Fe-S dually modulated adsorbate evolution and lattice oxygen compatible mechanism for water oxidation

Abstract

Simultaneously activating metal and lattice oxygen sites to construct a compatible multi-mechanism catalysis is expected for the oxygen evolution reaction (OER) by providing highly available active sites and mediate catalytic activity/stability, but significant challenges remain. Herein, Fe and S dually modulated NiFe oxyhydroxide (R-NiFeOOH@SO₄) is conceived by complete reconstruction of NiMoO₄·xH₂O@Fe,S during OER, and achieves compatible adsorbate evolution mechanism and lattice oxygen oxidation mechanism with simultaneously optimized metal/oxygen sites, as substantiated by in situ spectroscopy/mass spectrometry and chemical probe. Further theoretical analyses reveal that Fe promotes the OER kinetics under adsorbate evolution mechanism, while S excites the lattice oxygen activity under lattice oxygen oxidation mechanism, featuring upshifted O 2p band centers, enlarged d-d Coulomb interaction, weakened metal-oxygen bond and optimized intermediate adsorption free energy. Benefiting from the compatible multi-mechanism, R-NiFeOOH@SO₄ only requires overpotentials of 251 ± 5/291 ± 1 mV to drive current densities of 100/500 mA cm⁻² in alkaline media, with robust stability for over 300 h. This work provides insights in understanding the OER mechanism to better design high-performance OER catalysts.

The oxygen evolution reaction is crucial for energy conversion but faces challenges in catalyst optimization. Here, the authors present a dual-modulated NiFe oxyhydroxide (R-NiFeOOH@SO4) that enhances OER performance through optimized metal and lattice oxygen sites, achieving a compatible multi-mechanism.

Introduction

Electrocatalytic water splitting as promising technology for sustainable hydrogen production has attracted extensive attention^1–5. However, the anodic oxygen evolution reaction (OER) suffers from a sluggish kinetics, greatly limiting the efficiency of water splitting^6–10. Consequently, it is significant for in-depth understanding of the OER catalysis mechanism to conceive high-performing OER catalysts¹¹. Based on the conversion steps of key reaction intermediates during OER, there are two mainstream reaction pathways involved in the OER mechanism, the adsorbate evolution mechanism (AEM) and lattice oxygen mechanism (LOM)^12,13. In the AEM, the metal site serves as redox center and involves multiple highly related oxygen intermediates during OER, resulting in a thermodynamic limit on the minimum overpotential of 0.37 V^14–16. For the LOM, the oxygen site is activated as redox center and participates in formation of oxygen, which breaks the scaling relationship limit of the AEM with more favorable thermodynamics¹⁷. However, the dominant LOM pathway easily leads to structural turbulence and catalytic performance declination of catalysts due to involvement of lattice oxygen as well as formation and refilling of oxygen vacancies^18,19. Undoubtedly, to construct AEM-LOM compatible mechanisms by simultaneously triggering the redox of metal and lattice oxygen sites can combine the advantages of each pathway and reconcile the OER catalytic activity and stability^18,20,21. However, in current catalytic systems featuring a single component and coordination environment, the electron transfer process occurs either at the metal site or at the lattice oxygen site, depending on the positions of the metal and oxygen bands relative to the Fermi level²². Therefore, developing a coupled catalytic system that can concurrently activate metal and lattice oxygen redox couples still faces significant challenges.

Transition metal oxyhydroxides (MOOH, M = Fe, Co, or Ni), derived from irreversible structural reconstruction under OER electrooxidation conditions, are generally recognized as actual active species, in which NiFe-based oxyhydroxides are tending to replace commercial RuO₂ and IrO₂ as benchmark of OER catalysts^23–27. Particularly, some recent reports have found that OER reconstruction-derived oxyhydroxides with high valence metal centers have the potential to trigger LOM for improving the OER activity^28–31, which provides critical ideas for constructing catalysts with a symphonious OER pathway by electrocatalytic reconstruction strategies. However, electrocatalytic structural reconstruction of catalysts is usually limited to the near-surface nanoscale, resulting in low active component utilization^1,32. NiMoO₄ hydrates feature a three-dimensional network of four edge-shared NiO₆ octahedra connected to MoO₄ tetrahedra, and have satisfactory structural self-reconstructive properties through anodic potential-driven co-leaching of crystalline water and Mo, which can be selected as suitable self-sacrificial pre-catalyst to build oxyhydroxides with flexible coordination structures³³.

In this work, the NiMoO₄ hydrate is employed as pre-catalyst, while Fe and S as modulator are co-introduced by chemical etching, inducing abundant structural defects and promoting the complete reconstruction to R-NiFeOOH@SO₄ actives during the electrochemical activation. Characterization evidences reveal that anodic activation triggers the redox of both metal and lattice oxygen sites and involves the formation and refilling of oxygen vacancies. Furthermore, an AEM-LOM compatible OER catalytic mechanism can be confirmed by in situ attenuated total reflectance Fourier transform infrared spectra (ATR FTIR) and ¹⁸O isotope-labeling differential electrochemical mass spectrometry (DEMS), in which R-NiFeOOH@SO₄ is endowed with simultaneously optimized AEM and LOM pathways under the co-modulation of Fe and S species. Density functional theory (DFT) calculations reveal that the Fe introduction serving as active site of the AEM pathway optimizes the OER intermediate adsorption, while the S introduction significantly stimulates the lattice oxygen activity and increases the LOM pathway occupancy of OER. With the co-modulation of Fe and S, the R-NiFeOOH@SO₄ system achieves synergistic catalysis of AEM and LOM pathways, maximizing the utilization of surface metal and oxygen active sites and enhancing the OER catalytic activity.

Results

Synthesis and structural characterizations of catalysts

The NiFe oxyhydroxide with SO₄^2- modification (R-NiFeOOH@SO₄) was synthesized via a chemical etching and in-situ electrochemical self-reconstruction process as described in Fig. 1a. First, NiMoO₄ hydrate (NiMoO₄∙xH₂O) nanorods were grown on nickel foam (NF) substrate through a hydrothermal reaction (Fig. 1b and Supplementary Fig. 1), which were then used as template to introduce Fe and S species during the chemical etching process. The obtained NiMoO₄∙xH₂O@Fe,S pre-catalyst with roughened surfaces still predominantly exhibits the crystal structure of NiMoO₄∙xH₂O accompanied by attenuated crystallinity (Fig. 1c and Supplementary Figs. 2–4). Then, the R-NiFeOOH@SO₄ active was synthesized by anodic cyclic voltammetry (CV) activation to remove NiMoO₄∙xH₂O self-sacrificial templates. Accordingly, R-NiOOH and R-NiFeOOH as control sample were obtained by anodic activation of NiMoO₄∙xH₂O and NiMoO₄∙xH₂O@Fe.

Fig. 1. Preparation and structures of catalysts.

a Synthetic route of nanocoral-like NiMoO₄∙xH₂O@Fe,S, with the inset zoomed-in images presenting the surface structure evolution of the nanorods. FESEM, TEM, and High-resolution TEM images of b, f NiMoO₄∙xH₂O, c, g NiMoO₄∙xH₂O@Fe,S and d, e, h R-NiFeOOH@SO₄. (The inset shows a magnified view, and the inset in g is the corresponding selected area electron diffraction pattern). i HAADF-STEM image and corresponding EDS elemental mappings of Ni, Mo, O, Fe, S for R-NiFeOOH@SO₄.

Both field emission scanning electron microscopy (FESEM) and transmission electron microscopy (TEM) images show that all the catalysts inherit the initial nanorod architecture (Fig. 1d, e and Supplementary Figs. 5–7), and R-NiFeOOH@SO₄ exhibits obvious particle refinement on surfaces (Supplementary Fig. 8). After activation, the Brunauer-Emmett-Teller (BET) surface area of R-NiFeOOH@SO₄ increases (Supplementary Fig. 9 and Supplementary Table 1). The high-resolution TEM (HRTEM) image (Fig. 1f) reveals the high crystallinity of NiMoO₄∙xH₂O nanowires, while significant surface amorphization and more lattice defects are present for NiMoO₄∙xH₂O@Fe,S (Fig. 1g and Supplementary Fig. 4c) and NiMoO₄∙xH₂O@Fe (Supplementary Fig. 5c), further confirming the etching-induced decrease in crystallinity, which is more favorable for structural reconstruction during water oxidation. After anodic CV activation, the lattice fringes belonging to initial NiMoO₄∙xH₂O are hardly observed in HRTEM images (Fig. 1h and Supplementary Figs. 6b, 7c), and the exposed fringe spacings of 0.208 and 0.240 nm are corresponded to the (210) and (011) lattice planes of orthorhombic NiOOH (JCPDS No.27-0956), determining the successful transformation of NiMoO₄∙xH₂O as template to NiOOH actives. The powder X-ray diffraction (XRD) pattern (Supplementary Fig. 10) reveals that the diffraction peak of initial NiMoO₄∙xH₂O in R-NiFeOOH@SO₄, R-NiFeOOH, and R-NiOOH completely disappears, while no related diffraction peak of NiOOH is detected due to lower crystallinity. In addition, energy-dispersive X-ray spectroscopy (EDS) (Fig. 1i and Supplementary Figs. 6c, 7d and Supplementary Table 2) confirms the almost complete removal of elemental Mo from all catalysts, and the elements of Ni, Fe, S, and O are distributed homogeneously throughout the entire nanorod region for R-NiFeOOH@SO₄. The inductively coupled plasma-optical emission spectrometry (ICP-OES) results further substantiate the large-scale Mo dissolution in R-NiFeOOH@SO₄, and the content of Fe and S as modulator was determined to be 9.88 and 6.6 at % (Supplementary Table 3).

Structural evolution of catalysts induced by anodic activation was further explored. During the electrochemical CV activation, the OER activity of the catalysts present apparent dynamic self-optimization accompanied by the leaching of the Mo component and the transformation of oxyhydroxides (Supplementary Figs. 11–13). Furthermore, the introduction of Fe and S species can regulate the electro-oxidative reconstruction of Ni sites, accelerating the self-optimizing kinetics and Mo dissolution. In situ Raman spectra (Supplementary Figs. 14, 15) further reveal the conversion of the catalysts to oxyhydroxides at anodic potentials, in which NiMoO₄∙xH₂O@Fe,S has a modulated Ni-site redox and favorable Mo dissolution. This facilitates the complete reconstruction of NiMoO₄∙xH₂O@Fe,S into active oxyhydroxides, and provides more available reactive species to boost the OER process. Indeed, NiMoO₄∙xH₂O@Fe,S exhibits a larger increment in electrochemically active surface area (ECSA) induced by CV activation (Supplementary Figs. 16–19), indicating that anodic activation produces more available activity sites.

The electronic and coordination structures of catalysts were further analyzed by X-ray photoelectron spectroscopy (XPS) and X-ray absorption spectroscopy (XAS). The Ni 2p spectra of NiMoO₄∙xH₂O@Fe and NiMoO₄∙xH₂O@Fe,S significantly shift to lower binding energies relative to NiMoO₄∙xH₂O (Fig. 2a and Supplementary Fig. 20a, b), with a concomitant shift of the Mo 3d binding energy to higher energy (Supplementary Fig. 20c), demonstrating the migration of electrons from Mo to Ni sites. In Fe 2p XPS spectrum (Supplementary Fig. 20d), NiMoO₄∙xH₂O@Fe mainly presents the characteristic signal of Fe³⁺, while NiMoO₄∙xH₂O@Fe,S displays coexisting Fe³⁺ and Fe²⁺ due to the redox between the introduced S₂O₃^2- and Fe^3+34,. The O 1 s spectrum (Supplementary Fig. 20e) of NiMoO₄∙xH₂O@Fe and NiMoO₄∙xH₂O@Fe,S possesses additional M-OH and vacancy oxygen (O_v) characteristic signals, suggesting that the etching process induces the formation of O_v. Furthermore, the appearance of S-O and S-M signals evidences the introduction of S species into NiMoO₄∙xH₂O@Fe,S (Supplementary Fig. 20f).

Fig. 2. Electronic and coordination structures of catalysts.

High-resolution XPS of (a) Ni 2p, b O 1 s, c Fe 2p, and d S 2p for obtained samples before and after CV activation. e Normalized Ni K-edge, f Fe K-edge XANES spectra of obtained samples and selected reference materials. The inset shows a magnified view of the selected area, x-axis: Energy (eV), y-axis: Normalized absorption (a.u.). g Fourier transformed k3χ(R) Ni K-edge spectra of obtained samples before and after CV activation.

After CV activation, Mo signals are mostly difficult to detect in R-NiOOH, R-NiFeOOH, and R-NiFeOOH@SO₄ (Supplementary Fig. 21a-c). However, the Ni 2p binding energy of all catalysts presents an abnormal negative shift in Fig. 2a, which should originate from the conversion of unstable NiOOH to Ni(OH)₂ under ex-situ testing^{35, 36}. However, such a negative shift between NiMoO₄∙xH₂O@Fe,S and R-NiFeOOH@SO₄ is smaller (−0.24 eV), and R-NiFeOOH@SO₄ has a higher Ni oxidation state compared to R-NiOOH and R-NiFeOOH (Supplementary Fig. 22a), indicating less conversion of Ni²⁺, which comes from its more favorable complete reconstruction process. In particular, the metal-lattice oxygen (M-O) signal in the O 1 s spectrum (Fig. 2b) exhibits a clear positive shift for the derivatized oxyhydroxide, implying that the oxidation of lattice oxygen (M-O^(2-δ)-) is involved in the electrochemical activation process. Among them, the M-O offset of R-NiFeOOH@SO₄ is more significant (0.4 eV). Meanwhile, the vacancy oxygen content of R-NiFeOOH@SO₄ is reduced compared to NiMoO₄∙xH₂O@Fe,S due to OH^- refilling in the electrolyte. Furthermore, the Fe 2p signal peaks of R-NiFeOOH@SO₄ shift to the higher binding energy in Fig. 2c, indicating a transition of surface Fe species to a higher oxidation state, while the Fe 2p binding energy of R-NiFeOOH remains stable at +3 valence (Supplementary Fig. 22b). As shown in Fig. 2d, the S 2p signal of R-NiFeOOH@SO₄ is completely transformed into oxidized S species, which corresponds to the characteristic signal peak of sulfate (SO₄^2-)³⁷.

The adjustment of the electronic structure of the Ni site of NiMoO₄∙xH₂O by the introduction of Fe, S can be further confirmed by the X-ray absorption near edge structure (XANES) of Ni K-edge spectra in Supplementary Fig. 23a. After CV activation, the absorption edges of all catalysts shift to the lower energy, in which R-NiFeOOH@SO₄ has a lower degree of shift and is located at the highest energy position compared to R-NiOOH and R-NiFeOOH (Fig. 2e and Supplementary Fig. 23b), consistent with XPS analysis results. Compared to NiMoO₄∙xH₂O@Fe, the Fe K-edge XANES (Fig. 2f and Supplementary Fig. 25a) also displays a lower average Fe oxidation state in NiMoO₄∙xH₂O@Fe,S, which shifts to the higher energy for R-NiFeOOH@SO₄. In addition, the XANES spectra (Supplementary Fig. 26) of the S K-edge prove that the S species in R-NiFeOOH@SO₄ is presented in the form of SO₄^2-, further revealing the oxidation of the S species.

The local coordination structure of catalysts was investigated by Fourier transform extended X-ray absorption fine structure (FT-EXAFS) spectra. Two characteristic peaks at 1.59 and 2.72 Å in Ni K-edge FT-EXAFS spectra (Fig. 2g and Supplementary Fig. 23c) can be ascribed to the Ni-O first shell and the Ni-M second shell, respectively³⁸. In Supplementary Fig. 23c, the peak intensity of the Ni-O path of NiMoO₄∙xH₂O@Fe,S and NiMoO₄∙xH₂O@Fe is significantly reduced, and the reduced Ni-O coordination indicates etching-induced oxygen vacancy defects in the NiO₆ octahedron³⁹. In addition, the Ni-O path of NiMoO₄∙xH₂O@Fe,S shifts to the higher R direction due to S substitution. After CV activation, the peak intensity of the Ni-O path of R-NiFeOOH@SO₄ is enhanced (Fig. 2g), further confirming that the refilling of oxygen vacancies is involved in the OER activation process. The reduced oxygen vacancy content of R-NiFeOOH@SO₄ indicates that the refilling rate of OH^- is faster than the removal of lattice oxygen, beneficial to maintaining structural stability. Meanwhile, the peak intensity of the Ni-O path of R-NiOOH weakens, indicating that oxygen vacancy defects are formed during the activation process. For R-NiFeOOH, the change in Ni-O coordination is not obvious, accompanied by a shift toward higher R direction. In Fe K-edge FT-EXAFS spectra (Supplementary Fig. 25b), two main peaks can be associated with the Fe-O and Fe-M bonds, the typical shell is almost consistent with the Ni shell, indicating the incorporation of Fe. Obvious changes in the Fe site coordination environment after CV activation can be observed in Fe K-edge EXAFS and FT-EXAFS spectra (Supplementary Fig. 25c), in which the strengthening of Fe-M channels may originate from the dissolution and redeposition of Fe.

The above analysis indicates that the introduction of Fe and S species can regulate the crystal and electronic structure of NiMoO₄∙xH₂O, thus endowing NiMoO₄∙xH₂O@Fe,S with more favorable reconstruction kinetics toward R-NiFeOOH@SO₄. In particular, the OER activation simultaneously triggers the redox of metal and lattice oxygen sites, suggesting the possible existence of coupled AEM-LOM catalysis, in which R-NiFeOOH@SO₄ with Fe and S co-modulation exhibits more obvious metal/lattice oxygen oxidation and oxygen vacancy refilling.

Electrocatalytic OER performance

The OER performance of electrochemically stable reconstructed catalysts was further systematically evaluated with NiFe layered double hydroxide (NiFe LDH) as benchmark. The chemical etching condition for the synthesis of NiMoO₄∙xH₂O@Fe,S as pre-catalyst was explored (Supplementary Figs. 28, 29). The derivatized optimal R-NiFeOOH@SO₄ only requires an OER overpotential of 251 ± 5 mV to achieve a current density of 100 mA cm⁻², lower than that of R-NiOOH (367 ± 4 mV), NiFe LDH (286 ± 6 mV), and R-NiFeOOH (284 ± 7 mV) (Fig. 3a, b). In particular, R-NiFeOOH@SO₄ can drive high current densities of 500 and 1000 mA cm⁻² under competitive overpotentials of 291 ± 1 and 308 ± 1 mV, respectively. As shown in Fig. 3c and Supplementary Fig. 32a, it is also equipped with the lowest Tafel slope (56 mV dec⁻¹) and the smallest charge transfer resistance (R_ct), indicating its favorable OER kinetics and efficient charge transfer rate. As we previously analyzed, R-NiFeOOH@SO₄ (23.5 mF cm⁻²) holds the most favorable ECSA compared to R-NiOOH (11.8 mF cm⁻²), R-NiFeOOH (8.75 mF cm⁻²) and NiFe LDH (8.4 mF cm⁻²), emphasizing the modification effect of S species on the active surface (Supplementary Fig. 32b, 33). In addition, the ECSA normalized LSV curve (Supplementary Fig. 34) confirms the optimization of the anodic activation-induced R-NiFeOOH@SO₄ on the intrinsic OER activity of the catalyst, and the dissolved ion in the electrolyte as well as KOH etching has no obvious effect on the OER activity (Supplementary Fig. 35). Furthermore, NiFeOOH@SO₄ presents the optimal turnover frequency (TOF) for OER (Supplementary Fig. 36), with a near 98 ± 2% Faraday efficiency (Supplementary Figs. 37, 38).

Fig. 3. Electrocatalytic OER performance of catalysts.

a OER Polarization curves with 80% iR compensation (The electrode working area is 0.5 cm × 0.5 cm, and the solution resistance used for compensation is approximately 2.3 ± 0.2 Ω), and b corresponding overpotentials of catalysts at 100 and 500 mA cm⁻² (The standard deviations were obtained through three reduplicative measurements in Supplementary Fig. 31). c Tafel plots of catalysts. d LSV curves before and after 3000 CV cycles, and e Chronoamperometry curves without iR compensation of R-NiFeOOH@SO₄ in 1 M KOH at 0.65 and 0.85 V vs. Hg/HgO.

Also, the assembled Pt/C | | R-NiFeOOH@SO₄ electrode pair exhibits remarkable activity for overall water splitting (Supplementary Fig. 39), which can drive 10 and 100 mA cm⁻² current densities at only 1.441 and 1.618 V, respectively, without IR compensation, which significantly outperforms the commercial Pt/C | |RuO₂ electrode pair. On the other hand, for R-NiFeOOH@SO₄ (Fig. 3d), no obvious attenuation is observed before and after 3000 cycle CV scanning, and it even maintains stable operation for 300 h at 100 and 500 mA cm⁻² with only slight attenuation (Fig. 3e), fully illustrating the remarkable OER catalytic stability. After long-term stability test, the nanocoral-like architecture and composition of R-NiFeOOH@SO₄ still well maintain as evidenced by SEM, TEM, EDX, Raman and XPS characterizations, proving robust structural framework (Supplementary Figs. 40-43).

Compatible AEM-LOM OER catalysis

The synergistic catalysis mechanism involved in R-NiFeOOH@SO₄ and the origin of OER activity were further explored. In Fig. 4a and Supplementary Fig.44, the OER activity of all catalysts exhibits a certain degree of pH dependence, while the current density of R-NiFeOOH@SO₄ decreases more rapidly with the decrease of pH value, identifying that lattice oxygen is involved in the OER process¹⁴. In addition, R-NiFeOOH@SO₄ exhibits significantly attenuated OER activity in 1 M TMAOH due to the inhibited LOM under the strong binding of tetraalkylammonium cation (TMA⁺)^{29, 40}, while it has only a slight activity decay for R-NiFeOOH and R-NiOOH (Fig. 4b).

Fig. 4. Compatible AEM-LOM OER catalysis analysis.

a The relationship between the logarithm of current density of catalysts at the potential of 1.7 V versus RHE and pH. b OER Polarization curves of catalysts in 1 M KOH and 1 M TMAOH, respectively, without iR compensation. c Detected DEMS signals without any correction or subtraction of ¹⁶O¹⁶O, ¹⁶O¹⁸O and ¹⁸O¹⁸O for R-NiFeOOH@SO₄ relative to time and corresponding LSV curves, no iR compensation. In-situ Raman spectra of ¹⁸O-labelled (d) R-NiOOH, e R-NiFeOOH and f R-NiFeOOH@SO₄ measured at 1.624 V vs RHE in 0.1 M KOH with H₂¹⁶O. The Raman spectra of the ¹⁶O labeled sample is placed on the top for comparative analysis. In situ ATR-FTIR spectra of (g) R-NiOOH, h R-NiFeOOH, i R-NiFeOOH@SO₄.

To further validate the participation of lattice oxygen of catalysts during OER, in situ ¹⁸O isotope-labeling differential electrochemical mass spectrometry (DEMS) was manipulated. The catalysts were labeled with ¹⁸O to further detect the O₂ release during OER in an electrolyte containing ¹⁶O (Supplementary Fig. 45, Details are in the Methods). As shown in Fig. 4c and Supplementary Fig. 46, all the DEMS measurement results of ¹⁸O-labeled catalysts present the signals of m/z = 32, and m/z = 34, indicating the generation of ¹⁶O₂ and ¹⁶O¹⁸O gas, confirming that one lattice oxygen participates in O₂ of release⁴¹. The DEMS results (Supplementary Fig. 47) of R-NiFeOOH@SO₄ without ¹⁸O labeling only reveal ¹⁶O¹⁶O signals, and no obvious ¹⁶O¹⁸O is detected, indicating that the natural abundance of ¹⁸O isotopes has no significant interference with the measurement results, further substantiating that the detected ¹⁸O originates from the labeled lattice oxygen in the catalyst. Meanwhile, the measured ¹⁸O abundance (Supplementary Fig. 48) is significantly greater than the natural ¹⁸O abundance (0.2%), which can further exclude the interference of the natural abundance ¹⁸O on the test results⁴². Notably, R-NiFeOOH@SO₄ exhibits larger ¹⁶O¹⁸O peak area and a higher ¹⁶O¹⁸O /¹⁶O¹⁶O area ratio (Supplementary Figs. 49–52), indicating its higher lattice oxygen activity and more favorable LOM pathway⁴³. Meanwhile, compared with that of R-NiOOH, the ¹⁶O¹⁸O peak area of R-NiFeOOH increases, while its ¹⁶O¹⁸O/¹⁶O¹⁶O area ratio decreases, which means that the Fe modulation contributes more significantly to oxygen evolution under the AEM pathway.

Moreover, in situ Raman spectroscopy with ¹⁸O isotope labeling also confirms the involvement of lattice oxygen in the OER process. As shown in Fig. 4d–f and Supplementary Fig. 53, under the influence of oxygen mass on the vibration mode, the ¹⁸O-labeled catalysts present an obvious shift to a lower wavenumber²⁸. When a constant potential of 1.624 V vs RHE was applied in an electrolyte containing ¹⁶O, the Raman peak of the ¹⁸O-labeled catalysts gradually shifts back to the position of the conventional ¹⁶O-labeled catalyst as the ¹⁸O is consumed, among which R-NiFeOOH@SO₄ has faster lattice oxygen release, and the Raman peak shifts back within 1 min. In contrast, R-NiFeOOH and R-NiOOH require 20 min or even longer to release the labeled ¹⁸O. This result further substantiates the faster lattice oxygen release of R-NiFeOOH@SO₄ during the OER, revealing the effective activation of lattice oxygen by S species modulation.

In addition, in situ ATR FTIR was performed to investigate the adsorbed OER intermediate (Supplementary Fig. 54). In Fig. 4g-i, as the applied potential increases, distinct absorption peaks can be observed near 1200 and 1100 cm⁻¹, indicating the production of oxygenated intermediates. The peak located at 1207–1212 cm⁻¹ can be attributed to the O-O intermediate formed during the LOM pathway, further confirming the involvement of LOM⁴⁴. Meanwhile, the peak at about 1095 cm⁻¹ is ascribed to the *OOH produced by the AEM pathway, accompanied by a companion peak located at 1126 cm⁻¹, which may originate from the effect of the adjacent hydrogen bonding of the *OOH^44,45. Compared with R-NiOOH, R-NiFeOOH presents a significantly enhanced *OOH peak and a weakened O-O peak, implying an optimization of the OER in the AEM pathway induced by Fe introduction. In Fig. 4i, R-NiFeOOH@SO₄ exhibits a merged and higher intensity *OOH absorption peak at 1115 cm⁻¹, with further enhancement of the O-O peak, demonstrating the simultaneous optimization of AEM and LOM. In situ ATR FTIR results further confirm the coupled AEM and LOM of the catalysts, of which R-NiFeOOH@SO₄ facilitates the OER process with synergistically optimized AEM and LOM.

Theoretical calculation

DFT calculations were further carried out to gain deeper understanding of the connection between the OER synergy mechanism and the activity optimization. The actual OER active species NiOOH, NiFeOOH, and NiFeOOH@SO₄ was chosen as computational models. First, the density of states of the O 2p orbital was calculated to reflect the lattice oxygen activity in Fig. 5a. The O 2p band center (−2.626 eV) of NiFeOOH@SO₄ exhibits a upshift relative to the Fermi level (E_F) compared to NiOOH (−2.773 eV) and NiFeOOH (−2.814 eV), beneficial for the removal of electrons from oxygen sites, revealing the effective activation of lattice oxygen by S modulation^46,47. Besides, the Mott-Hubbard splitting in d-orbitals was further analyzed. Based on the molecular orbital theory, the overlap of oxygen 2p and metal d orbitals for transition metal oxyhydroxides generates M-O bonding bands (oxygen characteristics) and M-O* antibonding bands (metal characteristics), in which the M-O* antibonding band will split into empty upper Hubbard band (UHB) and electron-filled lower Hubbard band (LHB) under a strong d-d Coulomb interaction (U)^48,49. The energy difference between UHB and LHB can reflect the U parameter, whose increase can allow LHB to penetrate into the M-O bonding band, enabling the removal of electrons from the M-O bonding band, thereby achieving lattice oxygen oxidation^14,28,50. As illustrated in Fig. 5b, the calculated UHB/LHB center position of NiOOH, NiFeOOH and NiFeOOH@SO₄ is 1.05/−2.5, 1.12/−2.7 and 1.14/−2.86 eV, respectively. The larger U-value of NiFeOOH@SO₄ (4 eV) than that of NiFeOOH (3.82 eV) and NiOOH (3.55 eV) further verifies its facilitated lattice oxygen activation. Crystal orbital Hamilton populations (COHP) were calculated to evaluate the metal-oxygen bond strength. In Fig. 5c, the quantified Ni-O bond strength by integrating of -COHP up to the Fermi level (-ICOHP) was determined to be 0.752, 0.711, and 0.666 for NiOOH, NiFeOOH and NiFeOOH@SO₄, respectively. The lower -ICOHP value for NiFeOOH@SO₄ indicates that Fe,S co-modulation induces more electrons to be filled into the antibonding orbitals, which leads to a weaker Ni-O bond fueling the involvement of lattice oxygen and the formation of oxygen vacancies in the LOM pathway.

Fig. 5. DFT calculation analysis.

a Projected density of states, b Schematic band diagrams (UHB upper Hubbard band, LHB lower Hubbard band) and c Crystal orbital Hamilton populations (COHP) of the Ni-O bond. Gibbs free energy diagrams of the AEM pathway on (d) Ni sites and e Fe sites, and f LOM pathway. g The energy barrier of RDS for AEM and LOM. Above data are based on NiOOH, NiFeOOH and NiFeOOH-SO₄. h Schematic illustration of the AEM and LOM pathway on NiFeOOH@SO₄.

The Gibbs adsorption free energy diagram of NiOOH, NiFeOOH, and NiFeOOH@SO₄ was also computed founded the AEM and LOM pathways (Supplementary Figs. 55–62). The adsorption free energy of Ni and Fe atoms as adsorption sites in the AEM pathway is considered, respectively. As presented in Fig. 5d, the rate-determining step (RDS) energy barrier of the NiFeOOH@SO₄ model on the Ni site is optimized to 1.92 eV under Fe,S co-modification. Upon Fe introduction, the deprotonation energy barrier of *OH as an RSD at Fe sites dramatically decreases to 1.66 and 1.64 eV for NiFeOOH and NiFeOOH@SO₄, respectively, implying a switch of active sites to Fe (Fig. 5e), which theoretically explains the substantial enhancement of the OER activity by Fe introduction. The Gibbs free energy diagram of the LOM is exhibited in Fig. 5f, the formation of *O-OH is determined as the RDS for NiOOH, NiFeOOH, and NiFeOOH@SO₄, with energy barriers of 2.06, 3.18 and 2.01 eV, respectively. This calculation results suggest that the introduction of single Fe is not conducive to the LOM pathway, which may be related to the competitive AEM thermodynamics of Fe sites, while the modulation of S species (SO₄) effectively optimizes the OER activity in the LOM pathway.

Consequently, the Fe,S co-modulation achieves the simultaneous optimization of OER intermediate adsorption of NiFeOOH@SO₄ under AEM and LOM pathways, narrowing the energy barrier difference between AEM and LOM and improving the compatibility of multiple mechanisms (Fig. 5g, h).

Discussion

In summary, we design Fe,S species co-modulated NiMoO₄ hydrates (NiMoO₄∙xH₂O@Fe, S) as pre-catalyst, which regulates the crystal and electronic structure of NiMoO₄∙xH₂O and promotes its complete reconstruction into R-NiFeOOH@SO₄ actives. XPS and XAFS analyses reveal that anodic activation simultaneously trigger the redox of metal and lattice oxygen sites. Compared to the derived R-NiOOH and R-NiFeOOH, the R-NiFeOOH@SO₄ modified with S species is endowed with higher lattice oxygen activity and more favorable LOM tendency, as evidenced by electrochemical and chemical probe analysis. In situ ¹⁸O isotope-labeling DEMS and ATR FTIR spectroscopies confirm the aspired AEM-LOM compatible OER mechanism for R-NiFeOOH@SO₄, with simultaneous optimization of AEM and LOM pathways under the co-modulation of Fe and S. DFT calculation analysis further reveals the key roles of Fe and S as modulator and elucidates the origin of the improved OER activity. The Fe optimizes the adsorption free energy of OER intermediates under the AEM pathway by virtue of Fe active sites, while the S increases the lattice oxygen activity and reduces the energy barrier of the RDS under the LOM pathway, jointly optimizing the OER activity by improving the compatibility of the AEM and LOM pathways. As a result, the R-NiFeOOH@SO₄ active as derivative is equipped with high intrinsic activity and stability for OER in alkaline media. This work offers inspiration for designing highly stable and active non-noble metal catalysts via rational excitation of lattice oxygen and synergistic metal and oxygen sites.

Methods

Chemicals

Nickel nitrate hexahydrate (Ni(NO₃)₂·6H₂O), Sodium molybdate dihydrate (Na₂MoO₄·2H₂O), sodium thiosulfate pentahydrate (Na₂S₂O₃·5H₂O), Ferric nitrate nonahydrate (Fe(NO₃)₃·9H₂O), and Potassium hydroxide (KOH) were purchased from Sinopharm Chemical Reagent Co., Ltd. Hydrochloric acid (HCl), ethanol (C₂H₅OH) and acetone were purchased from Shanghai Chemical Reagent Company. Pt/C (20% Wt), RuO₂ and Nafion (5 wt %) were purchased from Sigma-Aldrich. All reagents were used directly without purification. Ultrapure water (18.25 MΩ cm) was used throughout the whole experiments.

Material preparations

NiMoO₄∙xH₂O nanowires were first synthesized on NF through a simple hydrothermal method. 1 mmol Ni(NO₃)₂·6H₂O and 1 mmol Na₂MoO₄·2H₂O were dissolved in 20 mL DI water and stirred for 30 min, the obtained solution was transferred to a 50 mL Teflon-lined stainless-steel autoclave with the pretreated NF (2 cm × 3 cm), then maintained at 150 °C for 6 h. After cooling down room temperature, the NiMoO₄∙xH₂O/NF was taken out and carefully cleaned and dried, and then immersed in 10 ml of mixed solution (2 mmol Fe(NO₃)₃·9H₂O and 1 mmol Na₂S₂O₃·5H₂O) preheated by a 60 °C water bath for another 10 minutes to obtain NiMoO₄∙xH₂O@Fe, S/NF. In addition, samples with different Na₂S₂O₃·5H₂O contents (0.5 mmol, 1 mmol) and holding time (1 min, 5 min, 15 min) are synthesized for comparison by the same method. Electrochemical CV activation was manipulated from 0.2 to 0.8 V vs. Hg/HgO for 20 cycles at a scan rate of 20 mV s⁻¹ to induce structural reconstruction of NiMoO₄∙xH₂O, NiMoO₄∙xH₂O@Fe, and NiMoO₄∙xH₂O@Fe, S, generating reactive R-NiOOH, R-NiFeOOH, and R-NiFeOOH@SO₄. The load of R-NiFeOOH@SO₄ on the Ni foam is determined to be about 6.85 mg cm⁻², which is obtained by high-precision electronic scale by the quality changes with bare Ni foam.

Characterization

The morphology and microstructures of as-prepared samples were characterized by scanning electron microscope (FESEM, Zeiss Ultra Plus) and transmission electron microscope (TEM, Talos F200S) equipped with X-ray energy dispersive spectroscopy (EDS). The crystalline structures information was obtained by X-ray powder diffraction (XRD, Empyrean) with Cu K α radiation (λ = 0.15406 nm). X-ray photoelectron spectroscopy (XPS) was performed on a Thermo Scientific K-Alpha instrument. Raman spectra were collected by using a 532 nm laser as excitation light (Renishaw, Invia). Fourier transform infrared spectroscopy (FT-IR) was conducted on Nicolet IS5 (Thermo Fisher Scientific Inc.). The inductively-coupled plasma optical emission spectrometry (ICP-OES) was conducted on Teledyne Leeman Labs Prodigy 7. X-ray absorption spectra (XAS) of Ni, Fe and S K-edge were collected at BL14W1 and BL16U1 station in Shanghai Synchrotron Radiation Facility, respectively.

Electrochemical measurements

All electrochemical measurements were conducted at room temperature (25 ± 2 °C) on a CHI 660E electrochemical workstation (Shanghai Chenhua) with a standard three-electrode system. The alkaline (1 M KOH) electrochemical measurements were performed using an Hg/HgO electrode and a graphite rod as the reference electrode and counter electrode, respectively, the synthesized self-supported electrodes (working area, 0.5 cm × 0.5 cm) were directly employed as working electrode. The 1 M KOH electrolyte used was freshly prepared (within two weeks) by adding 66 g KOH (85%) to 1000 ml deionized water and stored in a sealed volumetric flask away from light. And the pH value of the 1 M KOH is determined to be about 13.96 through multiple measurements (Supplementary Fig. 27).

The Hg/HgO reference electrode was calibrated by a H₂-saturated three-electrode system with Hg/HgO served as the reference electrode, Pt mesh and Pt wire were used as working electrodes and counter electrodes, respectively. All the electrochemical measurements were converted to the reversible hydrogen electrode (RHE) by the equation:

$$E_{\text{RHE}}=E_{\left(\text{Hg}/\text{HgO}\right)}\text{+}0.924\text{V}$$ 1

The polarization curves were obtained by using linear sweep voltammetry (LSV) at a scan rate of 2 mV s⁻¹ in the range of 0.2–1.2 V vs. Hg/HgO, with 80% iR compensation according to the formula:

$$E_{\text{compensated}}=E_{\text{measured}}-iRs$$ 2

where i is the measured current, and the Rs is the solution resistance, determined by the Nyquist plot in the high-frequency region (about 2.3 ± 0.2 Ω). Electrochemical impedance spectroscopy (EIS) was performed at the overpotential of 250 mV with the frequency range from 0.01 Hz to 100 kHz. The OER polarization curves before and after iR compensation are presented in Supplementary Fig. 30. The double layer capacitance (C_dl) was estimated by CV curves obtained at at various scan rates (20, 40, 60, 80, 100, 120 mV s⁻¹) in a non-faradaic region of 0.2–0.3 V (vs. Hg/HgO). Chronoamperometric curves without iR compensation were obtained at constant potential of 0.65 V and 0.85 V (vs. Hg/HgO) for evaluating the long-term stability of the catalyst. The Faradaic efficiency was calculated by the water drainage method to quantify the O₂ produced during the OER. Detailed information on the electrochemical measurements is provided in Supplementary Note 1. Electrochemical evaluation parameters such as overpotential, Faradaic efficiency, and TOF were tested three times to ensure repeatability, and other parameters were obtained by measurement once.

In situ Raman spectra

The in-situ Raman spectra were collected by a Raman spectrometer (LABRAM HR Evolution) with 532 nm excitation from an argon ion laser. NiMoO₄∙xH₂O/NF, NiMoO₄∙xH₂O@Fe/NF, and NiMoO₄∙xH₂O@Fe, S/NF were used as working electrodes, with a Hg/HgO and a graphite rod as reference and counter electrodes, respectively. During OER measurement in 1 M KOH, the Raman spectra were collected at different potentials by chronoamperometry test. For the in-situ Raman spectroscopy experiments with ¹⁸O isotope labeling, the catalysts were first activated in 1 M KOH electrolyte containing H₂¹⁸O by 30 cycles of CV in the 0–0.8 V vs Hg/HgO potential window for ¹⁸O isotope labeling. After careful washing with H₂¹⁶O several times, the ¹⁸O-labeled R-catalysts were placed in 0.1 M KOH electrolyte with H₂¹⁶O, and the in-situ Raman spectra were recorded for different time periods (1 min, 3 min, 5 min, 10 min, 20 min) by applying a constant potential of 1.624 (V vs. RHE).

In-situ ATR FT-IR

The in situ attenuated total reflectance Fourier transform infrared spectroscopy (ATR FTIR, Bruker INVENIO-S) was carried out by using a three-electrode setup in 1 M KOH. The catalysts are scraped off from the nickel foam to prepare the ink and then dripped onto the glassy carbon electrode (GCE) as the working electrode, while the Pt wire as the counter electrode, and Hg/HgO as the reference electrode. The FT-IR spectrometer with an extended range diamond ATR accessory (Shanghai Linglu Instrument Co. Ltd) was employed. The FT-IR spectra were recorded by chronoamperometry method with the applied potential range from 0 to 0.8 V vs Hg/HgO.

In-situ isotope-labeled DEMS measurements

Isotope-labeled DEMS measurements were performed on a QAS 100 device (Linglu Instruments, Shanghai). The H₂¹⁸O (¹⁸O abundance: 98%) was purchased from Wuhan Isotope Technology Co., Ltd. Hg/HgO and Pt wire were used as reference electrode and counter electrode, respectively. The porous polytetrafluoroethylene (PTFE) membrane supported samples were used as working electrode, in which the porous PTFE membrane was provided by Linglu Instruments, with a thickness of 60 μm, a pore size of 0.22 μm, and a porosity of 50%. First, the samples were labeled with ¹⁸O-isotopes by CV activation for 30 cycles in the 0–0.8 V vs Hg/HgO potential window in 1 M KOH solutions with H₂¹⁸O, with a scan rate of 20 mV/s. Subsequently, the ¹⁸O isotope-labeled samples were carefully rinsed with H₂¹⁶O at least five times to remove the residual H₂¹⁸O. Then, LSV tests were performed for the labeled catalysts in 1 M KOH containing H₂¹⁶O from 0–0.8 V vs Hg/HgO, and the produced oxygen with different molecular weights during OER was detected in real time by mass spectroscopy. The ¹⁶O¹⁸O/¹⁶O¹⁶O signal ratio was quantified by the signal peak area ratio, in which the baseline was subtracted. The atomic fraction of ¹⁸O can be quantified by the following formula⁴²:

$${18}_a=\frac{Q_{\text{MS}}\left(16\text{O}18\text{O}\right)}{2\left[\right)Q_{\text{MS}}\left(16\text{O}18\text{O}\right)+Q_{\text{MS}}\left(16\text{O}16\text{O}\right)}$$ 3

where the Q refers to the integrated DEMS charge.

DFT calculations

For all DFT calculations, the Vienna Ab initio Simulation Package (VASP) was employed, and the Perdew-Burke-Ernzerhof (PBE) functional and projector-augmented wave (PAW) method were applied to describe the exchange-correlation energy and electron-ion interactions^51–53. The cut-off energy for plane wave expansion was set to 400 eV, and the convergence criteria for force was below 0.02 eV/Å with the iterative convergence of energy of 10⁻⁵eV. The Brillouin zone was sampled with a 2 × 2 × 1 Monkhorst-Pack grid. The COHP were calculated by the Lobster code^54,55. The optimized structural model is provided in Supplementary Data 1. More detailed DFT calculations are shown in the Supplementary Note 2.

Author contributions

X.L., H.Z. and X.T. contributed equally to this work. X.L. designed and carried out the experiments, collected data, and wrote the manuscript. H.Z. Performed and analyzed DFT calculations., X.T. collected, analyzed the electrochemical work. S.L., K.Y., X.M., Z.T. collected, analyzed, and interpreted in situ Raman Spectra and ATR FT-IR data. P.J. guided the analysis of the partial characterization results. S.M. supervised the work and co-wrote the manuscript. All authors discussed the results and commented on the manuscript.

Peer review

Peer review information

Nature Communications thanks the anonymous reviewers for their contribution to the peer review of this work. A peer review file is available.

Data availability

The data that support the plots are available within this paper and its Supplementary Information. Figures 2–5, Supplementary Figs. 2, 3, 9-13, 15-26, 28-37, 39, 41-44, 46-53 data generated in this study are provided in the Source data files. Source data are provided with this paper.

Competing interests

The authors declare no competing interests.

Supplementary information

The online version contains supplementary material available at 10.1038/s41467-024-52682-y.