Stabilizing amorphous NiFe-based catalysts via borate bridging for water oxidation under industrial conditions

Abstract

The practical deployment of an anion exchange membrane water electrolyzer (AEMWE) relies on the exploration of active and durable electrocatalysts towards the sluggish oxygen evolution reaction (OER). Although amorphous NiFe-based catalysts (a-NiFeO_xH_y) emerge as the competitive candidate due to impressive intrinsic OER activity, their unique defective structure renders the metal sites more susceptible to over-oxidation and dissolution, leading to poor stability. To address this challenge, we incorporate borate groups (BO₃³⁻) into the a-NiFeO_xH_y lattice by occupying the oxygen vacancy sites. The bridged borates not only maintain the structural stability via filling the oxygen vacancies, but also assist electron transfer from Ni to Fe to suppress Fe ion dissolution, thereby enhancing the catalytic stability of a-NiFeO_xH_y. Moreover, the tailored electronic structure of Ni favors electrochemical reconstruction to high-valence Ni active species and optimizes adsorption of oxygen intermediates towards superior OER activity. Therefore, a-B-NiFeO_xH_y integrated into the AEMWE can deliver a noteworthy current density of 4.75 A cm⁻² at a voltage of 2.0 V and maintain stable operation at 0.5 A cm⁻² for 3000 hours. This study not only affords a promising electrocatalyst for the AEMWE, but also paves a new avenue to break the activity-stability trade-off of amorphous materials for the OER.

---

Broader context

With the global push toward carbon neutrality, hydrogen production via water electrolysis driven by renewable electricity is gaining increasing attention as a clean and sustainable energy solution. Anion exchange membrane water electrolysis (AEMWE) holds particular promise due to its potential to combine the low-cost catalysts of alkaline systems with the compact architecture of PEM electrolyzers. However, the slow kinetics and poor durability of the oxygen evolution reaction (OER) on the anode severely restrict the performance and long-term viability of AEMWE systems. Among the various catalysts explored, amorphous NiFe-based materials exhibit high OER activity but suffer from poor operational stability, especially under industrially relevant conditions such as high current density and elevated temperature. Breaking the long-standing trade-off between activity and stability remains a key challenge. This study addresses the stability issue of amorphous NiFe-based electrocatalysts for the OER by filling oxygen vacancies with borate anions (BO₃³⁻). The tailored electronic states and strengthened Ni–O–Fe structure enable industrial current density and long-term stability over 3000 h in a practical water electrolyzer.

Introduction

The increasing global energy demand and the limited availability of fossil resources have driven a surge of interest in the development of sustainable energy conversion and storage technologies.¹ Water electrolysis powered by renewable electricity stands out as an ideal solution for the utilization and storage of intermittent renewable energy sources, such as solar and wind power, into green hydrogen, which aligns with global carbon neutrality targets.² Among the various water electrolysis technologies, anion exchange membrane water electrolysis (AEMWE) attracts particular attention as it can combine the advantages of alkaline water electrolysis (AWE) and proton exchange membrane water electrolysis (PEMWE), i.e., utilize non-precious metal electrocatalysts and zero-gap configuration.³ However, the sluggish oxygen evolution reaction (OER) kinetics at the anode lead to a considerable voltage loss of AEMWE, thus significantly decreasing its energy conversion efficiency.⁴ In this regard, developing highly efficient electrocatalysts to accelerate the OER kinetics is crucial to advancing AEMWE as a viable technology for sustainable hydrogen production.

The past decades have witnessed a blooming development of non-precious metal OER electrocatalysts, among which nickel–iron layered double hydroxides (NiFe-LDH) emerge as the most promising candidate due to their natural abundance, cost-effectiveness, and high catalytic activity.⁵ To further improve their activity, a variety of strategies have been employed, including cation/anion doping, morphology engineering, hetero-interface construction, defect engineering and amorphization.⁶ Notably, amorphization has been proven effective in boosting the OER activity of NiFe-LDH because the disordered structure of amorphous materials not only increases active site density by exposing abundant unsaturated edge sites, but also contributes to increased flexibility in the electronic structure towards optimized binding energy of oxygen-containing intermediates for the OER.⁷ This structural flexibility also enables more efficient electron and proton transfer during the reaction, thus expediting the OER kinetics. Despite their improved catalytic activity, amorphous NiFe catalysts suffer from instability issues, primarily due to the oxidation and dissolution of metal ions, particularly Fe, at high anodic potential.⁸ In fact, insufficient stability is the main bottleneck for the practical application of NiFe-LDH, and the amorphous structure aggravates this issue. What's worse, rapid degradation can occur at a larger current density (>500 mA cm⁻²) and high temperature, which represents real operating conditions of AEMWE.⁹ Therefore, how to break the activity-stability trade-off of the amorphous NiFe remains a critical challenge and requires urgent advancement to make these materials viable for practical applications in AEMWE.

Motivated by this challenge, we herein propose a borate anion (BO₃³⁻) doping strategy for enhancing the stability of amorphous NiFe electrocatalysts. Owing to the intrinsic electron-deficient nature of boron atoms, they can act as effective electron transit centers to significantly promote the oxidation of Ni²⁺ to high-valent Ni species, thereby enhancing the OER activity. Meanwhile, the incorporation of BO₃³⁻ anions helps stabilize the metal–oxygen framework and fill oxygen vacancies, which effectively suppresses the dissolution of metal ions and improves the structural stability of the catalyst. Therefore, borate doping offers comprehensive advantages over conventional doping strategies in amorphous NiFe-based electrocatalysts. As a result, the optimized a-B-NiFeO_xH_y catalyst exhibits remarkable OER stability, with negligible potential decay after 170 hours of continuous testing at 10 mA cm⁻², whereas the a-NiFeO_xH_y catalyst shows a significant increase in voltage over the same period. More impressively, when integrated into a home-made AEMWE, a-B-NiFeO_xH_y can deliver a high current density of 4.75 A cm⁻² at 2.0 V and maintains long-term stability at 0.5 A cm⁻² for 3000 h, demonstrating the huge application potential in the practical device. Combining in situ characterization with theoretical calculations, we unveil that the presence of BO₃³⁻ not only facilitates electron transfer from Ni to Fe to suppress the oxidation and dissolution of Fe during the OER but also tailors the adsorption strength of OER intermediates to propel the reaction kinetics, thereby endowing the a-B-NiFeO_xH_y catalyst with significantly improved stability and activity.

Results and discussion

The amorphous borate-doped nickel–iron hydroxides (denoted as a-B-NiFeO_xH_y) were synthesized by a co-precipitation rapid-quenching method (Fig. 1a). In particular, a high-concentration solution of metal precursors with an optimal Ni/Fe ratio of 3 : 1 was quickly mixed with a 2 mol L⁻¹ NaBH₄ alkaline solution under stirring, followed by immediately quenching the mixture solution in liquid nitrogen (details are shown in the ESI†), which inhibits the further aggregation and alignment of metal hydroxide nuclei, resulting in the formation of an amorphous structure.¹⁰ Systematic optimization revealed that this specific Ni/Fe ratio and NaBH₄ concentration achieve the ideal balance between structural amorphization and catalytic performance, demonstrating optimal activity and stability among all tested compositions (Fig. S1 and S2, ESI†). And the borate-free amorphous nickel–iron hydroxides (a-NiFeO_xH_y) were synthesized using the similar method without the addition of NaBH₄ to the alkaline solution. For comparison, crystalline NiFe LDH was prepared via the conventional hydrothermal method.

Fig. 1 Morphological characterization of a-B-NiFeO_xH_y. (a) Schematic diagram of the synthesis of a-B-NiFeO_xH_y and a-NiFeO_xH_y. (b)–(d) TEM, (e) and (f) high-resolution TEM (selected area electron diffraction), and (g) elemental mapping images of a-B-NiFeO_xH_y.

The as-prepared catalysts display a nanosheet-assembled nanoflower structure (Fig. 1b–d and Fig. S3–S5, ESI†), revealed through scanning electron microscopy (SEM) and transmission electron microscopy (TEM). Notably, the amorphous structure characteristics of a-B-NiFeO_xH_y (Fig. 1e and f) and a-NiFeO_xH_y (Fig. S4, ESI†) are confirmed by high-resolution TEM (HRTEM), from which no discernible lattice fringes can be observed. Besides, neither obvious bright spots nor clear diffraction rings can be discerned in the selected area electron diffraction (SAED) patterns, manifesting the amorphousness of a-B-NiFeO_xH_y and a-NiFeO_xH_y (Fig. 1f).¹¹ In sharp contrast, the NiFe LDH counterpart exhibits clear lattice fringes and diffraction spots, suggesting the well-crystalline structure. Elemental mapping images (Fig. 1g) illustrate the uniform distribution of Ni, Fe, B, and O elements in a-B-NiFeO_xH_y, with the mass fraction of B being 2.98%, verifying the effective incorporation of B. The X-ray diffraction (XRD) patterns of a-NiFeO_xH_y and a-B-NiFeO_xH_y show similar diffraction peaks, which validate that the borate incorporation does not alter the lattice structure. (Fig. 2a). The broad diffraction peak reflects the lack of a long-range ordered structure, while the peak around 35° exhibiting broadened asymmetry indicates the presence of more defects and disordered regions in the sample. These XRD features are consistent with the characteristics of amorphous materials.¹² By contrast, the NiFe LDH shows typical diffraction peaks located at 11.4°, 22.9° and 33.5° corresponding to the (003), (006) and (012) planes of the layered double hydroxide crystalline structure.¹³

Fig. 2 Structural characterization of a-B-NiFeO_xH_y. (a) XRD patterns. (b) FTIR spectra. (c) B 1s XPS of a-B-NiFeO_xH_y. (d) EPR spectra. (e) Ni K-edge of XANES spectra. (f) Ni K-edge of Fourier-transform EXAFS spectra. (g)–(i) The WT contour plots of a-B-NiFeO_xH_y, a-NiFeO_xH_y and NiFe LDH.

To further investigate the chemical structure of a-NiFeO_xH_y and a-B-NiFeO_xH_y, Fourier transform infrared (FTIR) spectroscopy and Raman spectroscopy were employed. The FTIR (Fig. 2b) spectrum of a-B-NiFeO_xH_y exhibits absorption peaks at 975 cm⁻¹ and 1280 cm⁻¹, which are assigned as the vibrations of BO₃³⁻.¹⁴ Moreover, the Raman peak at 287 cm⁻¹ corresponding to the B–O bonds is also discerned for a-B-NiFeO_xH_y (Fig. S6, ESI†). These findings confirm the successful incorporation of BO₃³⁻ into the amorphous a-B-NiFeO_xH_y. The borate incorporation process is accompanied by release of hydrogen gas, which would enrich the pore structure via the foaming effect. Brunauer–Emmett–Teller (BET) analysis was performed to validate the hypothesis (Fig. S7, ESI†).¹⁵ The a-B-NiFeO_xH_y exhibits type-IV isotherms with a hysteresis loop at high relative pressure, suggesting the dominated mesoporous structure. Besides, the specific surface area of a-B-NiFeO_xH_y (224.83 m² g⁻¹) is significantly larger than that of a-NiFeO_xH_y (45.03 m² g⁻¹). The mesoporous structure and the large specific surface area are beneficial for exposing more active sites, facilitating mass transport, and thereby improving the OER performance.

The chemical composition and electronic structure were then examined by a series of characterization studies. The inductively coupled plasma optical emission spectrometry (ICP-OES) analysis reveals that the Ni/Fe ratio in all three catalysts is consistently close to 3 : 1, in agreement with the feeding ratio (Tables S1 and S2, ESI†). The successful incorporation of BO₃³⁻ in a-B-NiFeO_xH_y is confirmed by X-ray photoelectron spectroscopy (XPS), which shows a characteristic peak of B–O at ∼192 eV (Fig. 2c).¹⁶ Ni 2p XPS (Fig. S8a, ESI†) of a-B-NiFeO_xH_y exhibits two peaks at 855.52 eV and 873.14 eV, corresponding to Ni²⁺ 2p_3/2 and Ni²⁺ 2p_1/2,¹⁷ respectively, which represents a positive shift of 0.12 eV compared with a-NiFeO_xH_y (Fig. S9a, ESI†). While the Fe 2p XPS (Fig. S8b and S9b, ESI†) remains nearly unchanged upon doping BO₃³⁻, where four peaks at 710.52 eV, 713.51 eV, 724.55 eV and 727.85 eV, attributable to Fe²⁺ 2p_3/2, Fe³⁺ 2p_3/2, Fe²⁺ 2p_1/2 and Fe³⁺ 2p_1/2,¹⁸ respectively, are clearly identified. The XPS results show no detectable characteristic signals for metallic Ni or Fe, indicating that Ni and Fe predominantly exist in their oxidized states rather than as metallic particles. Notably, the O 1s spectra (Fig. S10, ESI†) reveal a lower proportion of oxygen vacancies in a-B-NiFeO_xH_y (36.8%) compared to a-NiFeO_xH_y (52.2%), in line with the weaker electron paramagnetic resonance (EPR) signals (Fig. 2d) observed for a-B-NiFeO_xH_y.¹⁹ Positron annihilation spectroscopy quantitatively confirms this vacancy occupation mechanism (Fig. S11 and Tables S3, S4, ESI†), showing a 36.9% reduction in oxygen vacancy concentration (I₂ intensity decrease from 68.36% to 43.11%) accompanied by characteristic lifetime elongation (0.3189 ns to 0.3371 ns), which reflects the modified electronic environment of residual vacancies due to BO₃³⁻ incorporation. These findings demonstrate that BO₃³⁻ is introduced into a-B-NiFeO_xH_y by occupying oxygen vacancies within the amorphous NiFe lattice.

The local structure of the materials was further investigated using X-ray absorption spectroscopy (XAS). The Ni K-edge X-ray absorption near-edge structure (XANES) spectra (Fig. 2e) show a positive shift of the absorption edge toward higher energy in a-B-NiFeO_xH_yversus a-NiFeO_xH_y counterpart, signifying a higher valence state of Ni through metal-to-ligand charge transfer.²⁰ In contrast, an insignificant change of the absorption edge is discerned in the Fe K-edge XANES spectra of a-B-NiFeO_xH_y and a-NiFeO_xH_y (Fig. S12, ESI†). We further performed a systematic correlation between the absorption edge position (E₀) and the known oxidation states of the reference compounds and constructed calibration curves for both Ni and Fe. Quantitative analysis reveals that the average oxidation states of Ni are +2.22 in a-B-NiFeO_xH_y and +2.06 in a-NiFeO_xH_y, while Fe oxidation states are +2.58 and +2.52, respectively (Fig. S13, ESI†), indicating that the introduction of BO₃³⁻ increased the oxidation state of Ni while leaving the oxidation state of Fe unchanged. This observation is consistent with the XPS results. The doping effect on the coordination environment was then checked. To be specific, Ni K-edge k³-weighted extended X-ray absorption fine structure (EXAFS) analysis reveals larger oscillation amplitudes of a-B-NiFeO_xH_y than that of a-NiFeO_xH_y, suggesting an increase in the average coordination number (CN) of Ni after the incorporation of BO₃³⁻ (Fig. S14, ESI†).²¹ The EXAFS spectra (Fig. 2f) disclose the presence of Ni–O and Ni–M coordination shells in both a-B-NiFeO_xH_y and a-NiFeO_xH_y,²² whereas a-B-NiFeO_xH_y displays a higher intensity of the Ni–O peak, further confirming the increase in the coordination number of Ni after the filling of oxygen vacancies by BO₃³⁻.²³ The noticeable elongation of Ni–O bonds in a-B-NiFeO_xH_y further confirms that the introduction of borate anions inhibits the contraction of metal–oxygen bonds typically associated with the formation of oxygen vacancies, thereby enhancing the structural stability of the catalyst. Similarly, the signal intensity of the second coordination shell (Ni–M) in a-B-NiFeO_xH_y is also significantly enhanced compared to that in a-NiFeO_xH_y, as further supported by wavelet transform (WT) analysis (Fig. 2g). In NiFe hydroxides, Ni and Fe atoms are bridged via lattice oxygen; thus, the enhanced Ni–M signal suggests that the filling of oxygen vacancies by BO₃³⁻ formed a more stable bridging structure with metal sites, leading to stronger intermetallic interactions.²⁴ A more robust Ni–O–M bridging oxygen structure is beneficial for facilitating electron transfer between Ni and Fe during the OER process, which can also suppress metal leaching or lattice degradation towards simultaneously improved activity and stability.²⁵

The OER performance of the prepared catalysts was systematically evaluated in 1.0 M KOH electrolyte. The polarization curves (Fig. 3a) reveal that a-B-NiFeO_xH_y exhibits significantly higher OER activity compared to a-NiFeO_xH_y and NiFe LDH. In particular, a-B-NiFeO_xH_y achieves current densities of 10 and 100 mA cm⁻² at low overpotentials (η) of 234 and 288 mV (Fig. 3b), respectively. In contrast, a-NiFeO_xH_y requires overpotentials of 280 and 337 mV to reach the same current densities, highlighting the superior catalytic performance of a-B-NiFeO_xH_y. The boosted OER kinetics of a-B-NiFeO_xH_y is also certified by the lower Tafel slope of 30.1 mV dec⁻¹ than those of a-NiFeO_xH_y (41.1 mV dec⁻¹) and NiFe LDH (57.2 mV dec⁻¹) (Fig. 3c). To investigate the origins of the enhanced catalytic activity on a-B-NiFeO_xH_y, the double-layer capacitance (C_dl) was estimated (Fig. S15, ESI†) to evaluate its electrochemical active surface area (ECSA), as C_dl is typically proportional to the ECSA of an electrocatalyst. The results show that a-B-NiFeO_xH_y has a significantly higher C_dl (Fig. 3d) value (2.7 mF cm⁻²) than a-NiFeO_xH_y (1.9 mF cm⁻²), indicating more accessible active sites of a-B-NiFeO_xH_y. Furthermore, the intrinsic activity of different catalysts was also compared by normalizing the OER current by ECSA (Fig. 3e). Obviously, across the whole potential window, a-B-NiFeO_xH_y exhibits superior intrinsic activity, which may be attributed to the optimized electronic structure of Ni active sites following the incorporation of BO₃³⁻. Moreover, Nyquist plots (Fig. 3f) reveal that the charge transfer resistance of a-B-NiFeOxHy is significantly lower than that of a-NiFeO_xH_y, indicating accelerated charge transfer kinetics on a-B-NiFeO_xH_y.

Fig. 3 Electrocatalytic performance of a-B-NiFeO_xH_y. (a) Polarization curves of various samples for the OER. (b) Overpotentials of a-B-NiFeO_xH_y, a-NiFeO_xH_y and NiFe LDH to achieve the current densities of 10 and 100 mA cm⁻². (c) Tafel plots of a-B-NiFeO_xH_y, a-NiFeO_xH_y and NiFe LDH. (d) Capacitance. (e) ECSA-normalized LSV curves. (f) Nyquist plots. (g) CP curves of a-B-NiFeO_xH_y and a-NiFeO_xH_y.

We further investigated the promotion effect of BO₃³⁻ anion incorporation on electrochemical stability through accelerated degradation tests. After 1000 cyclic voltammetry (CV) cycles (Fig. S17, ESI†), the a-B-NiFeO_xH_y electrocatalyst exhibits negligible degradation in OER performance, while a positive shift of 20 mV in overpotential at 20 mA cm⁻² is observed on a-NiFeO_xH_y, demonstrating enhanced stability upon BO₃³⁻ anion incorporation. The exceptional durability of the as-synthesized a-B-NiFeO_xH_y for the OER is further corroborated by chronoamperometry (CA) and chronopotentiometry (CP) tests (Fig. 3g and Fig. S16, S18–S20, ESI†). The decay rates for a-B-NiFeO_xH_y are 0.04 mA h⁻¹ and 0.03 mV h⁻¹ for CA and CP, respectively, significantly lower than those of a-NiFeO_xH_y (0.23 mA h⁻¹ and 1.20 mV h⁻¹). To simulate the potential application of a-B-NiFeO_xH_y in practical electrolyzers, its performance for overall water splitting (OWS) was evaluated in a two-electrode configuration (a-B-NiFeO_xH_y‖Pt/C). Remarkably, a-B-NiFeO_xH_y‖Pt/C only requires the cell voltages of 1.47, 1.65, and 1.82 V to deliver the current densities of 10, 50, and 100 mA cm⁻², respectively (Fig. S21, ESI†), lower than those of IrO₂‖Pt/C (1.50, 1.73 and 1.98 V). CP results of the a-B-NiFeO_xH_y‖Pt/C system demonstrate higher current densities at lower overpotentials for overall water splitting, along with no observable degradation after 30 hours of continuous operation, highlighting its long-term durability (Fig. S22, ESI†). These electrochemical measurements combined with characterization techniques collectively suggest that the introduction of BO₃³⁻ into the lattice oxygen vacancies endows a-B-NiFeO_xH_y with a more robust M–O–M bridging oxygen structure and enhanced Ni–Fe interaction, thereby leading to a substantial improvement in its OER stability compared to a-NiFeO_xH_y (Table S5, ESI†).

To comprehensively understand the influence of BO₃³⁻ on OER activity and stability, we monitored the dynamic structure evolution during the OER using XAS, XPS and Raman spectroscopy. The increase in white-line intensity and the positive shift in the absorption edge confirm the oxidation of Ni center during the OER process (Fig. S23, ESI†). Additionally, the enhanced Ni–M signal in the EXAFS spectra indicates a stronger coordination between Ni and Fe, as well as a tighter Ni–M metal bonding. These observations provide direct evidence for the transformation of NiFe hydroxide to NiFeOOH during the OER process.²⁶ Furthermore, the EXAFS curves (Fig. 4a and Fig. S24, ESI†) reveal a significant increase in the intensity of the Ni–O peak after the OER, indicating an increase in the coordination number (CN) of Ni in a-NiFeO_xH_y, consistent with the fitting results (increasing from 3.9 to 4.4, Fig. S25, ESI†). In contrast, the first coordination shell environment of Ni in a-B-NiFeO_xH_y remains nearly unchanged (Fig. 4b and Fig. S26, ESI†) due to the stable Ni–O–M oxygen-bridged structure, demonstrating that the incorporation of BO₃³⁻ into lattice oxygen vacancies enhances the structural stability of the catalyst during the OER. The chemical state change after the OER was then examined by XPS. Compared to OER-treated a-NiFeO_xH_y (Fig. S27–S29, ESI†), the Ni 2p_3/2 peak of OER-treated a-B-NiFeO_xH_y (Fig. 4c) exhibits a positive shift of 0.23 eV, while the Fe 2p_3/2 peak (Fig. 4d) shows a negative shift of 0.49 eV, consistent with the higher Ni absorption edge energy (Fig. S30, ESI†) and lower Fe absorption edge energy (Fig. S31, ESI†) observed in the XANES spectra of OER-treated a-B-NiFeO_xH_y. These results unambiguously reveal the preferential oxidation of borate-bridged Ni sites in a-B-NiFeOxHy to high-valent Ni³⁺ during the OER process. This is further supported by the lower onset potential and peak position of the Ni²⁺ to Ni³⁺ transition in a-B-NiFeO_xH_y (Fig. S32, ESI†).²⁷ Additionally, extensive studies have shown that under alkaline conditions and high anodic potentials, higher oxidation states of Fe are more susceptible to dissolution. The lower oxidation state of Fe in a-B-NiFeO_xH_y is promising to suppress its dissolution.

Fig. 4 Spectroscopic characterization of a-B-NiFeO_xH_y. (a) Ni K-edge of Fourier-transform EXAFS spectra in a-NiFeO_xH_y. (b) Ni K-edge of Fourier-transform EXAFS spectra in a-B-NiFeO_xH_y. (c) Ni 2p_3/2 XPS spectrum of the catalyst after the OER. (d) Fe 2p_3/2 XPS spectrum of the catalyst after the OER. (e) Electrochemical in situ Raman spectra of a-B-NiFeO_xH_y in 1 M KOH solution. (f) Electrochemical in situ Raman spectra of a-NiFeO_xH_y in 1 M KOH solution.

Operando electrochemical Raman spectroscopy, which provides more insights into the dynamic evolution of Ni species during the OER (Fig. 4e, f and Fig. S33, ESI†), was performed. Two characteristic peaks at 475 cm⁻¹ and 550 cm⁻¹ can be found, corresponding to the E_g bending vibration mode and the A_1g stretching vibration mode of Ni³⁺–O in NiOOH, respectively.²⁸ Compared to a-NiFeO_xH_y, the Ni³⁺–O signal in a-B-NiFeO_xH_y appears at a lower potential, indicating that the Ni species in a-B-NiFeO_xH_y are more easily oxidized after the introduction of BO₃³⁻. In the Raman spectrum, the A_1g peak corresponds to a more disordered vibrational mode, while the E_g peak represents a more ordered vibrational mode. Notably, the intensity ratio of the A_1g and E_g peaks (I₅₅₀/I₄₇₅) for a-B-NiFeO_xH_y is much smaller than that of a-NiFeO_xH_y, indicating that the incorporation of BO₃³⁻ has filled some of the oxygen vacancies, thereby reducing the disordered vibrations and enhancing the ordered vibrations (Fig. S34 and S35, ESI†). This suggests that a-B-NiFeO_xH_y possesses a more stable and ordered coordination structure.²⁹ Furthermore, the insignificant change in the I₅₅₀/I₄₇₅ value with increasing voltage evidences BO₃³⁻ mediated structural stabilization that kinetically suppresses dynamic reconstruction, whereas a-NiFeO_xH_y undergoes pronounced structure reconstruction.

The role of BO₃³⁻ incorporation into the host lattice in enhancing the OER electrocatalytic performance was further investigated using density functional theory (DFT) calculations. According to the experimental results, the models of a-NiFeOOH and BO₃³⁻-doped a-NiFeOOH (denoted as a-B-NiFeOOH) were constructed (Fig. S36, ESI†). Charge density difference analysis (Fig. 5a and Fig. S37, ESI†) reveals that the introduction of BO₃³⁻ into the lattice oxygen vacancies significantly promotes electron transfer between the metal and oxygen. A strong electron depletion around the Ni atoms is observed, indicating partial electron transfer from Ni to O.³⁰ This observation is in line with the Bader charge analysis (Fig. S38, ESI†), which demonstrates decreased electron density of Ni in a-B-NiFeOOH configuration. The electron localization function (ELF) analysis quantitatively discloses the enhanced electron density localization at hydroxyl moieties coordinated to Ni centers in the BO₃³⁻-incorporated sample, directly evidencing an increased oxidation state of Ni through ligand-induced charge polarization (Fig. 5b and Fig. S39, ESI†).

Fig. 5 DFT calculations. (a) Charge density difference diagrams (electron accumulation is in yellow and depletion in blue, respectively) of a-B-NiFeOOH. (b) ELF of a-B-NiFeOOH and a-NiFeOOH. The scale bar is located at the bottom, accompanied by a color scheme for viewing convenience. (Blue represents a fully delocalized electron state, while red represents a fully localized electron state.) (c) COHP analysis for a-B-NiFeOOH and a-NiFeOOH. (d) Polarization curves in 1 M KOH (without and with 1.0 mol L⁻¹ methanol). (e) Gibbs free energy diagrams of OER intermediates over a-B-NiFeOOH and a-NiFeOOH. (f) The dissolution simulation of Fe in a-B-NiFeOOH and a-NiFeOOH. (g) Dissolution of Fe after CV cycling. The silver, yellow, red, white, and green atoms represent Ni, Fe, O, H, and B, respectively.

The regulated electronic states would affect the adsorption/desorption properties of OER intermediates. We then performed crystal orbital Hamilton population (COHP) analysis to analyze the bonding–antibonding property of Ni–O bonds, where a more positive integrated COHP (ICOHP) value indicates weaker bond strength. As shown in Fig. 5c, a-B-NiFeOOH exhibits a more positive ICOHP value compared to a-NiFeOOH, indicating weakened interaction between the metal active sites and oxygen intermediates after introducing BO₃³⁻. Given the electrophilic nature of OH*, it can be probed through the reaction with nucleophilic alcohol molecules (e.g., methanol and ethanol).³¹ As shown in Fig. 5d, a-B-NiFeO_xH_y exhibits a lower faradaic current from methanol oxidation compared to a-NiFeO_xH_y (6.68 vs 8.11 mA cm⁻² V), quantitatively demonstrating weaker OH* adsorption on the borate-modified surface. The decreased OH* adsorption is further corroborated by Laviron analysis (Fig. S40 and S41, ESI†),³² which shows a smaller redox constant (K_s) of a-B-NiFeO_xH_y than that of a-NiFeO_xH_y. Operando electrochemical impedance spectroscopy (EIS) coupled with Bode phase analysis (Fig. S42, ESI†) quantitatively resolves the accelerated dehydrogenation kinetics of OH*/OOH* intermediates on a-B-NiFeO_xH_y, evidenced by the obvious reduction in low-frequency region phase angle.³³ Based on the four-electron transfer mechanism, the Gibbs free energy (ΔG) for each elementary step of the OER, including the adsorption/desorption of different oxygen intermediates (OH*, O*, OOH*, and O₂), was calculated (Fig. 5e and Fig. S43, ESI†). The Gibbs free energy diagrams show that the weakened binding affinity with OH* on a-B-NiFeOOH lowers the thermodynamic barrier for the potential-determining OOH* formation step from 2.17 eV to 1.77 eV, thereby enhancing the OER catalytic activity.

To shed light on the enhanced stability, the dissolution behavior of Fe in both a-B-NiFeOOH and a-NiFeOOH during the OER process was examined as the previous studies revealed that Fe over-oxidation and dissolution under high anodic potentials are key factors affecting the stability of NiFe-based catalysts (Fig. 5f). The formation energy barrier of soluble Fe in a-B-NiFeOOH was calculated to be −2.73 eV, which is significantly higher than that in a-NiFeOOH (−3.82 eV). This indicates that the introduction of BO₃³⁻ into the lattice oxygen vacancies inhibits the dissolution of Fe, thus improving the catalyst's stability. The suppressed Fe dissolution is also certified by quantitative analysis of dissolved Fe ions after stability testing (Fig. 5g and Fig. S44 and S45, ESI†). In particular, the dissolution of Ni is significantly lower than that of Fe. During the first 100 cycles of CV activation, the Fe dissolution in both a-NiFeO_xH_y and a-B-NiFeO_xH_y is comparable. However, after the subsequent 1000 cycles of CV stability testing, only a small amount of Fe dissolved from a-B-NiFeO_xH_y, while Fe continued to dissolve from a-NiFeO_xH_y. After the stability test, the Fe dissolution amounts from a-NiFeO_xH_y are 3.32 times greater than that from a-B-NiFeO_xH_y, affording experimental evidence for the inhibited degradation of Fe on the more stable BO₃³⁻ incorporated a-NiFeOOH structure.

Considering the high activity and stability of a-B-NiFeO_xH_y for the OER, we assembled an AEMWE device using a-B-NiFeO_xH_y as the anode catalyst to evaluate its potential for practical application. The steady-state polarization curve of a-B-NiFeO_xH_y shows significantly higher performance (Fig. 6a). In particular, industrial-level current densities (Fig. 6b) of 1, 2, and 3 A cm⁻² were achieved at low cell voltages of only 1.60 V, 1.72 V, and 1.84 V, respectively, which are markedly superior to those of a-NiFeO_xH_y (1.66 V at 1 A cm⁻², 1.83 V at 2 A cm⁻², and 1.95 V at 3 A cm⁻²). Notably, the electrolyzer can reach a high current density of 4.75 A cm⁻² at 2 V (Fig. S46, ESI†). Additionally, a-B-NiFeO_xH_y also exhibits unprecedented long-term stability by continuously operating at 0.5 A cm⁻² for 3000 h, with a voltage decay of only 0.11 mV h⁻¹, highlighting its significant potential for practical applications in the AEMWE (Fig. 6c and Fig. S47, ESI†). Overall, the a-B-NiFeO_xH_y anode electrolyzer outperforms the majority of reported non-precious metal catalyst-based anode electrolyzers in terms of both activity and, particularly, stability (Fig. 6d).

Fig. 6 AEMWE. (a) Polarization curves of the AEM electrolyzer obtained at 80 °C. (b) Comparison of the operational activity (voltage at 1 A cm⁻²) of a-B-NiFeO_xH_y with the reported anode catalysts. (c) CP curve of the AEM electrolyzer using the a-B-NiFeO_xH_y catalyst operated at 0.5 A cm⁻² at 60 °C. (d) Comparison of the operational stability (operation time and degradation rate) of a-B-NiFeO_xH_y with the reported anode catalysts.

Conclusions

In summary, this work presents an effective strategy for addressing the stability issue of amorphous NiFe-based OER electrocatalysts by incorporating BO₃³⁻ into the oxygen vacancy site. Mechanistic investigations reveal that the incorporated BO₃³⁻ not only optimizes the coordination environment of Ni atoms towards stabilizing the Ni–O–Fe oxygen-bridge framework but also facilitates electron transfer from Ni to Fe. This efficient electron transfer increases the dissolution energy barrier of Fe by 1.09 eV, thus suppressing its oxidation and dissolution for enhanced electrochemical stability. Operando spectroscopic analyses and theoretical calculations demonstrate that the electron redistribution promotes the formation of high-valence Ni³⁺ active species and optimizes oxygen intermediate adsorption towards faster OER kinetics. These attributes endow the a-B-NiFeO_xH_y catalyst with unprecedented activity and stability when used as the anode in the AEMWE, delivering a high current density of 4.75 A cm⁻² at 2.0 V cell voltage, while maintaining ultra-long-term stability over 3000 hours at 0.5 A cm⁻². This work provides fundamental insights into the rational design of high-performance amorphous electrocatalysts for sustainable water splitting applications.

Author contributions

W. Xing and J. Zhu supervised the project. J. Zhu proposed the concept. W. Qi performed the experiments and wrote the manuscript. M. Xiao and L. Wang revised the manuscript. K. Li conducted the DFT calculations. Y. Zeng helped with implementing the experiment. All authors participated in data analysis and manuscript discussion.

Conflicts of interest

There are no conflicts to declare.

Data availability

The data that support the findings of this study are available from the corresponding author upon reasonable request.

Acknowledgements

This work was financially supported by the National Natural Science Foundation of China (22279129) and the Jilin Province Science and Technology Development Program (20230101367JC and 20230201154GX). We thank the Shanghai Synchrotron Radiation Facility for conducting the X-ray absorption spectroscopy experiments at the BL17B1 station. The computational work was supported by the high-performance computing center of Jilin Province.

Notes and references

1. J. T. Li, Nano-Micro Lett., 2022, 14, 122.
2. M. Chen, N. Kitiphatpiboon, C. R. Feng, A. Abudula, Y. F. Ma and G. Q. Guan, eScience, 2023, 3, 100111.
3. J. Q. Wang, Y. Liu, G. C. Yang, Y. Q. Jiao, Y. M. Dong, C. G. Tian, H. J. Yan and H. G. Fu, Nat. Commun., 2025, 16, 1319.
4. B. H. Yao, Y. T. Chen, Y. Y. Yan, Y. Yang, H. H. Xing, Y. C. Xu, D. X. Jiao, Z. C. Xing, D. W. Wang and X. R. Yang, Angew. Chem., Int. Ed., 2024, 64, e202416141.
5. J. Zhao, J. J. Zhang, Z. Y. Li and X. H. Bu, Small, 2020, 16, 2003916.
6. L. Yang, Z. Liu, S. Zhu, L. Feng and W. Xing, Mater. Today Phys., 2021, 16, 100292.
7. W. Wang, Z. X. Wang, Y. C. Hu, Y. C. Liu and S. L. Chen, eScience, 2022, 2, 438–444.
8. H. X. Liao, G. H. Ni, P. F. Tan, K. Liu, X. Z. Liu, H. L. Liu, K. J. Chen, X. S. Zheng, M. Liu and J. Pan, Adv. Mater., 2023, 35, 2300347.
9. Y. H. Hu, T. Y. Shen, Z. H. Wu, Z. H. Song, X. L. Sun, S. Y. Hu and Y. F. Song, Adv. Funct. Mater., 2025, 35, 2413533.
10. Y. Duan, Z. Y. Yu, S. J. Hu, X. S. Zheng, C. T. Zhang, H. H. Ding, B. C. Hu, Q. Q. Fu, Z. L. Yu, X. Zheng, J. F. Zhu, M. R. Gao and S. H. Yu, Angew. Chem., Int. Ed., 2019, 58, 15772–15777.
11. C. Y. Guo, Y. M. Shi, S. Y. Lu, Y. F. Yu and B. Zhang, Chin. J. Catal., 2021, 42, 1287–1296.
12. Z. Li, Y. E. Zhou, M. H. Xie, H. Cheng, T. Wang, J. Chen, Y. Lu, Z. L. Tian, Y. Q. Lai and G. H. Yu, Angew. Chem., Int. Ed., 2023, 62, e202217815.
13. J. Y. Wang, Z. Y. Xing, R. Y. Kang, Y. J. Zheng, Z. Zhang, T. Ma, Y. Wang, B. Yin, Y. Z. Liao, L. Li, C. Cheng and S. Li, Adv. Funct. Mater., 2024, 35, 2418439.
14. H. X. Liao, G. H. Ni, P. F. Tan, Y. Liu, K. J. Chen, G. M. Wang, M. Liu and J. Pan, Appl. Catal., B, 2022, 317, 121713.
15. S. Q. Wang, W. Y. Huo, H. C. Feng, Z. H. Xie, J. K. Shang, E. V. Formo, P. H. C. Camargo, F. Fang and J. Q. Jiang, Adv. Mater., 2023, 35, 2304494.
16. Y. K. Bai, Y. Wu, X. C. Zhou, Y. F. Ye, K. Q. Nie, J. O. Wang, M. Xie, Z. X. Zhang, Z. J. Liu, T. Cheng and C. B. Gao, Nat. Commun., 2022, 13, 6094.
17. X. Li, Z. Xie, S. Roy, L. Gao, J. Liu, B. Zhao, R. Wei, B. Tang, H. Wang, P. Ajayan and K. Tang, Adv. Mater., 2024, 37, e2410295.
18. R. Li, R. Y. Wu, Z. B. Li, J. Wang, X. J. Liu, Y. R. Wen, F. K. Chiang, S. W. Chen, K. C. Chan and Z. P. Lu, Adv. Mater., 2023, 35, 2206890.
19. Y. Wang, P. Guo, J. Zhou, B. Bai, Y. F. Li, M. R. Li, P. Das, X. H. Wu, L. J. Zhang, Y. Cui, J. P. Xiao and Z. S. Wu, Energy Environ. Sci., 2024, 17, 8820–8828.
20. X. Z. Su, Y. Wang, J. Zhou, S. Q. Gu, J. Li and S. Zhang, J. Am. Chem. Soc., 2018, 140, 11286–11292.
21. J. T. Ding, D. Y. Guo, N. S. Wang, H. F. Wang, X. F. Yang, K. Shen, L. Y. Chen and Y. W. Li, Angew. Chem., Int. Ed., 2023, 62, e202311909.
22. Y. D. Yao, G. M. Zhao, X. Y. Guo, P. Xiong, Z. H. Xu, L. H. Zhang, C. S. Chen, C. Xu, T. S. Wu, Y. L. Soo, Z. M. Cui, M. M. J. Li and Y. Zhu, J. Am. Chem. Soc., 2024, 146, 15219–15229.
23. S. L. Zhao, C. H. Tan, C. T. He, P. F. An, F. Xie, S. Jiang, Y. F. Zhu, K. H. Wu, B. W. Zhang, H. J. Li, J. Zhang, Y. Chen, S. Q. Liu, J. C. Dong and Z. Y. Tang, Nat. Energy, 2020, 5, 881–890.
24. H. X. Liao, K. J. Chen, X. R. He, J. X. Tong, X. Z. Liu, P. F. Tan, X. Y. Guo and J. Pan, Nano Lett., 2024, 24, 15436–15443.
25. H. Wu, J. W. Chang, J. K. Yu, S. Y. Wang, Z. Hu, G. I. N. Waterhouse, X. Yong, Z. Y. Tang, J. B. Chang and S. Y. Lu, Nat. Commun., 2024, 15, 10315.
26. X. Luo, H. Zhao, X. Tan, S. Lin, K. Yu, X. Mu, Z. Tao, P. Ji and S. Mu, Nat. Commun., 2024, 15, 8293.
27. L. S. Peng, N. Yang, Y. Q. Yang, Q. Wang, X. Y. Xie, D. Sun-Waterhouse, L. Shang, T. R. Zhang and G. I. N. Waterhouse, Angew. Chem., Int. Ed., 2021, 60, 24612–24619.
28. M. Han, C. H. Wang, J. Zhong, J. R. Han, N. Wang, A. Seifitokaldani, Y. F. Yu, Y. C. Liu, X. H. Sun, A. Vomiero and H. Y. Liang, Appl. Catal., B, 2022, 301, 120764.
29. Y. J. Wu, J. Yang, T. X. Tu, W. Q. Li, P. F. Zhang, Y. Zhou, J. F. Li, J. T. Li and S. G. Sun, Angew. Chem., Int. Ed., 2021, 60, 26829–26836.
30. Q. Y. Wang, Y. J. Gong, X. Zi, L. Gan, E. Pensa, Y. X. Liu, Y. S. Xiao, H. M. Li, K. Liu, J. W. Fu, J. Liu, A. Stefancu, C. Cai, S. Y. Chen, S. G. Zhang, Y. R. Lu, T. S. Chan, C. Ma, X. Y. Cao, E. Cortes and M. Liu, Angew. Chem., Int. Ed., 2024, 63, e202405438.
31. H. B. Tao, Y. H. Xu, X. Huang, J. Z. Chen, L. J. Pei, J. M. Zhang, J. G. G. Chen and B. Liu, Joule, 2019, 3, 1498–1509.
32. Y. Huang, L. W. Jiang, B. Y. Shi, K. M. Ryan and J. J. Wang, Adv. Sci., 2021, 8, 2101775.
33. Y. Du, F. K. Xie, M. F. Lu, R. X. Lv, W. X. Liu, Y. D. Yan, S. C. Yan and Z. G. Zou, Nat. Commun., 2024, 15, 1780.

---

Footnote

† Electronic supplementary information (ESI) available. See DOI: https://doi.org/10.1039/d5ey00157a

---