Boosting oxygen evolution reaction by FeNi hydroxide-organic framework electrocatalyst toward alkaline water electrolyzer

Abstract

The oxygen evolution reaction plays a vital role in modern energy conversion and storage, and developing cost-efficient oxygen evolution reaction catalysts with industrially relevant activity and durability is highly desired but still challenging. Here, we report an efficient and durable FeNi hydroxide organic framework nanosheet array catalyst that competently affords long-term oxygen evolution reaction at industrial-grade current densities in alkaline electrolyte. The desirable high-intensity performance is attributed to three aspects as follows. First, two-dimensional nanosheet porous arrays with maximum specific surface facilitate mass/charge transfer to accommodate high-current-density catalysis. Second, in situ derived FeNi hydroxide motifs offer bimetallic synergistic catalysis centers with high intrinsic activity. Third, carboxyl ligands alleviate metal oxidation favorable for charge tolerability against peroxidation dissolution under strong polarization. As a result, this catalyst requires an overpotential of only 280 mV to deliver high current density up to 1 A/cm² with long durability over 1000 h. Moreover, an alkaline water electrolyzer with this catalyst alternative demonstrates an increased economic effectiveness compared to commercial levels at present.

Developing cost-efficient catalysts for oxygen evolution reaction is crucial for various modern energy technologies. Here the authors report an efficient and durable NiFe hydroxide organic framework catalyst for water oxidation at 1 A/cm² with long durability over 1000 h.

Introduction

The oxygen evolution reaction (OER) is an indispensable stepping stone to support various cathodic reactions^1–5, such as hydrogen evolution reaction (HER), carbon dioxide reduction reaction (CO₂RR) and nitrogen reduction reaction (NRR), for renewable energy conversion and storage. However, the OER efficiency generally suffers from the sluggish kinetics in the multistep electron-proton coupled transfer process accompanied with the H-O breakage and O-O formation^6,7. The RuO₂ and IrO₂ are considered as two benchmark OER catalysts, but the high cost and unsatisfactory stability impede their practical application⁸. Increasing study efforts have been devoted into the development of transition metal-based OER catalysts^9–16, including transition metal oxides, sulfides, and (oxy)hydroxides, and so on.

It is crucial for electrocatalysts to sustainably deliver high current densities at low overpotentials towards industrial applications with profitable markets^17–20. In addition to the material cost, the activity-stability balance at industrial-level current densities thus is a critical indexe that needs to be considered when designing OER catalysts for practical applications, while which has been much less emphasized for a long time. For instance, transition metal-based catalysts with excellent activity for alkaline OER have been explored in the past decade, especially the upmost ferro-nickel catalysts^21–25, but their performance at industrial-grade current densities over 0.5 A/cm² needs to be further improved in terms of low overpotentials and long-term durability.

Essentially, the activity and stability of anodic OER catalysts would encounter greater challenges under high current densities^26,27, different from the cases of low current densities or cathodic HER. Compared with the HER, the OER requires stronger anodic polarization, in which the active metal sites must be able to mediate the transitions of oxygen-containing intermediates against self-peroxidization aging²⁸. This requires the metal motifs have superior intrinsic activity and charge tolerance. Large current density means fast charge/mass transfer^29,30, so large active surface area, high electrical conductivity, and excellent hydrophilic/gas-phobic properties must be integrated into an electrocatalyst together to supply high-rate dynamics.

With these considerations in mind, we speculate that the metal-organic frameworks (MOFs) as an available platform for multifunctional integration design due to its unrivalled structural tunability^31,32. First, the MOFs endow with ordered structure and multistage porosity, capable of providing large specific surface area and valuable capillary effect for mass transfer acceleration. Second, various catalysts with synergistic ferro-nickel sites have shown super-high OER activity^33,34, thus ferro-nickel atoms might be designed as preferred metal nodes of MOFs to induce the atomic-scale synergy favorable for intrinsic kinetics. Third, the ferrocene (Fc) has recently emerged as an effective promoter in MOF-based catalysts based on its conductivity and redox properties^35,36, while its availability at high current densities remains uncertain. Moreover, its carboxylic ligands endow with the π–π interactions capable of improving bond links and electrical tolerance of metal sites³⁷. Therefore, developing Fc-based MOFs through rational structure design with targeted functions as robust OER catalysts for industrially relevant activity and durability should be feasible, but remains yet to be realized as a great challenge.

Here, we design the ferrocene-nickel organic framework (FcNiOF) nanosheet arrays and steer its reconstruction into the isomeric ferro-nickel hydroxide organic framework (FeNiHOF) nanosheet arrays as a highly active and durable OER catalyst. Benefiting from high specific surface and porosity, bimetallic synergistic active sites, and charge-flexible organic ligands together, the FeNiHOF shows high OER activity with low overpotentials of 280 and 284 mV at 1 and 2 A/cm², respectively, as well as long-term durability for over 1000 h at 1 A/cm², surpassing the most state-of-the-art OER catalysts reported with industrial-grade current densities so far. This work gains insights into the multifunctional integration principles towards high-current-density catalysis and offers a promising methodology for developing industrially usable electrocatalysts.

Results

Synthesis and structure of FcNiOF precursor

FcNiOF nanosheets grown on conductive Ni foam were synthesized using a facile hydrothermal method that is scalable toward industrial implementation (Fig. 1A). The color of Ni foam turns dark yellow, indicating the growth of FcNiOF nanosheets on the surface. The scanning electron microscopy (SEM) shows the uniform coverage of nanosheets standing densely on Ni foam substrate (Fig. 1B), constituting a vertical array structure (Fig. 1C). Such a unique array architecture endows with high porosity and large surface area favorable for dynamic transport of reactants and gaseous products, especially critical in drastic catalysis at high current densities. The transmission electron microscopy (TEM) reveals that these unit nanosheets have thin leaf-like structures and smooth surfaces, with an average thickness of around 28.9 nm (Supplementary Fig. 1) and several hundred nanometers in radial scale (Fig. 1D). The X-ray diffraction (XRD) of FcNiOF in Fig. 1E shows that all diffraction peaks except an accidental peak at 11.4° that can be related to the structure of carboxyl rings (PDF no. 50-2093) correspond to the simulated patterns basing on FcZn-MOF structure (CCDC no. 716347)³⁸. So, a crystal structure is determined that each Ni atom is coordinated by six O atoms of two carboxylate groups on different Fc units, and all the Fc units linked to each other with Ni atoms to form 2D lamella (Fig. 1A and Supplementary Fig. 2). However, the internal crystalline textures in these nanosheets are barely discernable under high-resolution TEM (HRTEM, Supplementary Fig. 3), similar to most observations of MOF texture^36,39. The energy-dispersive X-ray spectroscopy (EDS) analysis reveals the composition of Fe, Ni, C, and O elements in FcNiOF (Supplementary Fig. 4), and the corresponding elemental mapping visualizes their uniform distributions in an individual sheet (Fig. 1F).

Fig. 1. Structural characterizations of FcNiOF.

A Synthetic process of FcNiOF, (B, C) SEM images at different scale bars, (D) TEM image, (E) XRD pattern, and (F) EDS elemental maps of FcNiOF.

To probe the chemical composition and coordination, the Fourier-transform infrared (FT-IR), Raman and X-ray photoelectron spectroscopy spectra were further carried out on the as-synthesized FcNiOF nanosheets. In Supplementary Fig. 5, the FT-IR band located at 1670 cm⁻¹ that belongs to the C = O stretching vibration in the Fc-carboxyl groups vanishes completely in FcNiOF, which indicates that Ni atoms are successfully conjugated on the carboxyl groups of Fc units³⁵. The Raman bands at 605 and 345 cm⁻¹ are assigned to the characteristic ring-internal vibration mode and ring-external vibration mode of Fc units, respectively (Supplementary Fig. 6). The other bands assignments in the FT-IR and Raman spectra are listed in Supplementary Tables 1 and 2, respectively. The XPS survey analysis in Supplementary Fig. 7 indicates that FcNiOF contains four elements, Ni, Fe, C, and O, consistent with the EDS result. Moreover, two spin-orbit peaks located at 855.5 (Ni 2p_3/2) and 873.1 eV (Ni 2p_1/2) in high-resolution Ni-2p spectrum can be attributed to the Ni²⁺ ions coordinated with O atoms in carboxylic groups, besides the concomitant satellite peaks; the two spin-orbit peaks at 707.7 (Fe 2p_3/2) and 720.5 eV (Fe 2p_1/2) in high-resolution Fe-2p spectrum can be associated with Fe²⁺ ions in Fc units. The presence of carboxylic linkers is further verified by the XPS spectra of C-1s and O-1s cores.

Active species generation via CV activation

The cyclic voltammetry (CV) activation was first conducted to derive highly active and stable redox species adapted to large current densities. Interestingly, the CV curves show the gradual increase in the closed areas with scan cycles, reaching the steady maximum at the 10th cycle (Fig. 2A). The corresponding growth of metal Ni/Fe redox peaks indicates the increased pseudocapacitance by proliferous active sites, which implies the dynamic reconstruction occurring. By estimated from reverse scans, the overpotential at 0.2 A/cm² gradually decreases from 284 mV at the initial scan to 252 mV upon stabilization after 10 cycles (Supplementary Fig. 8), indicating a fast and effective OER activation. In situ electrochemical impedance spectroscopy (EIS) manifests that the charge transport resistance (${\mathrm{R}}_{\mathrm{ct}}$) gradually decreases from 11.02 to 1.57 ohm (Ω) as the CV proceeds (Fig. 2B), which reveals more favorable charge transfer after CV activation. The noticeable evolution of electrochemical properties induced by 10-cycle CV scans suggests a self-optimized reconstruction toward highly reactive species.

Fig. 2. Active-species generation during CV activation.

A CV curves of FcNiOF measured at 50 mV s⁻¹. Note: the geometric surface area ($\mathrm{GSA}$) is 0.25 cm²; series resistance (${\mathrm{R}}_{\mathrm{s}}$) is 2.1 Ω. B In situ EIS spectra with equivalent circuit model of FcNiOF at 1.524 V_RHE. C TEM and HRTEM images of FeNiHOF. D Raman spectra and XPS spectra of (E) Ni 2p and (F) Fe 2p cores, for FcNiOF and FeNiHOF. G Schematic diagram of reconstruction process from FcNiOF to FeNiHOF.

Further, systematic structural characterizations were conducted to disclose the reconstructed active species after CV activation. From the SEM images (Supplementary Fig. 9), the morphology of vertical nanosheet arrays remains unchanged, which is critical for ensuring mechanical adhesion and specific surface area for high-current-density catalysis. The TEM images of nanosheets display that their surfaces appear rough and uneven (Supplementary Fig. 10), distinguishable from the original smooth appearance. The HRTEM image (Fig. 2C) recognizes the emergence of some local small crystal domains, and their lattice fringes can be assigned to the (012) and (015) planes of Fe-Ni layered double hydroxides (FeNiLDH), respectively, corresponding to the periodic atom strength and SAED patterns (Supplementary Fig. 11). The XRD patterns show that the characteristic peaks of FcNiOF phase gradually weaken and then completely disappear along with CV activation (Supplementary Fig. 12), consistent with the quasi-amorphous texture observed by HRTEM. In the Raman spectra, two bands at 345 and 605 cm^-1 assigned to Fc units disappear completely while two new bands appear at 557 and 684 cm^-1 after CV activation, which can be assigned to Ni-O and Fe-O vibrations in FeNiLDH, and moreover the characteristic bands at 1465–1680 cm⁻¹ of carboxyl ligands retain (Fig. 2D). After activation, the Ni-2p XPS spectrum shows the slight bulge on the high binding energy side of characteristic peaks compared to pristine that (Fig. 2E) and reveals a mild increase in Ni³⁺ content based on the deconvolution analysis. Particularly, the Fe-2p XPS spectra before and after activation are totally different (Fig. 2F), where two characteristic peaks assigned to Fe²⁺ in Fc units transform into two broad peaks mixed with Fe²⁺ and Fe³⁺ in FeNiLDH. The FT-IR spectra show that the vibration bands of Fc groups vanish but the characteristic bands of carboxyl groups remain after CV activation (Supplementary Fig. 13). Meanwhile three extra bands emerge at 640, 880, and 2921 cm⁻¹, which correspond to the Ni-O-H and Fe-O-H bending modes as well as O-H stretching mode in LDH structure, respectively, elucidating the formation of FeNiLDH with carboxyl ligands. The XPS spectra on C−1s and O-1s cores further verify the existence of C = O and C-O bonds in carboxyl groups after CV activation (Supplementary Fig. 14). In addition, the EDS mapping images display the homogeneous distributions of Ni, Fe, C and O in an individual nanosheet (Supplementary Fig. 15), with slightly decreased metal contents with respect to the initial FcNiOF. These results together demonstrate that FcNiOF undergoes the self-reconstruction process, which consists of the distortion of Fc units and the formation of carboxyl-linked FeNiLDH motifs as truly active OER species (Fig. 2G). Thus, FcNiOF represents a pre-catalyst and the reconstructed FeNiLDH with organic framework becomes an actual active catalyst, which is referred to as FeNiHOF hereinafter. It needs to be emphasized that such an activation process can also be conducted by the chronopotentiometry (CP) at a current density of 1 A/cm² with decreasing overpotential in the initial 10 h (Supplementary Fig. 16). Thus, the activation process to derive the active species is compatible with the actual OER operation, without needs for additional procedure and cost, and here the study on CV activation separately from OER is only to capture the reconstruction information.

Note that various NiFe compounds have been generally found to transform into active (oxy)hydroxides (NiFeO_xH_y) during CV activation and OER processes^39–43. But, such anodic reconfiguration is usually accompanied by the dissolution of Fe species due to being peroxidized^33,44, resulting in the eventual failure of Ni-Fe synergies. Even the directly synthesized NiFeO_xH_y also generally suffers from activity attenuation during continuous OER process due to uncontrollable Fe dissolution⁴⁵, especially at high current densities. In contrast, there was negligible loss in Fe species in above CV activation over FcNiOF, leaving most Fe in the reconstructed NiFeO_xH_y, which promises the strong robustness toward high-current-density OER. To identify the cause of Fe fixation, we performed the same CV scans for Ni-MOF and Fc-MOF counterparts (Supplementary Fig. 17A), the finally stable products are denoted as NiHOF and FeHOF, respectively. The negligible change of OER activity upon scan cycling is observed for Ni-MOF, although its redox couple of Ni²⁺/Ni³⁺ is clearly visible and gradually larger. This means that more high-valence nickel species are generated via reconstruction but no new species to promote OER activity. In contrast, the redox couple of Fc-MOF is not prominent, but its oxidation peak brings closer to the OER onset without the potential gap (i.e., kinetic delay), suggesting that OER activity is higher at the high-valence iron site than at the nickel site. This comparison indicates that Fc component in FcNiOF adjusts the reconstruction product minimizing the kinetic difference between initial metal oxidation and subsequent OER (Supplementary Fig. 17B), responsible for the enhanced activity. Moreover, carboxyl ligands afford more flexible electronic structure for metal sites to avoid being over-oxidized under high anodic potential, consequently adaptive to the CV and OER with high current densities. In addition, we synthesized FeNiLDH via hydrothermal method to conduct similar CV activation (Supplementary Fig. 18), where no new species or ligands were observed after CV activation. It can be concluded that FeNiLDH did not undergo noticeable reconstruction, and its OER performance was inferior compared to FeNiHOF.

High-current-density OER performance of FeNiHOF

The electrocatalytic OER performance was assessed by quasi-steady LSV scans at 5 mV s^-1 in 1.0 M KOH for FeNiHOF, NiHOF and FeHOF, with commercial RuO₂ and Ni foam as references (Fig. 3A and Supplementary Fig. 19). The bare Ni foam shows low current output, indicative of its neglectable contribution to OER activity. Amongst contrast catalysts, FeNiHOF exhibits the highest OER activity with the minimum onset potential and fastest polarization behavior, and reaches an impressive current density up to 2 A/cm². Specifically, FeNiHOF requires low overpotentials of 273, 280, and 284 mV to deliver current densities of 0.5, 1 and 2 A/cm², respectively, which are much smaller those of FeHOF, NiHOF, and RuO₂ catalysts (Fig. 3B). Moreover, by comparison on overpotentials at high current densities, FeNiHOF outperforms the most of FeNi-based OER catalysts (Supplementary Table 3), showing an outstanding potential for industrial applications. In addition, the bimetallic-component-tuned Fe_xNi_1-xHOF (x = 0.3, 0.5, and 0.7) samples all show higher OER activity than monometallic FeHOF and NiHOF (Supplementary Fig. 20), and whereby the topmost activity is screened in FeNiHOF at x = 0.5. This indicates that the Ni-Fe synergy is responsible for the OER enhancement of FeNiHOF, which can also be suggested by the metal redox features prior to the OER in CV curves. As shown in Fig. 3C, NiHOF shows the maximum Ni²⁺/Ni³⁺ redox peak but the latest OER initiation and subsequent inferior activity, while the Fc addition enhances the difficulty of Ni oxidation but induces the barrier-free transition into OER and superior activity. Note that monometallic FeHOF does not exhibit obvious metal oxidation peak and commendable OER activity, meaning that single Fe sites are also not competent for OER catalysis. Therefore, bimetallic Ni-Fe synergy is suggested in favor of metal oxidation resistance and OER activity enhancement, consistent with previous findings^46–48.

Fig. 3. Electrochemical OER performance.

A LSV curves of FeNiHOF, FeHOF, and NiHOF, with RuO₂ and Ni foam as references. Note: $\mathrm{GSA}$ is 0.25 cm²; ${\mathrm{R}}_{\mathrm{s}}$ is 2.1 Ω; see non-iR corrected profiles in Supplementary Fig. 19. B Overpotentials at different current densities. C CV curves of Fe_xNi_1-xHOF (x = 0, 0.3, 0.5, 0.7, 1.0). Note: $\mathrm{GSA}$ is 0.25 cm²; ${\mathrm{R}}_{\mathrm{s}}$ is 2.1 Ω; see non-iR corrected profiles in Supplementary Fig. 20. D Tafel plots, (E) EIS Nyquist plots, and F $C_{\mathrm{dl}}$ values of FeNiHOF, FeHOF, NiHOF and Ni foam. G Gibbs free energy profiles along four-step OER pathway on FeNiHOF and FeNiLDH. H Schematic diagram of OER mechanism. I Charge density difference for FeNiHOF and FeNiLDH. Blue and yellow contours mark electron depletion and accumulation areas, respectively.

Tafel plots in Fig. 3D show that FeNiHOF has the smallest slope of 34.8 mV dec^-1, compared with FeHOF (58.7 mV dec^-1), NiHOF (95.4 mV dec^-1), and Ni foam (196.7 mV dec^-1), indicating the favorable OER kinetics of FeNiHOF. More importantly, FeNiHOF basically keep such a small Tafel slope unchanged with current density increasing, while other contrast catalysts show the notably increased Tafel slopes especially after current density exceeds 0.5 A/cm². This demonstrates that FeNiHOF can maintain intrinsic OER kinetics, not restricted by mass/charge transfer for high-current-density catalysis. The EIS Nyquist plots (Fig. 3E) show that FeNiHOF has much smaller ${\mathrm{R}}_{\mathrm{ct}}$ than FeHOF, NiHOF and Ni foam, indicating high charge transfer kinetics of FeNiHOF. Through non-Faraday CV curves (Supplementary Fig. 21), the double layer capacitances ($C_{\mathrm{dl}}$) were evaluated in Fig. 3F, where FeNiHOF exhibits the largest $C_{\mathrm{dl}}$ and electrochemical surface area ($\mathrm{ECSA}$) among contrast samples. Dynamic wetting images (Supplementary Fig. 22) manifest the superior hydrophilicity on FeNiHOF surface in comparison with Ni foam surface. Dynamic bubbling images (Supplementary Fig. 23) show numerous tiny O₂ bubbles detached from FeNiHOF at 1 A/cm², in which the most are within the size of 0.2–0.3 mm without adhesion on surface, distinct from the bubbling on Ni foam. These results demonstrate that FeNiHOF has the favorable mass/charge transfer kinetics for high-current-density catalysis, benefiting from the ordered porous array architecture. To determine the intrinsic activity of FeNiHOF, the LSV curves were normalized by the $\mathrm{ECSA}$ and active-site number (Supplementary Fig. 24), respectively. FeNiHOF exhibits higher $\mathrm{ECAS}$-normalized activity and turnover frequency ($\mathrm{TOF}$) than NiHOF and FeHOF, further verifying Ni-Fe synergistic contribution.

To understand the mechanism behind OER kinetics, we constructed and optimized the atomic structures of FeNiHOF and FeNiLDH systems (Supplementary Fig. 25) for theoretical calculations. The spin-resolved projected density of states in Supplementary Fig. 26 show that both FeNiHOF and FeNiLDH have the conductor-like property because the electron states pass through the Fermi levels. The Gibbs free energy profiles along four-step elementary pathways over two systems were calculated using the density function theory (DFT)^49,50, with the relaxed structures of different intermediates (Supplementary Fig. 27). The electric potential and solvation effects on free energy calculations were analyzed in Supplementary Fig. 28. The free energy steps shown in Fig. 3G indicate that the rate-determining step (RDS) is the transition of HOO* from O* intermediates for both FeNiHOF and FeNiLDH with different overpotentials (${\mathrm{U}}_{\mathrm{RDS}}$) of 0.56 and 1.13 V, respectively. This suggests that carboxylated metal sites reduce the energy barrier at the RDS (Supplementary Fig. 29), in which the O-O formation by absorbing one OH^- at O* intermediate is facilitated with an electron transfer, manifested as third-step acceleration in the OER process (Fig. 3H). The identification of the RDS also accords with the measured Tafel slope as low as 34.8 mV dec^{-1 51–53}. We further calculated the charge density difference when metal Fe site adsorbing HOO* intermediate in the highest oxidation states (Fig. 3I). The electronic interaction between the carboxyl ligand and the active metal is demonstrated in FeNiHOF, which mediates the electronic structure around the metal sites responsible for reducing the endothermic energy barrier of the RDS. Moreover, the electronic exchange in carboxyl-linked metal motifs can enhance the applied potential tolerability beneficial to the stability of metal sites. Therefore, the carboxyl conjugation is disclosed to play an important role in optimizing the electronic structure to improve OER kinetics and self-peroxide resistance.

The long-term stability of FeNiHOF at high current densities was studied to examine the practical feasibility toward industrial applications. As shown in Fig. 4A, FeNiHOF operates steadily water oxidation without any decay during long-term CP test at 1 A/cm² for 1000 h. There is no increase in overpotential but instead a slight decrease, probably due to a gain from temperature fluctuation. Real-time online inductively coupled plasma mass spectrometry (ICP) measurements during 100 h CP process at 0.5 A/cm² confirm the stability of active metal sites in FeNiHOF with considerable dissolution resistance (Fig. 4B, C). The initial Fe loss may be due to a small number of Fe atoms that are not bound by carboxyl groups. In contrast, the same CP test of typical FeNiLDH exhibits an obvious decay with 38% increase in overpotential at 0.5 A/cm² after 50 h operation (Fig. 4D), meanwhile accompanied by the leakage of metal sites especially Fe ions into the electrolyte (Fig. 4E). This indicates that the carboxyl conjugation in FeNiHOF can provide high charge flexibility to stabilize Fe sites with peroxidation resistance. Non-${\mathrm{iR}}_{\mathrm{s}}$ corrected voltammograms for Fig. 4A, B, D are presented in Supplementary Fig. 30. According to the online collection of gas products (Supplementary Fig. 31), the Faraday efficiency ($\mathrm{FE}$) was evaluated by comparing the number of electrons involved in the OER and that transmitted in the circuit (Fig. 4F) to be close to 100%, which means the specific selectivity of FeNiHOF to the OER without side reactions. In addition, the postmortem characterizations of FeNiHOF after durability test reveals that the nanoarray morphology, metal composition and organic linkers almost remain unchanged (Supplementary Fig. 32), further confirming the robust stability of FeNiHOF catalyst. It is worth noting that the catalytic stability at industrial-level high current densities is essential for practical applications, which is rarely reported for MOF-based or NiFe-based catalysts (Supplementary Table 4). Noticeably, the catalytic stability of FeNiHOF is superior to most reported OER catalysts in terms of the current density and duration of continuous operations.

Fig. 4. High-current-density durability tests.

A Long-term durability test for FeNiHOF in CP model at 1 A/cm² for 1000 h. Note: $\mathrm{GSA}$ is 0.25 cm²; ${\mathrm{R}}_{\mathrm{s}}$ is 2.1 Ω; see non-iR corrected profiles in Supplementary Fig. 30A. B CP curve at 0.5 A/cm² and (C) Fe/Ni ionic contents in electrolyte for FeNiHOF. Note: $\mathrm{GSA}$ is 0.25 cm²; ${\mathrm{R}}_{\mathrm{s}}$ is 2.1 Ω; see non-iR corrected profiles in Supplementary Fig. 30B. D CP curve at 0.5 A/cm² and (E) Fe/Ni ionic contents in electrolyte for FeNiLDH. Note: $\mathrm{GSA}$ is 0.25 cm²; ${\mathrm{R}}_{\mathrm{s}}$ is 2.1 Ω; see non-iR corrected profiles in Supplementary Fig. 30C.

Practical application in alkaline water electrolyzer

The hydrothermal method is amenable to scalable synthesis of catalyst electrode at low cost. We could coil and place the flexible Ni foam into a high-volume Teflon-lined stainless autoclave for the scale-up fabrication of FcNiOF electrode. A piece of as-synthesized FcNiOF electrode (9 cm × 10 cm) appears dark yellow color over the whole area, suggesting a uniform growth of FcNiOF nanosheets (Supplementary Fig. 33). Note that the reconstruction from FcNiOF into FeNiHOF would spontaneously take place under OER conditions and require no additional procedures. Based on cheap raw materials and simple preparation process, the price of FeNiHOF electrode is assessed to be only ~147 US$ m^-2 (Supplementary Table 5). Combined with its high performance in water oxidation, we believe that Ni foam-supported FeNiHOF as an OER anode has high cost-effectiveness for large-scale application in alkaline water electrolyzers (AWEs). In addition, the flexible Ni foam loaded with the catalyst was tested for mechanical strength through 100 times of bending tests (Supplementary Fig. 34), where no detectable mass loss indicates the reliable adhesion of in-situ grown catalyst on Ni foam, which is critical for gas-bubbling catalysis on electrode.

For the current AWE in industry, RANEY nickel (R-Ni) loaded onto nickel wire mesh (NWM) by thermal spray or coating with binders is commonly-used catalyst on the cathode for HER, and the bare NWM is directly used as the anode for OER. In contrast, orderly nanostructured catalysts self-supported on the conductive substrates without binders have been projected to be the most promising promoters for the innovation of AWEs¹³, but their activity-stability performance under industrial-grade water electrolysis with high current densities (≥0.5 A/cm²) remains uncertain towards practical applications^17,18. Herein, we assembled an AWE of FeNiHOF&R-Ni using FeNiHOF and commercial R-Ni as the anode and cathode, respectively, with a diaphragm of polyphenylene sulphide (PPS), to examine its feasibility in actual water electrolysis (Fig. 5A, B). The FeNiHOF&R-Ni electrolyzer requires a cell voltage as low as 1.87 V to achieve a high current density of 1 A/cm² in 1 M KOH at room temperature, far outperforming an electrolyzer with commercial NWM&R-Ni electrodes (Fig. 5C). Moreover, the cell voltage of the FeNiHOF&R-Ni to deliver a current density of 1 A/cm² decreases further to 1.81 V when operated in 6 M KOH, while the commercial NWM&R-Ni electrolyzer needs high cell voltages of 2.04–2.26 V to reach industrial-level current densities of 0.5–1 A/cm² under the same conditions. We further conducted the long-term CP test for the FeNiHOF&R-Ni electrolyzer operated at 1 A/cm² in 6 M KOH at room temperature with 20 times of ON/OFF switch (Fig. 5D), demonstrating the durability over 500 h with negligible charge in cell voltage. It is worth noting that the electrolyzer at the moment of shutdown usually generates a reverse current, which would deteriorate the durability of catalysts. But the stability study with ON/OFF switch is rarely involved for new-type catalysts so far^49,50, which should be concerned when designing catalysts towards industrial applications. The excellent performance suggests the potential of industrialized AWE with cost-effective FeNiHOF anode. The power consumption ($\mathrm{W}$) and energy efficiency ($\mathrm{ETH}$) of FeNiHOF&R-Ni are calculated to be ~4.23 kW h Nm^-3 and ~83.6% at 0.5 A/cm² at 1.77 V (Fig. 5E), superior to those of commercial NWM&R-Ni (4.88 kW h Nm^-3; 72.5%). The H₂ yield rate ($\mathrm{R}$) is around 3.18 Nm³ h⁻¹ m^-2 for the FeNiHOF&R-Ni at 1.8 V, higher than that for the NWM&R-Ni (0.45 Nm³ h^-1 m^-2). Moreover, for the FeNiHOF&R-Ni electrolyzer, the price of producing H₂ is estimated to be approximately US$ 1.01 per gallon of gasoline equivalent (GGE), which is much less than the 2026 technical goal (US$ 2.00) from the Department of Energy in United States⁵⁴. Finally, we comprehensively summarize the performance in terms of cell voltage, stability time, $\mathrm{W}$, $\mathrm{ETH}$, and $\mathrm{R}$ in Supplementary Table 6, where the AWE enabled by FeNiHOF catalyst stands out from the state-of-the-art AWEs reported in the literatures. Note that only the anode is replaced by FeNiHOF in our electrolyzer and the cathode is still the commercial standard R-Ni in order to highlight the effect of anode substitution, different from most advanced electrolyzers reported with dual electrode substitution (Supplementary Table 6). If meanwhile employing high-performance alternatives to R-Ni, the AWEs would have further room for performance improvement.

Fig. 5. Practical application in alkaline water electrolyzer.

A Stack structure and (B) photograph of FeNiHOF&R-Ni electrolyzer. C LSV curves of FeNiHOF&R-Ni and NWM&R-Ni electrolyzers with flowing electrolytes of 1 M and 6 M KOH. D Long-term CP curve of FeNiHOF&R-Ni electrolyzer at 1 A/cm² in 6 M KOH. E Energy efficiency contrast between FeNiHOF&R-Ni and commercial NWM&R-Ni electrolyzers.

Discussion

In summary, FeNiHOF nanosheet arrays were fabricated and demonstrated as a highly active and durable catalyst for industrial-grade OER in alkaline media. The structural characteristics of high specific surface and porosity, self-optimized active species, and charge-regulated organic ligands promote the charge/mass transfer dynamics, intrinsic OER activity, and metal segregation resistance, respectively, together contributing to high-current-density OER performance. As a result, FeNiHOF exhibits high OER activity with low overpotentials of 280 and 284 mV at 1 and 2 A/cm², respectively, as well as long-term stability for over 1000 h at 1 A/cm², superior to most OER catalysts reported previously. An actual AWE with FeNiHOF electrode demonstrates the industrial activity-stability balance and enhanced energy efficiency, confirming the application prospect of this electrode.

Methods

Chemicals

1,1’-ferrocene dicarboxylate (FcDA, 98%, Shanghai Maclin Biochemical Technology Co., Ltd), N,N dimethylformamide (DMF, 99.5%, Shanghai HUSHI Laboratory Equipment Co., Ltd), nickelous nitrate hexahydrate (Ni(NO₃)₂·6H₂O, 98%, Shanghai HUSHI Laboratory Equipment Co., Ltd), ferric nitrate nonahydrate (Fe(NO₃)₃· 9H₂O, 99.99%, Shanghai Aladdin Biochemical Technology Co., Ltd), carbamide (CO(NH₂)₂, 99.5%, Shanghai Aladdin Biochemical Technology Co., Ltd), potassium hydroxide (KOH, 85%, Shanghai HUSHI Laboratory Equipment Co., Ltd) were used without further purification. Ni foam (1.5 mm thick, Kunshan Luchuang Electronic Technology Co., Ltd), Hg/HgO electrode (1 M KOH, Shanghai Chuxi Industrial Co., Ltd), graphite rod electrode (6.0 mm diameter, Shanghai Chuxi Industrial Co., Ltd), NWM and R-Ni electrodes (0.5 mm thick, Hebei Ruiyun Silk Mesh Technology Co., Ltd), PPS membrane (0.85 mm thick, Japan Toray) were used as received. Deionized water made in our laboratory was used to prepare all aqueous solutions.

Synthesis of FcNiOF/Ni foam

FcNiOF was synthesized on Ni foam substrate by a one-step hydrothermal method. First, a piece of Ni foam (1.5 mm thick) was cut into several rectangle pieces of 2 cm × 2 cm and ultrasonically washed using hydrochloric acid, acetone, deionized water and ethanol each for 10 min in turn. Afterwards, 0.5 mmol FcDA and 0.5 mmol Ni(NO₃)₂·6H₂O were dissolved in 8 ml DMF and 4 ml deionized water, respectively. The two solutions were then mixed and transferred together with Ni foam into a 25 ml Teflon-lined autoclave and stored at 125 °C for 12 h. After that, the sample was extracted and rinsed with ethanol and deionized water several times, and then dried at 60 °C for 12 h in a vacuum oven to obtain FcNiOF/Ni foam. The catalyst loading on Ni foam was estimated to be about 0.9 mg cm⁻² by mass change measured by a high-precision electronic scale. The Fc_xNi_1-xOF samples were synthesized under the same conditions except that different ratios of Ni/Fc (x = 0, 0.3, 0.5, 0.7, 1.0) were used, labeled as Ni-MOF (x = 0), Fc_0.3Ni_0.7OF (x = 0.3), Fc_0.5Ni_0.5OF (x = 0.5), Fc_0.7Ni_0.3OF (x = 0.7) and Fc-MOF (x = 1.0), respectively.

Synthesis of FeNiLDH/Ni foam

Briefly, in a polytetrafluoroethylene lined autoclave, 0.5 mmol Fe(NO₃)₃·9H₂O, 4 mmol CO(NH₂)₂ and 0.5 mmol Ni(NO₃)₂·6H₂O were together dissolved in 35 mL deionized water and vigorously stirred at room temperature to form a homogeneous solution. A pre-treated piece of Ni foam was immersed vertically in the solution, ultrasonically treated and placed in an oven at 125 °C for 12 h. After that, the sample was taken and rinsed with deionized water and ethanol several times to obtain FeNiLDH/Ni foam after vacuum drying at 60 °C for 12 h.

Synthesis of RuO₂-loaded electrode

The RuO₂ electrode was prepared by adding 4.8 mg of commercial RuO₂ into dispersed solution containing 30 μL Nafion, 200 μL deionized water and 300 μL ethanol. The mixed sol was ultrasonically treated for 30 min, then coated on a pretreated Ni foam substrate and dried in air overnight.

Material characterization

XRD was conducted on a Shimadzu X-7000 diffractometer with a Cu Kα (λ = 1.54 Å) monochromatic beam at a scanning rate of 2° min⁻¹. SEM at 5 kV acceleration voltage was taken on a Hitachi S-4800II instrument. Atomic force microscopy (AFM) was taken on a SPM-9700HT instrument. TEM at 100 kV acceleration voltage was performed on the Tecnai F30 instrument with an energy dispersive X-ray module. Fourier transform infrared spectroscopy (FTIR) in the spectral range 400–4000 cm^-1 was recorded on the Varian 670-infrared spectrometer. Raman spectroscopy was conducted on an Oceanhood XS11639 spectrometer with a 532 nm laser. X-ray photoelectron spectroscopy (XPS) was taken on an Escalab 250Xi spectrometer under Al Kα (1486.60 eV) irradiation. Inductively coupled plasma mass spectrometry (ICP-MS) was measured using an Elan DRC-e (PerkinElmer). The pH value of the electrolyte was measured by a PHS-3C pH meter (Shanghai Yidian Scientific Instruments Co., Ltd).

Electrochemical tests

Electrochemical tests were performed on a CHI-1140D electrochemical workstation with a three-electrode cell, using the catalyst self-supported Ni foam as the working electrode, Hg/HgO and graphite rod as the reference and counter electrodes, respectively. The electrolyte is 1.0 M KOH solution at a constant temperature of 25 ± 2 °C, which was prepared by adding 33 g of KOH (85%) into 500 ml of deionized water with continuous stirring for ample dissolution. The effective geometric surface areas (GSA) of all working electrodes were controlled the same as 0.25 cm². OER performance was measured at a scanning rate of 5 mV/s by CV scanning. The durability test was carried out using the long-term CP model. The ${\mathrm{C}}_{\mathrm{dl}}$ values were calculated by non-Faraday CV curves at different scanning rates in the potential range of 1.044–1.144 V vs. RHE. The EIS was conducted in the frequency range of 0.01–100 kHz.

Potential calibration

Unless otherwise stated, the reported potentials ($E_{\mathrm{reported}}$) for three-electrode cells have been converted from measured potentials ($E_{\mathrm{measured}}$) to RHE and calibrated with 93% $i{R}_{\mathrm{s}}$ compensation by the following equations:

$$E_{\mathrm{reported}}=E_{\mathrm{measured}}-i{R}_{\mathrm{s}}$$ 1

$$E_{\mathrm{RHE}}=E_{\mathrm{Hg}/\mathrm{HgO}}^0+E_{\mathrm{Hg}/\mathrm{HgO}}+0.059\mathrm{pH}$$ 2

where $i$ is the measured current, $R_{\mathrm{s}}$ is the series resistance that is obtained directly by the electrochemical workstation, $E_{\mathrm{Hg}/\mathrm{HgO}}^0$ is the standard potential (0.098 V vs. RHE) of Hg/HgO reference electrode at 25 °C, and $\mathrm{pH}$ value of 1 M KOH is measured to be 13.98 by a pH meter. The relevant parameters used in calculations were obtained by measurement once.

OER activity evaluation

The Tafel slopes were obtained by Tafel plots from the linear portion fitting overpotential $\left(\mathrm{\eta }\right)$ against $\log ∣j∣$ by the following equation:

$$\eta =b\log \mid j\mid +a$$ 3

The $C_{\mathrm{dl}}$ were calculated by plotting differences in charging currents ($△j$) against scanning rates ($v$, 20, 40, 60, 80, 100, 120 mV s⁻¹) in non-faradic process based on the following equation:

$$C_{\mathrm{dl}}=△j/v$$ 4

The $\mathrm{ECSA}$ was evaluated by $C_{\mathrm{dl}}$ and standard reference ($C_{\mathrm{s}}=0.04\mathrm{mF}$ cm^-2) as follows:

$$\mathrm{ECSA}=C_{\mathrm{dl}}/C_{\mathrm{s}}$$ 5

The $\mathrm{TOF}$ values were determined via Eqs. 6 and 7:

$$\mathrm{TOF}=\frac{j\times A}{4\times F\times n}$$ 6

$$n=\frac{C_{\mathrm{dl}}\times △V\times A}{2F}$$ 7

where $j$ and $A$ are the geometric current density (A/cm²) and the electrode area (cm²), $F$ is the Faraday constant (C/mol), $n$ is the active-site mole number (mol), $△V$ is the potential range of CV scans (V), and the factors of 4 and 1/2 are one OER-involved electron number and charge-discharge times in one CV scan. The relevant parameters used in calculations were obtained by measurement once.

Fabrication of AWE electrolyzer

The AWE electrolyzer was assembled with an anode (9 cm²) and a cathode (9 cm²), separated by the PPS diaphragm, with sealing washers, bipolar plates, and pipelines. FeNiHOF catalyst grown on Ni foam with the loading mass of ~0.9 mg cm^-2 was directly used as a monolith anode and commercial R-Ni coated on Ni foam as a cathode to construct the FeNiHOF&R-Ni electrolyzer. For comparison, commercial NWM and R-Ni electrodes were used as an anode and a cathode to construct the NWM&R-Ni electrolyzer. The performance of the AWE electrolyzers was studied with a LW3030KD direct-current (DC) power supply in the flow-type 1 M KOH electrolyte at 25 ± 2 °C and 6 M KOH electrolyte at 25 ± 2 °C, respectively. The polarization curve was measured using LSV mode at a scan rate of 10 mV s⁻¹. The stability test was performed by CP mode at a current density of 1 A/cm^-2 for 500 h at 25 ± 2 °C. 20 times of ON/OFF switches in the stability test were implemented directly by turning the power on and off. The temperature of electrolyte was controlled by thermostatic chamber.

FE evaluation

Real-time monitoring of gas evolution over a sealed electrolyzer operated at a current density of 0.5 A/cm² was performed with gas chromatograph (Labsolar-IIIAG, Perfectlight Beijing). To evaluate the FE, the reacting electron number $\left(N_{\mathrm{re}}\right)$ and conducting electron number $\left(N_{\mathrm{ce}}\right)$ were calculated and compared based on the following equations:

$$N_{\mathrm{re}}=4M_0{N}_A$$ 8

$$N_{\mathrm{ce}}=jst/e$$ 9

$$\mathrm{FE}=\frac{N_{\mathrm{re}}}{N_{\mathrm{ce}}}\times 100\%$$ 10

where $M_0$ is the mole amount of evolved O₂, $N_A$ is the avogadro number, $e$ is the electron charge, as well as $j$, $s$, and $t$ is the current density, effective electrode area, and the time, respectively. The relevant parameters used in calculations were obtained by measurement once.

Energy efficiency calculations

When the typical current density in AWEs is 0.5 A/cm², the $W$ was calculated using Eq. 11 with unit electric quantity$\left(Q\right)$ and cell voltage $\left(V\right)$. In the standard state, the volume of hydrogen per mole is 22.4$\times$10⁻³ Nm³, so $Q$ was calculated with Eq. 12.

$$W\mathit{=}Q\times V/1000\mathrm{kW}\cdot \mathrm{h\; N}{\mathrm{m}}^{-3}$$ 11

$$Q=\frac{2N_{\mathrm{A}}e}{3600\times 22.4\times {10}^{-3}}=2390\mathrm{A}\cdot \mathrm{h\; N}{\mathrm{m}}^{-3}$$ 12

The standard thermoneutral voltage $\left(V_0=1.48\mathrm{V}\right)$ is used to calculate $\mathrm{ETH}$ by Eq. 13. At this voltage, there is no heat loss in the water electrolysis reaction.

$$\mathrm{ETH}=\frac{1.48\times Q}{W\times 1000}\times 100\%$$ 13

The H₂ yield rate $\left(R\right)$ was calculated from Eq. 14.

$$R=\frac{j}{Q}\mathrm{N}{\mathrm{m}}^3{\mathrm{h}}^{-1}{\mathrm{m}}^{-2}$$ 14

The price of H₂ per GGE ($P_{{\mathrm{H}}_2}$, $ GGE⁻¹) was estimated using Eq. 15.

$$P_{{\mathrm{H}}_2}=\frac{m\times W\times \mathrm{CT}}{\rho }$$ 15

where $m$ is the H₂ mass per GGE (0.997 kg GGE⁻¹), $W$ is the power consumption ($\mathrm{kW}\cdot \mathrm{h\; N}{\mathrm{m}}^{-3}$), $\rho$ is the H₂ density (0.083 kg Nm^-3) at normal atmosphere, and $\mathrm{CT}$ is the commercial tariff of about 0.02 $ (kW h)⁻¹, respectively. The relevant parameters used in calculations were obtained by measurement once.

DFT calculations

Spin-polarized density functional theory (DFT) were performed using the Vienna ab initio simulation package (VASP). The electron-ion interaction was described by projector-augmented wave (PAW) method, and the Perdew-Burke-Ernzerhof (PBE) within generalized gradient approximation (GGA) was employed to characterize the exchange-correlation energy. An energy cutoff of 400 eV was adopted for the plane-wave basis. The Brillouin zone were sampled by 7 × 7 × 1 and 11 × 11 × 1 k-points for structural optimizations and electronic calculations, respectively. The convergence thresholds were reached at 0.01 eV/Å and 1.0 × 10⁻⁵ eV for force and energy, respectively. The activation energy barriers were determined using the climbing-image nudged elastic band (CI-NEB) method⁵⁵. The DFT + U method⁵⁶ was used to describe Ni and Fe with U values of 3.8 and 4.3. The vacuum space of 20 Å was added to avoid periodic effects. The free energy of each elementary steps is defined by the equation as⁵⁷:

$$△\mathrm{G}=△\mathrm{E}+△\mathrm{ZPE}-\mathrm{T}△\mathrm{S}$$ 16

where $△\mathrm{E}$, $△\mathrm{ZPE}$, and $△\mathrm{S}$ are the changes in the energy, zero-point energy, and entropy contribution of the geometry, respectively. At 298.15 K, $△\mathrm{ZPE}$ and $△\mathrm{S}$ values for H₂O, H₂ and O₂ molecules are obtained from the NIST-JANAF thermodynamics table and those for OH*, O* and HOO* intermediates are obtained by calculating their vibrational frequencies in adsorbed states. The zero-point energies ($\mathrm{ZPE}$) and entropy corrections ($\mathrm{TS}$) for all involved species are shown in Supplementary Table 8. The calculation details with reaction equations and energy values are shown in Supplementary Fig. 28 and Table 7.

Author contributions

X.X. conceived the idea. X.X. and J.H. supervised the project. Y.C. and J.L. performed the experiments, collected and analyzed the data. Q.L., Y.L., and J.P. performed DFT calculations and analyzed the results. Y.C. and X.X. co-wrote the paper. All authors discussed the results and commented on the paper.

Peer review

Peer review information

Nature Communications thanks Rodrigo García-Muelas, Wenshuo Xu, and the other, anonymous, reviewer(s) for their contribution to the peer review of this work. A peer review file is available.

Data availability

The data generated in this study are provided in the Source data file 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/s41467-024-51521-4.