Introduction

The design of active, stable, and cost-effective catalysts for the oxygen evolution reaction (OER) is critical to improving the full process efficiency in electrocatalytic systems, such as water splitting and CO2/CO reduction reaction (CO2/CORR) coupled with OER1,2,3,4. In these systems, approaches that use alkaline electrolytes have the benefit of increasing efficiency toward CO2RR and CORR by suppressing the competing hydrogen evolution reaction (HER).

While there have been improvements in the activity of metal phosphides, sulfides, selenides, and borides/borates5,6,7,8, achieving high electrocatalytic activity (>1 A/cm2) united with long-term stability (>100 h) remains an unmet need9,10. The lack of highly active and stable OER catalysts limits the performance and stability of electrosynthesis processes that rely on an OER anode (i.e., CO2RR/CORR electrolysers) to production rates <100 mA/cm2 and durations of <100 h. It remains an important task for the field to increase catalytic activity while achieving long-term durable and earth-abundant OER catalysts that operate at alkaline pH.

Metal borides/borates exhibit promising OER performance in high-pH regimes: incorporating highly electronegative boron diminishes the metal oxidation reaction energy barrier under a positive bias and assists with charge transfer11,12. To date, the durability of metal borides has been limited to <60 h at a modest current density of ~20 mA/cm2,13,14,15. Thus far, to improve the activity and stability, significant efforts have been invested in charge-transfer enhancement and electronic structure optimization; however, we found herein that metal borides are prone to convert into metal borate under OER conditions, and thus to behave as the operando catalyst16,17,18.

This knowledge prompted us to design metal borides as precursors to construct, in-situ, the metal borates that would endow the catalysts with activity and stability.

We focused on a NiFe-Boride compound and probed its phase transformation using operando X-ray spectroscopy during OER. We witnessed the in-situ transformation of NiFe-Boride to NiB4O7 and FeBO3 phases on the catalyst surface during electrochemical activation. These results accord with DFT calculations which reveal that NiB4O7 species facilitate the step *O - > *OOH and enhance electrocatalytic activity toward OER.

Experimentally, we achieve an overpotential of 167 mV at 10 mA/cm2 and stable water splitting performance of 400 h at 1 A/cm2. Encouraged by its high activity and stability, we implement NiFe-Boride as the OER catalyst in a CORR membrane-electrode assembly (MEA) electrolyser that uses CO and H2O to produce C2H419. The CORR electrolyser based on the NiFe-Boride OER catalyst exhibited stable performance at 200 mA/cm2 for over 80 h—maintaining a C2H4 energy efficiency of 19%—whereas this value is 17% for the CORR electrolyser that relies on a noble IrO2 OER catalyst.

Results

Insight from computational studies into stable and active phases

Noting that metal borides undergo phase transformation at highly oxidative potentials, we sought to calculate Pourbaix diagrams to evaluate the phase composition of NiFe-Boride ternary system at high potentials in alkaline electrolytes. Compared to the Pourbaix diagram of NiFe (Fig. 1a), we observe two new stable phases, FeBO3 and NiB4O7, under a working potential (U = 1.4—2 V vs. RHE) at electrolyte pH (pH = 10—14) of alkaline OER in the Pourbaix diagram of NiFe-Boride (Fig. 1b), indicating that boron ions promote the new phases formation. The atomic configurations of these two new phases are shown in the Fig. 1c.

Fig. 1: Results of DFT calculations for NiFe-Boride catalysts.
figure 1

Pourbaix diagram of (a) NiFe and (b) NiFe-Boride; (c) The cell configurations of NiB4O7 and FeBO3; (d) The schematic illustration of the 4-step OER pathway; (e) Predicted OER reaction energy diagram for NiB4O7, FeBO3, γ-NiOOH, and γ-NiFeOOH in the alkaline electrolyte at 1.23 V vs. RHE.

Full size image

We carried out density functional theory (DFT) calculations to further explore the effects of these in-situ formed phases on the OER activity of NiFe catalyst. We first calculated the reaction energy of elementary steps and the overpotential for the OER based on the 4e-mechanism proposed by Nørskov et al. for water oxidation20,21,22 with the reaction pathway:

$$\ast +{{{{{{\rm{OH}}}}}}}^{-}\to \ast {{{{{\rm{OH}}}}}}+{e}^{-}$$
(1)
$$\ast {{{{{\rm{OH}}}}}}+{{{{{{\rm{OH}}}}}}}^{-}\to \ast {{{{{\rm{O}}}}}}+{{{{{{\rm{H}}}}}}}_{2}{{{{{\rm{O}}}}}}+{e}^{-}$$
(2)
$$\ast {{{{{\rm{O}}}}}}+{{{{{{\rm{OH}}}}}}}^{-}\to \ast {{{{{\rm{OOH}}}}}}+{{{{{{\rm{e}}}}}}}^{-}$$
(3)
$$\ast {{{{{\rm{OOH}}}}}}+{{{{{{\rm{OH}}}}}}}^{-}\to {{{{{{\rm{O}}}}}}}_{2}+{{{{{{\rm{H}}}}}}}_{2}{{{{{\rm{O}}}}}}+{{{{{{\rm{e}}}}}}}^{-}$$
(4)

We cleaved the (100) facet of these two new phases to study the adsorption energy of various intermediates for OER, and thus to determine the theoretical overpotentials as an indicator for catalytic activity (Fig. 1d and Supplementary Fig. 1). The largest reaction energy (Step 2, deprotonation of *OH) was identified as the potential-determining step (PDS)23, as shown in Fig. 1e, and the reaction energy was similar for both FeBO3 (0.53 V) and NiOOH (0.56 V) at 1.23 V. In contrast, NiB4O7 shows a thermodynamic overpotential of 0.43 V, lowest among NiOOH, NiFeOOH and FeBO3, which suggests that NiB4O7 phases are the active sites in NiFe-Boride that enable superior catalytic OER performance. We observed that the NiB4O7 shifts the PDS to the *OOH formation step at 1.23 V. We also obtained the similar calculation results at 1.40 V (Fig. 1e and Supplementary Fig. 2).

To account for other possible reaction mechanisms, we further studied the intramolecular oxygen coupling mechanism24: we found that Ni and Fe borate have increased activity compared to Ni and Fe (oxy)hydroxide (Supplementary Fig. 3). We also considered more sophisticated activity descriptors for multiple-electron processes, exploring the use of electrochemical-step symmetry index (ESSI) and the PDS at zero overpotential (Gmax)25,26,27: these results indicated higher OER activities of these NiFe-borate phases (Supplementary Table 1). Taken together, these results indicate that boron addition enables formation of new active phases, especially the Ni phase, in the pristine framework, thereby enhancing electrocatalytic activity toward OER.

Catalyst synthesis and characterization

To test these hypotheses experimentally, we synthesized NiFe-Boride catalysts by a chemical reduction method (see Methods for details). Transmission electron microscopy (TEM), scanning electron microscopy (SEM), and X-ray diffraction (XRD) were conducted to investigate the structure, composition, and morphology of the NiFe-Boride catalyst. TEM and SEM images (Supplementary Figs. 4 and 5) revealed that catalysts the consisted of agglomerated nanoparticles. Elemental mapping analysis indicated that Ni, Fe, B, and O elements distribute homogeneously, without obvious separation or aggregation (Fig. 2a and Supplementary Fig. 6). A broad XRD diffraction peak between 40° and 50° indicates low crystallinity in the catalyst (Fig. 2b).

Fig. 2: Characterization of the morphology and composition of NiFe-Boride catalyst.
figure 2

(a) Elemental mapping analysis, showing the uniform, uncorrelated spatial distribution of Ni, Fe, B and O. (b) XRD patterns. (c) B 1 s XPS spectrum of fresh NiFe-Boride sample. The Ni and Fe K-edge XANES (d) and (e) and EXAFS (f) and (g) data for fresh NiFe-Boride and control samples.

Full size image

We further carried out X-ray photoelectron spectroscopy (XPS) to study the chemical state of the NiFe-Boride catalyst. XPS survey spectra show the presence of Ni, Fe, B, and O, which agrees well with the elemental mapping results (Supplementary Fig. 7). The B 1 s spectrum obtained from the pristine sample was comprised of two peaks located at 192 eV and 187 eV, which are assigned to B-O and B0/B-M species (Fig. 2c), respectively, indicating boride. The Ni and Fe K-edge X-ray absorption near-edge structure (XANES, Fig. 2d, e) revealed that NiFe-Boride possesses an electronic structure that differs from that of the control samples, indicating that the incorporation of boride leads to phase and structural changes. To probe the origins of these changes, we obtained a Fourier transform (FT) of the extended X-ray absorption fine structure (EXAFS, Fig. 2f, g) spectrum. NiFe-Boride showed two FT features: (1) the peak at ~1.5 Å represents the single-scattering path of M-O and (2) the peak at ~2.5 Å results from the single-scattering path of the closest neighboring transition metals around the metal cations (M-M) (the locations of the peaks match the locations of those in the NiFe control sample)28. The altered electronic structure suggests that the boride ions were incorporated into the NiFe framework via the formation of BOx and NiFeBx species. Further investigations of the M-B peak remain a topic of interest.

Using inductively coupled plasma optical emission spectrometry (ICP-OES) analysis, we determined the molar ratio of Ni:Fe to be 1:1. Further analysis of the as-deposited system with the aid of energy-dispersive X-ray spectroscopy (EDS) indicated a 1:1 atomic ratio of Ni:Fe (Supplementary Fig. 8).

In-situ and ex-situ characterization

Using in-situ synchrotron radiation X-ray diffraction (SRXRD), we tracked changes in the crystal structure of catalysts during the OER process. We found that, prior to OER, the phase structure of the catalyst was NiFeBO4 (see Supplementary Fig. 9 for the model structure). We found that FeBO3 and NiB4O7 were formed under an applied potential (Fig. 3a, the standard patterns are provided in the Supplementary Fig. 10), consistent with in-situ formation of new phases in the NiFe-Boride catalyst during OER. We then carried out in-situ XAS to monitor the local electronic structure changes of the metal sites during the OER. The Ni and Fe K-edge XANES (Fig. 3b, c) spectra exhibited a positive energy shift, i.e., the valence states of Ni and Fe increased with the applied potential29. We analyzed the absorption edge energies in order to characterize the Ni/Fe valence states. Before OER, the valence states of Ni and Fe in NiFe-Boride were 1.91 and 2.73, respectively. During OER, the valence states of Ni and Fe increased to 1.98 and 2.95, respectively (Supplementary Figs. 11–13). The ex-situ L-edge XAS results were in good agreement with this change: the reacted catalyst – compared to the pristine one – had a higher Ni3+ ratio and a similar Fe2+/3+ ratio. The similarity in Fe2+/3+ ratios indicates that Fe ions return to their initial states upon release of the applied potential. These results accord with the higher Ni3+ ratio in the reacted catalyst observed via O K-edge XAS (Supplementary Fig. 14).

Fig. 3: Ex/in-situ characterization of as-prepared NiFe-Boride catalyst.
figure 3

a In-situ SRXRD patterns of the catalyst during the OER test in 1 M KOH electrolyte; The in-situ Ni and Fe K-edge XANES (b) and (c) and EXAFS (d) and (e) data of NiFe-Boride during the LSV process in 1 M KOH aqueous electrolyte. The in-situ applied voltages before and during the OER are 1.2 and 1.4 V vs. RHE, respectively. f Ni 2p and (g) Fe 2p XPS spectra for fresh and post-OER samples.

Full size image

We further analyzed the EXAFS spectra (Fig. 3d, e) to clarify the local structural variations associated with the Ni and Fe sites30,31. Under open circuit potential (OCP) in KOH electrolyte, the FT peaks demonstrated features similar to those obtained from the as-prepared catalysts. Once a positive bias was applied, the progressive FT peaks show that the original phase accomplishes some structural transformations. We conducted EXAFS fitting (Supplementary Tables 2–3 and Supplementary Figs. 15, 16) to determine the bond lengths of M-O and M-M (M is Ni or Fe). The interatomic distances of Ni-O and Ni-Ni(Fe) in as-prepared NiFe-Boride are 2.04 Å and 3.09 Å, respectively, while those measured during OER are 2.02 Å and 2.98 Å. The interatomic distances of Fe-O and Fe-Fe(Ni) in as-prepared NiFe-Boride are 2.23 Å and 3.22 Å, while those measured during OER are 1.92 and 3.05 Å. As oxidation and phase transformation progress, the M-O and M-M bonds become shorter than those at OCP. These results are in line with the generation of transition metal active sites with high valence29. The electronic structure of NiFe-Boride is different than that of Ni-Boride and Fe-Boride, indicating the presence of electronic communication between the Ni and Fe in M-M bonds (Supplementary Fig. 17).

We carried out XPS to study the local evolution of chemical state and electronic structure before vs. after OER, with all peaks calibrated using C 1 s (Supplementary Fig. 18). High-resolution XPS in Ni 2p region showed two main peaks of Ni2+ 2p3/2 and Ni2+ 2p1/2 located at 855.4 and 873.3 eV (Fig. 3f), accompanied by satellite peaks centered at 861.2 and 879.4 eV32. In the Fe 2p region, two main peaks located at 710.7 and 724.6 eV were assigned to Fe 2p3/2 and Fe 2p1/2 (Fig. 3g). We noted minor peaks centered at ~851.9 and 706.1 eV in the Ni 2p and Fe 2p spectra of fresh catalyst, suggesting the existence of Ni-B and Fe-B bonds in NiFe-Boride (Fig. 3f, g)33, consistent with EXAFS. The B 1 s spectrum consists of two peaks corresponding to B-O and B0/B-M species. Upon completion of electrochemical activation during OER, negative chemical shifts in the binding energy of Ni 2p and Fe 2p peaks were observed compared to pristine NiFe-Boride. These negative shifts indicated the modulation of the surface electronic structure through the oxidation process, in accordance with the positive shift in the peak position of B-O bonds (Supplementary Fig. 19). Harsh oxidative conditions during OER converted the M-B species into oxides and hydroxides, evidenced by the disappearance of Ni-B, Fe-B, and B0/B-M signals.

In sum, NiFe-Boride acts as a precursor that phase transforms at highly oxidative potentials, leading to the formation of stable FeBO3 and NiB4O7 phases during the OER process.

Electrocatalytic performance

The OER performance of NiFe-Boride and controls on Ni foam was examined in 1 M KOH electrolyte using a three-electrode cell at room temperature (see Methods). As shown in linear sweep voltammetry (LSV, Fig. 4a) curves, the overpotential of NiFe-Boride catalyst at 10 mA/cm2,34 and without iR compensation is 167 mV. This is lower than the 200 mV and 189 mV obtained for sol-gel NiFe and IrO2, respectively (Supplementary Table 4). Rotating ring disk measurements show a similar trend: NiFe-Boride enables higher-rate water oxidation performance than in control catalysts (Supplementary Fig. 20). To seek insight into the effect of in-situ active sites on performance, we compared the catalytic activity between annealed and pristine catalysts. The annealed catalysts have lower performance, consistent with an activity-promoter role for the in-situ formed active sites (Supplementary Figs. 21-22). A Ni:Fe ratio of 1:1 (Supplementary Fig. 23 and Table 5) leads to the lowest overpotential. We determined the Tafel slope (Fig. 4b), finding 25 mV/dec for the NiFe-Boride catalyst, lower than for IrO2 (41 mV/dec) and NiFe (66 mV/dec). The lower Tafel slope value indicating rate-determining step is shifted from the *O step to the *OOH formation step35,36. Compared to benchmark OER catalysts from literature, NiFe-Boride shows higher catalytic performance in terms of overpotential at 10 mA/cm2 and Tafel slope in the alkaline solution (Fig. 4c). From Nyquist plots based on electrochemical impedance spectroscopy (EIS, Fig. 4d) at applied potential 1.45 V vs. RHE reveal that NiFe-Boride exhibits a lower charge-transfer resistance (Rct) than does either NiFe or IrO2.

Fig. 4: Performance of NiFe-Boride catalyst and controls in a three-electrode configuration in 1M KOH aqueous electrolyte.
figure 4

a OER LSV polarization curves for catalysts loaded on Ni foam without iR correction. b The corresponding Tafel plot of catalysts. c Comparison of Tafel slope and overpotential required to achieve 10 mA/cm2, with references all measured in alkaline medium47,48,49,50,51,52,53,54,55,56,57. d EIS data for NiFe-Boride and controls in three-electrode configuration. The data were collected at 1.45 V vs. RHE. The inset provides the equivalent circuit: Rs series resistance, Rct charge-transfer resistance, and CPE constant-phase element related to the double-layer capacitance. e Chronopotentiometric curves obtained from the NiFe-Boride catalyst on Ni foam electrode at a constant current density of 20 mA/cm2. f Comparison of stability at different current density, with references all measured in alkaline medium13,14,15,33,58. g Operating voltage and O2 FE at constant 1 A/cm2 current density in a three-electrode configuration in 1 M KOH aqueous electrolyte. h The operating voltage and ethylene FE were monitored at constant 200 mA/cm2 in a membrane-electrode assembly device. NiFe-Boride and IrO2 supported on titanium felt were used as the anodes. The high surface area Cu catalyst on hydrophobic carbon paper acted as cathode. Humidified CO was flowed through the gas channels in the cathode, and 2 M aqueous KOH solution was flowed through channels in the anode.

Full size image

We then used electrochemically active surface area (ECSA) obtained via the double-layer capacitance (Cdl, Supplementary Fig. 24) technique to estimate a normalized current density37. This for NiFe-Boride is 7.5 mA/cm2, >2x greater than that in IrO2, and 4x greater than in NiFe catalysts, at 1.45 V vs. RHE (Supplementary Fig. 25). IR-corrected LSV excludes the effect of ionic conductivity on performance: the NiFe-Boride catalyst showed an iR-corrected overpotential of 160 mV at 10 mA/cm2 on Ni foam (Supplementary Fig. 26), which is 35 and 23 mV lower than for NiFe and IrO2.

We measured the stability of the best-performing NiFe-Boride catalyst at 20 mA/cm2 and found that the increase in potential was less than 10 mV after 100 h of continuous water splitting (Fig. 4e), more stable than the control and reference samples (Fig. 4f). We obtained stability at 1 A/cm2, finding that catalysts enable stable operation of 400 h with an overpotential of 460 mV without degradation (Fig. 4g). B 1 s XPS results indicate that the B concentration gradually decreases over the first 100 h but then remains constant over the ensuing 300 h (Supplementary Fig. 27). The Faradaic efficiency (FE) toward O2 remains 98 ± 2% over the first 400 h, excluding a major current contribution associated with materials corrosion.

To investigate further the potential relevance of these catalysts to future systems, we implemented NiFe-Boride as the OER catalyst in a CORR MEA electrolyser that produces C2H4 from CO and H2O (Supplementary Fig. 28). The CORR electrolyser exhibited stable performance at 200 mA/cm2 for over 80 h, maintaining a cell potential of −2.85 V, a C2H4 FE of 45%, and a C2H4 energy efficiency of 19%. Under similar operating conditions, the cell potential for the IrO2 OER catalyst-based electrolyser was −3.06 V, which is ~200 mV higher than in the NiFe-Boride OER catalyst-based system (Fig. 4h and Supplementary Fig. 29), corresponding to a C2H4 energy efficiency of 17%. Using XPS and EDS line scanning, we found that Ni, Fe, and B elemental signals remained following the stability study (Supplementary Figs. 30 and 31)., i.e., the borate remained in the NiFe.

Discussion

Starting from DFT calculations, we saw that the phase transformation of NiFe-Boride to NiFe-Borate can be activated under OER potentials in OER. The in-situ formed new phase (NiB4O7) can function as active site to facilitate *O - > *OOH, enhancing the electrocatalytic activity of NiFe-Boride in OER. We evaluated the electrochemical performance of the catalyst and found that NiFe-Boride possesses a Tafel slope of 25 mV/dec and an overpotential of 167 mV at current density 10 mA/cm2 in 1 M KOH solution, outperforming benchmark NiFe oxide. The in-situ formed NiFe-Borate phases are confirmed via operando spectroscopies. The catalyst demonstrates stable performance over 400 h at 1 A/cm2 in water splitting. We further showcased 80 h of stable C2H4 electrosynthesis using a NiFe-Boride OER catalyst in a CORR MEA electrolyser, with a ~200 mV lower cell potential compared to MEA electrolysers based IrO2 OER catalysts.

Methods

I.1. Chemicals

Iron (III) chloride (FeCl3), nickel (II) chloride hexahygrate (NiCl2·6H2O), Sodium borohydride (NaBH4), ethanol (≥99.5%) and acetone (≥99 %) were purchased from Sigma-Aldrich. Iridium (IV) oxide dehydrate (99.99%) was purchased from Alfa Aesar®. All chemicals were used without further purification.

I.2. Synthesis of NiFe-Boride and control catalysts

Th NiFe-Boride catalysts were synthesized using a modified chemical reduction method. NiCl2·6H2O (332.8 mg) and anhydrous FeCl3 (227.1 mg) were first dissolved in deionized water (DI, 2 mL) in a vial. A solution of NaBH4 (2 mL, 5 M) in water was prepared in a separate vial. All solutions mentioned above were cooled in an ice bath for 10 min to prevent uncontrolled hydrolysis and consideration which may lead to the formation of the precipitate. The Ni and Fe precursors were then added dropwise to NaBH4 to form a black solution. Finally, the prepared solution was further washed and centrifuged with water and ethanol three times before vacuum drying. The samples of different Ni:Fe ratios were synthesized by changing the Ni and Fe precursors concentration.

NiFe catalysts were synthesized using a modified aqueous sol-gel technique. NiCl2·6H2O (2.1 mmol) and anhydrous FeCl3 (0.35 mmol) were first dissolved in ethanol (2 mL) in a vial. A solution of deionized water (DI) (0.18 mL) in ethanol (2 mL) was prepared in a separate vial. All solutions mentioned above were cooled in an ice bath for 2 h to prevent uncontrolled hydrolysis and consideration, which may lead to the formation of precipitate rather than gel formation. The Ni and Fe precursors were then mixed with an ethanol-water mixture to form a clear solution. Propylene oxide (≈ 1 mL) was then slowly added to form a gel. Finally, the prepared NiFe wet-gel was aged for 24 h to promote network formation and immersed in acetone for 5 days before the gel was vacuum dried.

I.3. Characterization

The morphologies of the catalysts were characterized by SEM (Hitachi S-4800) imaging, and energy-dispersive X-ray spectroscopy mapping operated at 1.5 kV was used to characterize the surfaces of all samples. TEM images and corresponding elemental mapping analysis were conducted using a JEM-2100F. The electronic structure measurements were carried out using XPS (ESCALAB 250Xi). To determine the best Ni/Fe ratio in the NiFe-Boride system, we first dissolved 50 mg of the catalyst into 5 ml aqua regia through microwave digestion. Then, we took out 1 ml of sample and diluted it to 10 mL to test ICP-OES (5110).

In-situ XAS was carried out in the same condition as electrocatalytic characterization case using a modified single cell with an opening sealed by Kapton tape. The XAS signal was collected by Vortex detector in 9BM beamline of Advanced Photon Source (APS) in Argonne National Laboratory. X-ray absorption near-edge spectra (XANES) of Ni and Fe K-edge were collected by silicon drift detector at ambient air in 9BM beamline of Advanced Photon Source (APS) located in the Argonne National Laboratory.

In-situ SRXRD measurements on NiFe-Boride in a typical three-electrode setup as the same condition in electrochemical characterization case were conducted at beamline 01C2 at National Synchrotron Radiation Research Center using a specially designed acrylic container with a 1 mL of gap, sealed by Kepton tape. 16 keV of the incident X-ray (λ = 0.7749 Å) was allowed to perpendicularly transmit through the sample and electrolyte, so that the signal of synchrotron X-ray diffraction of transmission mode would be collected in a large Debye-Scherrer camera and processed using the GSAS-II software 45 to obtain plots of intensity versus 2θ.

I.4. Catalysts loaded on the electrodes

To load the catalyst on the Ni foam (thickness: 1.6 mm, Sigma), 20 mg of catalyst was dispersed in a mixture consisting of 2 mL water and 2 mL ethanol, followed by the addition of 100 μL Nafion solution. The suspension was sonicated for 30 min to prepare a homogeneous ink. Ni foam with a fixed area of 0.5 × 0.5 cm2 coated with water-resistant silicone glue was drop-casted with 20 μL of the catalyst ink. Ag/AgCl (with saturated KCl as the filling solution) and platinum foil were used as the reference and counter electrodes, respectively.

I.5. Electrochemical characterization

Electrochemical measurements were performed using a three-electrode system connected to an electrochemical workstation (Autolab PGSTAT302N), equipped with built-in EIS analyzer. The working electrode was Ni foam (thickness: 1.6 mm, Sigma), which was prepared by dispersing 20 mg of catalyst in a mixture containing 1 mL water and 1 mL ethanol, followed by the addition of 100 μL Nafion solution. The suspension was sonicated for 30 min to prepare a homogeneous ink. Cyclic voltammetry (CV) measurements at 50 mV/s were performed prior to recording LSV at 1 mV/s for each sample. EIS measurements were conducted in a static solution at 1.45 V (vs. RHE). The amplitude of the sinusoidal wave was 10 mV, and the frequency scan range was between 100 kHz and 0.01 Hz. All experiments were performed at ambient temperature (23 ± 2 °C), and electrode potential was converted to the RHE scale using Eq. (5)

$$E\left({vs}.{RHE}\right)=E({vs}.{Ag}/{AgCl})+{E}_{{Ag}/{AgCl}}({vs}.{NHE})+0.059\times {pH}$$
(5)

MEA test

The MEA cell is a complete CO electrolyser (Dioxide Materials) consisted of a titanium anode flow field, 904 L Stainless Steel cathode flow field, and associated nuts, bolts and insulating kit. The geometric area of each flow field is 5 cm2, of which 45% is the channel while the rest 55% is the land area. The catalyst electrode was attached on the cathode by a copper tape, which was protected by insulating Kapton tapes to avoid electrical contact with membranes or electrolytes. A Sustainion® AEM membrane (Fumasep) was activated in 1 M aqueous KOH solution for 24 h, washed with water prior to use. The anode consisted of NiFe-Boride supported on titanium felt was prepared by a dip-coating and thermal decomposition method. The cathode is the high surface area Cu catalyst on hydrophobic carbon paper. Aqueous 2 M KOH electrolyte was used as the anolyte and was circulated using a peristaltic pump. The flow rate of the anolyte was kept at 20 mL/min. The flow rate of the CO gas flowing into the cathodic gas chamber was kept at 80 sccm by a mass flow controller. The CORR was initiated by applying a constant current density of 200 mA/cm2, and the current density was kept constant over the course of electrolysis.

IR correction

All the polarization curves on different supports were corrected for ohmic losses (including the wiring, substrate, catalyst, and solution resistances). The iR-corrected data were given by the following equation:

$${E}_{{Co}{rrected}}=E-{iR}$$
(6)

where R is the series resistance of measurement, which can be obtained from an EIS Nyquist plot as the first intercept of the main arc with the axis.

ECSA

The ECSA of each catalyst was determined by measuring the electrochemical double-layer capacitances (Cdl) from the scan rate CV-dependence plot. The CV cycle potential window was between 0.7 and 0.8 V vs. RHE, and the scan rates were 20, 40, 60, 80, 100, and 120 mV/s. The Cdl was estimated \(\triangle j=\frac{1}{2}(j_{{charge}}-{J}_{{off\; charge}})\) at the average potential in the selected range against the scan rates. The slope of the linear fit was calculated as the Cdl. A specific capacitance of 40 μF/cm2 was used here. The ECSA of the catalyst was calculated by using the following equation,

$${ECSA}=\frac{{C}_{{dl}}}{40\mu F/{{cm}}^{2}}{{cm}}_{{ECSA}}^{2}$$
(7)

DFT calculations

We performed the DFT calculations using the Vienna Ab Initio Simulation Package (VASP) code38,39. The exchange-correlation energy was modelled using Perdew-Burke-Ernzerhof (PBE) functional40 within the generalized gradient approximation (GGA). The projector augmented-wave (PAW) pseudo-potentials41 were used to describe the ionic cores. The cut-off energy of 450 eV was adopted after a series of tests. A Methfessel–Paxton smearing of 0.05 eV to the orbital occupation was applied for the geometry optimization and total energy computations. In all calculations, atoms at all positions were assumed to possess Hellmann-Feynman forces lower than 0.02 eV/Å, and the electronic iteration convergence criterion was 10-5 eV using the normal algorithm. Pourbaix diagrams were calculated using Atomic Simulation Environment (ASE)42 with input formation energy values from Materials project43. Surface models for NiB4O7 and FeBO3 were simulated by five-layer (2 × 2) NiB4O7 (100) and FeBO3 (100) supercells (Supplementary Fig. 32), accompanied with a sufficient vacuum gap of 15 Å. Structural optimizations were performed on all modified slab models. We considered 4-steps intermediates adsorption to simulate the OER process (details in below & Supplementary Fig. 33). During adsorption calculations, the top two layers were fully relaxed, while the other layers were fixed at the tested lattice position.

The reaction energies (ΔG) were obtained by

$$\varDelta {{{{{\rm{G}}}}}}=\varDelta {{{{{{\rm{E}}}}}}}^{{{{{{\rm{DFT}}}}}}}+\varDelta {{{{{\rm{ZPE}}}}}}-{{{{{\rm{T}}}}}}\varDelta {{{{{\rm{S}}}}}}$$
(8)

where ΔEDFT is the reaction energy calculated from DFT; ΔZPE is the zero-point energy, which was neglected here; ΔS is the change in entropy, and for gaseous molecules ΔS values are obtained from the standard database in the NIST web-book44.

Taking the reaction (1) as an example, the reaction energy is equals to

$$\varDelta {{{{{{\rm{G}}}}}}}^{1}={{{{{\rm{G}}}}}}(\ast {{{{{\rm{OH}}}}}})\mbox{-}{{{{{\rm{G}}}}}}(\ast )\mbox{-}G({{{{{{\rm{OH}}}}}}}^{-})$$
(9)

where G(*OH) is the total energy of *OH adsorption configuration; G(*) is the energy of catalyst surface; G(OH-) is the energy of hydroxyl ion, to calculate this value, we assume the equilibrium

$${H}_{2}{{{{{\rm{O}}}}}}\leftrightarrow {H}^{+}+{{OH}}^{-}$$
(10)

which relates the chemical potentials as

$${\mu }_{{H}^{+}}+{\mu }_{{{OH}}^{-}}={\mu }_{{H}_{2}{{{{{\rm{O}}}}}}}$$
(11)

Then,

$${\mu }_{{H}^{+}}+{\mu }_{{e}^{-}}+{\mu }_{{{OH}}^{-}}-{\mu }_{{e}^{-}}={\mu }_{{H}_{2}{{{{{\rm{O}}}}}}}$$
(12)

Thus,

$${\mu }_{{{OH}}^{-}}-{\mu }_{{e}^{-}}={\mu }_{{H}_{2}{{{{{\rm{O}}}}}}}-({\mu }_{{H}^{+}}+{\mu }_{{e}^{-}})$$
(13)

here, \({\mu }_{{H}^{+}}+{\mu }_{{e}^{-}}\) can be calculated using computational hydrogen electrode (CHE) model developed by Nørskov and co-workers45,46. Then, the theoretical overpotential \(\eta\) for OER can be calculated by

$$\eta ={{{{{\rm{max }}}}}}\left(\Delta {G}^{i}\right)-1.23$$
(14)

\(\Delta {G}^{i}\) stands for the reaction energy for reaction (8) – (11), i = 1, 2, 3, 4.

Electrochemical-step symmetry index (ESSI) and maximum energy Gmax can act as a sophisticated activity descriptors for multiple-electron processes. We calculated ESSI using the following formula,

$${ESSI}=\frac{1}{n}\mathop{\sum }\limits_{k=1}^{n}\left(\frac{\triangle {G}_{k}^{+}}{e}\right)$$
(15)

where \(\triangle {G}_{k}^{+}\) means all positive values of reaction energies, while negative values are not accounted.

The Gmax is the free-energy spanning from intermediate with the smallest free energy to the intermediate with the highest energy during OER.