As an efficient and pollution-free energy source, hydrogen can effectively solve the excessive consumption of fossil energy and the ensuing environmental pollution problems [1]. Water splitting is considered to be one of the most promising clean hydrogen production technologies, but the very slow proton-coupled electron transfer process and high overpotential seriously hinder the overall efficiency of the oxygen evolution reaction (OER) in the anode region [2,3]. Noble metal oxides (IrO₂ and RuO₂) are often used as electrocatalysts in OER processes; however, the scarcity, instability and high-cost of these materials limit their mass production and industrial applications [4]. Therefore, low-cost transition metal (TM)-based materials are favored by researchers due to their terrestrial abundance and regulated electrocatalytic activity [5,6]. Unfortunately, single TM-based compounds have poor stability and low conductivity, so it is attractive to combine TM compounds with carbon materials with strong conductivity and high stability [7,8]. However, preparing the two materials separately and then recombining them is time-consuming and complex, which greatly reduces the application value of this strategy [9,10].

The metal-organic frameworks (MOFs) formed by the coordination of organic ligands and metal ions/clusters has the advantages of easy adjustment of structure, large surface area and high porosity, but it also has the shortcomings such as low stability and poor conductivity [11,12]. Since the derivatives of MOFs can not only provide TM, carbon and heteroatoms required for catalysis, but also have the advantage of the structural stability after pyrolysis, high-temperature calcination of MOFs is an important strategy for preparing efficient TM/N—C electrocatalysts. In the carbonization process, organic ligands can be converted into porous N doped carbon (N—C), and metal ions can be converted into TM nanoparticles or TM-based compounds and embedded in carbon materials [13-15]. In recent years, MOFs have become effective templates for the preparation of TM-based oxides, sulfides, phosphates, selenides, and other unique ordered nanostructures.

The adsorbate evolution mechanism (AEM) indicates that three oxygen-related intermediates (*OH, *O and *OOH) are involved in the OER process, but the overpotential of OER in AEM is higher than 370 mV due to the linear scaling relation (LSR) between *OH and *OOH intermediates [16,17]. That is, the competition between the key steps in *OOH formation and O–H bond cleavage results in the smallest theoretical overpotential (0.4 eV) in AEM [18]. Although the above composite strategy can achieve good catalytic activity of OER, it is sometimes limited by the AEM, and the overpotential is still not satisfactory. In recent years, heteroatom doping has been shown to alter the OER pathway from AEM to lattice oxygen mechanism (LOM) [19]. Importantly, LOM is not bound by this LSR due to the lack of *OOH intermediates, thus breaking the overpotential limit of AEM (370 mV) [20]. In summary, heteroatom doping is an effective strategy to strengthen transition metal matrix composite carbon matrix materials.

Herein, the beaded stream-like N and P-codoped carbon-coated Fe₃O₄ nanocomposite (N,P-Fe₃O₄@C) is prepared by two-step annealing using MIL-88A as precursor. The unique nanostructure is beneficial to increase the specific surface area of the catalyst and preventing the agglomeration of the TM nanoparticles. During the pyrolysis process, organic ligands serve as carbon and nitrogen sources, and sodium hypophosphite is introduced to form a layer of amorphous porous carbon co-doped with N and P to tune the intrinsic electronic structure and activity of the catalyst, which greatly enhances the conductivity of the catalyst. Most importantly, P doping causes the OER pathway to transform from AEM to LOM. Taking into account the above three advantages, N,P-Fe₃O₄@C displays excellent OER catalytic activity, with a current density of 10 mA/cm², an overpotential of 201 mV, a Tafel slope of 57.1 mV/dec, and can maintain a current density of 10 mA/cm² for stable operation for 100 h. This study provides a new direction for constructing TM/carbon matrix composites doped with heteroatoms.

The specific synthesis process of N,P-Fe₃O₄@C is shown in Fig. 1. Firstly, MIL-88A NR is synthesized by a simple hydrothermal method, and the X-ray diffraction (XRD) peaks are consistent with the previous results (Fig. S1 in Supporting information) [21]. The scanning electron microscopy (SEM) images show that MIL-88A NR is rod-like with an average length of 5 µm and an average width of 300 nm (Fig. S2 in Supporting information). N-Fe₂O₃@C is then obtained by annealing of MIL-88A NR precursor. In Fig. 2a, the XRD peaks of N-Fe₂O₃@C match well with the standard XRD card (No. 97–008–2135) of Fe₂O₃. The SEM (Figs. 2b and c) and TEM (Figs. 2d and e) images show that N-Fe₂O₃@C and N,P-Fe₃O₄@C present a beaded stream-like morphology, which proved that the morphology remained intact after P doping. This unique morphology can effectively prevent material aggregation in OER process, increase specific surface area and accelerate electron transport rate. The high-resolution TEM (HRTEM) image of N-Fe₂O₃@C (Fig. 2f) shows a lattice spacing of 0.23 nm, which corresponds to the (006) plane of the Fe₂O₃. Moreover, an amorphous carbon coating can be seen on the catalyst surface in Fig. 2f. In addition, C, N, Fe and O are homogeneously dispersed in the elemental mapping of N-Fe₂O₃@C (Figs. S3a-e in Supporting information), indicating that the catalyst is successfully synthesized and N element is uniformly doped into the catalyst.

Fig. 2a shows that the XRD peaks match well with Fe₃O₄ standard card (No. 97–002–0596). The lattice spacing of 0.485 nm in the HRTEM image corresponds to the (111) plane of Fe₃O₄ (Fig. 2g), and it can be clearly seen that the surface of the target catalyst is also coated with amorphous carbon. The selected area electron diffraction (SAED) pattern of target sample in Fig. S4 (Supporting information) indicates a single crystalline characteristic. Compared with N-Fe₂O₃@C, the element mapping image (Figs. 2h-m) and the EDS spectrum (Fig. S3f in Supporting information) of N,P-Fe₃O₄@C adds a uniform distribution of P element, which proves that P is successfully doped into the catalyst. In order to prove the conversion of Fe₂O₃ to Fe₃O₄ after P doping, the P-doped product with annealing time of 30 min (N,P-Fe₃O₄@C-30 min) is collected for XRD testing. The results show that the diffraction peaks of Fe₂O₃ and Fe₃O₄ are simultaneously observed on the XRD spectra of N,P-Fe₃O₄@C-30 min (Fig. S5 in Supporting information), which proves the reduction effect of P doping in the conversion process.

XPS is tested to further investigate the composition and chemical state of the catalyst surface. Fig. 3a shows that there are four elements C, N, O and Fe in both N-Fe₂O₃@C and N,P-Fe₃O₄@C catalysts, and there is an additional P element in the latter catalyst, which proves the successful doping of P. The XPS spectrum of C 1s in the two samples both can be fitted into four peaks located at 284.8, 285.3, 286.25 and 289.1 eV, which are assigned to the C–C, C–N, C–O and O–C = O (Fig. 3b) [22,23], respectively. The O 1s core-level spectra of N-Fe₂O₃@C and N,P-Fe₃O₄@C displayed in Fig. 3c have two obvious peaks, which include two characteristic peaks related to M-O and O—C = O (located at 529.8 and 531.9 eV), with additional two peaks (centered at 531.3 and 533.3 eV, correspond to OH species adsorbed on oxygen vacancy sites and P-O) in the latter [22,24]. Fig. 3d demonstrates the N 1s core-level spectrum of precursor and target sample, the peaks located at 398.3, 399.7, 400.8 and 402.3 eV are attributed to the pyridinic N, pyrrolic N, graphitic N and oxidized N [25]. Successful doping of N can improve electron transport and thus enhance OER performance [26]. The Fe 2p core-level spectrum (Fig. 3e) of N-Fe₂O₃@C can be divided into 711.1, 724.6, 719.1 and 733.0 eV, which correspond to Fe 2p_3/2 and Fe 2p_1/2 of Fe³⁺ species and two satellite peaks (designated Sat.), respectively [27]. In Fig. 3e, the Fe 2p XPS spectrum of N,P-Fe₃O₄@C can be resolved into six peaks, where the peaks at 710.4 and 723.9 eV belong to Fe²⁺ species, those at 712.1 and 725.8 eV are attributed to Fe³⁺ species, the peaks centered at 707.2 and 719.6 eV are assigned as Fe-P [28,29]. In the P 2p XPS spectrum (Fig. 3f), three peaks are observed at 129.5 eV (2p_1/2), 130.4 eV (2p_3/2), and P-O at 133.6 eV [30].

The graphite-like carbon structure and defect sites in the target catalyst are studied by Raman spectroscopy. Fig. 4a shows that the Raman spectra of the P-doped catalyst show obvious disordered (D) bond and graphitic (G) appear at 1335 and 1595 cm⁻¹, and the I_D/I_G value is 1.02, which proves the existence of graphitized carbon in N,P-Fe₃O₄@C, and is conducive to better conductivity of the catalyst [31]. Brunauer-Emmett-Teller (BET) result in Fig. 4b shows that N,P-Fe₃O₄@C has a large specific surface area of 254.87 m²/g and a large pore volume of 0.365 cm³/g, which is conducive to full contact of the active site with the electrolyte to achieve excellent OER performance.

The OER catalytic performance of catalysts is evaluated by LSV polarization curve. In Fig. 5a, the LSV curves recorded at 5 mV/s in 1 mol/L KOH electrolyte demonstrate that the N,P-Fe₃O₄@C requires 201 mV to drive a current density of 10 mA/cm², which lower than RuO₂ (254 mV), N-Fe₂O₃@C (394 mV), and MIL-88A NR (455 mV). Fig. 5b lists the overpotential of all the catalysts in this study under different current densities. Compared with similar catalysts, the target catalyst has excellent OER catalytic performance (Table S1 in Supporting information). As shown in Fig. 5c, the Tafel slope of N,P-Fe₃O₄@C is 57.1 mV/dec, which is lower than RuO₂ (65.9 mV/dec), N-Fe₂O₃@C (104.1 mV/dec), MIL-88A NR (168.2 mV/dec) and CP (245.7 mV/dec). The Nyquist plots (Fig. S6 in Supporting information) fitted using the equivalent circuit in Fig. S7 (Supporting information) shows that the semi-circle radius of N,P-Fe₃O₄@C is smaller than that of the other three catalysts, indicating faster electron transfer at interface the electrolyte between the electrolyte and the target catalyst.

The turnover frequency (TOF) is one of the important factors in the inherent catalytic activity of the catalyst to OER [32,33]. In Figs. S8a and b (Supporting information), the slopes of N-Fe₂O₃@C and N,P-Fe₃O₄@C are 0.026 and 0.038, respectively. When η is 300 mV, the TOF value of N,P-Fe₃O₄@C is 2.467 s⁻¹, which is much higher than that of N-Fe₂O₃@C (0.035 s⁻¹) (Fig. S8c in Supporting information), indicating that P doping greatly enhances the inherent activity of the catalyst. The electrochemical active surface area (ECSA) is used to evaluate the active area and intrinsic activity of the catalysts [34,35]. The capacitance current density shows a good linear relationship with the scanning rate, and the calculated double-layer capacitance (C_dl) of N-Fe₂O₃@C and N,P-Fe₃O₄@C is 3.3 and 9.2 mF/cm², respectively, which proves that the latter has more active sites than the former (Figs. S9a-c in Supporting information). Moreover, through the formula: ECSA = C_dl/C_s, the ECSA normalization curves in Fig. S9d (Supporting information) can also prove that N,P-Fe₃O₄@C has better inherent activity.

The multicurrent step chronopotentiometry is used to evaluate the bubble dissipation rate of the catalyst (Fig. S10 in Supporting information), proving that the target catalyst has good mass transport, electrical conductivity and mechanical robustness [36]. To investigate the electrochemical stability of N,P-Fe₃O₄@C, a long-term stability test is performed at a current density of 10 mA/cm² (Fig. 5d). The results show that the current density maintains 90.4% of the initial value after about 100 h electrochemical testing. LSV curves (Fig. S11 in Supporting information), SEM (Fig. S12a in Supporting information) and XRD (Fig. S12b in Supporting information) before and after electrochemical tests show not significant changes, indicating that the catalyst has good stability.

The key step of P doping has a great influence on the performance of OER, so the OER mechanism is studied. LSV curves of N-Fe₂O₃@C and N,P-Fe₃O₄@C at different pH values are recorded in Figs. 6a and b. The function graph of current density on pH (Fig. 6c) proves that the OER catalytic activity of the target catalyst is highly dependent on the pH of the electrolyte, while the OER catalytic activity of the precursor is less dependent on the pH of the solution. These results suggest that the P-doping step may trigger LOM pathway through uncoordinated proton-electron transfer processes [19,37]. As one of the important intermediates in the LOM path, peroxo-like (O₂²⁻) can be used as an indicator of whether OER passes through the LOM pathway [17,38]. At the same time, tetramethylammonium cation (TMA⁺) has a strong interaction with O₂²⁻, so TMA⁺ can affect the activity of the catalyst passing through the LOM pathway, but cannot affect the catalyst passing through the AEM pathway [19]. Obviously, compared with LSV curves containing and without TMA⁺ in Fig. 6d, TMA⁺ has little influence on N-Fe₂O₃@C. However, due to the addition of TMA⁺, the catalytic activity of N,P-Fe₃O₄@C is greatly reduced, which proves that the latter is via LOM pathway and the former via the AEM pathway.

Density functional theory (DFT) has been applied in computational analysis to investigate the effect of P doping on the mechanism of OER. The density of state (DOS) of N-Fe₂O₃@C and N,P-Fe₃O₄@C are calculated in Figs. 7a and b. After P doping, the O 2p orbital energy center increased by 0.34 eV, while the Fe 3d orbital energy center decreased by 0.15 eV (Fig. 7c). This change in energy results in an overlap between O 2p orbital band and Fe 3d orbital band, which significantly enhances the covalent between metal and oxygen and further activates the LOM mechanism. The schematic diagram of the reaction processes of AEM and LOM is shown in Fig. 7d. Fig. 7e displays that the energy barrier of N-Fe₂O₃@C passing through AEM (1.01 eV) is much lower than that of LOM (1.33 eV), confirming that N-Fe₂O₃@C is more preferred to the AEM pathway. In contrast, the free energy diagram of N,P-Fe₃O₄@C passing through LOM shows energy barrier of 0.90 eV, which is lower than AEM (1.06 eV), indicating that N,P-Fe₃O₄@C is more inclined to the LOM pathway (Fig. 7f).

In summary, the unique beaded structure of N,P-Fe₃O₄@C is obtained by using MIL-88 frame material as the precursor through a two-step pyrolysis treatment. The unique 3D structure avoids the aggregation of Fe₃O₄ nanoparticles, and heteroatom doping regulates the electronic structure of the catalyst, improving the conductivity and specific surface area of the catalyst. In addition, DFT calculations shows that P doping is a key step to achieve the transition from AEM to LOM pathway during OER. N,P-Fe₃O₄@C has a low overpotential of 201 mV at 10 mA/cm² and long-term stability (100 h). This strategy sheds a new light on constructing new transition metal matrix composite carbon matrix materials.

Declaration of competing interest

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.

CRediT authorship contribution statement

Lin Zhang: Writing – original draft, Methodology, Formal analysis, Conceptualization. Jianlong Li: Investigation, Formal analysis, Data curation. Maoyuan Hu: Formal analysis, Data curation. Yao Xu: Visualization. Xiaoli Xiong: Writing – review & editing, Funding acquisition, Conceptualization. Zhaoyu Jin: Writing – review & editing, Project administration, Methodology.

Acknowledgments

The authors thank the National Nature Science Foundation of China (No. 22304021), National Key Research and Development Project (No. 2022YFA1505300) and Sichuan Department of Science and Technology Program of China (No. 2022YFG0312) for financial support.

Supplementary materials

Supplementary material associated with this article can be found, in the online version, at doi:10.1016/j.cclet.2025.111123.