. However, their industrialization remains limited, mainly due to the absence of catalysts that exhibit high activity, stability, and corrosion resistance under prolonged OER conditions. Research on acidic OER catalysts has predominantly focused on precious metals, although some studies have explored nonprecious alternatives such as MnOx, lead oxides, strontium titanate, and polyoxometalates. A fundamental question remains: whether catalysts undergo structural reconstruction under acidic conditions. Addressing this issue is essential, as it may contribute to the development of more efficient and stable OER catalysts under acidic conditions.
2) The exploration of dynamic processes, such as surface reconstruction, transient intermediate evolution, and phase transitions in the OER heavily relies on
insitu/operando characterization techniques. However, variations in spatial and temporal resolution, along with inherent limitations of each technique, pose significant challenges to fully capturing the complexity of these dynamic processes [
44], as tabulated in Table 2. For instance, XAS is highly effective for monitoring changes in oxidation states, yet it may overlook short-lived intermediates, HRTEM provides detailed structural information at the atomic-level but involves high-energy electron beams that can alter sensitive materials during observation, potentially inducing structural modifications. Raman spectroscopy detects vibrational features with high sensitivity but is often hindered by fluorescence interference; furthermore, broadening or overlap of Raman peaks can lead to misinterpretation of phase transitions or intermediate species. Due to these inherent limitations, ambiguities often arise in the interpretation of structural evolution “in real-time,” as the collected data may not accurately reflect the actual, undisturbed dynamics of the system. To address these challenges, the following mitigation strategies are proposed: (1) Correlative techniques: Combine XAS with volume-sensitive techniques such as XRD or neutron scattering to bridge surface and bulk information. For example, XAS can trace surface oxidation state changes, while XRD simultaneously tracks bulk phase transitions during electrocatalysis; (2) Minimizing beam damage: Reduce electron flux or exposure time in TEM using high-sensitivity detectors (e.g., direct electron detectors) to preserve signal-to-noise ratio while limiting radiation-induced artifacts. Additionally, using
insitu environmental control, such as gas or liquid cells, can help stabilize materials under observation and mitigate beam-induced damage; (3) Cross-validation of fast dynamics: Integrate Raman spectroscopy with high temporal resolution methods (e.g., ultrafast XRD or transient absorption spectroscopy) to cross-validate ultrafast dynamic events. For example, coupling Raman spectroscopy with femtosecond XRD can link vibrational signatures to structural changes at the atomic-scale in real time.
3) The limitations inherent in the various characterization methods necessitate multimodal approaches to comprehensively elucidate catalyst behavior, as no single technique can simultaneously resolve structural phase transitions, valence state changes, and compositional dynamics under operating conditions. This fundamental constraint underscores the urgent need to develop integrated operando platforms that synergistically combine complementary spectroscopic, microscopic, and diffraction-based methodologies. However, in practice, the development of such platforms faces significant technical challenges in terms of coupling, especially with regard to the compatibility of electrolytic cells across different modalities. For example, liquid-phase Raman spectroscopy requires optically transparent windows, such as quartz or calcium fluoride, to facilitate laser penetration. In contrast, X-ray-based techniques such as XAS and XRD require window materials with low X-ray absorption and high transmittance, such as beryllium, Kapton films, or silicon nitride membranes. To reconcile these requirements, composite window materials, such as bilayer films composed of Kapton and silicon nitride, can be used to strike a balance between optical transparency and X-ray compatibility. Additionally, surface modifications, for instance, gold nanoparticle coatings, can be introduced to enhance the intensity of the Raman signal. Another major challenge lies in integrating techniques that operate at different spatial and temporal resolutions. For example, atomic-scale spectroscopies like XAS and XANES typically provide high temporal resolution (typically milliseconds to seconds), whereas nanoscale imaging tools such as HRTEM or atomic force microscopy (AFM) often operate at slower time scales (seconds to minutes). Furthermore, the micrometric penetration depth of X-rays and the nanometric surface sensitivity of Raman spectroscopy can lead to dimensional mismatches in the collected data. Addressing this issue may require three-dimensional data reconstruction or layer-by-layer peeling approaches to correlate subsurface and surface-level dynamics.
4) Harnessing artificial intelligence (AI) in OER catalyst research facilitates data-driven breakthroughs in catalyst design, mechanistic understanding, and performance optimization. Given the complexity of the OER, which involves complex multi-step proton-electron transfers and the intermediate dynamic evolution of active sites during catalyst reconstruction, conventional in situ/operando techniques alone are often insufficient to fully capture these processes. AI offers a powerful framework for integrating device experimental and theoretical datasets to address these complexities. Machine learning potential energy surfaces (MLPES), for example, can effectively replace traditional DFT calculations to efficiently simulate adsorption energies of intermediates on catalyst surfaces, energy barriers of transition states, and identify rate-determining steps. For example, during the OER-induced transformation of TMCs into oxyhydroxides, AI can quantify the relationship between reconstruction rates and intermediate adsorption strength, thereby providing insight into the formation mechanisms of dynamic active sites.
PDF
(2975KB)
