The rapid development of the global economy has increased the consumption of energy materials. It is urgent to explore alternative sustainable energy sources to reduce the dependence on fossil fuels [1, 2]. Hydrogen generation by electrochemical water splitting is considered to be one of the most exciting approaches for energy storage and conversion, owing to the fact that hydrogen has a high energy density and represents a promising renewable and environmentally friendly energy resource [3-5]. In order to achieve efficient and economical water electrolysis, it is essential to prepare suitable high-activity, low-cost oxygen evolution reaction (OER) electrocatalysts. The OER is a half-reaction of water splitting and is thermodynamically unfavorable. Additionally, due to its four-electron-proton-coupled transfer process, it has a much higher overpotential than the other half-reaction, the so-called hydrogen evolution reaction (HER). Therefore, the OER determines the overall efficiency of the total water splitting system [6, 7]. At present, various electrocatalysts have been developed and evaluated for use in the OER. Notably, some precious metals and their oxides (IrO₂ or RuO₂) provide high OER activity and relatively low overpotential values; however, the scarcity and high costs of these metals have severely limited their large-scale application [8, 9]. In contrast, the use of transition metals and their derivatives, such as Co, Ni, and Fe, is desirable owing to their low-cost and high earth-abundance [10, 11].

Generally, there are two main channels to fabricate highly active OER electrocatalysts. One approach is to design catalysts with structural advantages, such as large specific surface areas, reasonable pore size, and accessible active sites; the other is to regulate the electronic environment of the reaction site to enhance the electrochemical activity [12]. The development of metal-organic frameworks (MOFs) provides new opportunities as platforms for the design of non-noble metal catalysts for water splitting, as the naturally porous structure of MOFs could promote the diffusion of the substrate or product and significantly accelerate charge transfer [13]. At the same time, the large specific surface area of MOFs exposes their active sites uniformly, making them good templates or precursors for the preparation of high-performance OER electrocatalysts [14]. However, most of the MOF systems reported at present still suffer from significant problems, including poor electrical conductivity and structurally obscured metal centers, which severely hinder their direct utilization in electrochemical processes [15].

In this regard, polymetallic electrocatalysts have been studied extensively for total water splitting [16]. For example, Hao and co-workers demonstrated trimetallic Ni₅Co₃Mo-OH nanosheets that showed high electrocatalytic performance in the OER [17]. Additionally, Guo and co-workers successfully synthesized a novel CoSe/FeSe₂ catalyst by taking advantage of the synergistic effects among the different components and the electron interactions between Fe, Co, and Se to enhance the OER activity [18]. Nevertheless, further breakthroughs in OER activity using transition metals remain challenging. Lanthanide ions have been used in various applications because of their strong magnetic anisotropies and large spin capacities [19]. Some research suggests that the Dy and Ni spins are ferromagnetically coupled [20]. Additionally, it could increase the magnetism of Fe and change the roughness of transition-metal-based catalyst surfaces [21, 22]; however, relatively few works have focused on its OER applications.

Based on the above considerations, we first constructed a unique Fe-Dy-mediated MOF-Ni structure with exposed metal centers for the OER. The FeDy@MOF-Ni on carbon cloth (CC) catalyst was designed and fabricated through a simple two-step hydrothermal method to construct Dy₂O₃ and FeSO₄ doped with MOF-Ni on CC. In addition to the close contact between the MOF-Ni nanosheets and Fe-Dy, the obtained FeDy@MOF-Ni forms a 3D rhombus-like hierarchical structure in which Fe closely interacts with the adjacent Ni in the presence of Dy. As a result, the FeDy@MOF-Ni/CC exhibits impressive OER activity, with an overpotential of 251 mV at a current density of 10 mA cm^-2, which is on par with the performance of commercial RuO₂. When FeDy@MOF-Ni/CC is used in conjunction with CC modified with commercial Pt/C in a two-electrode system, it shows excellent activity for water splitting, requiring only 1.57 V to achieve a current density of 10 mA cm^-2. The unique structure of FeDy@MOF-Ni/CC creates a robust synergetic effect between Fe/Dy and the Ni active sites. It is expected to provide excellent electron transport and expose more active sites, which are responsible for the superior OER activity. These results suggest the great capacity of FeDy@MOF-Ni/CC as an efficient, economical, and durable OER electrocatalyst, demonstrating the unique electrochemical improvement in the catalyst induced by Dy-doping. Therefore, this new attractive avenue of utilizing dysprosium species to adjust the coupling effect between transition metals may provide an approach for the rational design of OER electrocatalysts.

First, CC was cut into pieces with a size of 1 cm × 1 cm. The CC was then successively sonicated in 0.5 M hydrochloric acid, deionized water, and ethanol three times, followed by drying under vacuum at 60 ℃ before use.

In the facile synthesis procedure, 0.58 g of Ni(NO₃)₂·6H₂O and 0.05 g of CTAB were dispersed ultrasonically in 30 mL of methanol to obtain solution A. Meanwhile, 0.66 g of 2-methylimidazole was dissolved in another 30 mL of methanol to obtain solution B. Subsequently, solution A was slowly injected into solution B under continuous stirring at room temperature. After 30 min, the resulting solution and CC were transferred into a Teflon-lined stainless autoclave (90 mL) and reacted at 180 ℃ for 8 h. After cooling to room temperature, the MOF-Ni-modified CC was removed, washed with abundant deionized water, and vacuum-dried at 60 ℃ for 12 h. The obtained sample is denoted as MOF-Ni/CC.

For the synthesis of FeDy@MOF-Ni/CC, a facile method involving the direct mixing of FeSO₄·7H₂O and Dy₂O₃ in a 3:2 molar ratio was designed. First, 0.18 mmol Dy₂O₃ and 0.27 mmol FeSO₄·7H₂O were dissolved into 10 mL deionized water and 10 mL ethanol. After stirring for 1.0 h, 3 mL of hydrazine hydrate was gradually added to the above solution. Ten minutes later, the resulting mixture and the as-prepared MOF-Ni/CC were transferred into a Teflon-lined stainless autoclave (50 mL) and reacted at 180 ℃ for 20 h. After cooling to room temperature, the black sample was removed, thoroughly washed with deionized water, and vacuum-dried at 60 ℃ for 3.0 h. The obtained sample was named FeDy@MOF-Ni/CC.

The preparation process for Dy@MOF-Ni/CC, Fe@MOF-Ni/CC, and DyFe/CC was similar to that of FeDy@MOF-Ni, except that only 0.18 mmol Dy₂O₃ or 0.27 mmol FeSO₄·7H₂O were used for the preparation of Dy@MOF-Ni or Fe@MOF-Ni, respectively. For comparison, Fe-Dy/CC was also prepared using pure CC instead of MOF-Ni/CC using the approach described above.

The preparation process was similar to that of MOF-Ni. Specifically, 0.58 g of Ni(NO₃)₂·6H₂O and 0.05 g of CTAB were dispersed ultrasonically in 30 mL of methanol. The resulting solution and CC were then transferred into a Teflon-lined stainless autoclave (50 mL) and reacted at 180 ℃ for 8 h. The resulting product was named Ni(OH)₂/CC. Subsequently, the Ni(OH)₂/CC and a solution containing FeSO₄·7H₂O and Dy₂O₃ were transferred into a Teflon-lined stainless autoclave (50 mL) and reacted at 180 ℃ for 20 h, as discussed above. The resulting composite was called FeDy@Ni(OH)₂/CC.

Electrochemical measurements were carried out in a standard three-electrode system operated using a VMP3B-2x2 electrochemical analyzer (CH Instruments, Inc, France). The catalysts loaded on the CC were used as the working electrode (1 cm × 1 cm), a saturated calomel electrode (SCE) was used as the reference electrode, and a Pt foil was used as the counter electrode. Electrochemical measurements of the catalysts were conducted in 1.0 M KOH solution. Polarization curves were acquired using linear sweep voltammetry (LSV) at a scan rate of 5 mV s^-1. Electrochemical impedance spectroscopy (EIS) measurements were carried out in the frequency range of 200 kHz-0.1 Hz. Chronopotentiometry was utilized for long-term stability tests. All measured potentials were calibrated to the RHE using the following equation: E(RHE) = E(SCE) + 0.2415 V + 0.059·pH (Fig. S1). All electrochemical data were corrected against ohm potential drop. CV tests of the catalysts were conducted at various scan rates in 1 M KOH solution to obtain the electrochemical double-layer capacitance (C_dl).

A schematic diagram of the fabrication process of FeDy@MOF-Ni/CC using the facile two-step hydrothermal strategy is illustrated in Fig. 1a. First, a piece of clean CC was immersed in a methanol solution containing Ni(NO₃)₂, 2-methylimidazole, and CTAB, and then hydrothermally heated to 180 ℃ for 8 h in a Teflon-lined stainless autoclave. A series of MOF-Ni-coated CC composites were prepared, and referred to as MOF-Ni/CC. Subsequently, the obtained MOF-Ni/CC composites were immersed in a mixed solution of FeSO₄, Dy₂O₃, and hydrazine. The resulting mixture was hydrothermally heated to 180 ℃ to acquire the final product, Fe-Dy-oxide-modulated MOF-Ni/CC. Unless otherwise specified, this is abbreviated as FeDy@MOF-Ni in the following discussion.

X-ray diffraction (XRD) patterns were obtained and used to analyze the phase composition of the materials. As shown in Fig. S2a, the crystal structure of MOF-Ni gave rise to six strong diffraction peaks, which is consistent with previous reports for MOF-Ni [23]. In addition, the XRD patterns of FeDy@MOF-Ni, together with those of Fe@MOF-Ni and Dy@MOF-Ni, are shown in Fig. 1b and Fig. S2b. The XRD data of FeDy@MOF-Ni, Fe@MOF-Ni, and Dy@MOF-Ni match well with the diffraction peaks of Dy₂O₃ (JCPDS: 43-1006) [24, 25] and Fe₂O₃ (JCPDS: 39-1346). The crystal structure of MOF-Ni was gradually destroyed during the FeDy@MOF-Ni synthesis due to ion exchange resulting in structural deformation [26, 27].

The microstructures of the catalysts were examined using scanning electron microscopy (SEM) and transmission electron microscopy (TEM). The SEM image in Fig. 1c shows that MOF-Ni was successfully grafted onto the CC surface, and the high-resolution SEM image in Fig. 1d displays a lamellar structure composed of two-dimensional nanosheets. However, after treatment with the Fe-Dy precursor, the morphologies of the synthesized products underwent dramatic changes, from uniformly distributed lamellar structures to irregular rhombus-like structures (Figs. 1e-f). The changes in their morphologies may have been due to the formation process, namely, dissolution-regrowth during the construction of the final material. To explore the dissolution-regrowth mechanism, we investigated the thermal stability of the as-prepared MOF-Ni/CC and monitored it via SEM. The SEM image (Fig. S3) shows that the MOF-Ni was completely dissociated after 20 h at 180 ℃ in water/methanol mixture. On the other hand, FeDy@MOF-Ni was formed under the same conditions. These results indicated that the ion displacement occurs simultaneously with MOF-Ni dissociation to create a relatively stable nanocomposite on the surface of the CC. In addition, elemental analysis (Table S1) indicated great changes in the element N before and after the hydrothermal treatment of MOF-Ni, further confirming the dissolution-regrowth mechanism.

The rhombus-like structure of FeDy@MOF-Ni was further investigated using TEM (Fig. 2a). It can be observed that the composite is a bulk material composed of many stacked lamellar layers. This unique structure was believed to facilitate the transport of electrolytes and gas emissions. The high-resolution TEM image shows that FeDy@MOF-Ni exhibits two distinct types of lattice fringes with spacings of 0.31 and 0.37 nm, respectively, corresponding to the (222) plane of Dy₂O₃ and the (210) plane of Fe₂O₃ (Fig. 2b). In addition, the HAADF-STEM image of FeDy@MOF-Ni demonstrates that the elements Ni, Dy, Fe, C, O, and N were uniformly distributed throughout the structure (Fig. 2c).

The surface functional groups of the synthesized materials were investigated using Fourier-transform infrared spectroscopy (FTIR). As shown in Fig. 3a, FeDy@MOF-Ni exhibits three characteristic peaks typical of FeDy@MOF-Ni and Dy@MOF-Ni, where the two peaks at ca. 550 and 726 cm^-1 are the stretching vibrations of Dy–O [28, 29], and the peak at ca. 3609 cm^-1 can be ascribed to the stretching vibration of Dy–OH. The FTIR spectra of Fe@MOF-Ni reveals three characteristic peaks at 1623, 3160, and 3247 cm^-1, which correspond to the stretching vibrations of adsorbed H₂O and of O-H and N-H in the imidazole functional groups, respectively [30-31], implying the presence of the MOF-Ni structure in the composite. Raman spectroscopy was applied to further verify the formation of FeDy@MOF-Ni. As shown in Fig. 3b, the Raman spectrum of FeDy@MOF-Ni also corresponds to those of Dy@MOF-Ni and Fe@MOF-Ni; the three bands at 314, 397, and 501 cm^-1 were assigned to Dy–O vibration [32] and that at 1065 cm^-1 was attributed to Ni-O vibration [33]. More importantly, the additional Raman characteristic bands centered at 603 and 662 cm^-1 clearly confirmed the existence of partial Fe₂O₃ species in the composite [34-36]. All the above phenomena indicated that the engineering of the FeDy@MOF-Ni sample was successfully actualized using a simple two-step hydrothermal reaction.

Thermogravimetric analysis (TGA) was used to explore the mass variation during the pyrolysis of FeDy@MOF-Ni, Fe@MOF-Ni, and Dy@MOF-Ni in an O₂ atmosphere in the temperature range of 25 ℃ to 800 ℃ with a heating rate of 10 ℃ min^-1 (Fig. 3c and Fig. S4). In Fig. 3c, the mass loss at temperatures below 230 ℃ was primarily due to the evaporation of adsorbed water [37, 38]. The sharp exothermic peak indicates that the MOF appears to be pyrolyzed when the temperature reaches approximately 263.8 ℃ [39]. At higher temperatures near 836.1 ℃, a slight weight loss occurs due to continuous crystallization and catalyst aggregation [40]. Interestingly, no mass loss is observed at such high temperatures for the Dy@MOF-Ni material, while a weight loss appears at 379.3 ℃ due to the dehydration of the Dy metal hydroxides [13, 32]. Moreover, the mass loss of FeDy@MOF-Ni was 16.2%, which was much lower than that of Fe@MOF-Ni (47.4%), but higher than that of Dy@MOF-Ni (5.2%) in a similar temperature range (Fig. S3). This noticeable difference proved once again that the Dy and Fe species had been successfully doped into the MOF-Ni.

The Brunauer-Emmett-Teller (BET) gas-sorption measurements demonstrated that the FeDy@MOF-Ni material showed the characteristics of a type III isotherm with a pore size range of 30–50 nm and a corresponding BET specific surface area of approximately 19.3 m² g^-1 (Fig. 3d). However, the relative BET particular surface area of the Ni-MOF was approximately 229.0 m² g^-1, further indicating the dissociation of the Ni-MOF precursor and the successful preparation of FeDy@MOF-Ni. Although its BET value was smaller than that of Ni-MOF, FeDy@MOF-Ni exhibited excellent OER performance due to the strong synergistic effects between the three metals and as a result of its high-valence Ni³⁺ and Fe³⁺ active sites.

The compositions and surface chemical states of the as-prepared catalysts were further characterized using X-ray photoelectron spectroscopy (XPS). The survey spectrum (Fig. S5a) reveals the presence of C, N, O, Ni, Fe, and Dy in the composite; the high-resolution C 1s spectrum was referenced to C=O (284.0 eV), C–C (284.8 eV), and C–O (286.0 eV) as calibration standards (Fig. S5b) [41]. Meanwhile, for the MOF-Ni precursor, the ratio of Ni²⁺/Ni³⁺ in the high-resolution Ni 2p region was 2.0 (Fig. S5c). Only pyridinic-N (398.5 eV) and pyrrolic-N (399.9 eV) were observed in the high-resolution N 1s spectrum (Fig. S5d) [42, 43], which was consistent with the theoretical results.

The high-resolution Ni 2p spectrum of the as-prepared catalysts exhibits characteristic Ni 2p_3/2 and Ni 2p_1/2 peaks as well as the corresponding shake-up satellites, with the peaks at 855.1 and 872.8 eV being associated with the Ni²⁺ species, while those at 856.6 and 873.7 eV are associated with Ni³⁺ species (Fig. 4a) [44, 45]. In addition, the other pairs of peaks were assigned as satellite peaks. It is worth noting that the high-resolution Ni 2p spectrum of Dy@MOF-Ni shows a Ni⁰ peak, because Ni²⁺ is more easily reduced than Dy³⁺ (Ni²⁺ + 2e^-→ Ni, φ^Ɵ = -0.23 V; Dy³⁺ + 3e^-→ Dy, φ^Ɵ = -2.295 V). However, the metallic Ni species in FeDy@MOF-Ni and Fe@MOF-Ni are hard to detect because Fe species has a higher standard redox potential Fe³⁺/Fe²⁺ (φ^Ɵ = +0.77) than Ni²⁺/Ni⁰ (φ^Ɵ = -0.23 V), thereby oxidizing the Ni⁰ species. Interestingly, the Ni²⁺/Ni³⁺ ratios decrease when both Fe and Dy species are added, with FeDy@MOF-Ni having significantly lower values than Fe@MOF-Ni (2.12), Dy@MOF-Ni (2.52), and MOF-Ni (2.0), suggesting a strong coupling effect for FeDy@MOF-Ni during the hydrothermal reactions.

At the same time, the Fe species of the various catalysts were also explored (Fig. 4b). A series of peaks can be fitted in the high-resolution Fe 2p XPS profiles, of which those with binding energies at approximately 709.9 eV and 711.7 eV are assignable to Fe²⁺ and Fe³⁺, respectively [46]. The Fe²⁺/Fe³⁺ ratio of FeDy@MOF-Ni was lower than that of Fe@MOF-Ni, implying that Fe²⁺ can be converted to a higher valence state in the presence of Dy species. In the high-resolution Dy 3d_5/2 spectrum of FeDy@MOF-Ni in Fig. 4c, a peak corresponding to Dy₂O₃ was fitted at 1297.0 eV [47, 48]; this peak was shifted 0.70 eV toward lower binding energy compared with that of Dy@MOF-Ni, which could be due to the effect of the surface enrichment of Ni and Fe species on the charge density around the Dy species. These results indicate that the content of high-valence Ni and Fe species was remarkably improved in FeDy@MOF-Ni by the participation of Dy species. The positive effects of high-valent transition metal species on OER performance have been widely reported in the literature [49-50]. Additionally, inductively coupled plasma spectrometry (ICP) indicated the Ni/Fe ratio of FeDy@MOF-Ni was approximately 1:1 (Table S2).

The electrocatalytic activity of all the OER catalysts was evaluated using linear sweep voltammetry (LSV) at a scan rate of 5 mV s^-1 in 1.0 M KOH solution. All the LSV polarization curves were processed using iR compensation (Fig. S6) and RHE correction due to the effect of ohmic resistance. Furthermore, FeDy@MOF-Ni showed the lowest ohmic resistance, suggesting that it had the highest conductivity among the catalysts. For comparison, the OER performance of the catalyst RuO₂ (Aladdin Industrial Corporation, 2.0 mg cm^-2 loading on CC) was also measured. The CV curves overlapped after 3–5 cycles in all the measurements. As shown in the polarization curves in Fig. 5a, a pair of sharp Ni²⁺/Ni³⁺ redox peaks appeared in the overpotential range of 0.1 to 0.2 V, which were ascribed to the formation of NiOOH species [2]. Notably, FeDy@MOF-Ni displays superior catalytic activity (overpotential of 251 mV) at a current density of 10 mA cm^-2; this overpotential was much smaller than those of Fe@MOF-Ni (288 mV), Dy@MOF-Ni (382 mV), MOF-Ni (361 mV), and Fe-Dy (401 mV), and on par with that of commercial RuO₂ (253 mV). The corresponding Tafel plots are a critical parameter to estimate OER reaction kinetics. In Fig. 5b, the plot of FeDy@MOF-Ni is presented. Its Tafel slope of 51.8 mV dec^-1 is much lower than those of Fe@MOF-Ni (54.6 mV dec^-1), Dy@MOF-Ni (105.1 mV dec^-1), MOF-Ni (96.9 mV dec^-1), Fe-Dy (80.8 mV dec^-1), and RuO₂ (62.6 mV dec^-1). This merit indicates that dysprosium could optimize the coupling effect between transition metals to enhance OER performance. The OER activity of FeDy@MOF-Ni was further highlighted by comparing our best catalyst to other reported high-performance polymetallic catalysts reported in alkaline media (Fig. 5c) [51-55].

In order to further determine the intrinsic OER activities of the catalysts, their turnover frequency (TOF) values were calculated using the formula below:

Where J corresponds to the current density, A is the area of the electrode with the number of moles of the catalysts, F is the Faraday constant (96, 485 C mol^-1) and n is the number of moles of all metal atoms. Note that all the metal atoms were assumed to be active species [56]. As shown in Fig. 5d, FeDy@MOF-Ni exhibited a high TOF value of 22.9 s^-1 at 300 mV, which was higher than those of Fe@MOF-Ni (6.61 s^-1), and Dy@MOF-Ni (0.71 s^-1), indicating the superior OER activity of FeDy@MOF-Ni compared to the other catalysts.

In addition to the TOF value, the electrochemical active surface area (EASA) is another parameter that indirectly indicates OER activity. In general, the higher the double-layer capacitance (C_dl), the smaller the overpotential will be at a given current density. The EASA is proportional to the C_dl, which can be acquired from CV curves obtained at different scan rates in a non-faradaic potential region [57, 58]. As shown in Fig. S7, the C_dl of FeDy@MOF-Ni was obtained from cyclic voltammogram curves (CVs) at scan rates of 100, 120, 140, 160, 180, and 200 mV s^-1. The results show that the catalyst FeDy@MOF-Ni had a C_dl of 0.71 mF cm^-2, which was higher than those of Fe@MOF-Ni (0.55 mF cm^-2), Dy@MOF-Ni (0.68 mF cm^-2), and Fe-Dy (0.56 mF cm^-2) (Fig. S8).

There are several benefits to the use of MOF-Ni as a nickel precursor. Firstly, it can improve chemical absorption through the hydroxyl group on the catalyst surface due to the stronger OH–Ni bond [49]. Secondly, its unique 2D-structured nanosheets expose an abundance of coordinatively unsaturated metal sites and structural defects [59, 60]. Finally, its OER performance can be significantly improved by doping it with other metal species [61, 62]. Therefore, although the MOF-Ni and Ni(OH)₂ precursors have similar lamellar structures (Figs. S9, 10), the OER performance of FeDy@MOF-Ni is far superior to that of the catalyst FeDy@Ni(OH)₂ under the same conditions. This result implies the significant role of the MOF-Ni precursor in the overall process.

To investigate the OER performance for potential industrial applications, the overall water splitting reaction was explored under industrially relevant conditions using a two-electrode configuration. FeDy@MOF-Ni was used as the anodic oxygen-generating catalyst, and CC modified with commercial Pt/C modified was used as the cathodic hydrogen-producing catalyst in 1.0 M KOH electrolyte (Fig. 6a). As we expected, FeDy@MOF-Ni exhibited excellent catalytic performance, affording a current density of 10 mA cm^-2 at a cell voltage of only 1.57 V. Furthermore, the stability of FeDy@MOF-Ni was measured in a three-electrode system, and the overpotential increased only 7 mV at a current of 10 mA cm^-2 current after 80 h (Fig. 6b). The SEM images and XPS data for FeDy@MOF-Ni after the long-term stability test showed that it had retained its irregular morphology and richness in Fe³⁺ and Ni³⁺ species (Figs. S11, 12). These results demonstrated that FeDy@MOF-Ni has excellent mass transport (inward movement of OH^- and outward diffusion of oxygen gas) as well as high electrical conductivity.

In summary, a unique FeDy@MOF-Ni-CC composite with a hierarchical rhombus-like structure was fabricated using a two-step approach followed by a hydrothermal method. The catalytic results indicate that the optimum catalyst FeDy@MOF-Ni-CC has an excellent OER activity and requires an overpotential of only 251 mV to drive a current density of 10 mA cm^-2. Moreover, the catalyst also exhibits a small Tafel slope of ~52 mV dec^-1 as well as reasonable stability in 1.0 M KOH. The high OER performance is mainly due to the 3D rhombus-like structure, which induces high electrical conductivity, and the synergies between the multiple metal centers. This work provides an alternative for the design and fabrication of low-cost, highly efficient, and robust anodic catalysts by introducing dysprosium to regulate the effects of transition metals.

Figs. S1–S12 and Tables S1–3 give more details on characterization of our synthesized materials and their catalytic performance data; additional photograph, XRD, SEM, XPS and catalytic performance data (PDF).