Colloidal-nanoparticle-derived nickel/ferric oxide heterointerfaces for promoting the alkaline oxygen evolution reaction

Abstract

Heterointerface engineering is an effective strategy to enhance electrocatalytic activity for water splitting by binding two different materials. Specifically, it offers prospects in creating viable transition-metal-based anode materials for catalyzing the oxygen evolution reaction (OER). However, despite its promise, metal/metal oxide heterointerface engineering has not been comprehensively explored for maximizing the electrocatalytic performance, that is, for minimizing the OER overpotential and maintaining the physicochemical characteristics of the catalyst during operation. Therefore, to resolve these issues, the present study was aimed at synthesizing colloidal nanoparticles (NPs) in mixed form at various Ni/Fe ratios (100 : 0 to 0 : 100) to create heterointerfaces between metallic Ni and Fe₂O₃ NPs. Among the prepared NPs, the 75 : 25 sample annealed under H₂/Ar atmosphere recorded the highest alkaline water oxidation activity (199 mV at 10 mA cm⁻²) with long-term durability for five days, indicating that the 75 : 25 ratio was the optimal composition for generating abundant Ni/Fe₂O₃ heterointerfaces. The practical applicability of the 75 : 25 sample was confirmed by its remarkable performance as an anode catalyst for an anion exchange membrane water electrolyzer (AEMWE). The outstanding electrocatalytic performance of the 75 : 25 specimen was induced not only by the interparticle interactions between Ni and Fe₂O₃ NPs, but also by the intraparticle interactions within the heterostructured NPs. Overall, this report illustrates the benefits of Ni/Fe₂O₃ heterointerface engineering in obtaining highly efficient anode catalysts for AEMWEs.

---

1 Introduction

Water-electrolysis-driven power-to-gas technology is an ideal scheme for storing renewable electricity and converting it to H₂ without generating carbon emissions, thereby helping to achieve net-zero emissions to combat the climate crisis.^1–4 Water electrolysis involves two half-reactions: the hydrogen evolution reaction (HER) at the cathode and the oxygen evolution reaction (OER) at the anode. As the OER is more kinetically sluggish than the HER owing to its four-electron mechanism, OER-catalyzing materials have been developed using transition metals under alkaline conditions, thereby imparting low toxicity, long-term durability, and economic feasibility to the cell system.^5–9 However, transition-metal-based OER catalysts require further scrutiny for operation in industrial water electrolysis systems (such as anion exchange membrane water electrolyzers (AEMWEs)) with high activity and stability. Consequently, transition-metal-based anode materials with outstanding alkaline OER performance should be developed for single-cell systems and half-cell reactions.^10,11

Recently, heterostructures of transition metals and their oxides have been found to achieve high alkaline OER performance by modulating the electronic structure.^12–15 Suryanto et al. reported that Ni/Fe oxide heterostructured nanoparticles exhibited outstanding OER activity (210 mV at 10 mA cm⁻²) owing to the strong electronic coupling at Ni/Fe oxide heterointerfaces.¹⁶ Luo et al. synthesized Ni_0.95/Mn_0.05O nanoparticles anchored on carbon nanotubes, and realized an OER overpotential of 293 mV at 10 mA cm⁻² owing to the strong electronic interactions at the heterointerface.¹⁷ Additionally, Ni/NiO_x heterointerfaces have been actively studied as alkaline OER catalysts.^18–21 Notably, in metal/metal oxide heterointerfaces, metallic Ni plays a critical role in promoting OER activity when it is attached to transition metal oxides at the nanoscale.

Heterointerfaces have been created by physical mixing without using complex synthesis procedures. For instance, Görlin et al. reported a facile strategy for generating a heterointerface between Ni(OH)₂ and FeOOH species through physical mixing.²² The heterointerface between the two inferior catalysts exhibited OER overpotential (298 mV at 10 mA cm⁻²), confirming that physical-mixing-derived heterointerfaces could enhance the OER activity through the exposure of surface and edge sites. Bhutani et al. synthesized a heterostructure of FeP and CoP via physical mixing, which exhibited a low OER overpotential of 220 mV at 10 mA cm⁻² and remarkable stability for 200 h (at 200 mA cm⁻²).²³ These outstanding performance metrics were attributed to the charge transfer between Fe and Co metals and the reduction in the energy barrier of the OER-determining step by the oxygenated surface of the heterointerface. These mixture-based heterointerface studies have demonstrated that physical mixing is an attractive strategy for conveniently preparing transition-metal-based heterointerfaces. However, despite their simplicity, physical-mixing-derived heterointerfaces have not yet been comprehensively studied for maximizing and optimizing their electrocatalytic performance. Specifically, a new synthesis strategy should be devised to minimize the OER overpotential and preserve the physicochemical characteristics of the metal/metal oxide heterointerface during operation by rebuilding it via physical mixing with nanoscale particles.

In this study, colloidal nanoparticles (NPs) with unique physical mixing states were prepared using different Ni/Fe ratios (100 : 0 to 0 : 100), which varied according to the amounts of Ni and Fe precursors. In particular, Ni–Fe intermediate (NFI) NPs with heterointerfaces were created between metallic Ni and Fe₂O₃ NPs (Ni/Fe₂O₃), resulting in high OER activity with a low overpotential (199 mV @ 10 mA cm⁻²; the 75 : 25 sample annealed under H₂/Ar mixture gas) and long-term durability for 5 d. In addition to the presence of heterointerfaces, the generation of heterostructured NPs that intrinsically contained heterointerfaces and the abundant formation of NiOOH at the applied potential were presumably responsible for the outstanding OER performance of the 75 : 25 sample. Furthermore, the effects of the oxidation state of Ni on the OER were revealed by comparing the OER activity of NiO/Fe₂O₃ with that of Ni/Fe₂O₃; the results proved that metallic Ni could induce stronger electronic interactions than those of NiO when forming a heterointerface with Fe₂O₃, as the OER activity of Ni/Fe₂O₃ was higher than that of NiO/Fe₂O₃. The 75 : 25 sample with Ni/Fe₂O₃ heterointerfaces also exhibited superior performance in AEMWE single-cell and alkaline OER half-cell systems, resulting in an impressive current density of 5.13 A cm⁻² at 2 V_cell and long-term durability for 8 d at 1 A cm⁻², demonstrating its potential as an anode catalyst for AEMWEs. These results indicate that the construction of Ni/Fe₂O₃ heterointerfaces aids significantly in rapidly catalyzing the OER, with the 75 : 25 sample showing particular promise for alkaline water oxidation.

2 Methods

2.1. Chemicals

Nickel(II) chloride hexahydrate (NiCl₂·6H₂O; 98%) and oleic acid (OA; technical grade, 90%) were supplied by Alfa Aesar. Iron(III) chloride hexahydrate (FeCl₃·6H₂O; ACS reagent grade, 97%) and 1-octadecene (ODE; technical grade, 90%) were obtained from Sigma Aldrich. Sodium oleate (NaOL; >97%) was purchased from Tokyo Chemical Industry.

2.2. Colloidal nanoparticle synthesis

The five distinct colloidal NPs (100 : 0 to 0 : 100) were synthesized using a modified version of previously reported methods.^24–26 To prepare the Ni–Fe oleate precursor, the metal precursors (x and y mmol of Ni and Fe chlorides, respectively; 4 mmol) were mixed with ethanol (6 mL), deionized water (8 mL), and hexane (14 mL) in a round-bottom flask containing NaOL (2x + 3y mmol). The resulting mixture was heated to 70 °C, forming the Ni–Fe oleate precursor within the hexane phase through a 4 h reaction. The Ni–Fe oleate in hexane was collected using a separatory funnel by discarding the aqueous NaCl solution. Subsequently, hexane was evaporated, and the resulting dried waxy-solid Ni–Fe oleate was dissolved in ODE (20 mL) and OA (1 mL). The obtained mixture was vacuum-dried at 120 °C for 1 h to eliminate impurities, and then reacted at 310 °C for 2 h under N₂ to synthesize the Ni–Fe oxide NPs. The products were precipitated and collected via centrifugation using chloroform and acetone. The NPs were then dispersed in hexane to fabricate working electrodes.

2.3. Material characterization

XRD analysis was conducted using an AXS New D8 Advance diffractometer (Bruker, Billerica, MA, USA) with a Cu K-α radiation source and a LynxEye line detector. NP films—that is, the XRD samples—were fabricated by drop-casting the NPs onto zero-background quartz. (HR)TEM images were acquired using a JEM-2100 microscope (JEOL) operating at 200 kV. EDS analysis was conducted using a Carl Zeiss SIGMA microscope. XPS profiles were obtained using a K-alpha+ spectrometer (Thermo Fisher Scientific) with an Al Kα radiation source. In situ Raman analysis was conducted with home-made electrochemical cell comprising a working electrode (75 : 25), Hg/HgO reference electrode, and Pt wire counter electrode in 1 M KOH electrolyte. The working electrode in cell on stage of Commercial micro-Raman instrument (XperRam 200, Nanobase) was analyzed using a 532 nm laser source (5 mW) and low-magnification objective lens (5×) during catalysis.

2.4. Electrochemical measurements

Ti foil with dimensions of 0.5 × 4 cm² was treated with an aqueous oxalic acid solution (5 wt%) at 70 °C for 30 min under vigorous stirring.²⁷ Subsequently, the working electrode for the OER was fabricated by adding the solution containing NPs (200 μg) dropwise onto the end of the acid-treated Ti foil (0.5 × 0.5 cm²). To remove organic species, thermal treatment at 400 °C was performed for 1 h under H₂/Ar mixed gas or air. Alkaline OER experiments were conducted in a 1 M KOH aqueous solution, using Hg/HgO and Pt wire as the reference and counter electrodes, respectively. To stabilize and optimize the OER performance of the samples, cyclic voltammetry (CV) was conducted 20 times in the potential range of 1.0–1.6 V (vs. RHE) prior to linear sweep voltammetry (LSV) measurements. The LSV curves were acquired at a scan rate of 5 mV s⁻¹ over a potential range of 1.0–1.6 V (vs. RHE). Electrochemical impedance spectroscopy (EIS) analysis was conducted at 0.670 V (vs. Hg/HgO) over a range of frequencies from 1000 to 0.01 kHz using an AC signal amplitude of 10 mV. Double layer capacitance (C_dl) values were obtained from CV measurements (in the nonfaradaic region from 0.05 to 0.2 V vs. Hg/HgO) at scan rates of 80, 120, 160, and 200 mV s⁻¹, thereby helping determine the electrochemically active surface area (ECSA). The iR correction was conducted for all the electrochemical performance data by considering the solution resistance value at a whole potential range depicted in figures.

2.5. AEMWE single-cell operation

The 75 : 25 sample (2 mg cm⁻², ionomer 10 wt%) and Pt/C (0.5 mg_Pt cm⁻², TANAKA TEC10E50E 46.9 wt%, ionomer/C = 0.5) were used as the anode and cathode, respectively, in the AEMWE system. PiperION Anion Exchange Dispersion (5 wt%) and PiperION AEM 20 microns were used as the ionomer and membrane, respectively. On the anodic and cathodic sides, SS fiber paper and Sigracet 39 BB (SGL Carbon), respectively, were employed as the porous transport layer (PTL). The reactant 1 M KOH was only supplied to the anode side at a flow rate of 3 mL min⁻¹; moreover, it was used during the single-cell durability test at a flow rate of 30 mL min⁻¹. The operating cell voltage (V_cell) range was 1.3–2.1 V_cell, the cell temperature of the AEMWE was 80 °C, and the active area was 2.2 × 2.3 cm².

3 Results and discussion

Colloidal NPs were synthesized by heating a mixture of the waxy-solid metal oleate precursors, ODE, and OA using a conventional Schlenk line system (Fig. 1a). Five pristine (p-) samples were prepared by varying the amounts of Ni and Fe precursors (by metal molarity) (denoted p-100 : 0 (Ni only), p-75 : 25, p-50 : 50, p-25 : 75, and p-0 : 100 (Fe only), respectively). To confirm the formation of colloidal NPs, the crystal structures of the five samples were analyzed by X-ray diffraction (XRD) (Fig. 1b); the results indicated that all NP samples had mixed phases that varied with the metal precursor ratio. The XRD pattern of the p-100 : 0 (Ni only) sample displayed strong peaks that matched well with the reference spectra of metallic Ni with cubic and hexagonal structures, whereas the pattern of the p-0 : 100 (Fe only) sample was similar to that of iron oxides (Fe₂O₃ and FeO) with peak shift at ∼43° which might be due to the structural distortion or phase interaction. The XRD spectra of the NFI NPs (that is, NPs synthesized using Ni–Fe oleates in 75 : 25, 50 : 50, and 25 : 75 ratios) confirmed that these specimens were physical mixtures of Ni metal and Fe₂O₃ NPs rather than a single phase corresponding to each synthesis condition. The incorporation of Fe induced a leftward shift of the major Ni-related XRD peak at ∼44° from the p-100 : 0 sample to the p-25 : 75 specimen; this was likely due to the increase in lattice spacing caused by the introduction of Fe into Ni metal, given their differences in atomic size and strains, and the decrease in the amount of hexagonal Ni with increasing content of Fe within the NPs. The Ni peak shift due to the introduction of Fe and the reduction in the hexagonal Ni portion established the complexity of the synthesized mixture-based NP system.

Fig. 1 (a) Schematic illustration of colloidal synthetic procedures to obtain the NPs. (b) XRD patterns of the five pristine (p-) colloidal NPs from p-100 : 0 to p-0 : 100 depending on the metal amount ratio of Ni and Fe, respectively (red bars: PDF 03-065-2865, orange bars: PDF 01-089-7129, pink bars: PDF 00-047-1049, grey bars: PDF 00-039-1346, purple bars: PDF 01-076-0957, navy bars: PDF 00-066-0245, and black bars: PDF 00-046-1312).

The morphological characteristics of the five samples were revealed by transmission electron microscopy (TEM) (Fig. 2a–k). The mean size of the synthesized p-100 : 0, p-75 : 25, p-50 : 50, p-25 : 75, and p-0 : 100 NPs was determined by TEM to be 16.73 ± 15.51, 18.09 ± 7.64, 10.85 ± 5.66, 10.98 ± 3.34, and 16.76 ± 1.74 nm, respectively (Fig. 2a–f). The TEM analysis indicated that despite the use of different precursor ratios, no significant disparities were observed in the NP size from the p-100 : 0 sample to the p-0 : 100 specimen; however, the NP uniformity changed considerably depending on the metal ratio. The p-100 : 0 sample (Ni only) was highly polydisperse with a large standard deviation of 15.51 nm, whereas the p-0 : 100 (Fe only) sample was monodisperse with a small deviation of 1.74 nm. The NFI samples exhibited moderate uniformity owing to their mixed Ni and Fe contents. Notably, an increase in the Fe content gradually enhanced the uniformity and reduced the size deviation from 15.51 nm (p-100 : 0) to 1.74 nm (p-0 : 100) (Fig. 2a–f), presumably owing to the variation in the Fe oleate amount in the presence of oleic acid in the different samples, which kinetically determined the NP shape.²⁴ Additionally, the five pristine samples were probed by high-resolution TEM (HRTEM) to investigate their interplanar lattice spacing (Fig. S1†). Lattice analysis of the p-100 : 0 colloidal sample validated the presence of Ni metal NPs by revealing the Ni (002) with a d-spacing of 0.217 nm (Fig. S1a†). HRTEM analysis of the p-0 : 100 sample revealed the FeO phase with the (200) facet and a d-spacing of 0.216 nm, corroborating the corresponding XRD results, which revealed the FeO phase in addition to Fe₂O₃ (Fig. S1e†). The NFI samples mainly contained both cubic Ni and Fe₂O₃ phases (Fig. S1b–d†), substantiating the observation that the synthesized colloidal NP products were complex physical mixtures with various phases.

Fig. 2 TEM images of the five pristine (p-) colloidal NP samples depending on the certain metal precursor ratio of (a) 100 : 0, (b) 75 : 25, (c) 50 : 50, (d) 25 : 75, and (e) 0 : 100. (f) The NP average size and standard deviation of the five NP samples. The TEM EDS mapping data of (g) p-100 : 0, (h) p-75 : 25, (i) p-50 : 50, (j) p-25 : 75, and (k) p-0 : 100. (l) Atomic ratios of five distinct specimens from 100 : 0 to 0 : 100.

TEM energy-dispersive spectroscopy (EDS) mapping analysis was conducted to identify the elemental configurations of Ni, Fe, and O within the five NP samples (Fig. 2g–k). The p-100 : 0 and p-0 : 100 samples were the non-uniform Ni metal NPs and even Fe oxide NPs, respectively (Fig. 2g and k). The NFI NPs were homogeneously mixed phases comprising Ni metal and Fe₂O₃ NPs, which enabled the facile formation of multiple physical contacts between the Ni and Fe₂O₃ NPs (Fig. 2h–j); these findings are identical to those obtained by XRD, which implied the mixing of various phases. The atomic ratios of Ni, Fe, and O within the NPs were determined by scanning electron microscopy (SEM) EDS to avoid analyzing the Ni- or Fe-concentrated area. The Ni/Fe/O atomic ratio for the five samples was evaluated to be 75 : 0 : 25, 47 : 20 : 33, 24 : 24 : 52, 19 : 30 : 51, and 0 : 48 : 52, respectively (Fig. 2l); these values were proportional to the amounts of Ni and Fe precursors loaded into the flask. The proportion of O in the 100 : 0 sample (25 at%) was lower than that in the 0 : 100 specimen (52 at%) because the former was mainly in the metallic Ni phase, indicating that the oxygen content gradually increased from the p-100 : 0 to p-0 : 100 samples.

Prior to examining alkaline OER performance, working electrodes were prepared by loading each p-NP sample (200 μg) onto acid-treated Ti foil (0.25 cm²) considering mass activity (Fig. S2†). Subsequently, thermal treatment was conducted at 400 °C in the presence of mixed H₂/Ar gas for 1 h, thereby eliminating electrochemically inactive organic species, robustly binding NPs to each other and to the Ti substrate, and strengthening the contacts between Ni and Fe₂O₃. The OER activities of the five samples annealed under H₂/Ar gas (denoted 100 : 0–0 : 100, no prefix) and bare Ti foil were evaluated by acquiring LSV curves after 20 CV cycles performed to stabilize and optimize the OER performance in 1 M KOH (Fig. 3a). As LSV curve of bare Ti foil showed no significant OER current (black line), the LSV data of five samples from 100 : 0 to 0 : 100 presented their own catalytic performances without Ti substrate effect. Among the specimens investigated, the 75 : 25 sample exhibited the lowest OER overpotential, requiring values of only 199, 247, and 285 mV to achieve current densities of 10, 50, and 100 mA cm⁻², respectively. These OER overpotentials are significantly lower than those of several recently reported Ni/Fe-based catalysts (Table S1†),^28–32 highlighting the potential of the 75 : 25 Ni/Fe₂O₃ system as an anode material in practical water electrolyzer systems operating under alkaline conditions. The other NFI samples, 50 : 50 and 25 : 75, exhibited slightly higher overpotentials at 10 mA cm⁻² (238 and 239 mV, respectively) than that of the 75 : 25 specimen. The monometallic samples (100 : 0 and 0 : 100) showed inferior OER activity, exhibiting overpotentials of 359 and 440 mV, respectively, at 10 mA cm⁻². The oxidation peak of 100 : 0 sample at ∼1.3 V still remained after both CV and LSV runs owing to its metallic Ni phase, indicating that the electrochemical reaction could not fully oxidize the metal Ni phase of 100 : 0 sample. Contrarily, the LSV curves of the other samples (75 : 25–0 : 100) displayed no remarkable oxidation peak because these samples contained Fe₂O₃ NPs and a smaller amount of Ni metal NPs than 100 : 0 specimen. These LSV results confirmed that the NFI NPs were considerably more active for water oxidation than the 100 : 0 (Ni only) and 0 : 100 (Fe only) specimens, presumably owing to the synergistic effect of the metal/metal oxide heterointerfaces between Ni and Fe₂O₃ NPs.

Fig. 3 Electrochemical performance measurement data of (a) LSV curves (inset) the overpotentials of the five samples showing volcano trend, (b) Tafel slopes, (c) ECSA data, and (d) Nyquist plots of the five main samples. (e) Long-term durability test of 75 : 25 sample under alkaline condition at 10 mA cm⁻² for 5 days. The iR-correction was done for all the electrochemical performance data.

The OER overpotential of the five samples at 10 mA cm⁻² showed a volcano trend (Fig. 3a, inset), which suggested that the 75 : 25 sample was optimal for conducting water oxidation in an alkaline medium. The intrinsic OER activity of the five samples was also explored by obtaining polarization curves normalized by the C_dl, which showed that the 75 : 25 sample exhibited the highest intrinsic OER activity among the specimens investigated (Fig. S3a†). Fig. S3b† showed the turnover frequency (TOF) values of 100 : 0–0 : 100 samples, also proving the highest OER intrinsic activity of 75 : 25 sample among them. Tafel slope analysis (Fig. 3b) indicated that the 75 : 25 sample exhibited the fastest OER kinetics among the samples as it recorded the lowest gradient of 47.4 mV dec⁻¹. Furthermore, among the specimens, the 75 : 25 sample also exhibited the largest ECSA (Fig. 3c) and the lowest charge-transfer resistance (R_ct), as evaluated by EIS (Fig. 3d). To clarify the sustainability of the 75 : 25 sample, XRD and (HR)TEM analyses were performed after both CV activation and LSV measurements, which showed that the crystallinity, morphologies, and unique mixed states of the specimen were maintained (Fig. S4 and S5†). The post-OER results indicated that the 75 : 25 sample showed remarkable durability under alkaline OER conditions. Chronopotentiometry analysis was performed at 10 mA cm⁻² to evaluate the stability of the 75 : 25 sample under alkaline OER conditions for 5 days (Fig. 3e). The result revealed a marginal increase in the overpotential (59 mV) over five days, corroborating the superiority of the 75 : 25 sample as an anode material for water electrolysis.

The impressive OER activity of the 75 : 25 sample likely originated from the formation of Ni/Fe₂O₃ heterointerfaces, which might facilitate strong electronic interactions between the Ni and Fe₂O₃ domains.^16,33 To clarify the effects of the Ni/Fe₂O₃ heterointerfaces on the alkaline OER, an additional sample was prepared by mixing and annealing the metallic Ni NPs (p-100 : 0, Ni only) and Fe₂O₃ NPs (p-0 : 100, Fe only) with ratio of 3 : 1 (denoted Ni + Fe₂O₃; Fig. 4a and b), which was identical metallic ratio of 75 : 25 sample. The original 75 : 25 sample exhibited higher OER activity than that of Ni + Fe₂O₃ (Fig. 4a). When Ni (100 : 0) and Fe₂O₃ (0 : 100) were simply mixed, the resulting Ni + Fe₂O₃ system was not a homogeneous mixture, leading to fewer interactions between the Ni and Fe₂O₃ NPs (Fig. 4b). In contrast, the 75 : 25 sample exhibited a more homogeneous physical mixing state than that of Ni + Fe₂O₃, according to TEM EDS analysis (Fig. 2h and 4b), suggesting that the contact between Ni and Fe₂O₃ was greater in the former than in the latter. Interestingly, the 75 : 25 sample also comprised heterostructured NPs containing intrinsic Ni/Fe₂O₃ heterointerfaces generated during colloidal heat-up process (Fig. 4c and d). These results indicated that the heterointerfaces evolved to a greater degree in the sample obtained by simultaneously mixing the Ni and Fe precursors (that is, the (p-)75 : 25 NPs) than in Ni + Fe₂O₃. Considering the TEM and EDS results, the higher OER activity of the 75 : 25 specimen than that of Ni + Fe₂O₃ was primarily due to the Ni/Fe₂O₃ heterointerfaces and the formation of heterostructured NPs (Fig. 2b and 4).

Fig. 4 (a) Polarization curves of 75 : 25 (dark green solid line) and Ni + Fe₂O₃ (mixture of p-100 : 0 and p-0 : 100 annealed under H₂/Ar mixture gas atmosphere, black dashed line). (b) TEM and EDS mapping images of Ni + Fe₂O₃ sample. (c and d) HRTEM and EDS mapping images of heterostructured NPs in p-75 : 25 sample. The heterostructured NPs were fabricated by colloidal heat-up process, no physical contact.

In general, the catalytic activity of transition-metal-based catalysts depends significantly on the oxidation state of the Ni component; for example, Ni-(oxy)hydroxides with high oxidation states are considered efficient OER catalysts.^34–37 Therefore, the dependence of surface oxidation on OER activity was investigated by loading each of the five as-synthesized samples (p-100 : 0 to p-0 : 100) onto Ti foil, and then annealing the resulting systems in air instead of heating them in the H₂/Ar gas mixture, thereby forming Ni oxides (NiO/Fe₂O₃). XRD analysis of the air-annealed (a-) samples (denoted a-100 : 0 to a-25 : 75; Fig. 5a) revealed a significant decrease in the Ni metal peak intensity and the emergence of NiO peaks, indicating the successful oxidation of Ni. Compared to the p-0 : 100 specimen, the a-0 : 100 sample showed no changes in crystal structure after air oxidation owing to the stability of the Fe₂O₃ phase, implying the selective oxidation of metallic Ni without a phase change of Fe₂O₃.

Fig. 5 (a) XRD patterns of air-oxidized NPs on the acid-treated Ti foil (green star reference of hexagonal Ti: PDF 00-044-1294). TEM images of the five air-annealed (a-) NP samples of (b) a-100 : 0, (c) a-75 : 25, (d) a-50 : 50, (e) a-25 : 75, and (f) a-0 : 100. (g) HRTEM image of a-75 : 25, showing NiO/Fe₂O₃ heterointerfaces.

TEM analysis was also performed to explore the morphological characteristics of the five air-annealed samples (Fig. 5b–f). Compared to the original standard samples, the air-annealed specimens showed slightly changed morphologies for the Ni-containing NPs as Ni metal readily transformed into the NiO phase through air oxidation at 400 °C for 1 h. HRTEM analysis of the a-75 : 25 sample with the optimal Ni/Fe ratio revealed the formation of the NiO phase with a d-spacing of 0.208 nm for the NiO (200) plane (Fig. 5g). Moreover, the Fe₂O₃ phase in the a-75 : 25 NPs was also maintained, as suggested by the interplanar spacing of 0.351 nm for the Fe₂O₃ (311) plane. Furthermore, the HRTEM-based lattice spacing analysis confirmed the selective oxidation of Ni into NiO without changes in the crystal structures and morphologies of the Fe₂O₃ NPs. Additionally, air oxidation produced NiO/Fe₂O₃ heterointerfaces facilitating charge transfer (Fig. 5g), similar to the generation of Ni/Fe₂O₃ heterointerfaces via H₂/Ar annealing.

The Ni oxidation effect of the heterointerface on the OER activity was confirmed by obtaining LSV curves for six samples—that is, the five air-annealed samples and the original 75 : 25 sample (Fig. 6; dark green solid line corresponds to the profile in Fig. 3a). Among the air-annealed specimens, a-75 : 25 exhibited the highest OER activity as it achieved an overpotential of 266 mV at 10 mA cm⁻², further demonstrating that the Ni/Fe ratio of 75 : 25 was optimal for alkaline OER operation with the mixed NP systems. The a-100 : 0 and a-50 : 50 specimens required a considerably higher overpotential (341 and 411 mV, respectively) than that of a-75 : 25 to achieve a current density of 10 mA cm⁻², whereas a-25 : 75 and a-0 : 100 did not show noticeable OER activity. Furthermore, the air-annealed samples did not exceed the OER activity of the 75 : 25 specimen from the onset potential to the end, and TOF values of air-annealed samples were lower than that of 75 : 25 (Fig. S3b and c†), indicating that the Ni/Fe₂O₃ heterointerfaces were much more effective for the alkaline OER than those of NiO/Fe₂O₃. The higher OER activity of Ni/Fe₂O₃ than that of NiO/Fe₂O₃ also implied that the electronic interactions might be more critical for the alkaline OER when metallic Ni, rather than the NiO phase, bonded with Fe₂O₃.

Fig. 6 LSV curves to investigate the oxidation effect of Ni, showing OER activities of five air-annealed (a-) samples from a-100 : 0 to a-0 : 100 (dotted lines) and the original 75 : 25 (dark green solid line) annealed under H₂/Ar mixture gas. The iR-correction was conducted for these LSV curves.

During the OER, the abundant formation of OER adsorbates—particularly oxyhydroxides—is vital for producing O₂ gas on the anode.^8,38 Therefore, to ascertain the manner in which the formation of O-containing intermediates depended on the Ni state, X-ray photoelectron spectroscopy (XPS) was performed on the 75 : 25 and a-75 : 25 samples (Fig. 7). Notably, the Ni 2p spectra were significantly affected by the annealing protocol, whereas the Fe 2p and O 1s profiles showed negligible changes in the surface oxidation states. The Ni 2p spectra of the 75 : 25 sample mainly contained peaks related to metallic Ni (at 853.0 and 870.0 eV),^16,39 NiO (at 853.8 and 870.8 eV),^40,41 and Ni(OH)₂ (at 855.9 and 872.9 eV) (Fig. 7a).^40,42,43 In contrast, the a-75 : 25 specimen contained a larger portion of NiO (854.3 and 871.3 eV) and featured Ni(OH)₂ (855.9 and 872.9 eV) on the surface without Ni⁰.^36,44,45 Only the a-75 : 25 sample contained the Ni³⁺ species (857.5 and 874.5 eV),^46,47 which likely originated from Ni₃O₄ or Ni₂O₃ generated via air oxidation. Overall, the Ni 2p spectra (Fig. 7a) confirmed the oxidation of Ni through air annealing.

Fig. 7 XPS spectra of 75 : 25 sample (above one of each figure) and a-75 : 25 sample (below one of each figure). (a) Ni 2p, (b) Fe 2p, and (c) O 1s spectra obtained after thermal treatment. (d) Ni 2p, (e) Fe 2p, and (f) O 1s spectra obtained after activation of twenty CV cycles. (g) Ni 2p, (h) Fe 2p, and (i) O 1s spectra obtained after OER (LSV measurement after activation).

The remarkable disparity in the valence state of Ni between the 75 : 25 and a-75 : 25 samples caused the difference in OER activity as much as the OER overpotential of 67 mV at 10 mA cm⁻² (Fig. 6), which was probably due to the effect of metallic Ni in the Ni/Fe₂O₃ heterointerfaces, as discussed earlier. Another potential factor was likely the difference in adsorption free energy between Ni and NiO at the rate-determining step (RDS) involving the conversion of *O to *OOH during the alkaline OER (with “*” representing surface-adsorbed components).⁴⁸ Considering the free energy for O-containing intermediates, Ni (111) exhibited a value that was much closer to 0 eV than that of NiO (111) at the other conversion steps as well as the RDS. Additionally, the hydrogen-adsorption-related free energy of pure Ni (−0.58 eV) was also closer to zero than that of NiO (0.97 eV).⁴⁹ As hydrogen and O-containing intermediates are more favorable to be adsorbed on Ni metal domain than NiO, utilization of metallic Ni phase is reasonable to achieve higher electrocatalytic activity than usage of NiO phase for electrocatalyst towards water splitting.

Moreover, numerous studies have combined the Fe species with Ni-based materials to attenuate the free energy at RDS during OER requiring both deprotonation and oxygenation.⁵⁰ Liu et al. revealed that Fe–NiOOH showed lower barrier energy at RDS (1.72 eV) and higher electron localization function (0.68) than NiOOH (1.89 eV and 0.52, respectively).⁵¹ Similar to our work, electronic interaction at Ni/Fe₂O₃ heterointerface was also analyzed in previous works through density functional theory analysis, describing the electron accumulation and consumption on Fe and Ni sites, respectively.^16,52 The electronic interaction between Ni and Fe₂O₃ domains diminished OER RDS barrier and optimized the d-band center of Ni/Fe₂O₃ heterointerfaces (−3.10 eV) as medium value of metallic Ni (−1.60 eV) and Fe₂O₃ (−6.40 eV).⁵² Additional to the Ni/Fe₂O₃, Hao et al. calculated the d-band center value of NiO/Fe₂O₃ (−3.14 eV),⁵³ demonstrating that stronger bonding of O-containing adsorbates on Ni/Fe₂O₃ than NiO/Fe₂O₃ might induce exceptional OER performances. Therefore, combining Ni with Fe₂O₃ could be more effective for electrocatalysis than those of NiO/Fe₂O₃.

The Fe 2p XPS profiles (Fig. 7b) indicated that Fe³⁺ was the only component, given the presence of peaks at 710.8 and 724.6 eV and the absence of those corresponding to Fe⁰ and Fe²⁺ species (Fig. 7b),^24,40,54 corroborating the fact that the Fe oxides mainly existed as Fe₂O₃. In particular, the oxidation state of Fe³⁺ did not vary with the annealing process, further substantiating the selective modulation of the Ni state and dependence of the OER performance on metallic Ni. The O 1s spectra of the samples indicated that water molecules, metal oxides, and metal hydroxides were present on the film surfaces after the thermal treatments (Fig. 7c). Moreover, the type of gas used for thermal treatments did not induce any remarkable changes in the O 1s spectra. Considering the static oxidation state of Fe, Ni—rather than Fe—was the key component that dictated the OER performance of the prepared samples, particularly the 75 : 25 NPs.

The XPS analysis conducted after CV activation (Fig. 7d–f) and after both CV and LSV runs (Fig. 7g–i) revealed a correlation between the OER performance and the chemical state of Ni. After CV activation, Ni³⁺ was produced via the formation of NiOOH—that is, the OER active site (Fig. 7d, top). Although the subsequent LSV run conducted towards oxidation direction diminished the peak intensities of Ni 2p due to the formation of adsorbates such as (oxy)hydroxides or Ni leaching, peak position of Ni 2p was retained after LSV (Fig. 7g, top), indicating that CV activation stabilized and optimized the catalyst surface to enhance the OER activity and durability by forming NiOOH and even preserving Ni⁰. However, in the a-75 : 25 sample, the Ni states were maintained regardless of the process (air annealing, activation, and OER) once the NiO phase was established, indicating that the NiOOH species might not be formed enough to exceed the OER activity of the 75 : 25 sample (Fig. 7a, d and g).

In the Fe 2p spectra of the 75 : 25 and a-75 : 25 specimens obtained after the CV pre-test and LSV measurements (Fig. 7e and h), only the Fe³⁺ species appeared steadily, even after the CV and LSV runs and regardless of the annealing conditions, indicating that the Fe₂O₃ phase was retained. Considering the lack of peak position changes for Fe³⁺, the formation of abundant FeOOH species on the surfaces of the two catalysts after the CV and LSV tests was somewhat implausible (Fig. 7e and h). The O 1s spectra confirmed the formation of M–OOH bonds, which likely originated from the NiOOH species under both H₂/Ar exposure and air annealing after the CV and LSV tests (Fig. 7f and i). The XPS analysis confirmed the formation of NiOOH on the surface of the 75 : 25 catalyst, which may have significantly contributed to its competitive OER activity and long-term durability. The XPS-analysis-based comparisons between metallic Ni and NiO revealed that the metallic Ni domains within the 75 : 25 sample played a critical role in the evolution of oxygen molecules by not only interacting with Fe₂O₃ for creating a synergistic effect toward OER catalysis, but also efficiently forming NiOOH species on their surfaces, thereby maximizing the OER performance of the 75 : 25 sample through CV pre-runs.

Additional to the XPS, in situ Raman spectroscopy analysis was performed to elucidate the surface reconstruction of 75 : 25 sample during OER by applying anodic potential from 1.05 to 1.65 V_RHE after twenty CV cycles (Fig. S6†). The formation of NiOOH on 75 : 25 sample was verified after CV activation at the peaks of 474 and 548 cm⁻¹ corresponding to E_g and A_1g vibrations by oxidation and hydroxylation on Ni domains of 75 : 25 specimen,^8,55,56 further proving that CV activation process could optimize and stabilize the OER performance of 75 : 25. A LSV run with potential increase of oxidation direction made those peaks more pronounced continuously, which provided a direct evidence of surface Ni conversion process towards higher oxidation states of Ni³⁺ species during OER. The in situ Raman data was identical with XPS results of NiOOH formation, suggesting that Ni is more critical than Fe³⁺ to evolving oxygen molecules.⁵¹ Additional to the in situ Raman result, both higher OER activity of Ni/Fe₂O₃ than NiO/Fe₂O₃ and marginal peak shift of Fe³⁺ towards lower binding energy (Fig. 7b and e) indicated that the electronic interaction was stronger at Ni/Fe₂O₃ than NiO/Fe₂O₃, and the interface electrons might be transferred from Ni to Fe sites in both cases.

The practical applicability of the 75 : 25 sample for water splitting was evaluated by using it as an anode material with a Pt/C cathode in an AEMWE single-cell system (Fig. 8a). The AEMWE was operated within a cell voltage (V_cell) range of 1.3–2.1 V_cell at a cell temperature of 80 °C by feeding 1 M KOH reactant to the anode side, which resulted in an outstanding current density of 5.13 A cm⁻² at 2 V_cell (Fig. 8b). Notably, the performance of the 75 : 25 sample in the AEMWE cell is considerably higher than that of other recently reported transition-metal-based anode materials (Table S1†). Furthermore, a durability test of the AEMWE was conducted by measuring the change in V_cell when a current density of 1 A cm⁻² was consistently applied to the cell (Fig. 8c). The AEMWE cell presented a highly stable state during cell operation for 8 d by maintaining a value of ∼1.85 V_cell, despite showing a minimal increase (∼0.2 V_cell) at the initial stage, which was probably due to the increased resistance of cell components such as the membrane, ionomer, and current collector rather than the catalyst, as indicated in previous studies on AEMWEs.^57,58 The AEMWE cell performance as well as the half-cell-scale capability confirmed the superiority of the 75 : 25 sample containing Ni/Fe₂O₃ heterointerfaces in catalyzing water oxidation under alkaline conditions. The results of the I–V curve and durability analyses (Fig. 8b and c) underscored the utility of the 75 : 25 sample containing Ni/Fe₂O₃ heterointerfaces as a noteworthy anode catalyst for practical, economically viable AEMWE systems, owing to the synergistic interactions between the metallic Ni and Fe₂O₃ domains. Considering the recently reported literature, further study of Ni/Fe₂O₃ could considerably develop the heterointerface catalyst design as practical electrocatalysts for AEMWE by doping certain elements having high electronegativity or generating oxygen defects on the catalyst with simplified synthetic procedures for large-scale catalyst preparation.^59–61

Fig. 8 (a) Illustration of the AEMWE single-cell system. (b) Polarization curve and (c) stability test result of single-cell.

4 Conclusion

Five colloidal NP samples in complex mixed states with various Ni/Fe ratios—ranging from 100 : 0 (Ni only) to 0 : 100 (Fe only)—were prepared. In particular, the NFI NPs (p-75 : 25 to p-25 : 75) comprised metallic Ni and Fe₂O₃ NPs, which contained not only heterointerfaces between Ni metal and Fe₂O₃ NPs, but also heterostructured NPs with Ni/Fe₂O₃ heterointerfaces. Among the five samples that were annealed under H₂/Ar atmosphere, the 75 : 25 NPs exhibited the lowest OER overpotential of 199 mV at 10 mA cm⁻² and excellent long-term durability for five days, owing to the strong electronic interactions at the Ni/Fe₂O₃ heterointerfaces. The importance of the metallic Ni phase in alkaline OER catalysis was clarified by exploring the OER activity of NiO/Fe₂O₃, whose electronic interactions were found to be less significant than those in Ni/Fe₂O₃. Furthermore, more NiOOH species—the OER active phase—were generated on the Ni sites of Ni/Fe₂O₃ (75 : 25) than on those of NiO/Fe₂O₃ (a-75 : 25), indicating the superiority of the 75 : 25 sample for the alkaline OER. When the 75 : 25 sample with numerous Ni/Fe₂O₃ interfaces was applied to the anode of an AEMWE, it functioned as a potential catalyst material for practical water electrolysis systems as it achieved a high current density of 5.13 A cm⁻² @ 2 V_cell with remarkable durability for 8 d. Overall, this study implemented a strategy to construct Ni/Fe₂O₃ heterointerfaces with remarkable alkaline OER performance without using Pt-group metals, with the 75 : 25 catalyst showing significant promise as an anode material for practical water electrolysis systems operating under alkaline conditions.

Abbreviations

HER	Hydrogen evolution reaction
OER	Oxygen evolution reaction
AEMWE	Anion exchange membrane water electrolysis
NP	Nanoparticle
NFI	Ni–Fe intermediate
ODE	1-Octadecene
OA	Oleic acid
LSV	Linear sweep voltammetry
CV	Cyclic voltammetry
ECSA	Electrochemically active surface area
EIS	Electrochemical impedance spectroscopy
RDS	Rate-determining step
NaOL	Sodium oleate
PTL	Porous transport layer

Data availability

Data will be made available on request.

Author contributions

W. J. and Y.-K. H. conducted experiments and wrote the manuscript (original draft & review). H. K. performed formal analysis and data curation. Y. P., Y. J. H., and Y. L. assisted with material characterization. G. K. and H. S. conducted the formal analysis. J. L. supervised the formal analysis and experiments. D.-H. H. supervised the overall experiments, performed experimental validation, and wrote the paper (review & editing).

Conflicts of interest

The authors declare no competing financial interest.

Acknowledgements

This study was supported by the National Research Foundation of Korea (NRF) grants funded by the Korean Government (MSIT) (grant numbers: RS-2020-NR049537 and RS-2024-00467234). W. J. and Y.-K. H. contributed equally to this study.

References

1. C. Niether, S. Faure, A. Bordet, J. Deseure, M. Chatenet, J. Carrey, B. Chaudret and A. Rouet, Nat. Energy, 2018, 3, 476–483.
2. J. C. Ehlers, A. A. Feidenhans'l, K. T. Therkildsen and G. O. Larrazábal, ACS Energy Lett., 2023, 8, 1502–1509.
3. Z.-Y. Wu, F.-Y. Chen, B. Li, S.-W. Yu, Y. Z. Finfrock, D. M. Meira, Q.-Q. Yan, P. Zhu, M.-X. Chen, T.-W. Song, Z. Yin, H.-W. Liang, S. Zhang, G. Wang and H. Wang, Nat. Mater., 2023, 22, 100–108.
4. H. Ozcan, R. S. El-Emam, S. Celik and B. A. Horri, Cleaner Chemical Engineering, 2023, 8, 100115.
5. R. R. R. Sulaiman, W. Y. Wong and K. S. Loh, Int. J. Energy Res., 2022, 46, 2241–2276.
6. D. Li, A. R. Motz, C. Bae, C. Fujimoto, G. Yang, F.-Y. Zhang, K. E. Ayers and Y. S. Kim, Energy Environ. Sci., 2021, 14, 3393–3419.
7. L. Han, S. Dong and E. Wang, Adv. Mater., 2016, 28, 9266–9291.
8. E. Hwang, M. Kim, W. Lee, Y. Park, W. Jeong, Y. J. Hwang, Y. Lee, S. Bang, G. Kim, L. M. Moreau, H. Son, S. Back and D.-H. Ha, Chem. Eng. J., 2024, 494, 152903.
9. Y. Lee, W. Jeong, Y. J. Hwang, B. An, H. Lee, H. Jeong, G. Kim, Y. Park, M. Kim and D.-H. Ha, J. Mater. Chem. A, 2024, 12, 28574–28594.
10. Q. Li, A. M. Villarino, C. R. Peltier, A. J. Macbeth, Y. Yang, M.-J. Kim, Z. Shi, M. R. Krumov, C. Lei, G. G. Rodríguez-Calero, J. Soto, S.-H. Yu, P. F. Mutolo, L. Xiao, L. Zhuang, D. A. Muller, G. W. Coates, P. Zelenay and H. D. Abruña, J. Phys. Chem. C, 2023, 127, 7901–7912.
11. M. Klingenhof, P. Hauke, M. Kroschel, X. Wang, T. Merzdorf, C. Binninger, T. N. Thanh, B. Paul, D. Teschner, R. Schlögl and P. Strasser, ACS Energy Lett., 2022, 7, 3415–3422.
12. T. Kwon, M. Jun, J. Joo and K. Lee, J. Mater. Chem. A, 2019, 7, 5090–5110.
13. N. Zhang and R. Jiang, ChemPlusChem, 2021, 86, 1586–1601.
14. Y. Yang, J. Li and Y. Sun, Chem. Eng. J., 2022, 440, 135868.
15. A. P. Tiwari, M. S. Rahman and W. J. Scheideler, Adv. Mater. Technol., 2024, 9, 2400160.
16. B. H. R. Suryanto, Y. Wang, R. K. Hocking, W. Adamson and C. Zhao, Nat. Commun., 2019, 10, 5599.
17. L. Luo, H. Huang, Y. Yang, S. Gong, Y. Li, Y. Wang, W. Luo and Z. Li, Appl. Surf. Sci., 2022, 575, 151699.
18. G.-Q. Han, Y.-R. Liu, W.-H. Hu, B. Dong, X. Li, X. Shang, Y.-M. Chai, Y.-Q. Liu and C.-G. Liu, Appl. Surf. Sci., 2015, 359, 172–176.
19. S. N. Faisal, E. Haque, N. Noorbehesht, H. Liu, M. M. Islam, L. Shabnam, A. K. Roy, E. Pourazadi, M. S. Islam, A. T. Harris and A. I. Minett, Sustainable Energy Fuels, 2018, 2, 2081–2089.
20. M. S. A. Akbari, R. Bagheri, Z. Song and M. M. Najafpour, Sci. Rep., 2020, 10, 8757.
21. H. Radinger, P. Connor, S. Tengeler, R. W. Stark, W. Jaegermann and B. Kaiser, Chem. Mater., 2021, 33, 8259–8266.
22. M. Görlin, P. Chernev, P. Paciok, C.-W. Tai, J. Ferreira de Araújo, T. Reier, M. Heggen, R. Dunin-Borkowski, P. Strasser and H. Dau, Chem. Commun., 2019, 55, 818–821.
23. D. Bhutani, S. Maity, S. Chaturvedi, D. Chalapathi, U. V. Waghmare, C. Narayana, V. C. Prabhakaran and E. Muthusamy, J. Mater. Chem. A, 2022, 10, 22354–22362.
24. J. A. Bau, P. Li, A. J. Marenco, S. Trudel, B. C. Olsen, E. J. Luber and J. M. Buriak, Chem. Mater., 2014, 26, 4796–4804.
25. J. Park, K. An, Y. Hwang, J.-G. Park, H.-J. Noh, J.-Y. Kim, J.-H. Park, N.-M. Hwang and T. Hyeon, Nat. Mater., 2004, 3, 891–895.
26. Y.-K. Hong, H. T. Kim, Y. Park, W. Jeong, M. Kim, E. Hwang, Y. J. Hwang, M.-H. Lee and D.-H. Ha, Nanoscale, 2023, 15, 4604–4611.
27. K.-R. Yeo, K.-S. Lee, H. Kim, J. Lee and S.-K. Kim, Energy Environ. Sci., 2022, 15, 3449–3461.
28. W. Wan, Y. Zhao, S. Wei, C. A. Triana, J. Li, A. Arcifa, C. S. Allen, R. Cao and G. R. Patzke, Nat. Commun., 2021, 12, 5589.
29. Y.-l. Wang, T.-h. Yang, S. Yue, H.-b. Zheng, X.-p. Liu, P.-z. Gao, H. Qin and H.-n. Xiao, ACS Appl. Mater. Interfaces, 2023, 15, 11631–11641.
30. Y. Han, J. Wu, L. Tang, X. An, X. Yang, T. Li, Q. Wang and X. Wu, Appl. Surf. Sci., 2023, 610, 155288.
31. C.-F. Li, H.-B. Tang, J.-W. Zhao and G.-R. Li, J. Mater. Chem. A, 2023, 11, 5841–5850.
32. J. Fan, X. Zhang, M. Han, X. Xiang, C. Guo, Y. Lin, N. Shi, D. Xu, Y. Lai and J. Bao, Small, 2024, 20, 2303927.
33. M. Asnavandi, Y. Yin, Y. Li, C. Sun and C. Zhao, ACS Energy Lett., 2018, 3, 1515–1520.
34. M. Ledendecker, S. K. Calderón, C. Papp, H.-P. Steinrück, M. Antonietti and M. Shalom, Angew. Chem., Int. Ed., 2015, 54, 12361–12365.
35. F. Dionigi, Z. Zeng, I. Sinev, T. Merzdorf, S. Deshpande, M. B. Lopez, S. Kunze, I. Zegkinoglou, H. Sarodnik, D. Fan, A. Bergmann, J. Drnec, J. F. d. Araujo, M. Gliech, D. Teschner, J. Zhu, W.-X. Li, J. Greeley, B. R. Cuenya and P. Strasser, Nat. Commun., 2020, 11, 2522.
36. Y. Zhai, X. Ren, Y. Sun, D. Li, B. Wang and S. Liu, Appl. Catal., B, 2023, 323, 122091.
37. K. Chen, J. Qian, W. Xu and T.-T. Li, Inorg. Chem., 2024, 63, 642–652.
38. S. Jung, R. A. Senthil, A. Min, A. Kumar, C. J. Moon, G. H. Jeong, T. W. Kim and M. Y. Choi, J. Mater. Chem. A, 2024, 12, 8694–8706.
39. L. Pasquali, B. P. Doyle, F. Borgatti, A. Giglia, N. Mahne, M. Pedio, S. Nannarone, A. K. Kaveev, A. S. Balanev, B. B. Krichevtsov, S. M. Suturin and N. S. Sokolov, Surf. Sci., 2006, 600, 4170–4175.
40. C. Roy, B. Sebok, S. B. Scott, E. M. Fiordaliso, J. E. Sørensen, A. Bodin, D. B. Trimarco, C. D. Damsgaard, P. C. K. Vesborg, O. Hansen, I. E. L. Stephens, J. Kibsgaard and I. Chorkendorff, Nat. Catal., 2018, 1, 820–829.
41. J. Qi, W. Zhang, R. Xiang, K. Liu, H.-Y. Wang, M. Chen, Y. Han and R. Cao, Adv. Sci., 2015, 2, 1500199.
42. S. Liu, H. Zhang, E. Hu, T. Zhu, C. Zhou, Y. Huang, M. Ling, X. Gao and Z. Lin, J. Mater. Chem. A, 2021, 9, 23697–23702.
43. Y. Park, K.-R. Yeo, Y. J. Hwang, M. Kim, W. Jeong, S.-K. Kim and D.-H. Ha, J. Alloys Compd., 2024, 993, 174649.
44. F. Bao, E. Kemppainen, I. Dorbandt, F. Xi, R. Bors, N. Maticiuc, R. Wenisch, R. Bagacki, C. Schary, U. Michalczik, P. Bogdanoff, I. Lauermann, R. van de Krol, R. Schlatmann and S. Calnan, ACS Catal., 2021, 11, 10537–10552.
45. L. Lin, R. Xin, M. Yuan, T. Wang, J. Li, Y. Xu, X. Xu, M. Li, Y. Du, J. Wang, S. Wang, F. Jiang, W. Wu, C. Lu, B. Huang, Z. Sun, J. Liu, J. He and G. Sun, ACS Catal., 2023, 13, 1431–1440.
46. P. Salunkhe, A. A. V. Muhammed and D. Kekuda, Mater. Res. Express, 2020, 7, 016427.
47. S. Zhang, C. Tan, R. Yan, X. Zou, F.-L. Hu, Y. Mi, C. Yan and S. Zhao, Angew. Chem., Int. Ed., 2023, 62, e202302795.
48. F. Qiao, X. Wang, C. Sun, Y. Chen, J. Xu, B. Ouyang, J. Zhu and E. Kan, J. Mater. Sci., 2022, 57, 16437–16447.
49. H. Wang, M. Cui, G. Fu, J. Zhang, X. Ding, I. Azaceta, M. Bugnet, D. M. Kepaptsoglou, V. K. Lazarov, V. A. de la Peña O'Shea, F. E. Oropeza and K. H. L. Zhang, Sci. China:Chem., 2022, 65, 1885–1894.
50. Y. Shi, L. Song, Y. Liu, T. Wang, C. Li, J. Lai and L. Wang, Adv. Energy Mater., 2024, 14, 2402046.
51. X. Liu, X. Zhao, S. Cao, M. Xu, Y. Wang, W. Xue and J. Li, Appl. Catal., B, 2023, 331, 122715.
52. C. Xuan, T. Shen and B. Hou, Chem. Eng. J., 2024, 479, 147723.
53. Y. Hao, D. Yu, S. Zhu, C.-H. Kuo, Y.-M. Chang, L. Wang, H.-Y. Chen, M. Shao and S. Peng, Energy Environ. Sci., 2023, 16, 1100–1110.
54. C. Zhao, J. Wang, Y. Gao, J. Zhang, C. Huang, Q. Shi, S. Mu, Q. Xiao, S. Huo, Z. Xia, J. Zhang, X. Lu and Y. Zhao, Adv. Funct. Mater., 2024, 34, 2307917.
55. Q. Zhang, W. Xiao, H. C. Fu, X. L. Li, J. L. Lei, H. Q. Luo and N. B. Li, ACS Catal., 2023, 13, 14975–14986.
56. L. Sun, Z. Zhou, Y. Xie, J. Zheng, X. Pan, L. Li and G. Zhao, Adv. Funct. Mater., 2023, 33, 2301884.
57. Y. Hong, S. C. Cho, S. Kim, H. Jin, J. H. Seol, T. K. Lee, J.-k. Ryu, G. M. Tomboc, T. Kim, H. Baik, C. Choi, J. Jo, S. Jeong, E. Lee, Y. Jung, D. Ahn, Y.-T. Kim, S. J. Yoo, S. U. Lee and K. Lee, Adv. Energy Mater., 2024, 14, 2304269.
58. Y. Hong, S. Jeong, J. H. Seol, T. Kim, S. C. Cho, T. K. Lee, C. Yang, H. Baik, H. S. Park, E. Lee, S. J. Yoo, S. U. Lee and K. Lee, Adv. Energy Mater., 2024, 14, 2401426.
59. N. Song, K. Lv, B. Feng, Z. Zhang, X. Tan, Y. Wei, Z. Yan, Y. Zhang, X. Yu, L. Li, X. Zhang, X. Yang, Y. Li and Z. Lu, Chem. Eng. J., 2025, 505, 158921.
60. Y. Zhang, B. Wang, C. Hu, M. Humayun, Y. Huang, Y. Cao, M. Negem, Y. Ding and C. Wang, Chin. J. Struct. Chem., 2024, 43, 100243.
61. T. Ni, X. Hou, H. Wang, L. Chu, S. Dai and M. Huang, Chin. J. Struct. Chem., 2024, 43, 100210.

---

Footnotes

† Electronic supplementary information (ESI) available. See DOI: https://doi.org/10.1039/d4ta09280e

‡ These authors contributed equally to the present study.

---