Substrate-adaptive sacrificial corrosion strategy enables 700 mV oxygen evolution window for enhanced seawater electrolysis

Abstract

Seawater electrolysis for hydrogen production offers a sustainable solution to the energy crisis and freshwater scarcity. The presence of chloride ions triggers competitive chlorine oxidation at the anode, which rivals the oxygen evolution reaction and leads to pronounced reductions in overall efficiency and long-term stability. This challenge underscores the urgent need for highly efficient, durable, and scalable anode materials to accelerate the transition of seawater electrolysis from laboratory research to industrial application. In this work, we introduce a substrate-adaptive sacrificial corrosion strategy that enables the universal growth of highly active corrosion products on a wide range of conductive substrates, notably including stainless steel. The optimized electrode achieves an overpotential of 182 mV at 10 mA/cm², and sustains 500 mA/cm² for 1000 h in 10 m KOH seawater. To fill the gap in describing anode selectivity, an oxygen evolution window is proposed and measured. The measured value of 700 mV, far exceeding the thermodynamic limit of 480 mV, provides compelling experimental evidence that kinetic regulation can break the thermodynamic framework. This work provides a scalable synthesis platform and mechanistic insights for designing industrial seawater electrolyzers with extended durability and selectivity.

Seawater electrolysis for green hydrogen is hindered by corrosive chloride ions. Here, the authors report a substrate-adaptive sacrificial corrosion strategy for creating highly active anodes that resist seawater corrosion and break thermodynamic selectivity limits.

Introduction

Owing to its high energy density and environmental friendliness, hydrogen is regarded as an optimal energy carrier for the storage of renewable energies like solar and wind¹. Water electrolysis is a highly efficient technique to convert renewable energy into hydrogen, providing high-purity hydrogen and effectively mitigating the intermittency and fluctuations of renewable energy². However, current water electrolysis systems for hydrogen production mostly rely on high-purity freshwater. Given the global freshwater scarcity and the advancement of offshore wind and solar power, developing seawater-based electrolysis technologies aligns better with future energy demands³. Unfortunately, seawater contains a significant amount of impurity ions (e.g., Ca²⁺, Mg²⁺, and Cl⁻) and suspended particles, making its electrolysis significantly more complex than freshwater electrolysis and imposing higher demands on catalytic materials^4,5. Among these, the effect of Cl⁻ ions on the water electrolysis reaction is notably significant. At the anode, Cl⁻ ions exert a dual threat: their small size enables deep penetration and strong corrosive attack⁶, while their oxidation competes directly with the oxygen evolution reaction (OER), lowering the overall efficiency of water splitting. The concomitant generation of potent oxidants, including ClO⁻ species and Cl₂ gas, further accelerates electrode degradation and the breakdown of electrolyzer components⁷. These challenges highlight the urgent need for catalytic materials with superior OER selectivity relative to the chlorine evolution reaction (CER) to ensure both high activity and long-term stability in seawater electrolysis.

At present, four mainstream strategies have been proposed to improve the anodic selectivity in seawater electrolysis⁸. (i) Thermodynamic control: taking advantage of the theoretical 480 mV potential window between oxygen evolution and Cl⁻ oxidation at pH > 7.5⁹, catalysts are designed to deliver high OER activity within this overpotential range while maintaining operation strictly inside this window. (ii) Electrostatic exclusion: functional layers incorporating anions such as CO₃²⁻, SO₄²⁻, or PO₄³⁻ are employed to repel Cl⁻ and thereby suppress surface CER¹⁰. (iii) Lewis acid-base regulation: tuning surface metal sites to behave as strong Lewis acids weakens their interaction with the weak Lewis base Cl⁻, enriching the electrode surface with strong Lewis base OH⁻ species¹¹. (iv) In situ consumption of Cl⁻ and its oxidative products: coupling auxiliary reactions to convert these species into value-added chemicals such as α,α-dichloroketones, HCl, or chlorinated polymers¹². Despite encouraging progress, current investigations remain largely qualitative in evaluating OER selectivity across different catalytic structures¹³. The lack of a robust quantitative descriptor hampers direct benchmarking, precludes the construction of a reliable structure-selectivity database, and complicates data-driven catalyst design. Furthermore, additional strategies that effectively suppress CER remain urgently needed.

Considering the future industrialization of seawater electrolysis, the development of scalable synthesis methods for seawater-compatible electrode materials is essential. In 2018, Zou’s group first introduced the concept of a corrosion-based synthesis and successfully fabricated highly active NiFe-LDH structures using this approach¹⁴. Specifically, the method exploits the spontaneous corrosion of Fe species, where optimization of the corrosion medium enables the transformation of corrosion products from Fe₂O₃ to NiFe-LDH with superior OER catalytic activity. The NiFe-LDH produced in this manner not only outperforms counterparts prepared by hydrothermal or electrodeposition routes but can also be synthesized on a large scale under simple, controllable conditions, offering strong prospects for industrial deployment. Subsequent studies extended this strategy by tailoring the corrosion solution to access a variety of high-performance catalysts, including CoFe(OH)_x¹⁵, (Ni, Fe)₃S₂¹⁶, Ni_0.75Fe_2.25O₄¹⁷, and Ni(Fe)OOH/Ni(Fe)S_x¹⁸. However, these corrosion-derived electrodes have so far been restricted to Fe-, Ni-, or NiFe-based substrates¹⁹. This limitation arises from the internal requirements of the corrosion method: the substrate itself must participate in the corrosion reaction, which (i) constrains the structure of the corrosion products to that of the substrate and (ii) prevents the growth of active phases on substrates that are inherently corrosion-resistant. Such constraints not only narrow the applicability of the corrosion strategy but also hinder its use in seawater electrolysis, where substrate stability is critical.

In this study, based on our notable sacrificial corrosion strategy, a highly active NiFe-LDH corrosion product, breaking the internal limitation of the conventional corrosion engineering, is successfully grown on a series of conductive substrates. By combining this strategy with stainless steel substrates, we develop advanced anode materials resistant to seawater corrosion. The optimized electrode exhibits an overpotential of only 182 mV to achieve a current density of 10 mA/cm², maintains stable operation for over 1000 h at 500 mA cm⁻² in 10 m KOH seawater electrolyte. Remarkably, when stored in the coastal climate of Hainan, China, the electrode retains its initial catalytic activity for 3.5 years, underscoring its promising environmental stability. Through systematic experimental analysis, we identify an expanded OER-only potential window of 700 mV for our corroded layer electrode, confirming our earlier kinetic hypothesis that the thermodynamic 480 mV window can be broadened, further offering a practical metric to quantify OER selectivity. Finally, comprehensive calculations hypothesize a dual-site cooperative strategy capable of suppressing CER: by introducing auxiliary sites within the catalytic phase, the highly active oxygen-evolving centers could be shielded from Cl* intermediates, thereby endowing the overall material with pronounced selectivity toward OER. This study not only offers a valuable direction to explore the OER selectivity of materials, but also presents a perspective on expanding the application of specific materials through the corrosion method.

Results

Synthesis strategy and structural characterization of ACCE

Our sacrificial corrosion strategy involves the initial deposition of a corrosion layer and the subsequent growth of catalytically active corrosion products. The composition of the corrosion layer should be a metal layer prone to corrosion, which serves as the soil for growing the corrosion product layer through specific corrosion processes (Fig. 1a). The introduction of the corrosion layer substantially relaxes the stringent requirements on the composition and type of the substrate for all corrosion-based syntheses, theoretically enabling growth on any desired substrates^14,19. To validate the efficacy and scalability of the sacrificial corrosion strategy, we integrate it with our Fe-derived corrosion method. Specifically, an Fe layer serving as the corrosion layer is first electrodeposited onto the conductive substrate by applying a 0.2 A current for 2 min in a 0.1 m FeCl₂ solution (please see the “Methods” section for details). Due to the Fe layer’s susceptibility to oxygen corrosion, the potential difference between the Fe/Fe²⁺ and O₂/OH⁻ couples can drive the rapid growth of NiFe-LDH nanoarrays on the corrosion layer’s surface under corrosion solution (Fig. 1b). Five conductive substrates with different compositions (such as Ni foam, Fe foam, Cu foam, Carbon cloth, and Ti mesh) are selected for verification. The scanning electron microscopy (SEM) images (Fig. 1c–g) combined with Raman spectra (Fig. S2a) confirm that the presence of the corrosion layer enables uniform NiFe-LDH growth across various substrates. The main reactions involved can be summarized as follows:

$$\mathrm{F}{\mathrm{e}}^{2+}+2{\mathrm{e}}^{-}\to \mathrm{Fe}\left(\mathrm{s}\right)$$ 1

$${\mathrm{O}}_2+2{\mathrm{H}}_2\mathrm{O}+4{\mathrm{e}}^{-}\to 4\mathrm{O}{\mathrm{H}}^{-}$$ 2

$$\mathrm{F}\mathrm{e}-2{\mathrm{e}}^{-}\to \mathrm{F}{\mathrm{e}}^{2+}$$ 3

$$\mathrm{F}{\mathrm{e}}^{2+}-{\mathrm{e}}^{-}\to \mathrm{F}{\mathrm{e}}^{3+}$$ 4

$$\mathrm{F}{\mathrm{e}}^{3+}+{\mathrm{Ni}}^{2+}+\mathrm{O}{\mathrm{H}}^{-}+{\mathrm{S}{\mathrm{O}}_4}^{2-}\stackrel{\mathrm{\Delta }}{\to }\mathrm{NiFe}\text{-}\mathrm{LDH}$$ 5

Fig. 1. Overview and universality of the sacrificial corrosion strategy.

a Schematics and b involved mechanism of the sacrificial corrosion strategy. SEM images of corrosion product grown on c carbon cloth, d Cu foam, e Ti mesh, f Ni foam, and g Fe foam based on a sacrificial corrosion strategy.

The corresponding Linear sweep voltammetry (LSV) result (Fig. S2b–f) further supports the successful growth of highly OER-active NiFe-LDH on these substrates, where the observed variations in oxidation peaks possibly result from differences in substrate structure and the contribution of minor corrosion products.

Next, the applicability of the sacrificial corrosion method to stainless steel-based substrates is investigated. The optimized alloy structure of stainless steel endows it with outstanding corrosion resistance, making conventional corrosion-based growth of modified products particularly difficult. Nevertheless, we attempt to fabricate highly OER-active NiFe-LDH on stainless steel sheet (SS) using the sacrificial corrosion strategy (Fig. 2a). The corresponding SEM images (Fig. S3) show that the Fe corrosion layer is successfully grown on SS with a microstructure of fern-like nanoclusters, denoted as Fe/SS, and the corresponding Fe layer thickness is determined to be around 1 μm. Only 3 h are required for our modified corrosion method (optimization procedures are presented in the “Methods” section, Fig. S4), further improving overall synthesis efficiency. Figure 2b shows that vertical nanoarrays uniformly cover the surface of the SS, and the thickness of nanosheet is evaluated to be ca. 8 nm based on transmission electron microscopy (TEM) image (Fig. 2c). The high-resolution TEM (HRTEM) image in Fig. 2d depicts that a set of lattice fringes with d = 0.198 nm can be assigned to the (018) planes of NiFe-LDH¹⁴, together with the powder X-ray diffraction (XRD) result (Fig. S5), proving that the crystal structure of nanosheet arrays can be attributed to rhombohedral NiFe-LDH structure²⁰. The cross-sectional SEM image (Fig. S6a) shows that the thickness of the as-grown film is around 3.8 μm, which is thicker than that of the Fe corrosion layer. Additionally, the corresponding element mappings (Fig. 2e–h, Fig. S6b–e) reveal that the Fe, Ni, and O species are distributed uniformly throughout the film, and the molar ratio of Ni:Fe in NiFe-LDH nanosheet is evaluated to be 3:1 based on energy dispersive X-ray spectroscopy (EDS) results. Thus, compared to the traditional corrosion method that cannot grow any corrosion products on a stainless steel substrate (Fig. S7), corrosion-engineering-derived NiFe-LDH nanosheet arrays have been successfully grown on the anticorrosive stainless-steel sheet through our simple strategy. In view of the substrate’s anticorrosive properties, the as-prepared material is named as anticorrosive corrosion-layer electrode (ACCE).

Fig. 2. Synthesis strategy and morphology characterization of ACCE.

a Schematic of the synthesis of ACCE followed by the Cyclic Voltammetry activation getting ACCE-a. b SEM images, c TEM image and d HRTEM image of ACCE. e STEM image of ACCE and f–h the corresponding elemental mapping images.

Considering that the material structure is prone to transformation under high oxidation potential conditions at the anode during water electrolysis, ex situ Raman spectroscopy is first performed to identify possible structure transition under catalytic condition for ACCE. The Raman spectra of ACCE in the range from 200 to 800 cm⁻¹ is shown in Fig. 3a. A pair of Raman bands located at 455 and 533 cm⁻¹ can be attributed to the E_g bending and A_1g stretching vibrations of Ni–O bonds derived from the NiFe(OH)₂ layers in the NiFe-LDH structure²¹. Different cyclic voltammetry (CV) cycles under catalytic conditions are subsequently conducted over ACCE (see the “Methods” section for details). These two characteristic bands disappear after 2 CV cycles, and a new pair of bands appears at 472 and 548 cm⁻¹, respectively. These two bands can be assigned to the E_g Ni–O bending vibration and A_1g Ni–O stretching vibration modes in Ni-based oxyhydroxides^22,23, demonstrating that the NiFe-LDH structure actually transforms into NiFeOOH through the CV cycles. Additionally, no other changes are observed with an increasing number of CV cycles, indicating that no other phase evolution occurs after NiFeOOH forms on the surface of ACCE. For simplicity, ACCE with 10 CV cycles (named as ACCE-a) is regarded as exhibiting a stable phase for OER, and the corresponding structure characterizations are performed for both ACCE and ACCE-a for comparable structure identification.

Fig. 3. Properties related to crystal structure and electronic configuration.

a Raman spectra of ACCE and relevant samples after different CV cycles. High-resolution b Ni 2p, c Fe 2p and d O 1s core level XPS spectra for ACCE and ACCE-a. e EPR patterns for ACCE, ACCE-a and NiFe-LDH-H. XANES spectra of f Fe K-edge, g Ni K-edge collected for ACCE and ACCE-a. h FT-EXAFS k³χ(R) spectra at Fe K-edge (up) and Ni K-edge (down) collected for ACCE and ACCE-a. i EXAFS wavelet transform (WT) of Fe K-edge for ACCE and ACCE-a.

X-ray photoelectron spectroscopy (XPS) is carried out to figure out the valence state transition of the surface species from ACCE to ACCE-a. As revealed by Fig. 3b, c, the Ni and Fe species on the surface of ACCE are dominated by Ni²⁺ and Fe³⁺, respectively, which is consistent with corresponding oxide states in the NiFe-LDH structure. Additionally, both metallic Ni⁰ and Fe⁰ species can also be observed on ACCE. The existence of Fe⁰ should originate from the as-deposited corrosion layer, and the generation of Ni⁰ might be due to a simultaneous displacement reaction (Ni²⁺ + Fe → Ni + Fe²⁺) accompanying the corrosion reaction²⁴. As for ACCE-a, both Fe⁰ and Ni⁰ species disappear, indicating that the surface of ACCE-a has been fully oxidized, and a new peak attributed to Ni³⁺ can also be identified, in accord with the formation of NiFeOOH on ACCE-a. Moreover, the O species in ACCE and ACCE-a are also carefully considered for recognizing local geometric structure transition. In the O 1s core-level spectra (Fig. 3d), the signals located at 529.6 eV (O1), 531.3 eV(O2), and 532.7 eV (O3) are assigned to lattice oxygen in M–O bond, surface oxygen vacancy, and adsorbed H₂O or hydroxy groups, respectively^25,26. ACCE exhibits a rather high concentration of oxygen vacancies (45.9%), indicating its low coordinated nature. Intriguingly, the proportion of oxygen vacancies in ACCE-a increases further to 54.3%, demonstrating that a portion of oxygen vacancies is generated in situ during the phase transformation.

Electron paramagnetic resonance (EPR) and X-ray absorption spectroscopy (XAS) are further performed to evaluate the condition of O vacancies in ACCE and ACCE-a. Another NiFe-LDH structure synthesized by a representative hydrothermal method is chosen to serve as a control sample for EPR, which is denoted as NiFe-LDH-H (see the “Methods” section for details). In Fig. 3e, the intensity of the EPR peak at g = 2.004 follows a trend of ACCE-a > ACCE > NiFe-LDH-H^27,28. This result indicates (i) the defect-rich nature of ACCE derived from the corrosion reaction compared with NiFe-LDH-H, (ii) an increase in O vacancies during the structure evolution, in agreement with the aforementioned XPS analysis. ACCE and ACCE-a are further studied by XAS, and the samples are physically stripped from the substrates. Due to the fact that the corrosion layers are not completely converted into the NiFe-LDH structure during synthesis, it is difficult to single out the NiFe-LDH structure from the samples. Accordingly, both ACCE and ACCE-a exhibit characteristic signals of metal features for both Fe and Ni species in the X-ray absorption near-edge structure (XANES) spectra. The absorption threshold positions of both Fe and Ni in ACCE-a are higher than those in ACCE (Fig. 3f, g), indicating an increase in average valence of both Fe and Ni species during structure transformation. In extended X-ray adsorption fine structure (EXAFS) spectra (Fig. 3h), prominent peaks indexed to Fe–Fe/Ni and Ni–Ni/Fe should originate mainly from the metallic corrosion layer. Fortunately, the signals of the first shell Ni–O and Fe–O bonds from surface species are not fully covered in R-space, which are located at 1.56 Å and 1.57 Ų⁹, respectively. By comparing the intensities of the first shell M–O peaks between ACCE and ACCE-a, the Fe–O signal decreases after the structure evolution, while the Ni–O bond is unchanged. This phenomenon suggests that low-coordinated structure generates around Fe sites in ACCE-a. The same conclusion can also be drawn from the WT-EXAFS of Fe K-edge shown in Fig. 3i. Together with the XPS and EPR results, it is reasonable to propose that the O vacancy structure tends to form adjacent to Fe sites during the phase transformation process. Additionally, the intensities of the second shell M–M peaks (Fe–Fe/Ni and Ni–Ni/Fe) for ACCE-a both increase compared with those for ACCE, suggesting that surface reconstruction might also occur during structure evolution, which would lead to a stabilization of the atomic structure for both Fe and Ni species in ACCE-a.

Electrochemical oxygen evolution properties of ACCE-a

The electrocatalytic performance of ACCE-a for OER is first evaluated in 1 m KOH solution (see details in the “Methods” section). The electrocatalytic activities of Fe/SS, IrO₂/SS, and SS are also provided as internal standards. As shown in polarization curves (Fig. 4a), the ACCE-a requires an overpotential of only 232 mV to achieve a current density of 10 mA/cm², while the noble metal-based IrO₂/SS with optimized loading amount (Fig. S8) requires a larger overpotential of 314 mV to reach the same current density. These results indicate that ACCE-a exhibits promising catalytic activity for OER even better than that of the IrO₂-based catalyst, and the high catalytic activity should originate from surface NiFeOOH phase, because both Fe/SS and SS show poor OER activities, requiring overpotentials of 328 and 359 mV, respectively, to reach 10 mA/cm². These results are further confirmed by the corresponding Tafel plots (Fig. 4b) and Nyquist plots (Fig. S9). ACCE-a exhibits the smallest Tafel slope (42.1 mV/dec) and charge-transfer resistance (R_ct) value (0.42 Ω cm²) among the samples, demonstrating its rapid reaction kinetics and fast charge-transfer property for ACCE-a over OER^30,31. Moreover, LSV measurements over a broader potential range are conducted to characterize the high current density activity of ACCE-a. As shown in Fig. S10, ACCE-a exhibits notable catalytic performance with overpotentials of 348 mV and 432 mV at current densities of 500 mA/cm² and 1000 mA/cm², respectively.

Fig. 4. Electrochemical behaviors for alkaline OER.

a Polarization curves of ACCE-a, IrO₂/SS, Fe/SS, and SS (working area: 0.5 cm × 0.5 cm) in 1 m KOH electrolyte with 95% iR-compensations (solution resistance: 2.49 ± 0.17 Ω). b Tafel plots for OER over ACCE-a, IrO₂/SS, Fe/SS and SS. c In situ Raman spectra collected at different potentials (from 1.368 to 1.468 V vs. RHE) with ACCE (without activation) as the anode, in 1 m KOH solution. Bode plots from in situ EIS over d ACCE-a and e NiFe-LDH-H (V vs. RHE). f Nyquist plots of ACCE-a under various potentials (V vs. RHE). Simulated (g) R_o/R_ct values and (h) Q₁/Q₂ values for ACCE-a and NiFe-LDH-H (V vs. RHE).

To investigate the impact of the synthesis method on the catalytic performance, the catalytic activities of ACCE-a and NiFe-LDH-H are further compared. As presented in Fig. S11a, NiFe-LDH-H needs a relatively large overpotential of 268 mV to obtain 10 mA/cm² current density. Moreover, the larger Tafel slope (73.3 mV/dec) of NiFe-LDH-H indicates its slower reaction kinetics compared to ACCE-a. The rate-determining step (RDS) can be attributed to the adsorption and energy optimization of OH⁻ reactants (M + OH⁻ → M–OH + e⁻ and M–OH → M–OH*, where M represents the catalytic active site) (Fig. S11b). As for ACCE-a, the OER process is limited by the second electron/proton step (OH* + OH⁻ → O* + H₂O + e⁻), resulting in a Tafel slope near 40 mV/dec. By comparing the RDS between NiFe-LDH-H and ACCE-a, it can be inferred that the defect-rich ACCE-a could facilitate the adsorption and activation of oxygenated reactants at the unsaturated sites. Additionally, NiFe-LDH-H exhibits a higher R_ct value (0.99 Ω cm²) than ACCE-a (0.42 Ω cm²), demonstrating the effectiveness of our corrosion method (Fig. S11c). Five additional parallel tests are performed (Fig. S11d, e), all of which fall within the confidence interval, further strengthening the credibility of the conclusions.

The electrochemical active surface area (ECSA) differences between ACCE-a and NiFe-LDH-H are further compared. As shown in Figure S11f, ACCE-a exhibits a double-layer capacitance (C_dl) of 3.84 mF/cm², corresponding to an ECSA of 24 cm². In contrast, the C_dl and ECSA values for NiFe-LDH-H are 1.33 mF/cm² and 8.31 cm², respectively. The 2.89-fold increase in ECSA highlights that the short and dense nanosheets of ACCE-a expose substantially more active sites than the larger nanosheets of NiFe-LDH (Fig. S11g, h). Meanwhile, the specific activity is calculated by normalizing the current density to ECSA (Fig. S11i). At an overpotential of 290 mV, ACCE-a performs a specific activity of 1.29 mA/cm², while NiFe-LDH-H only reaches 0.74 mA/cm². The above characterizations confirm the simultaneous increase in both site density and intrinsic catalytic activity for ACCE-a.

In situ Raman spectroscopy is further carried out to identify the phase evolution behavior under OER conditions. A series of in situ Raman spectra is recorded from 1.368 V to 1.468 V (vs. reversible hydrogen electrode (RHE)) for 120 s in 1 m KOH. In Fig. 4c, three kinds of Raman bands can be observed, which are divided into three regions marked by different colors. More specifically, from 1.368 V to 1.408 V vs. RHE (red region), the peaks at 455 and 529 cm⁻¹ are assigned to the Ni–O vibrations in Brucite-like NiFe-LDH structure, consistent with the ex situ Raman results. At a potential of 1.428 V vs. RHE (pale blue part), the peaks become broader, suggesting the partial conversion of NiFe(OH)₂ structure to NiFeOOH occurring under this oxidizing potential. From 1.448 V to 1.568 V vs. RHE (blue region), only peaks at 472 and 553 cm⁻¹ deriving from oxyhydroxide structure can be identified, indicating that the NiFe-LDH structure totally converts into NiFeOOH structure, which acts as the actual catalytic phase under OER operating conditions. As for NiFe-LDH-H, no significant change in the Raman bands is observed from the corresponding in situ Raman spectroscopy when the potential is increased to 1.528 V vs. RHE (Fig. S12). When the potential is further raised to 1.548 V vs. RHE, the signal intensity decreases due to interference from gas bubbles. This indicates that the generation of the catalytically active phase occurs more readily in ACCE-a, which may also contribute to its promising OER catalytic activity.

To elucidate the kinetic origin of the notable OER activity, in situ electrochemical impedance spectroscopy (EIS) is performed on ACCE-a and NiFe-LDH-H. The Bode plots presented in Fig. 4d, e display the impedance characteristics of ACCE-a and NiFe-LDH-H. Specifically, the phase angle observed in the low-frequency region indicates a non-homogeneous charge distribution within the electrode. In our system, this feature is attributed to the formation of oxygenated intermediates (such as M–OH and M–O) on the catalyst surface. At lower potentials in the low-frequency region, ACCE-a begins to exhibit a phase angle transition, occurring approximately 25 mV lower than for NiFe-LDH-H. This indicates that reaction intermediates form more easily on the surface of ACCE-a, representing a faster occurrence of the OER. Subsequently, the Nyquist plots acquired for both ACCE-a and NiFe-LDH-H at various potentials are fitted to an R_s(Q₁R_o)(Q₂R_ct) equivalent circuit model³² (Figs. 4f and S13). In this model, R_s is attributed to the solution resistance, R_o and Q₁ together describe the oxidation process within the catalytic structure, and R_ct and Q₂ reflect the catalytic reaction process between the catalytic layer and the solution. As shown in the radar chart (Fig. 4g, h), the smaller R_o and larger Q₁ for ACCE-a compared to NiFe-LDH-H indicate enhanced intrinsic conductivity for ACCE-a, along with a faster transition to the NiFeOOH phase, which is consistent with the in situ Raman results. For the fitted values of R_ct and Q₂, ACCE-a also exhibits a smaller R_ct and a larger Q₂ than NiFe-LDH-H at all potentials, reflecting better catalytic activity for ACCE-a and a faster charge transfer process at the catalyst/electrolyte interface, which is in agreement with the Tafel slope analysis.

Based on the above electrochemical analysis, ACCE-a, synthesized via the sacrificial corrosion method, exhibits four key advantages over NiFe-LDH-H produced by the hydrothermal approach: (ⅰ) The smaller and denser nanosheet morphology, coupled with unsaturated coordination structures, facilitates the exposure of more active sites. (ⅱ) It demonstrates better intrinsic conductivity, which allows for rapid charge transfer between the substrate and the catalytic layer, as well as between the catalytic layer and the electrolyte. This enhanced conductivity facilitates the easier transformation of NiFe-LDH to the NiFeOOH active phase during the OER process. (ⅲ) In situ phase transformation introduces more oxygen vacancy structures, which enhances the intrinsic activity of the active sites and results in a lower Tafel slope. (ⅳ) The synergy between a high density of active sites and optimized intrinsic activity facilitates the rapid consumption and conversion of reaction intermediates, resulting in efficient oxygen evolution at significantly lower overpotentials. These distinct structural and kinetic characteristics, induced by the sacrificial corrosion method, endow ACCE-a with promising OER catalytic activity and substantiate the advanced nature of our synthesis strategy.

Exploration of the practical uses of ACCE-a in seawater electrolysis

The catalytic activity of ACCE-a is subsequently evaluated in the seawater environment. To simplify the system, 1 m KOH solution containing 0.5 m NaCl is employed as the electrolyte to simulate seawater condition (denoted as simulated seawater). By evaluating the catalytic activities of ACCE-a and IrO₂ in 1 m KOH and simulated seawater (Figs. 5a and S14), ACCE-a exhibits a slight degradation in the chloride ions containing electrolyte, affording a current density of 10 mA/cm² at 235 mV. This behavior might originate from structural features, which is consistent with other works relevant to Ni-based (oxy)hydroxides^33–35. Additionally, the multi-current step curves of ACCE-a are performed in simulated seawater (Fig. S15). The potential changes immediately with the instantaneous change in current density, indicating that ACCE-a has good conductivity together with mass transport properties.

Fig. 5. Electrochemical behaviors of ACCE-a for seawater electrolysis.

a Polarization curves for OER under 95% iR-compensations by using ACCE-a and IrO₂ as the electrode (working area: 0.5 cm × 0.5 cm) in 1 m KOH with and without 0.5 m NaCl (solution resistance: 2.12 ± 0.09 Ω). b UV–Vis absorption spectra of the DPD test solutions for ACCE-a, IrO₂ and reference electrolyte collected under different measurement conditions. Inset: Digital graphs of the test solutions prepared from the reaction electrolyte for ACCE-a and IrO₂ collected at the overpotential of 480 mV and 700 mV, respectively. c Derived non-CER overpotential and OER-only potential window based on the onset potentials of OER and CER for IrO₂ and ACCE-a in 1 m KOH with 0.5 m NaCl. d Polarization curve of 3D-ACCE-a with 95% iR-compensations in 1 m KOH (solution resistance: 2.36 ± 0.01 Ω). Inset: digital image of 3D-ACCE. e Chronopotentiometric curve of 3D-ACCE-a at a current density of 500 mA/cm² without iR-compensations in 10 m KOH seawater.

To accurately identify the actual OER-only potential window of ACCE-a, appropriate testing methods and protocols are carefully explored. Given the difficulty in observing subtle color changes with the naked eye, the N,N-Diethyl-1,4-phenylenediamine (DPD) reagent is selected due to its high sensitivity. When the DPD reagent binds to ClO⁻ ions, it forms a pink-colored oxidation product with characteristic peaks at 515 nm and 551 nm in the UV–Visible (UV–Vis) absorption spectrum^36,37. The linear correlation between different ClO⁻ concentrations and adsorption strength is shown in Fig. S16. We first use an IrO₂ electrode as the reference electrode to conduct stability tests at different voltages and currents, and analyze the residual chlorine detection results. As depicted in Fig. S17, even when the IrO₂ electrode reaches the voltage threshold for chlorine evolution, ClO⁻ ions could not be detected if the current value is insufficient. To address the issue of insufficient ClO⁻ accumulation from short test durations, the chronoamperometry (i-t) tests are conducted for 10 h to guarantee adequate ClO⁻ accumulation. Subsequently, a suitable testing current and duration (280 mA, 10 h) are identified, and the OER-only ranges for both IrO₂ and ACCE-a electrodes are measured. In Fig. 5b, it is evident that the OER-only potential range of ACCE-a extends to 700 mV, which is considerably wider than that of IrO₂ (480 mV). The corresponding Pourbaix Diagram is summarized in Fig. 5c, indicating that ACCE-a, compared to IrO₂, has a larger safe potential window where CER do not occur, making it more suitable for seawater electrolysis applications.

The catalytic stability of ACCE-a under seawater electrolysis is carefully assessed. The chronopotentiometric curve (Fig. S18a) shows that the OER activity of ACCE-a is maintained for 100 h at the current density of 100 mA/cm² without an obvious potential rise. A sample grown on Fe foam via the traditional corrosion method is also tested. After operating in alkaline simulated seawater at 100 mA/cm² for 6.3 h, it shows obvious potential fluctuations and morphological degradation (Fig. S18b). The surface gradually forms a brown precipitate, indicating that the material is unsuitable for seawater electrolysis. Accelerated aging tests on ACCE-a are then performed by conducting 10,000 CV cycles in alkaline simulated seawater. The CV curve of ACCE-a shows no significant performance degradation (Fig. S18c), and the nanosheet morphology remains largely intact (Fig. S18d, e). The measured interplanar spacing corresponds to the (018) crystal plane of NiFe-LDH (Fig. S18f). These results confirm that ACCE-a has notable stability and chlorine corrosion resistance in simulated seawater. The contact angle test (Fig. S18g, h) demonstrates that ACCE-a has good hydrophilicity (contact angle = 22°) and gas repellency (bubble contact angle = 156°). This reveals that the nanosheet morphology of ACCE-a promotes contact with the solution and aids in the separation of products during the catalytic reaction, and thereby helps maintain high catalytic and structural stability.

Considering the practical application scenarios of seawater electrolysis, such as offshore wind power integration, the structural stability of electrode materials under non-operational conditions requires further evaluation. Specifically, in the high-salinity and high-humidity atmospheric environment induced by seawater, water molecules would adsorb onto the material surface, forming a stable thin water film. This film, along with atmospheric CO₂, O₂, and Cl⁻, subsequently establishes a weakly acidic electrochemical corrosion environment. Evidently, electrodes would be more susceptible to corrosion reactions under such conditions compared to alkaline solutions, leading to phase transformations and subsequent deactivation. Therefore, the corrosion resistance under saline-humid conditions should be a key criterion for assessing practical applicability. Accordingly, the corrosion resistance of ACCE is tested in the air of Hainan, China—a coastal region with elevated humidity levels. As shown in Fig. S18i, no obvious changes are noticed in the LSV polarization curves of ACCE after being put in the air for 3.5 years, which verifies the compatibility of our synthesis method with seawater electrolysis applications.

Given that three-dimensional substrates better match practical applications, a corrosion layer electrode loaded onto three-dimensional stainless-steel mesh (3D-ACCE) is also synthesized. The inset in Fig. 5d shows a photograph of 3D-ACCE, indicating that the NiFe-LDH synthesized by this method grows uniformly on the complex network substrate. In the LSV curve shown in Figs. 5d and S19a, 3D-ACCE-a achieves current densities of 10 and 50 mA/cm² with overpotentials of 182 and 232 mV, respectively. This indicates that with higher loading and catalytic active area, 3D-ACCE can achieve better catalytic performance. Compared with similar structures reported in recent years, the 3D-ACCE-a electrode demonstrates strong competitiveness in terms of catalytic activity and stability (Table S1). Then, the OER performance of 3D-ACCE-a is studied under real seawater conditions. In 1 m KOH seawater and 10 m KOH seawater solutions, the 3D-ACCE-a catalyst achieves a current density of 500 mA/cm² at overpotentials of 420 mV and 294 mV, respectively (Fig. S19b). The stability of the electrode in real seawater is further investigated (Figs. 5e and S19c). Remarkably, the electrode maintains its performance for 1000 h at a current density of 500 mA/cm², indicating that the material has high practical application value.

The electrocatalytic performance of 3D-ACCE-a toward overall water splitting is further evaluated in an alkaline flow electrolyzer (working area: 1 cm²), using Raney Ni as the cathode. As shown in Fig. S20a, b, the 3D-ACCE-a achieves a current density of 10 mA/cm² at 1.477 V with promising stability, lower than that of the commercial Raney Ni (1.556 V). At a potential of 1.7 V, the current density achieved by 3D-ACCE-a is 2.23 times that of the Raney Ni benchmark. The Faraday efficiency of the 3D-ACCE-a||Raney Ni electrode couple reaches 92.86% (Fig. S20c). To evaluate its potential for industrial applications, a larger-area alkaline flow electrolyzer is assembled (working area: 25 cm²). As shown in Fig. S20e, the 3D-ACCE-a||Raney Ni couple achieves currents of 1 A and 10 A at potentials of 1.633 V and 2.319 V, respectively, which are lower than those required by the Raney Ni||Raney Ni couple (1.726 V, 2.496 V). The electrolyzer also demonstrates promising stability, maintaining stable operation at 5 A for at least 100 h (Fig. S20f). In addition, a 3D-ACCE electrode with an area of approximately 30 cm × 30 cm is fabricated via a scale-up experiment (Fig. S21). The microscopic morphology of five regions is examined, and all regions display similar nanosheet morphologies, proving that the sacrificial corrosion method is well-suited for the uniform synthesis of large-sized electrodes. These results demonstrate that 3D-ACCE-a can effectively replace Raney Ni in industrial water electrolysis for hydrogen production.

Examination of the origins of catalytic activity and selectivity in ACCE-a

Theoretical calculations are carried out to investigate the promotion mechanism of ACCE-a for both OER and CER. Considering that the catalytically active phase is identified as γ-NiFeOOH, a γ-NiFeOOH model (NiFeOOH) is constructed. Given the substantial proportion of oxygen vacancy structures in ACCE-a, the preferred formation site and impact on the OER activity of NiFeOOH are first examined. In the NiFeOOH structure, the calculated formation energies of oxygen vacancies adjacent to Fe and Ni sites are 0.285 eV and 1.173 eV, respectively. Because both values are positive, the creation of oxygen vacancies should be an endothermic process, requiring sufficient energy input to overcome the formation barrier. This result rationalizes the experimental observation that the oxygen-vacancy concentration increases under high potential conditions. Additionally, the lower formation energy near Fe indicates that oxygen vacancies preferentially form around Fe sites, consistent with the EXAFS results. Therefore, the structure with oxygen vacancies (Vₒ) constructed around the Fe sites is chosen to describe ACCE-a (NiFeOOH-Vₒ).

Both Ni and Fe sites are considered to be potential active sites, and the calculated OER catalytic activities at such sites for pristine NiFeOOH are depicted in Fig. 6a, b. As for the Fe site, the theoretical overpotential for OER is 1.14 eV, and the potential determining step (PDS) is identified as the step from OH⁻ to OH* (OH⁻ → OH* + e⁻, where * denotes adsorption sites), indicating weak binding between oxygen-containing intermediates and Fe sites. In case of the Ni site, the PDS is determined to be the step from OH* to O* (OH* + OH⁻ → O* + H₂O + e⁻), with a theoretical overpotential of 1.36 eV. In the pristine NiFeOOH structure, Fe sites could exhibit higher activity than Ni sites, which is consistent with other reported works^38,39. The OER activity of NiFeOOH-V_o is also examined. Regarding the Fe site, after introducing V_o structure around Fe sites, the OH* intermediate still adsorbs at the Fe top site, while the O* and OOH* intermediates intend to fill the V_o structure, forming bridging bonds between Ni and Fe sites (see Fig. S22). Moreover, the PDS for NiFeOOH-V_o transitions to the OOH* formation step (O* + OH⁻ → OOH* + e⁻), accompanied by enhanced OER catalytic activity with a theoretical overpotential of 0.19 eV. For the Ni site, the three oxygen intermediates all adsorb at the Ni top site, with the transition from OH⁻ to OH* as the PDS, and the theoretical overpotential is 0.68 eV. Accordingly, the intrinsic activity ranking is NiFeOOH-V_o-Fe > NiFeOOH-V_o-Ni > NiFeOOH-Fe > NiFeOOH-Ni, and the intrinsic activities of the Fe sites always surpass those of the Ni sites. Additionally, both Ni and Fe sites in NiFeOOH-V_o exhibit higher activity compared to those in the pristine NiFeOOH structure, which is consistent with the experimental findings. By comparing the ΔG_O*-ΔG_OH* values, the intrinsic OER activity of the NiFeOOH-V_o structure is improved by increasing the adsorption of oxygen intermediates at both the Fe and Ni sites. The enhanced hybridization between oxygen intermediates and Fe, Ni sites in the NiFeOOH-V_o, as shown in Figs. 6c, S23, and Table S2, further support the above conclusion. In summary, the above calculation results demonstrate that the V_o structure could positively affect OER by adjusting the coordination structures of Fe and Ni, which in turn optimize their interaction with oxygen-containing intermediates and promote the overall catalytic process.

Fig. 6. Theoretical insights into OER/CER mechanism of the catalysts.

The OER free energy diagrams for NiFeOOH structure with and without O vacancy structure at both a Fe site and b Ni site. c Calculated projected density of states (pDOS) of Fe d band for Fe adsorption site in NiFeOOH and NiFeOOH-V_o structure. d The maximum free energy difference (ΔG_max) under thermodynamic consideration of IrO₂ and NiFeOOH-V_o structure for OER (bule bars), CER (red bars) and hypoCER (orange bars). The e hypoCER and f CER free energy diagrams for NiFeOOH-V_o and IrO₂ structure. g The calculated Pourbaix diagram for chlorine-water system with total chlorine species concentration of 0.5 m (pH > 7.5) based on the corresponding theoretical overpotential of NiFeOOH-V_o and IrO₂ structure. The CHE represents a computational hydrogen electrode⁴⁷. h The charge density difference of Cl adsorbed NiFeOOH-V_o structure and IrO₂, the yellow and blue regions represent electron accumulation and depletion, respectively. i Schematics of promoted OER activity and selectivity over NiFeOOH-V_o structure.

The difference in OER selectivity between the NiFeOOH-V_o and IrO₂ structures is investigated. Concerning the Cl⁻ oxidation reaction, both CER and hypochlorite evolution reaction (hypoCER) pathways are considered, with calculation details provided in the “Methods” section. Given the differences in equilibrium potentials among the three reactions, it is unclear to evaluate selectivity based on overpotential values. Therefore, the energy barriers of the PDS are directly compared (Fig. 6d). As for the IrO₂ structure, according to the relevant literature, Cl⁻ tends to adsorb on the oxygen sites of IrO₂⁴⁰. The energy barrier for Cl⁻ oxidation is smaller than that for water oxidation (Fig. S24), indicating that IrO₂ exhibits poor OER selectivity and favors Cl⁻ oxidation. Regarding the NiFeOOH-V_o structure, the OER selectivities for both the Ni and Fe sites are evaluated. At the Ni site, the energy barrier for OER is slightly smaller, but the energy barriers for the three reactions are similar in magnitude. At the Fe site, the trend for energy barriers is OER < CER < hypoCER, and the OER energy barrier is significantly lower compared to the two Cl⁻ oxidation reactions. This suggests that Cl⁻ oxidation is more difficult at Fe sites in NiFeOOH-V_o. Subsequently, the PDSs for Cl⁻ oxidation on the two structures are analyzed (Fig. 6e, f). For the IrO₂ structure, the PDSs for CER and hypoCER are the formation of Cl₂ and HOCl, respectively, indicating that the IrO₂ structure has a relatively strong affinity for Cl⁻ adsorption. For the NiFeOOH-V_o structure, the CER and hypoCER at the Ni site are limited by Cl* adsorption, reflecting weaker intermediate-surface interaction. As for the Fe site, the CER and hypoCER are determined by the Cl₂ and HOCl formation, respectively, which is similar to IrO₂. Furthermore, the Fe site exhibits stronger Cl* adsorption, leading to a higher energy barrier than that on IrO₂. The catalytic activities of the two sites in NiFeOOH-V_o are further integrated, and the theoretical Pourbaix diagrams for IrO₂ and NiFeOOH-V_o, which include pH effects, are illustrated in Fig. 6g. Both OER and hypoCER are pH-dependent, while CER is not affected by pH. In contrast to the IrO₂ structure, where the reaction potential trend is hypoCER < OER < CER, the hypoCER potential for NiFeOOH-V_o is 520 mV higher than OER potential. This creates a substantial OER-only potential window, indicating that ACCE-a has notable selectivity for OER, which aligns with experimental findings.

The electron transfer processes between IrO₂/NiFeOOH-V_o and Cl* are further analyzed to elucidate the possible selectivity mechanism for ACCE-a (Fig. 6h). In the IrO₂ structure, the Cl intermediate adsorbs at the O site and transfers 0.439 electrons to O, resulting in positively charged Cl*. Regarding the NiFeOOH-V_o structure, Cl intermediates at the Fe and Ni sites result in negatively charged Cl*, which gain 0.43 and 0.16 electrons, respectively, consistent with the aforementioned adsorption strength trend. Specifically, at the Fe site, Cl* stabilizes in a bridge position between the Fe and the adjacent Ni atom, accounting for the stronger interaction. At the Ni site, Cl* adsorbs at the top site of the Ni atom. Given that the Ni site exhibits a lower CER/hypoCER theoretical overpotential than the Fe site, the selectivity of the NiFeOOH-V_o structure is primarily governed by the Ni site. Since Cl⁻ oxidation at the Ni site for NiFeOOH-V_o is limited by the Cl adsorption process, reducing the interaction between Ni and Cl⁻ should be an effective strategy for designing NiFe-based catalytic structures with higher OER selectivities. Comprehensively, the proposed catalytic mechanism for the NiFeOOH-V_o structure is summarized in Fig. 6i, illustrating the distinct roles of the Ni and Fe sites. The Fe sites, which are highly active for the water oxidation reaction, could be effectively protected by the Ni sites, which engage in the competing Cl⁻ oxidation reactions. Additionally, based on the negative charge feature of Cl adsorption on the NiFeOOH-V_o structure, the adsorbed Cl intermediates may create a negatively charged surface layer, potentially repelling Cl⁻ ions through electrostatic repulsion and thereby improving OER selectivity of ACCE-a.

Discussion

In conclusion, a notable sacrificial corrosion strategy is developed to synthesize highly active NiFe-LDH nanoarray films on various conductive substrates. Specifically, the 3D-ACCE-a electrode grown on a stainless steel mesh reaches a current density of 10 mA/cm² at an overpotential of only 182 mV, and it can stably operate at 500 mA/cm² for 1000 h in 10 m KOH seawater solution, with no performance degradation even after 3.5 years in high humidity. An OER-only potential window of 700 mV, far exceeding the 480 mV thermodynamic framework, is quantitatively measured, confirming that kinetic regulation can extend beyond thermodynamic limits. Benefiting from the structural advantages imparted by sacrificial corrosion, ACCE-a exhibits a markedly higher density of active sites, promising intrinsic activity, enhanced conductivity, and improved phase conversion ability compared with hydrothermally synthesized counterparts. DFT calculations reveal that the introduction of V_o shifts the PDS of catalytically active NiFeOOH phase from the OH* formation step to the OOH* formation step, reducing the theoretical overpotential at the Fe sites in NiFeOOH-V_o to 0.19 eV, thereby enhancing the OER catalytic activity. Moreover, theoretical analysis of the CER/hypoCER pathways indicates that Ni sites could safeguard Fe sites, giving rise to a previously unrecognized dual-site cooperative mechanism that accounts for the high oxygen-evolution selectivity and offers a guiding principle for future catalyst design. To sum up, this work establishes a scalable synthesis platform for constructing highly active corrosion-derived products suited to industrial operation, further laying a fresh perspective for the directed design of high-performance seawater electrolysis anodes.

Methods

Chemicals

Stainless steel sheet (denoted as SS, type: 304, thickness: 0.2 mm) was purchased from Shenzhen Hongdu Metal Co., Ltd. Stainless steel mesh (denoted as SSM, type: 316, mesh number: 300, wire diameter: 0.04 mm) was purchased from Shenzhen Oriental Sieve Co., Ltd. Ni foam (thickness: 0.3 mm) and Cu foam (thickness: 1 mm) were purchased from Kunshan Guangjiayuan New Materials Co., Ltd. Fe foam (thickness: 1 mm) was purchased from Kunshan Electronic Co., Ltd. Raney Ni was purchased from Shanghai Tengqing Testing Instruments Co., Ltd. Carbon cloth (thickness: 0.31 mm, WOS1009) was purchased from Taiwan CeTech Co., Ltd. Titanium mesh (thickness: 0.1 mm, 100 meshes) was purchased from Kangwei Metal Wire Mesh Co., Ltd. Before using, all substrates except Raney Ni and Fe foam were cleaned several times with dilute nitric acid (3 m), ethanol and deionized water. Raney Ni was cleaned with ethanol and water. Fe foam was treated with 1 m oxalic acid solution, ethanol and deionized water. Ferrous chloride (FeCl₂·4H₂O, 98%), potassium hydroxide (KOH, 95%), urea (CH₄N₂O, AR), oxalic acid (H₂C₂O₄, 99%) and Iridium dioxide (IrO₂, 99.9%) were purchased from Shanghai Macklin Biochemical Co., Ltd. Nickel sulfate hexahydrate (NiSO₄·6H₂O, AR). Nitric acid (HNO₃, AR) was purchased from Guangzhou Chemical Factory. Iron nitrate nonahydrate (Fe(NO₃)₃·9H₂O, AR), Sodium chloride (NaCl, AR), Ammonium fluoride (NH₄F, 98%) and Nafion 117 solution (5%) were purchased from Shanghai Aladdin Bio-Chem Technology Co., Ltd. Sulfuric acid (H₂SO₄, AR), Nickel nitrate hexahydrate (Ni(NO₃)₂·6H₂O, AR) and Ethanol (C₂H₅OH, 99.5%) were purchased from Xilong Chemical Factory. DPD residual chlorine indicator (Determination range: 0.05–1.0 mg/L) was purchased from Beijing Lingze Technology Co., Ltd. Alkaline electrolyzer membrane (ZF-500, 0.5 ± 0.05 mm) was purchased from Hangzhou Huamo Technology Co., Ltd and immersed in deionized water for storage. Deionized water (>18 MΩ cm resistivity) was produced by a Water Purification System of Heal Force. Seawater was taken from the South China Sea near the Boundary Island.

Synthesis of ACCE and 3D-ACCE

ACCE and 3D-ACCE were synthesized via an electroplating procedure followed by a modified corrosion reaction¹⁴. For the construction of a two-electrode system for the electrodeposition process, we chose stainless steel substrates (SS or SSM, 1 cm × 4 cm) to serve as the cathode, a piece of SS (1 cm × 4 cm) was set as the anode, and the composition of the electrolyte was 0.1 m FeCl₂ solution. For the detailed electrodeposition process, the distance between cathode and anode was set as 1.6 cm, and the deposition current and deposition time were set as 0.2 A and 2 min, respectively. After the electrodeposition process, the cathode was taken out of the electrolyte and cleaned using deionized water, denoted as Fe/SS or Fe/SSM.

As for the modified corrosion reaction, the 10 mL of 0.1 m NiSO₄ solution was first added into a 12 mL glass bottle to serve as corrosion solution, then the Fe/SS or Fe/SSM was immersed into the solution at 80 °C for 3 h. After the reaction, the as-obtained electrode was washed with deionized water and placed in a vacuum drying oven at 60 °C for 12 h, denoted as ACCE (from Fe/SS) or 3D-ACCE (from Fe/SSM). The loading amounts of ACCE and 3D-ACCE were determined to be 1.767 and 2.206 mg/cm², respectively, based on inductively coupled plasma-mass spectrometry (ICP-MS). Specifically, ICP-MS measurements showed that the total masses of Ni ions in the catalytic layer of 1 cm² ACCE and 3D-ACCE were 0.9813 mg and 1.2172 mg, respectively, while the total masses of Fe ions were 0.1896 mg and 0.2384 mg, respectively. Based on the molar ratio of Ni:Fe determined by X-ray energy dispersive spectra (EDS) results (3:1), and referring to the corresponding NiFe-LDH structural formula Ni_0.75Fe_0.25(CO₃)_0.125(OH)₂·0.38H₂O, the relative atomic mass percentages of Ni and Fe in the structure were calculated to be 41.40% and 13.12%, respectively. Since both metallic Ni and Fe are present in ACCE, the XPS peak fitting results indicated that the proportions of Ni⁰ and Ni²⁺ in ACCE were 24.00% and 76.00%, respectively, while the proportions of Fe⁰ and Fe³⁺ were 9.16% and 90.84%, respectively. Assuming all Fe³⁺ originates from NiFe-LDH, the mass of NiFe-LDH can be calculated as 0.1896 mg × 90.84%/13.12% = 1.3127 mg. Consequently, the mass of metallic Ni is 0.9813 mg − 1.3127 mg × 41.40% = 0.4365 mg, and the mass of metallic Fe is 0.1896 mg × 9.16% = 0.0173 mg. The loading of the ACCE catalytic layer was obtained by summing the three components (NiFe-LDH, metallic Ni, and metallic Fe), resulting in 1.767 mg/cm². Similarly, the loading of 3D-ACCE was calculated to be 2.206 mg/cm². Additionally, we also estimated the loading amounts of ACCE and 3D-ACCE using a high-precision electronic balance by measuring the mass difference between the as-prepared electrode and the bare substrate with the same area. The values obtained were 1.095 mg/cm² and 1.770 mg/cm², respectively. Considering the inherent errors in both estimation methods, and based on the principle of conservative estimation, we selected the larger estimated loading values of 1.767 mg/cm² and 2.206 mg/cm² for ACCE and 3D-ACCE, respectively. Meanwhile, the electrodes (from Fe/SS) prepared by varying the corrosion time (1, 2, 4, 5 h) were denoted as ACCE-1h, ACCE-2h, ACCE-4h, and ACCE-5h, respectively.

Synthesis of IrO₂/SS

The optimized loading amount of powdered IrO₂ on stainless steel sheet (SS) was evaluated through corresponding comparison experiments (Fig. S8), which demonstrates that the optimal catalytic performance of IrO₂ is achieved at a loading amount of 2 mg/cm². Accordingly, 4 mg of IrO₂ catalyst was added into 760 μL of isopropyl alcohol, followed by the addition of 40 μL of Nafion solution. The resulting mixture was sonicated for 30 min to obtain a homogeneously dispersed solution. Portions of the mixture (20, 40, 60, 80, 100, and 150 μL) were dropped onto a 0.25 cm² stainless steel sheet and allowed to air-dry to form the IrO₂ electrodes, with IrO₂ loadings of 0.4, 0.8, 1.2, 1.6, 2, and 3 mg/cm², respectively.

Synthesis of NiFe-LDH-H

A mixture of 0.247 g of Ni(NO₃)₂·6H₂O, 0.15 g of urea, 0.0463 g of NH₄F, and 0.09 g of Fe(NO₃)₃·9H₂O was added to 15 mL of deionized water and stirred magnetically for 15 min. The mixed solution, along with a cleaned stainless steel sheet (SS), was then placed in a 25 mL Teflon-lined autoclave and subjected to hydrothermal treatment at 120 °C for 8 h. After cooling, the SS with the hydrothermally grown NiFe-LDH film was obtained and denoted as NiFe-LDH-H.

Synthesis of NiFe-LDH-C/FF

NiFe-LDH-C/FF electrodes were synthesized on a Fe foam substrate via the traditional corrosion method. A 1 cm × 4 cm piece of Fe foam was ultrasonically cleaned in 1 m oxalic acid, ethanol, and deionized water for 5 min each to remove surface oxides. It was then transferred to a 0.1 m NiSO₄ solution for 12 h. Afterward, it was rinsed with deionized water and dried, denoted as NiFe-LDH-C/FF.

Electrolyte preparation

The 1 m KOH solution was prepared by adding 59.06 g of KOH (95%) to 1 L of deionized water. After thorough mixing and cooling, the solution (pH = 13.94 ± 0.01) was transferred to a plastic volumetric flask for storage. The simulated seawater consisted of 1 m KOH and 0.5 m NaCl, prepared by adding 59.06 g of KOH (95%) and 29.37 g of NaCl to 1 L of deionized water. Following mixing and cooling, the solution (pH = 13.94 ± 0.01) was stored in a plastic volumetric flask.

Raw seawater was treated to get buffered seawater. According to the literature, the concentration of Mg²⁺ and Ca²⁺ in seawater is 0.412 g/kg and 1.23 g/kg, respectively⁴¹. Therefore, compared with 1 m KOH solution obtained from purified water, more KOH is needed to precipitate Mg²⁺ and Ca²⁺ ions in raw seawater. In detail, 150 mL of raw seawater was first transferred into a beaker, followed by the addition of 9.98 g of KOH under magnetic stirring. After standing for more than 12 h, the supernatant solution denoted as buffered seawater (pH = 13.92 ± 0.01) was taken out to use as the electrolyte for seawater electrolysis. 10 m KOH seawater was prepared by adding 53.15 g of KOH (95%) to 100 mL of buffered seawater.

Materials characterizations

The powder X-ray diffraction (XRD) patterns were obtained from a Panalytical Empyrean X-ray diffractometer with Cu Kα radiation (λ = 1.5418 Å). The scanning electron microscope (SEM) images were recorded on a JEOL JSM-7800F electron microscope at an accelerating voltage of 5 kV. The Energy Dispersive X-ray (EDX) analysis and transmission electron microscope (TEM) images were collected on a JEOL JSM-2800 microscope equipped with a field emission gun operating at 200 kV. The X-ray photoelectron spectroscopy (XPS) was performed on an KRATOS Axis Supra X-ray photoelectron spectrometer with a monochromatic X-ray source (Al Kα hυ = 1486.6 eV). All the XPS spectra have been calibrated by the binding energy of C1s at 284.8 eV. The Raman spectra of the materials were obtained with a Renishaw Raman system model InVia operating with a 50 mW argon-ion laser (514 nm) as the excitation light source. EPR was performed on a Bruker A300 EPR spectrometer. Fe and Ni K-edge XAS measurements were executed in transmission mode using a Si (111) double-crystal monochromator at the BL14W1 beamline of the Shanghai Synchrotron Radiation Facility, with the facility operating at 3.5 GeV and maintaining a maximum current of 230 mA. The UV–Vis absorption spectra were measured by UV–Vis spectroscopy (Persee, TU-1901). The contact angle and underwater bubble contact angle were measured by contact angle measurement (SINDIN, SDC-80).

Electrochemical measurements

Electrochemical measurements were performed using a standard three-electrode configuration with a Zahner Instrument (Model ZENNIUM), unless otherwise stated. 1 m KOH (pH = 13.94 ± 0.01), 1 m KOH with 0.5 m NaCl (simulated seawater, pH = 13.94 ± 0.01) and buffered seawater (pH = 13.92 ± 0.01) were used as electrolyte. A glass cell with a diameter of 6.5 cm and a height of 7 cm acted as the electrochemical cell. The materials (Fe/SS, ACCE, and 3D-ACCE, etc.) were used as the working electrode. Unless otherwise specified, the working surface area for all electrochemical tests was 0.25 cm². Hg/HgO electrode and Pt foil (1 cm × 1 cm) were used as reference and counter electrode, respectively.

The Hg/HgO electrode was calibrated according to a method reported in the literature⁴². Specifically, H₂ gas was bubbled into the electrolyte (1 m KOH or 10 m KOH seawater) until saturation. The Hg/HgO electrode served as the reference electrode, and the platinum wire was used as both the working and counter electrode. LSV was conducted at a scan rate of 1 mV/s. The results showed that the potential at 0 mA current was −0.9236 V vs. Hg/HgO in 1 m KOH and −0.9824 V vs. Hg/HgO in 10 m KOH seawater (Fig. S1). Based on the conservative approximation, the relation between the Hg/HgO electrode and the RHE can thus be established using the following equations:

In 1 m KOH:

$$E_{\mathrm{RHE}}=E_{\mathrm{Hg}/\mathrm{HgO}}+0.924\mathrm{V}$$ 6

In 10 m KOH seawater:

$$E_{\mathrm{RHE}}=E_{\mathrm{Hg}/\mathrm{HgO}}+0.983\mathrm{V}$$ 7

In addition, the catalytic activity of powdered IrO₂ was measured by loading it on the SS with an optimal loading amount (0.5 mg for 0.25 cm², Fig. S8), and a polymer binder (Nafion) was also introduced for the purpose of fixation. The current densities were normalized with the geometric surface areas of the electrodes. For LSV measurements, the scan rate was set to 2 mV s⁻¹ and compensated by 95% iR-compensations, and the resistances of the test system were estimated from corresponding single-point impedance measurements unless otherwise stated. Chronopotentiometric measurements were performed to evaluate the stability of the catalyst without iR-compensation. EIS was measured on the materials under a constant overpotential of 294 mV with an amplitude of 10 mV. The frequency of EIS ranged from 100 kHz to 0.1 Hz.

Before performance testing, all electrodes were activated by CV for ten cycles within a potential range of 0.924–1.924 V vs. RHE at a scan rate of 100 mV/s. The activated ACCE and 3D-ACCE electrode were denoted as ACCE-a and 3D-ACCE-a, respectively.

In order to get the effective electrochemical active surface area (ECSA) of a material, a series of CV measurements was performed first at various scan rates (20, 40, 60, 80, 100 mV s⁻¹) in the potential window between 0.924 to 1.024 V (vs. RHE), and the sweep segments of the measurements were set to 10 to ensure consistency. The geometric double-layer capacitance (C_dl) was determined by calculating the slope of the linear regression line derived from the current density differentials (ΔJ = J_anodic − J_cathodic) versus scan rates at 0.974 V vs. RHE. Specifically, C_dl corresponds to half of the obtained slope. Subsequently, the ECSA was computed using the equation below,

$$\mathrm{ECSA}=\frac{C_{\mathrm{dl}}}{C_s}\times \mathrm{ASA}$$ 8

where C_s and ASA denoted the specific capacitance and the actual surface area of the electrode, respectively. For the calculations in this study, C_s was approximated as 0.04 mF cm^-2.

The alkaline flow electrolyzer was composed of an anode (Raney Ni, IrO₂ or 3D-ACCE-a), cathode (Raney Ni), membrane (ZF-500), and other components (endplates, flow channel, current collectors, fluorine rubber gaskets and PTFE gaskets). The loading amount of IrO₂ on stainless steel mesh was adjusted to 2 mg/cm², and the working areas of the electrodes were 1 cm² or 25 cm². The alkaline flow electrolyzer was operated in 1 m KOH as the electrolyte at 25 °C. Electrochemical measurements of alkaline flow electrolyzers with working areas of 1 cm² and 25 cm² were conducted using a Zahner electrochemical workstation (Model ZENNIUM) and a DC power source (KUAIQU, model R-SPS3030), respectively.

Faradaic efficiency (FE) was measured under galvanostatic mode. The volume of evolved gas was collected using the water drainage method and compared with the theoretical value. Accordingly, the FE was calculated using the following formulas:

Actual molar amount of O₂:

$$n_{\mathrm{measured}}=\left(p-p_{\mathrm{liquid}}\right)V/\left(RT\right)$$ 9

Theoretical molar amount of O₂:

$$n_{\mathrm{theoretical}}=It/\left(nF\right)$$ 10

$$\mathrm{FE}=\left(n_{\mathrm{measured}}/n_{\mathrm{theoretical}}\right)\times 100\%$$ 11

where p is atmospheric pressure, p_liquid is liquid column pressure, V is the measured gas volume, R is the ideal gas constant (8.314 J mol⁻¹ K⁻¹), T is the test temperature, I is the test current, t is the test time, n is the number of electrons transferred per O₂ molecule generated (n = 4), and F is the Faraday constant (96485 C/mol).

Workflow for OER-only potential window evaluation

A simple workflow is conducted to evaluate the OER-only potential window for seawater electrolysis. DPD spectrophotometry was used to detect the ClO⁻ formation for checking the OER selectivity towards CER in Cl⁻ involved conditions. More specifically, a commercial DPD residual chlorine tablet was used to conduct the corresponding measurement. In order to accumulate enough ClO⁻ ions for detection, 10 h OER measurement using target material as working electrode was carried out before taking the testing electrolyte (5 mL from 40 mL) from the electrochemical cell. The testing electrolyte was then mixed well with about 5 mL 0.5 m H₂SO₄ to adjust the pH of solution to 7. Then the solution after standing was put into the DPD tablet and the mixed solution was left to stand for 30 min, further characterized by UV–Vis spectrum with absorbance measured in a wavelength range of 300–800 nm. If ClO⁻ ions are present in the solution, the solution would change from colorless to pink, and the absorption peaks would appear around 515 nm and 551 nm. For comparison, a control experiment was conducted using a 1 m KOH + 0.5 m NaCl solution (pH adjusted to 7) via DPD testing, which was denoted as electrolyte in the Figs. 4b and S17.

Computation details

All spin-polarized density functional theory (DFT) calculations were carried out through the Vienna ab initio simulation package (VASP). The Perdew-Burke-Ernzerhof (PBE) functional together with the generalized gradient approximation was used for describing the exchange-correlation interactions among electrons, and the projector-augmented wave pseudopotentials were used to describe the electron-ion interactions. A uniform cutoff energy of 600 eV was applied, with the energy and force convergence criteria set to 10⁻⁵eV and 0.02 eV/Å. Hubbard U corrections were applied to Ni (6.6 eV) and Fe (3.5 eV) species⁴³. K-point sampling followed a K-spacing parameter of 0.04. All slab structures contained a 15 Å vacuum layer to eliminate periodic-image effects. The dipolar corrections were considered, and the long-range weak van der Waals interactions were corrected using the DFT-D3 approach. The symmetry of the slab models was all switched off. All presentation images of the theoretical models were rendered using the VESTA software⁴⁴.

The oxygen vacancy formation energy (ΔE_o-form) was calculated by following equation:

$$\mathrm{\Delta }{E}_{\mathrm{o}-\mathrm{form}}=E_{\mathrm{vac}}+1/2E_{\mathrm{O}2}-E_{\mathrm{per}}$$ 12

where E_vac and E_per are the total energies of structure with and without one oxygen vacancy, respectively. E_O2 is the energy of the O₂ molecule.

The orbital overlap ratio (H_pd) presented in this work quantifies the extent of p-d orbital hybridization between the Fe/Ni d orbitals of Fe/Ni active sites in NiFeOOH and the O p orbital of the adsorbed oxygen-containing intermediates. Specifically, it is defined as the fraction of the p-d orbital overlap relative to the total p orbital population, which can be calculated using the following equation:

$$H_{pd}=\frac{\int \min \left({\rho }_p,{\rho }_d\right)d\epsilon }{\int {\rho }_{p}d\epsilon }$$ 13

where H_pd represents the orbital overlap ratio, ρ_d and ρ_p denote the d- and p-projected densities of states, and ε is the energy relative to the Fermi level.

Calculation methods for OER and CER

We adopted the general adsorbate evolution mechanism to evaluate the OER activity. The corresponding reaction Gibbs free energies for each step are given as:

$$\mathrm{\Delta }{G}_1=E\left(\mathrm{H}\mathrm{O}*\right)-E\left(*\right)-E_{\mathrm{H}2\mathrm{O}}+1/2E_{\mathrm{H}2}+{\left(\mathrm{\Delta }ZPE-T\mathrm{\Delta }S\right)}_1-eU$$ 14

$$\mathrm{\Delta }{G}_2=E\left(\mathrm{O}*\right)-E\left(\mathrm{H}\mathrm{O}*\right)+1/2E_{\mathrm{H}2}+{\left(\mathrm{\Delta }ZPE-T\mathrm{\Delta }S\right)}_2-eU$$ 15

$$\mathrm{\Delta }{G}_3=E\left(\mathrm{HOO}*\right)-E\left(\mathrm{O}*\right)-E_{\mathrm{H}2\mathrm{O}}+1/2E_{\mathrm{H}2}+{\left(\mathrm{\Delta }ZPE-T\mathrm{\Delta }S\right)}_3-eU$$ 16

$$\mathrm{\Delta }{G}_4=E\left(*\right)-E\left(\mathrm{HOO}*\right)+E_{\mathrm{O}2}+1/2E_{\mathrm{H}2}+{\left(\mathrm{\Delta }ZPE-T\mathrm{\Delta }S\right)}_4-eU$$ 17

where E(*), E(HO*), E(O*), E(HOO*) denote the total energy of the clean surface and the intermediate-adsorbed surface, respectively. E_H2O, E_H2 and E_O2 are the computed energies for the H₂O, H₂, and O₂ molecules, respectively. Zero-point energy corrections (ΔZPE) were derived from the computed vibrational frequencies, and the bias term (−eU) introduces the influence of the external potential U. Instead of calculating O₂ by DFT directly, the free energy change for H₂O → 1/2O₂ + H₂ was set at the experimental value of 2.46 eV per H₂O molecule.

Due to the fact that our electrochemical characterizations for water electrolysis are conducted under alkaline conditions, two Cl⁻ oxidation reactions, including CER and hypochlorite OCl⁻ formation (hypoCER) are considered, which are both two-electron-transfer process. It should be noted that CER is pH-independent and hypoCER is pH-dependent. The DFT calculations were carried out based on the Volmer-Heyrovsky mechanism, which is outlined below:

Volmer step (CER):

$$\mathrm{M}+\mathrm{C}{\mathrm{l}}^{-}\to \mathrm{M}-\mathrm{Cl}+{\mathrm{e}}^{-}$$ 18

Heyrovsky step (CER):

$$\mathrm{M}-\mathrm{Cl}+\mathrm{C}{\mathrm{l}}^{-}\to \mathrm{M}+\mathrm{C}{\mathrm{l}}_2+{\mathrm{e}}^{-}$$ 19

Volmer step (hypoCER):

$$\mathrm{M}+\mathrm{C}{\mathrm{l}}^{-}\to \mathrm{M}-\mathrm{Cl}+{\mathrm{e}}^{-}$$ 20

Heyrovsky step (hypoCER):

$$\mathrm{M}-\mathrm{Cl}+{\mathrm{H}}_2\mathrm{O}\to \mathrm{M}+\mathrm{Cl}{\mathrm{O}}^{-}+2{\mathrm{H}}^{+}+{\mathrm{e}}^{-}$$ 21

where M represents the catalytic active sites on the surface of catalysts, M can be either metal sites or oxygen sites. For hypoCER, hypochlorous acid (HOCl) are considered to avoid computational errors for direct OCl⁻ calculation⁴⁵. Based on computational standard hydrogen electrode (CHE), the equilibrium state of Cl₂ and HOCl formations can be expressed as:

$$1/2{\mathrm{Cl}}_2+{\mathrm{e}}^{-}\rightleftharpoons \mathrm{C}{\mathrm{l}}^{-}\left(\mathrm{aq}\right)+1.36\mathrm{eV}$$ 22

$$\mathrm{HClO}+{\mathrm{H}}^{+}+{\mathrm{e}}^{-}\rightleftharpoons 1/2\mathrm{C}{\mathrm{l}}_2+{\mathrm{H}}_2\mathrm{O}+1.61\mathrm{eV}$$ 23

Accordingly, the reaction free energy of each step could be expressed as follows:

As for CER:

$$\mathrm{\Delta }{G}_1=E\left(\mathrm{C}\mathrm{l}*\right)-E\left(*\right)-1/2E_{\mathrm{Cl}2}+{\left(\mathrm{\Delta }ZPE-T\mathrm{\Delta }S\right)}_1+1.36-eU$$ 24

$$\mathrm{\Delta }{G}_2=E\left(*\right)-E\left(\mathrm{Cl}*\right)+1/2E_{\mathrm{Cl}2}+{\left(\mathrm{\Delta }ZPE-T\mathrm{\Delta }S\right)}_2+1.36-eU$$ 25

As for hypoCER:

$$\mathrm{\Delta }{G}_1=E\left(\mathrm{C}\mathrm{l}*\right)-E\left(*\right)-1/2E_{\mathrm{C}\mathrm{l}2}+{\left(\mathrm{\Delta }ZPE-T\mathrm{\Delta }S\right)}_1+1.36-eU$$ 26

$$\mathrm{\Delta }{G}_2=E\left(*\right)-E\left(\mathrm{C}\mathrm{l}*\right)+1/2E_{\mathrm{Cl}2}+{\left(\mathrm{\Delta }ZPE-T\mathrm{\Delta }S\right)}_2+1.61-eU$$ 27

where E(*), E(Cl*) are the total energy of the clean surface and the intermediate-adsorbed surface with intermediate, respectively. E_Cl2 is the computed energies for the Cl₂ molecules. Zero-point energy corrections (ΔZPE) were derived from the computed vibrational frequencies, and the bias term (−eU) introduces the influence of the external potential U.

Author contributions

S.L., X.Zou, and Y.L. conceived the idea and directed the project. X.Zhang performed the experiments and analyzed the results. Q.H. and X.S. supported the in situ experiments. X.L. and X.B. provided advice about the experimental analysis. T.L. and Y.L. conducted the theoretical computation. C.-Z.W. provides guidance on the theoretical calculation. X.Zhang, T.L., and Y.L. co-wrote the manuscript. All authors reviewed and edited the manuscript.

Peer review

Peer review information

Nature Communications thanks Jianyong Feng, P. Tsiakaras and the other anonymous reviewer(s) for their contribution to the peer review of this work. A peer review file is available.

Data availability

Source Data file has been deposited in Figshare under accession code DOI link⁴⁶. All other relevant data that support the findings of this study can be obtained from the corresponding authors upon request.

Competing interests

The authors declare no competing interests.

Supplementary information

The online version contains supplementary material available at 10.1038/s41467-025-67439-4.