Electrolysis of water is one of the most promising methods for storing and utilizing renewable energy sources in the form of clean chemicals [1,2]. However, hydrogen evolution reaction (HER) and oxygen evolution reaction (OER) as two half-reactions for the electrochemical water splitting may be kinetically sluggish due to involved multiple proton-coupled electron transfer [3]. Utilizing the interaction between the metal and the support, so-called metal-support interaction (MSI), is one of the most essential strategies to enhance electrocatalytic efficiency due to structural and synergetic promotion [4]. To be specific, MSI contains the electronic, geometric and bifunctional effects [5], among which electronic effects play the most important role in exploring bifunctional electrocatalysts to accelerate the process of HER and OER effectively. However, the performance of bifunctional electrocatalysts and the utilization efficiency of noble metals still have much room for improvement so far. Therefore, it is urgent to take advantage of MSI to further enhance the catalytic properties.

Currently, non-platinum noble metals including Ru, Rh and Ir have always been attracted in electrochemical hydrogen production research [6]. Among them, Ru has attracted much attention because of its low cost (1/3 of the price of Pt), high catalytic activity and excellent stability [7]. But, due to the large cohesive energy, Ru nanoparticles (NPs) are prone to aggregate, resulting in reduced activity [8]. So, uniformly dispersing or loading Ru NPs on low-cost support materials with high electrical conductivity is considered as a promising strategy to further reduce the cost of Ru-based electrocatalyst while maintaining its high HER activity [9]. In previous studies, Sun et al. [10] proved that high dispersion of the ultrasmall Ru NPs in Ru/S-rGO could increase the active surface area of the electrocatalyst, enhance the MSI, and thus improve the catalysis activity for HER under alkaline conditions. Liu et al. [11] loaded Ru nanoclusters on porous Co₃O₄ nanowires (NWs), and the obtained Ru/Co₃O₄ NWs have better HER performance in alkaline medium.

Transition metal oxide Co₃O₄ is a spinel oxide, which is considered as a promising non-noble metal electrocatalyst with the advantages of low cost, abundant reserves, and good durability [12-15]. It is often used for OER in alkaline media. However, due to its poor intrinsic activity, the catalytic performance is lower than most advanced OER catalysts. Therefore, it is promising to modify Co₃O₄ to improve its catalytic performance in OER. Making the efforts to modify the electronic structure of catalysts is one of the most efficient strategies to systematically improve the intrinsic OER activity [16-19], this is due to 3d orbital electronic configuration plays a decisive role in the influence of binding energy of OER intermediates on the oxide surface. Yan et al. [20] engineered the electronic structure of Co₃O₄ by carbon-doping. As a result, the as-prepared C–Co₃O₄ exhibited low overpotential along with good stability in alkaline media for both HER and OER. Yan et al. [21] doped Ag ion into Co₃O₄ nanosheets to improve the ratio of the active sites (Co²⁺) and obtained high efficiency OER electrocatalyst in acidic electrolyte. Furthermore, as the redox characteristics of Fe is switchable between +3 and +2 oxidation states, Fe-doping may contribute better electrochemical performance to the Co₃O₄ [22]. Many studies have reported that Fe³⁺ incorporation in cobalt oxide could trigger a spin state change at the neighboring Co³⁺ and elongation of the Co³⁺-O bond and resulting better activity in OER. Budiyanto et al. [23] proved that incorporating a small amount of iron in cobalt oxide was beneficial to enhance the OER activity by inducing lower charge transfer resistance, an increase of the Co²⁺_(Td) sites, and formation of open mesopores structures. Gao et al. [24] introduced high spin state Fe³⁺ into octahedral sites controllably to regulate the valence state of Co³⁺ to optimized OER activity.

In the view of above considerations, Fe-doped Co₃O₄ in an ordered mesoporous morphology as an efficient solid support for hosting ultra-small Ru NPs (~1.48 nm) for the HER and OER was shaped. Our research revealed that the Ru/FeCo, as a bifunctional electrocatalyst, not only shows excellent overpotentials of 155 mV (10 mA/cm²) for HER and 283 mV (10 mA/cm²) for OER, respectively, but also has excellent long-term electrochemical durability. More importantly, we found that incorporating Fe in Co₃O₄ resulted in a decrease in the NWs' array length and increase in the number of active sites Co²⁺ and defective oxygen species. Furthermore, comparing to the Ru/Co₃O₄, the MSI in Ru/FeCo improves electron transfer between Ru NPs and supports, which helps Ru atoms keep in electron-rich state and is beneficial to enhance its catalytic activity. All in all, this work provides an innovative direction for the design of high-efficiency bifunctional electrocatalysts by virtue of the interactions between active metals and metal-oxide supports.

Co₃O₄, FeCo, Ru/Co₃O₄ and Ru/FeCo with pure phase and good crystallinity were examined by XRD. The typical diffraction peaks (Fig. S1a in Supporting information) of Co₃O₄ at 2θ = 31.27^°, 36.85^°, 44.81^°, 59.35^° and 65.23^° can be indexed to (220), (311), (400), (511) and (440) planes (JCPDS No. 74-2120). There were no Fe-related peaks were detected excepted high lattice expansion in spinel structures, as evidenced by the 0.20° shift to left of the (440) peaks in Fig. S1b (Supporting information). Furthermore, the crystalline phase of Co₃O₄ and FeCo did not change after the introduction of Ru through cation exchange and the subsequent reduction at 125 ℃ in 10% H₂/Ar atmosphere.

TEM characterization was performed to visualize the morphology and topography of the samples (Figs. 1a–d) [25,26]. It can be seen that Co₃O₄ showed mesostructured NW morphology with a high degree of order. With the incorporation of Fe, the length of the NWs decreased while the disorder increased. As shown in Figs. 1c and d, ultra-small Ru NPs with the size of ca. 1.48 nm are uniformly dispersed on the pure Co₃O₄ and Fe-doped Co₃O₄ porous NWs clearly, indicating MSI induces a great reduction in the agglomeration of Ru NPs and promotion of the accessibility of active metallic Ru sites. In addition, the corresponding lattice fringe of the samples were measured using HRTEM images as shown in Figs. 1e–h. It was found that a slight increase of the lattice fringe in the FeCo and Ru/FeCo, such as, the FeCo interplanar distance corresponding to (220) was slightly larger than that of Co₃O₄ and the Ru/FeCo (311) interplanar distance was larger than that of Ru/Co₃O₄. The results further proved the doping of Fe cations enter into the Co₃O₄ lattice, which is consistent with the analysis of XRD. Meanwhile, the lattice distortion and expansion of Co₃O₄ can be ascribed to the larger radius of high-spin Fe³⁺ (0.645 Å) compared to high-spin Co³⁺ (0.61 Å) in the newly formed Co²⁺_(Td)-O-Fe³⁺_(Oh) bonds [24]. Fig. 1i shows the HAADF-STEM and EDX elemental maps of Ru/FeCo. All expected elements including Co, Fe and Ru are uniformly distributed within Ru/FeCo, further suggesting that Ru/FeCo catalyst was successfully obtained.

The topography of the catalysts was further investigated by SEM [27,28]. Secondary electron (SE) imaging for Co₃O₄ and Ru/Co₃O₄ shown in Figs. S2a and c (Supporting information) reveal that these two samples presented short rod-like structures, indicating the Ru-loading has very little effect on the morphology. In contrast, smaller and less ordered mesostructured domains were observed in FeCo and Ru/FeCo (Figs. S2b and d in Supporting information), which was in accord with TEM results. The distribution of Co, Fe, O and Ru was also confirmed by the elemental distribution diagram (Fig. S3 in Supporting information), confirming the successful preparation of Ru/FeCo again.

The chemical species and valence states of Co₃O₄, FeCo, Ru/Co₃O₄ and Ru/FeCo were characterized by XPS. The full-spectrum of Ru/FeCo confirms the presence of Ru, Fe, Co and O elements (Fig. S4 in Supporting information), which was consistent with STEM-EDX and SEM-EDX. In the high-resolution XPS spectrum of Co 2p (Fig. 2a), Co 2p_3/2 and Co 2p_1/2 core levels appear at 777.0–785.0 eV and 792.5–801.0 eV, respectively [29,30], and the peaks of Ru/FeCo shift to high binding energy. According to the Gaussian fitting method, the two main peaks can be fitted into four peaks. Among them, the fitting peaks at 797.2 eV and 795.6 eV are attributed to Co²⁺ and Co³⁺ corresponding to the Co 2p_1/2. However, the other two fitting peaks at 781.5 eV and 780.2 eV are associated with the Co²⁺ and Co³⁺ of Co 2p_3/2 [31]. The spin-orbital splitting 15.1 eV and 15.5 eV are characteristic of the existence of both Co³⁺ and Co²⁺, respectively [32]. In addition, the Co²⁺/Co³⁺ ratio (Fig. 2e) of Co₃O₄, FeCo, Ru/Co₃O₄ and Ru/FeCo is 0.859, 0.901, 1.255 and 1.313, respectively. It can be concluded that Fe³⁺ mainly occupies the octahedral field of Co³⁺, in accordance with the decreased proportion of Co³⁺ [23,24], and the MSI of Ru/Co₃O₄ and Ru/FeCo leads to the formation of more defective Co²⁺ sites. In the spectrum of the Fe 2p (Fig. 2b), two distinct peaks are ascribed to Fe 2p_3/2 and Fe 2p_1/2, respectively. The peaks located at around 711.5 eV and 713.8 eV are associated with the Fe²⁺ and Fe³⁺ cations of 2p_3/2 whereas the peaks occurred at around 722.1 eV and 725.8 eV were the characteristic peaks of Fe²⁺ and Fe³⁺ cations of 2p_1/2 [33,34]. Additionally, the satellite peaks ascribed to Fe²⁺ and Fe³⁺ were also observed. Thus the valence of iron in Fe-doped cobalt oxide in this work should be the coexistence of divalent and trivalent. Meanwhile, Ru⁰ were detected in both Ru/Co₃O₄ and Ru/FeCo (Fig. 2c), the according peaks of Ru 3p were located at 461.9 eV and 484.2 eV [35]. This positive shift of Ru 3p binding energy is consistent with the positive shift of Co 2p spectrum for Ru/FeCo indicating Fe-doping improves electron transfer between Ru NPs and supports, which helps Ru atoms keep in electron-poor state and is beneficial to the improvement of catalytic activity. The O 1s spectra of all catalysts are presented in Fig. 2d, which is mainly de-convoluted into two peaks corresponding to different oxygen species on the catalysts. As previously reported, the peak at BE of 528.9–530.2 eV and 530.9–531.7 eV were characteristic of lattice oxygen (O_α, i.e., O²⁻) and defective oxygen (O_β, i.e., O⁻, or O₂²⁻), respectively. Sometimes with a shoulder peak at 532.3–533.1 eV ascribed to hydroxyl species (OH) or adsorbed water species present as contaminants (O_γ) [36-38]. The defective oxygen species O_β/(O_α + O_β + O_γ) ratio (Fig. 2f) was 0.308 in Co₃O₄ catalysts, whereas the Ru/FeCo catalysts presented the ratio with 0.436. The increase in defective oxygen, which indicated that the generation of more oxygen vacancies, can be attributed to the MSI.

The electrocatalytic HER activities of various catalysts including Co₃O₄, FeCo, Ru/Co₃O₄ and Ru/FeCo were evaluated in 1 mol/L KOH in a three-electrode system. It should be noted that all potentials in this work were calculated with iR correction and reported at 10 mA/cm² reversible hydrogen electrode (RHE). First, it was explored that the effect of Fe-doping on the electrocatalytic performance of Co₃O₄ NWs for the HER. It can be clearly seen from the linear sweep voltammetry (LSV) curves (Fig. 3a) and the required overpotential to achieve a current density of 10 mA/cm² (Fig. 3b and Table S1 in Supporting information) that the introduction of Fe showed a remarkable decrease from 484 mV to 361 mV in the overpotential values of Co₃O₄. After the loading of Ru NPs, the catalytic performances of Ru/Co₃O₄ and Ru/FeCo enhanced significantly owing to their proven efficiency for the HER. In particular, the overpotential of Ru/FeCo was at 155 mV for 10 mA/cm², which was much lower than that of Ru/Co₃O₄ (226 mV) and FeCo (361 mV), demonstrating the effectiveness of combining iron doping and Ru anchoring onto Co₃O₄ NWs. Similarly, Fig. 3c compares the current densities obtained at a constant potential of 250 mV. As can be seen from these figures, the pure Co₃O₄ without any modification exhibited very poor electrocatalytic activity toward HER while the Ru/FeCo catalyst showed outstanding performance.

The HER kinetics were probed from the linear fitting of the Tafel plots (Fig. 3d) [39]. As known, a smaller Tafel slope indicates the fast catalytic kinetics due to a great increase of the electrocatalytic current density [40-42]. The Tafel slope of the FeCo was measured to be 88.35 mV/dec, which was lower than that of the Co₃O₄ (95.47 mV/dec), indicating better catalytic kinetics of Fe-doped Co₃O₄. As expected, Ru/FeCo had the lowest Tafel slope value (82.81 mV/dec), indicating the relatively rapid HER kinetics among the as-prepared samples. Meanwhile, the electrochemical impedance spectroscopy (EIS) tests were used to evaluate charge transfer kinetics of different catalysts [43,44]. Small R_ct indicates excellent charge transfer ability of catalysts [45]. As shown in the Nyquist plots (Fig. 3e), FeCo has a lower charge transfer resistance compared to pure Co₃O₄. This is due to incorporating iron is beneficial to enhance electron transfer on the electrical double layer [46]. Promisingly, the Ru/FeCo showed the lowest R_ct (~17 Ω) in Table S1. This result is in excellent agreement with the findings from the LSV and Tafel plots, suggesting excellent electrocatalytic activity for the HER.

To further evaluate the intrinsic HER activity of the Ru/FeCo catalyst, it was measured that the electrochemical double-layer capacitance (C_dl) from the cyclic voltammetry (CV) as a function of scan rate (20–100 mV/s) (Fig. S5 in Supporting information) [47,48]. It is well established that C_dl is linearly proportional to the electrochemically surface area (ECSA) of the catalysts [47,49,50]. The C_dl value of Ru/FeCo was determined to be 86.24 mF/cm² (Fig. 3f), which was significantly higher than those of Co₃O₄, FeCo and Ru/Co₃O₄, revealing the outstanding electrochemically active area with highly exposed catalytic active sites of the Ru/FeCo for the HER.

OER activities were also examined by observing the overpotential at 10 mA/cm² in 1 mol/L KOH electrolyte under the same condition. The LSV curves of four catalysts above are displayed in Fig. 4a. The trend in the overpotential for OER is the same as that for HER, which can be concluded as Ru/FeCo (283 mV) < Ru/Co₃O₄ (311 mV) < FeCo (375 mV) < Co₃O₄ (429 mV) (Fig. 4b and Table S2 in Supporting information), suggesting that the Ru/FeCo catalyst has the most excellent OER activity among four catalysts. The current densities obtained at a constant potential of 350 mV (Fig. 4c) were compared, it can be seen that Ru/FeCo has the highest current density with the value of 30.9 mA/cm². In Fig. 4d and Table S2, the Tafel slope of Ru/FeCo was 79.08 mV/dec, which is smaller than that of other three catalysts, suggesting a favorable OER kinetics on Ru/FeCo. Regarding to the charge transfer kinetics of catalysts for OER, Ru/FeCo displays the smallest R_ct (~25 Ω) in Fig. 4e and Table S2, indicating the MSI between Ru and FeCo enhanced the charge transfer ability of Ru/FeCo. The intrinsic OER activity of Ru/FeCo catalyst was also evaluated by the C_dl from the CV (Fig. S6 in Supporting information). As shown in Fig. 4f and Table S2, the C_dl of Ru/FeCo (77.62 mF/cm²) goes up dramatically, suggesting that MSI makes the Ru/FeCo have a lager electrochemically active area, which can expose more catalytic active sites for OER.

The stability of catalysts is a key factor for their practical application. It was examined by current-time curve method at 0.8 V (vs. RHE) for Ru/FeCo catalyst in alkaline solution (1 mol/L KOH) (Figs. S7a in Supporting information). The i-t curve shows negligible attenuation after a long period of 20 h test, demonstrating that the Ru/FeCo catalyst has excellent long-term electrochemical durability. The oscillation phenomenon of the current density shown in the partial enlarged detail (insert of Fig. S7a) was due to the large number of bubbles generated on the electrode surface during the i-t test. Furthermore, TEM and HRTEM images of the used catalysts were performed in the locations with less-visible amorphous phase of Nafion binder. As shown in Figs. S7b-e (Supporting information), the mesostructured NW morphology of Ru/FeCo are maintained, and no structural collapse occurs. Compared to the fresh catalysts, the bulk region of used samples maintains the high crystallinity with similar lattice spacing. It indicates that Ru/FeCo catalyst has good structural stability and reusability, mainly thanks to the strong interactions developed between Ru NPs and FeCo support.

In summary, the optimized Ru/FeCo catalyst is developed as a bifunctional electrocatalyst for efficient water splitting. A series of physical characterization and chemical tests show that the ultra-small Ru NPs (~1.48 nm) are uniformly distributed on the supports, and the overpotential of HER at 10 mA/cm² is 155 mV, while the overpotential of OER at 10 mA/cm² is 283 mV. In addition, the Ru/FeCo catalyst has excellent long-term electrochemical durability in 1 mol/L KOH solution, and no significant changes were found in the structural comparison of the catalysts before and after the reaction. These excellent properties can be attributed to the abundant metal-support interactions in Ru/FeCo, which enhance the charge transfer between Ru and FeCo support resulting in modified electronic structure of Ru NPs to facilitate the electrochemical reactions. This work provides a guideline for the design of high-efficiency bifunctional electrocatalysts.

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

The authors gratefully acknowledge the financial support provided by the National Natural Science Foundation of China (Nos. 52161145403, 22072164, 51932005), Liaoning Revitalization Talents Program (No. XLYC1807175), and the Research Fund of Shenyang National Laboratory for Materials Science.

Supplementary material associated with this article can be found, in the online version, at doi:10.1016/j.cclet.2022.108085.